Diamond based Heatsinks (Highest Thermal Conductivity Available!)

H

Hvatum

Gawd
Joined
Apr 21, 2005
Messages
882
Just noticed this older issue of Wired had an article about synthetic diamonds (2003). Does anyone know the progress of this? Having a Heatsink or substrate made out of diamond would obviously be pretty sweet. Diamond has the highest thermal conducvitiy of any element on the entire planet.

Imagine the OC you could get with a solid diamond heatsink and phase change cooling setup... *drool*

http://www.wired.com/wired/archive/11.09/diamond.html
 
More on this here.

Diamond is just too expensive atm.
Thx De Beers :mad:

There is also talk of using diamond in stead of silicon in chips.
 
Dunno said:
More on this here.

Diamond is just too expensive atm.
Thx De Beers :mad:

There is also talk of using diamond in stead of silicon in chips.
Click to expand...

Naturally occuring diamond has too many impurities to be used for large scale chip production anyway. De Beers could care less about the industrial diamond market. They're only out to make your wife believe that blood-soaked hunks of carbon from the earth mined by poor Africans working in slave-like conditions are the representation of true love. They do such a good job too. :mad:

Diamonds as sythesized by Apollo can be made in Wafer forms at up almost a foot square (Large enough for chips) and cost $5 a caret right now. If they could get the price down to $1 a caret it would be concievable to have absolute top end heatsinks made out solid diamond for around $120. They could also probably sell the dust leftover from shaping the diamonds and put it in Arctic silver! Just think:

Arctic Diamond Thermal Paste (Made from actual diamond dust).
 
Hvatum said:
Naturally occuring diamond has too many impurities to be used for large scale chip production anyway. De Beers could care less about the industrial diamond market. They're only out to make your wife believe that blood-soaked hunks of carbon from the earth mined by poor Africans working in slave-like conditions are the representation of true love. They do such a good job too. :mad:

Diamonds as sythesized by Apollo can be made in Wafer forms at up almost a foot square (Large enough for chips) and cost $5 a caret right now. If they could get the price down to $1 a caret it would be concievable to have absolute top end heatsinks made out solid diamond for around $120. They could also probably sell the dust leftover from shaping the diamonds and put it in Arctic silver! Just think:

Arctic Diamond Thermal Paste (Made from actual diamond dust).
Click to expand...

DeBeers definitely cares about synthetic diamond technology, because anything that lends itself to making cheaper diamond is a threat to their business. Recent technology has led to the synthesis of diamonds which are indistinguishable from naturalk ones (except for the fractional cost). Check out this article, it's an interesting read...
Anyhow, aside from bulk cost, forming/machining heatsink shapes will not be cheap anytime very soon. I can see diamond heatspreaders becoming available for very high-performance applications, however. Think huge power densities and huge budgets, and companies like NASA come to mind...
 
yes but synthetic is still synthetic. for the big bucks of debeers the snobbery is gigantic, and none of their customers would ever be caught wearing 'synthetic'.... they have HUGE stockpiles round the back in order to keep the diamond price up, and just dribble it out. also, you cant buy a diamond, you have to buy a pouch with some good and some crap in them, again in order to keep the price up.

still girls best friend etc....
 
This would be very interesting to get your hands on.. The possibilities are quite intriquing to think about. I wonder what a completely diamond cooled computer would cost..

(thinks about something else) I wonder what overclocks you would get on water with diamonds as the blocks... *drools*
 
heh.....
a while ago i priced out some cvd diamond 'thermal windows'.

They are actually capable of tailoring the thermal conductivity : they offer 3 grades.

A 1 inch diameter .5 mm thick diamond thermal window of the top grade was about 8000$ US.
 
That's fantastic. I'd love to see diamond semiconductors. Imagine what kind of speeds you could hit if 500df+ wouldn't damage the core. Not to mention the diamond HSF's. Hopefully we'll see some more on this in the near future, when they can actually mass produce the diamond wafers.
 
Keep in mind the LARGEST they can make is ~ 3 carats. I'm sure it will be a while before they can make one large enough to carve a heatsink out of..
 
whats the most common gemstone ocurring in nature?
diamond. even more common than garnet.

without debeers in the way real diamonds would be cheaper to mine than synthetic is to manufacture.

now who says you'd need solid sheet pieces to make a heatsink or heat spreader. just encase or contain the low grade diamond pieces in a ceramic grid where the diamonds surface is lower than the ceramic container as it applies to the cpu surface.

or better yet use a conatiner structure made of buckyballs to contain the diamonds. sure you'd lose some thermal benefit but you'd get a far stronger container that would be able to act like a semi conductor on its own. one way electrical flow through the structure could be possible with this as well for a peltier effect with diamond "spikes" radiating heat into a thermal absorbtion mass like water. add tubes around the upward spikes and make heat pipes on top of it all. you then have a super tough peltier heatspreader feeding heatpipes. no sink needed. no worry about tightening down the latch too much and cracking the die either.

diamond dust wouldn't make a good thermal compound. the mass youd need to suspend the granules in to withstand the heat they could would kill the thermal benefit. plus the granules would be small enough to deteriorate under heat turning back to carbon and lose thermal properties till you end up with a fried cpu.
 
wetware_interface said:
whats the most common gemstone ocurring in nature?
diamond. even more common than garnet.

without debeers in the way real diamonds would be cheaper to mine than synthetic is to manufacture.

now who says you'd need solid sheet pieces to make a heatsink or heat spreader. just encase or contain the low grade diamond pieces in a ceramic grid where the diamonds surface is lower than the ceramic container as it applies to the cpu surface.

or better yet use a conatiner structure made of buckyballs to contain the diamonds. sure you'd lose some thermal benefit but you'd get a far stronger container that would be able to act like a semi conductor on its own. one way electrical flow through the structure could be possible with this as well for a peltier effect with diamond "spikes" radiating heat into a thermal absorbtion mass like water. add tubes around the upward spikes and make heat pipes on top of it all. you then have a super tough peltier heatspreader feeding heatpipes. no sink needed. no worry about tightening down the latch too much and cracking the die either.

diamond dust wouldn't make a good thermal compound. the mass youd need to suspend the granules in to withstand the heat they could would kill the thermal benefit. plus the granules would be small enough to deteriorate under heat turning back to carbon and lose thermal properties till you end up with a fried cpu.
Click to expand...

Diamond powder in a metal matrix would be more likely to work than in a ceramic matrix, however I don't know if it would end up being better or worse than just pure metal. You'd have to think about things like interface quality which is never perfect and hence will degrade thermal conductivity.
Also, diamond is metastable, but will remain in diamond form at CPU temperatures so thermal degradation is not an issue. I don't understand what you mean about the suspension - diamond is much less dense than silver, which is suspended in AS products. Make a thick enough goop and it won't be a problem.
 
Diamonds as heatsinks is pretty crazy idea, but don't forget that diamond is really just carbon made up in certain formation, I think instead of diamond heatsinks there will be more move to the use of things like carbon nanotubes, which are a great invention once they can mass produce them, cuz they also have really high thermal conductivity properties, probably up there with diamond, and make great semiconductors, so they'll probably replace silicon, in a matter of time. Besides they may be used in many many applications cuz their tensile and compressive stress properties are huge compared to that of steal, so way stronger way lighter materials for building, imagine super strong paperweight cases with great thermal conducting properties. boo yeah
 
Bling-Bling for the cpu?

wouldnt the whole CPU die have to me made out of this stuff to do any good?
 
ubckid2050 said:
Diamonds as heatsinks is pretty crazy idea, but don't forget that diamond is really just carbon made up in certain formation, I think instead of diamond heatsinks there will be more move to the use of things like carbon nanotubes, which are a great invention once they can mass produce them, cuz they also have really high thermal conductivity properties, probably up there with diamond, and make great semiconductors, so they'll probably replace silicon, in a matter of time. Besides they may be used in many many applications cuz their tensile and compressive stress properties are huge compared to that of steal, so way stronger way lighter materials for building, imagine super strong paperweight cases with great thermal conducting properties. boo yeah
Click to expand...


Carbon nanotubes /= diamond, you cannot assume that properties are comparable. Nanotubes are anisotropic in many properties (not the same in every direction). For example, electrical or thermal conductivity may be very high along the axis of the tube, but very low from side to side. Every advanced material has its niche in the tech world, so although there are places where carbon nanotubes may be advantageous, my opinion is that there is no practical use for them in cooling or structural applications at the moment.

A side note on the strength issue that you mentioned - many materials in fiber form are superior to steel, and many composite materials in bulk form are superior to bulk steel. Composites having epoxy/plastic matrices and glass/carbon/other ceramic reinforcements are increasingly common in the world today, in products ranging from aerospace components to plastic patio chairs. Even toothpaste could be considered a nano-material :p
 
indeed you are correct diamond is not equal to carbon nanotubes, never said so, and you are quite right insaying across different axis the material properties of carbon nanotubes behave differently, However, I wish to disagree at least in the structural application area because there are large moves to use carbon nanotube ribbon to make a space elevator, So I believe that the structural applications will no doubt come in time. The cooling applications all I have to go with is that the thermal conductivity of carbon nanotubes, especially the single walled version have high thermal conductivity, in the ranges if 6600 W/mK in the photovoltaic applications based on some research done by Sheila G. Bailey et al. Anyways, I'm sure for a long time Diamond will have more potential for cooling applications.
 
ubckid2050 said:
indeed you are correct diamond is not equal to carbon nanotubes, never said so, and you are quite right insaying across different axis the material properties of carbon nanotubes behave differently, However, I wish to disagree at least in the structural application area because there are large moves to use carbon nanotube ribbon to make a space elevator, So I believe that the structural applications will no doubt come in time. The cooling applications all I have to go with is that the thermal conductivity of carbon nanotubes, especially the single walled version have high thermal conductivity, in the ranges if 6600 W/mK in the photovoltaic applications based on some research done by Sheila G. Bailey et al. Anyways, I'm sure for a long time Diamond will have more potential for cooling applications.
Click to expand...


All in good time... :D The space elevator sounds pretty sweet. Then again, whether or not space travel can yet be considered a 'practical application' is another debate. I might have to check out Bailey's work, that sounds pretty interesting.
 
Someone brought up that the core might need to be made out of diamond as well as the heatsink for any of this to do any good at all. This gives me an idea, what if both your CPU core and heatsink were just one big chunk of diamond (or one big chunk of any material for that matter)?

Someone also brought up that a CPU core made out of diamond could withstand much higher temperatures without being damaged, if this proves true we might very well end up with a blast from the past with passively cooled processors, or chips that don’t even need a heatsink at all.

Think about that for a second, how strange would it feel to be running heatsink-les processors again?...Although this time they could get hot enough to melt lead :p Overdoing your overclock could actually end with you dragging a hose down to your basement to cool off the core of your processor enough to pick it up, while light pours down from the tiny hole where the hot diamond core burned through the case, desk, and floor above :eek:
 
Unknown-One said:
Someone brought up that the core might need to be made out of diamond as well as the heatsink for any of this to do any good at all. This gives me an idea, what if both your CPU core and heatsink were just one big chunk of diamond (or one big chunk of any material for that matter)?

Someone also brought up that a CPU core made out of diamond could withstand much higher temperatures without being damaged, if this proves true we might very well end up with a blast from the past with passively cooled processors, or chips that don’t even need a heatsink at all.

Think about that for a second, how strange would it feel to be running heatsink-les processors again?...Although this time they could get hot enough to melt lead :p Overdoing your overclock could actually end with you dragging a hose down to your basement to cool off the core of your processor enough to pick it up, while light pours down from the tiny hole where the hot diamond core burned through the case, desk, and floor above :eek:
Click to expand...


so many things that you said will never happen...won't see a sinkless cpu ever, because the raw heat made would melt anything within the vicinity of the CPU (mobo pcb, case, etc)...also, i don't think processors would ever be passively cooled, because it's insanely hard for a chunk of diamond to be milled into shapes like we have now.

also, the conductivity of diamond doesn't mean that it will release its heat faster than, say, copper, it would be like aluminum and just soak up the heat without doing much to dissipate it.

diamond in any cooling application would be best used as an interface between the source and the dissipation medium (this case, the heatsink), not specifically as one, the other, or both simultaneously.
 
Actually.. using Diamond inplace of silicon might be the best use for it, since it has a much higher heat tollarence, which would allow for alot of headroom in overclocking.
 
RaphaelVinceti said:
Actually.. using Diamond inplace of silicon might be the best use for it, since it has a much higher heat tollarence, which would allow for alot of headroom in overclocking.
Click to expand...


Does anyone have evidence that diamond-based semiconductors would allow for higher operating temperatures? Elemental silicon itself is not remotely close to being thermodynamically unstable/molten at 70 or even 200C... I'm just wondering where this notion came from, does anbody with background in electrical eng. or semiconductors have any input on elevated-temp related device failure?
 
zer0signal667 said:
Does anyone have evidence that diamond-based semiconductors would allow for higher operating temperatures? Elemental silicon itself is not remotely close to being thermodynamically unstable/molten at 70 or even 200C... I'm just wondering where this notion came from, does anbody with background in electrical eng. or semiconductors have any input on elevated-temp related device failure?
Click to expand...

Hmmm that's a good observation. I think that silicon-based CPUs become unstable far before they actually become thermodynamically unstable because of the sensitivity of the transisters, and such. Perhaps, since hundreds of millions of transisters are packed into a piece of silicon, a decent amount of heat would cause it to function incorrectly, though the silicon itself is fine. I can't think of a good example, but think about a solid aluminum bat. Pretty hard and sturdy right? How about a bunch of aluminum foil stacked atop of each other? Same material, but the foil is much more sensitive to bending, and deformation etc. Not the greatest example, but I hope it fits in the situation.

I think maybe diamond-based cores will tolerate much more heat before the transisters start to function incorrecly, but of course this is all speculation on my part.
 
Yoshiyuki Blade said:
Hmmm that's a good observation. I think that silicon-based CPUs become unstable far before they actually become thermodynamically unstable because of the sensitivity of the transisters, and such. Perhaps, since hundreds of millions of transisters are packed into a piece of silicon, a decent amount of heat would cause it to function incorrectly, though the silicon itself is fine. I can't think of a good example, but think about a solid aluminum bat. Pretty hard and sturdy right? How about a bunch of aluminum foil stacked atop of each other? Same material, but the foil is much more sensitive to bending, and deformation etc. Not the greatest example, but I hope it fits in the situation.

I think maybe diamond-based cores will tolerate much more heat before the transisters start to function incorrecly, but of course this is all speculation on my part.
Click to expand...

Let's say that the component that is actually malfunctioning is a random transistor/semiconductor, not the silicon or diamond substrate that is it within. Perhaps the higher thermal conductivity of diamond will allow cooler operating temperatures even with higher power consumption, and in that case the benefit is not an increase in some magical property called "heat tolerance" - it is increased heat transfer efficiency.

On the other hand, let's say that diamond-based semiconductors offer some distinct advantage, on a quantum mechanics level, which allows them to operate at higher temperatures than silicon-based ones. In this case, the benefit is possibly the magical "heat tolerance" property, which is undoubtedly an agglomeration of many much more complex properties. But in the end, the benefit is an increase in allowable operating temperature.

What I'm wondering is which scenario, or maybe both or neither, is true. I don't mean to come off as an ass about it, but I can't stand when people make statements/generalizatons with seemingly no evidence to back it up or even explain it (I'm referring to people in general, not you). My inquisitive mind just wants to know why
:D
 
zer0signal667 said:
Let's say that the component that is actually malfunctioning is a random transistor/semiconductor, not the silicon or diamond substrate that is it within. Perhaps the higher thermal conductivity of diamond will allow cooler operating temperatures even with higher power consumption, and in that case the benefit is not an increase in some magical property called "heat tolerance" - it is increased heat transfer efficiency.

On the other hand, let's say that diamond-based semiconductors offer some distinct advantage, on a quantum mechanics level, which allows them to operate at higher temperatures than silicon-based ones. In this case, the benefit is possibly the magical "heat tolerance" property, which is undoubtedly an agglomeration of many much more complex properties. But in the end, the benefit is an increase in allowable operating temperature.

What I'm wondering is which scenario, or maybe both or neither, is true. I don't mean to come off as an ass about it, but I can't stand when people make statements/generalizatons with seemingly no evidence to back it up or even explain it (I'm referring to people in general, not you). My inquisitive mind just wants to know why
:D
Click to expand...

Haha, I know how you feel. Well all I can say is "who knows?" Until we can find some scientific evidence on this, let's just hope that diamond-based CPUs > Silicon-based :p. I'm pretty confident that diamonds are better though, since in the article they imply that diamonds were just way too expensive and rare to be used this way. Otherwise, we probably would have used them by now.
 
I can think of two factors that could help a Diamond CPU's heat tolerence over a Silicon one.

1. The atoms are smaller.
Therefore more atoms to carry the volages across the narrowest gates, etc.
At the moment the smallest parts are only a few atom across/deep.
Just having an extra atom may make the whole thing more robust.

2. Stronger bonds.
Because the atoms are smaller the bond between them are stronger.
This will probably help to cut down on thermal migration of atoms acoss a gate.
Also making it more rebust.

Luck........ :D
 
wetware_interface said:
whats the most common gemstone ocurring in nature?
diamond. even more common than garnet.

without debeers in the way real diamonds would be cheaper to mine than synthetic is to manufacture.

now who says you'd need solid sheet pieces to make a heatsink or heat spreader. just encase or contain the low grade diamond pieces in a ceramic grid where the diamonds surface is lower than the ceramic container as it applies to the cpu surface.

or better yet use a conatiner structure made of buckyballs to contain the diamonds. sure you'd lose some thermal benefit but you'd get a far stronger container that would be able to act like a semi conductor on its own. one way electrical flow through the structure could be possible with this as well for a peltier effect with diamond "spikes" radiating heat into a thermal absorbtion mass like water. add tubes around the upward spikes and make heat pipes on top of it all. you then have a super tough peltier heatspreader feeding heatpipes. no sink needed. no worry about tightening down the latch too much and cracking the die either.

diamond dust wouldn't make a good thermal compound. the mass youd need to suspend the granules in to withstand the heat they could would kill the thermal benefit. plus the granules would be small enough to deteriorate under heat turning back to carbon and lose thermal properties till you end up with a fried cpu.
Click to expand...

A late entry in the "Newbie Post of the Year" contest. You're not supposed to be new and smart :p
 
Well going on what that other guy said, lets not forget that in tenth grade chemisty topic Bonding. Networks solids have the highest boiling points and the lowest boiling points, which would suggest it is a great conductor of heat. The bonds between the atoms of a network solid (aka a diamond) are among the strongest. It takes lots of pressure and high temperatures to turn network solids to a molten state, and are good conductors of heat at all times, but would making heatsinks like this really be worth it because of the mass production of copper/silver heatsinks? A diamond would be great as a heat conductor, but mixed with metal, I am not sure of. They would have to be pretty big for them to make a difference, because copper/silver does do a good job. And yes, the atoms would alow less thermal migration, making it an even better solution, but the diamonds would probably the footing connecting to the CPU to dissipate the heat straight away to be at maximum effect, because I really dont think it would help or at least minimal help being scattered throughout the heatsink.

I may have made some mistakes because my knowledge is limited, just trying to help the conversation.
 
Relentless3O said:
Well going on what that other guy said, lets not forget that in tenth grade chemisty topic Bonding. Networks solids have the highest boiling points and the lowest boiling points, which would suggest it is a great conductor of heat. The bonds between the atoms of a network solid (aka a diamond) are among the strongest. It takes lots of pressure and high temperatures to turn network solids to a molten state, and are good conductors of heat at all times, but would making heatsinks like this really be worth it because of the mass production of copper/silver heatsinks? A diamond would be great as a heat conductor, but mixed with metal, I am not sure of. They would have to be pretty big for them to make a difference, because copper/silver does do a good job. And yes, the atoms would alow less thermal migration, making it an even better solution, but the diamonds would probably the footing connecting to the CPU to dissipate the heat straight away to be at maximum effect, because I really dont think it would help or at least minimal help being scattered throughout the heatsink.

I may have made some mistakes because my knowledge is limited, just trying to help the conversation.
Click to expand...


You just scrambled my brain... :confused:

I'll assume that by network solids you mean covalent bonded materials... these don't necessarily have high thermal conductivity.

Diamond in metal would act the same way as silver particles in silicone (thermal paste). The thermal conductivity would be affected by volume percentage rather than particle size, and volume percentage can be most effectively increased by engineering the proper particle size distribution (particles of different sizes can be designed to pack well).

I'm not sure why you think that diamond/diamond composites would be better suited for heat transfer straight away from the CPU... there's no reason that would be true unless the material had some anisotropy that caused such an effect.

edit - I apologize if this sounded harsh, but since my life is basically composite materials with a strong interest in heat transfer I hate to see people displaying questionable information on these topics :D
 
No, I meant that a network solid, such as a diamond, has/may have a very high conductivity for heat and can stand much more then a metal can. I dont know the melting/freezing point of a diamond, and I really dont think anyone has tried that before. Maybe because its a dumb idea, or maybe that its so hard to achieve? I understand the more volume something has the more heat conductivity it has (as in thermal paste? am I understanding you right?) Im just saying that having maybe diamond plates or a large concentration of diamonds at the bottom of the heatsink would be better for it, and having diamonds disipated throughout much better.
Yes, network solids are bonded covalently and are immobile in the structure which would allow one of two things- A) The heat transfered throught out the heatsink and thus cooling the CPU or B) Not conducting at all, doing absolutely nothing. I do not know if the diamond would conduct heat, but in theory of a network solid, it should.
 
Relentless3O said:
No, I meant that a network solid, such as a diamond, has/may have a very high conductivity for heat and can stand much more then a metal can. I dont know the melting/freezing point of a diamond, and I really dont think anyone has tried that before. Maybe because its a dumb idea, or maybe that its so hard to achieve? I understand the more volume something has the more heat conductivity it has (as in thermal paste? am I understanding you right?) Im just saying that having maybe diamond plates or a large concentration of diamonds at the bottom of the heatsink would be better for it, and having diamonds disipated throughout much better.
Yes, network solids are bonded covalently and are immobile in the structure which would allow one of two things- A) The heat transfered throught out the heatsink and thus cooling the CPU or B) Not conducting at all, doing absolutely nothing. I do not know if the diamond would conduct heat, but in theory of a network solid, it should.
Click to expand...


Just remember that all generalizations are false ;)

I won't go back into materials theory re: bonding and thermal properties... however I'll try to clear up the volume fraction/thermal paste things. Volume and thermal conductivity do not go together - thermal conductivity is not dependent on dimensions or area or volume. You suggested that for a diamond composite to have good heat transfer capabilities, they diamond particles would have to be large. I was disputing this, and suggesting that the particles should be sized appropriately so that they pack densely within the composite.

For example, a jar of marbles, all the same size: assuming you can pack them as efficiently as possible, you'll still have almost 30% of the jar filled with air. Add some more smaller marbles to fill in some of the spaces inbetween, and you are left with only maybe 20% air. Repeat with smaller marbles, approach zero air volume.

This works for thermal paste because you want lots of silver in there. This works for particle-reinforced composites, to an extent, because you desire some property of the filler material for the bulk material. This doesn't work for gumball machines because some people would get ripped off paying $.25 for a millimeter-sized gumball. :p
 
Hvatum said:
Just noticed this older issue of Wired had an article about synthetic diamonds (2003). Does anyone know the progress of this? Having a Heatsink or substrate made out of diamond would obviously be pretty sweet. Diamond has the highest thermal conducvitiy of any element on the entire planet.

Imagine the OC you could get with a solid diamond heatsink and phase change cooling setup... *drool*

http://www.wired.com/wired/archive/11.09/diamond.html
Click to expand...

Diamond is not an element.
Ribs1
 
*points above him* Nope.. Its a carbon based compound created by intense heat/pressure near the earth's core. Or in a lab.. one of the two..

(o.o... Cheap engagement ring with a 2 carat pink diamond... only.. lab created.. oh well.. she wont know the difference)
 
before I go home, I just wanted to mention that there are quite a few papers on using emical-vapor-deposited (CVD) diamond as an interface for Laser LEDs. I also remember seeing an article concerning the use of diamond as the packaging substrate for multi-chip packages in order to better spread heat around.

Ah here it is http://www.red-river.com/Publicatio...f a Miniaturized Commercial Supercomputer.pdf

look at the bottom of page 2, showing the difference between an aluminum packaging and the diamond packaging. Isn't is pretty?

*ahhh* *oooh* *uuuh*
 
You must log in or register to reply here.
Back
Top