Print 88 comment(s) - last by Archmaille.. on Jun 16 at 4:29 PM

The OCZ Hydrajet cooler, featured upside-down to reaveal the carbon nanotube conductor. (Source DailyTech, Anh Huynh)
The OCZ Hydrojet cooler uses an advanced heatsink material

The first heatsink to make use of directional carbon nanotubes, the OCZ Hydrojet, was on display at Computex 2007. Carbon nanotubes, an allotrope of carbon, are widely regarded as the next major thermal interface material because of their superior thermal conduction properties.  

The contact base of the OCZ Hydrojet is made completely of carbon-nanotubes, which OCZ claims are five times more efficient than copper. Carbon nanotubes have been looked upon as a strong alternative to traditional copper based heatsinks. They are ideal for application in heat transfer products because of their impressive heat-conduction properties.  Carbon nanotube based interfaces have been shown to conduct more heat than conventional thermal interface materials at the same temperatures. In addition, they have shown to be ballistic conductors at room temperature, which means electrons can flow through CNTs without collisions.

Carbon nanotubes are small wire-like structures made out of a sheet of graphene.  The sheet of graphene used to construct CNTs is roughly one-atom thick, and is rolled up into a cylinder. The diameter of the cylinder ranges in the nanometers.

Unlike most other thermal materials, carbon nanotubes are able to move heat in one direction. On the other hand, copper, which is looked upon as one of the more superior thermal materials, moves heat radially. In the case of CNTs, heat is moved along the alignment of the nanotubes.

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

In my understanding
By Treckin on 6/9/2007 3:55:25 PM , Rating: 2
I believe the main bottleneck in cooling now is no longer the heat dissipation rate of the air coolers; The best out there (tuniq, 120, etc) already move more heat than is transfered to them. The main bottleneck is the bridge between the core --> die. As the process shrinks, more heat is generated in a smaller space, and more and more core layers are added, generating a hot-spot where the core on die is located. Essentially, you could cool to your hearts content, but unless there is a more effective way to move the heat away from the core and into the cooler, than the rate at which the CPU can be cooled will stagnate.

Perhaps CNT's that extend from/through (possible?) the core out of the top of the die, working in consort with tubes on base of your cooler?

Also, I wanted to point out why downward facing heat pipe coolers are less effective than laterally cooled towers:
The way heat pipes work is through the heat transference principle of convection. When cooling the pipes in a downward fashion, the cool air moving down the length of the heat pipes disrupts the convection occurring in the pipes, essentially making sure that the convecting heat never reaches the top of the pipes and is conducted to the fins ---> air.

RE: In my understanding
By Cogman on 6/9/2007 7:52:20 PM , Rating: 1
move more heat than is transfered to them

Wrong.. they move as much heat as is transfered to them, it is impossible to move more heat then is available. there is still room for advancements and I think CNT are a step in the right direction.

As the process shrinks, more heat is generated in a smaller space

Semi Wrong, when there is a process shrink, heat produced tends to go down per transistor, however, more transistors are packed together generating a fair amount of heat.

and more and more core layers are added, generating a hot-spot where the core on die is located.

you are forgetting that your layers are getting thiner and thinner. as well there are really no gaps in between layers. the heat transfer will be very good in this part, until we are talking about 100's of layers there really is no problem of heat transfer from them.

Perhaps CNT's that extend from/through (possible?) the core out of the top of the die, working in consort with tubes on base of your cooler?

This, however, is a good idea. Heck, constructing the entire cpu out of CNTs would be ideal as they can work at much higher temps then normal silicon can (and near indestructible). I think that CPU manufactures should look into changing the heat spreaders on the cores from nickel (I believe) to Carbon-nano tubes, however that might make lapping impossible :P

Basically, you make some assumptions that are just wrong. if we where working with very large things they might be right, but not when you are getting as small as we are. The proof? Pelter + water cooling. it is able to transfer much more heat out of the core then regular air cooling. Until air cooling gets to the point where it cools the CPU to room temps (or within a couple of *C), there is still room for improvement.

RE: In my understanding
By Treckin on 6/9/2007 9:35:55 PM , Rating: 2
Thanks for the feedback.
As it may that my understanding of the mechanics behind the limitations of air cooling are somewhat flawed, consider this:
Why is it then that the temperature of the fins on the air cooler are not the same temperature as the core?
Obviously there is a block between the heat coming out of the core, and the final destination of the core. And as to die shrinks, I was right. You were simply nitpicking (badly). If we were cooling individual transistors, than your bringing them into it would have relevance. However, as we are not, you shouldn't have. More transistors per square millimeter = more heat, albeit each individual T has a lower thermal output.
My lack of micro-processor design knowledge will start here, so forgive me and feel free to correct me-
If the CNT's are ballistic connectors ion addition to being one of the best thermal conductors around, wouldn't manufacturing the connections (wires?) inside a core allow for the pathways to functions exactly as a great conductor in addition to a heat pipe? I don't think one would preclude the other from working; the electrons are conveyed along the atoms and bonds themselves, while the heat is transmitted via conduction though the center of the tube.

RE: In my understanding
By jtok202 on 6/10/2007 12:53:14 AM , Rating: 2
Check out a couple of articles on the design and energy usage for C.P.U's I feel that will give you a better understanding of the need for cooling within a die. Interestingly in millions of transistors there are significantly more transisters in a C2D versus a 939/AM2 however the actual Thermal output is decreased and energy usage is down this goes hand in hand with more efficient transistors even after a increase of density per mm. The idea of a more efficient cover for the CPU is intriguing and there I see a real application for the carbon nanotubes. (if heat transfer is one way they will kill the idea of peltiers :( ) Lapping is not only to remove the nickel which is a relatively poor heat conductor but also to smooth out the insane bows that are present in most CPUS AMD's and INTELS. (read anandtech for a good article on C2D imo and check out if you want to see a community for all things overclock/watercooling/lapping so on imo)

RE: In my understanding
By Cogman on 6/10/2007 8:57:35 AM , Rating: 1
For the first questions, that is simple, it is because the heat transferred from core - heatspreader - heatsink bottom - fins is not a perfect one thermally. That's all there is too it. However, we know that air cooling is not at its best, you just have to look at a water cooling system to know you can do better. A wc system will get better and lower temps then an air cooling system will, yet their is no change in core, heatspreader, or the likes.

as stated above, due to processor design, the Core 2 Duo actually produces less heat while having more transistors to its predecessors.

and for the next question, You are perfectly right in saying that one would not preclude the other from working, however, you really would not need that great of cooling with a CNT processor as they would withstand much higher temps without causing instability (they don't have problems with electron leakage or burnout that occurs on modern cpus when they overheat) the reason it is not used is because the manufacturing process is not there. I don't know how it is now, but just a couple of years ago CNT where some of the most expensive substance on earth, manipulating them is a fair amount different from the ability to manipulate silicon, that is why they are not used (yet) for microprocessor, but trust me, they would be far Superior to current silicon.

"Nowadays you can buy a CPU cheaper than the CPU fan." -- Unnamed AMD executive

Most Popular ArticlesSmartphone Screen Protectors – What To Look For
September 21, 2016, 9:33 AM
UN Meeting to Tackle Antimicrobial Resistance
September 21, 2016, 9:52 AM
Walmart may get "Robot Shopping Carts?"
September 17, 2016, 6:01 AM
5 Cases for iPhone 7 and 7 iPhone Plus
September 18, 2016, 10:08 AM
Update: Problem-Free Galaxy Note7s CPSC Approved
September 22, 2016, 5:30 AM

Copyright 2016 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki