Maker Pro
Maker Pro

Synthetic diamonds transfer heat better than metals.

S

Skybuck Flying

Jan 1, 1970
0
Some questions:

1. How much percentage are synthetic diamonds better at transferring heat
than metals like copper and other metals in todays cooling devices for
computerchips ? A rough percentage estimate is good enough for me.

2. Does it make sense to use synthetic diamonds for cpu coolers or is there
another limiting factor, like maybe the ammount of heat that can be
transfered to the air ?

Bye,
Skybuck.
 
L

Lorne Mower

Jan 1, 1970
0
Skybuck Flying said:
Some questions:

1. How much percentage are synthetic diamonds better at transferring heat
than metals like copper and other metals in todays cooling devices for
computerchips ? A rough percentage estimate is good enough for me.

Synthetic diamond generally around 250% better than copper, pure diamond
around 500% better
2. Does it make sense to use synthetic diamonds for cpu coolers or is
there another limiting factor, like maybe the ammount of heat that can be
transfered to the air ?

In an ideal world it would be the best solution, however I would swop my
diamond heatsink for a fanned copper heatpipe plus a brace of 8800GTX's any
day!!!

Lorne
 
R

Robert Baer

Jan 1, 1970
0
Skybuck said:
Some questions:

1. How much percentage are synthetic diamonds better at transferring heat
than metals like copper and other metals in todays cooling devices for
computerchips ? A rough percentage estimate is good enough for me.

2. Does it make sense to use synthetic diamonds for cpu coolers or is there
another limiting factor, like maybe the ammount of heat that can be
transfered to the air ?

Bye,
Skybuck.
I have no numbers, but i understand that diamond is a better thermal
conductor than copper.
Actuall, if there is a lot of heat to transfer, it makes sense (and
this has been done) to mount the IC directly on a diamond chip, which
then is mounted on a large heatsink.
And you are right; moving thermal energy from the heatsink to cooler
air can be a problem.
Basically, you do not want lamanar flow at the heatsink surface, as
that boundary can become (relatively) insulating.
Once the transfer is made, lamanar flow is desirable to get that heat
away as efficently as possible.
Can become a nasty tradeoff...
 
M

mike

Jan 1, 1970
0
| Skybuck Flying wrote:
|
| > Some questions:
| >
| > 1. How much percentage are synthetic diamonds better at
transferring heat
| > than metals like copper and other metals in todays cooling devices
for
| > computerchips ? A rough percentage estimate is good enough for me.
| >
| > 2. Does it make sense to use synthetic diamonds for cpu coolers or
is there
| > another limiting factor, like maybe the ammount of heat that can
be
| > transfered to the air ?
| >
| > Bye,
| > Skybuck.
| >
| >
| >
| >
| >
| I have no numbers, but i understand that diamond is a better
thermal
| conductor than copper.
| Actuall, if there is a lot of heat to transfer, it makes sense
(and
| this has been done) to mount the IC directly on a diamond chip,
which
| then is mounted on a large heatsink.
| And you are right; moving thermal energy from the heatsink to
cooler
| air can be a problem.
| Basically, you do not want lamanar flow at the heatsink surface,
as
| that boundary can become (relatively) insulating.
| Once the transfer is made, lamanar flow is desirable to get that
heat
| away as efficently as possible.
| Can become a nasty tradeoff...

Someday we will be able to make large diamonds cheaply. Then we will
make solid diamond heat sinks with internal water channels and
dissipate 10 kw in a few cubic centimeters. Pentium 4's will come
back in stile and run at 10GHz!

Mike Sicilian
 
B

Boris Mohar

Jan 1, 1970
0
Someday we will be able to make large diamonds cheaply. Then we will
make solid diamond heat sinks with internal water channels and
dissipate 10 kw in a few cubic centimeters. Pentium 4's will come
back in stile and run at 10GHz!

Mike Sicilian
Won't work. To suck up 10KW from few cc the water would have to flow so
fast that it would likely heat up the heatsink. Of course I didn't do the
math :)
 
J

John Larkin

Jan 1, 1970
0
Some questions:

1. How much percentage are synthetic diamonds better at transferring heat
than metals like copper and other metals in todays cooling devices for
computerchips ? A rough percentage estimate is good enough for me.

2. Does it make sense to use synthetic diamonds for cpu coolers or is there
another limiting factor, like maybe the ammount of heat that can be
transfered to the air ?

Bye,
Skybuck.

Pure diamond is roughly twice as good a heat conductor as pure copper.
Isotopically pure diamond is several times better again.

The problem with diamond is that it's not available in big chunks. A
thin film of diamond doesn't spread heat laterally, so interposing it
between a chip and a heat sink makes the thermal situation worse, not
better. Diamond would only be useful if you could make the whole
heatsink out of it, or mount the chip on a big and *thick* hunk of
diamond embedded in a copper heatsink. I could tell you more except
for that damned NDA.

There are diamond-filled greases and epoxies. Their thermal resistance
isn't much better than other types, since the thermal resistance is
dominated by the grease or epoxy between the fill grains. In that the
diamond fill is usually big grains, the diamond stuff may well be
worse.

Nowadays, probably the best heatsink is a water-cooled copper block.

John
 
M

Mike Monett

Jan 1, 1970
0
Won't work. To suck up 10KW from few cc the water would have to
flow so fast that it would likely heat up the heatsink. Of course
I didn't do the math :)
Boris Mohar

How about taking a page from the nuclear power industry? A
pressurized boiling water reactor should have some impressive power
density numbers. We should be able to do the same thing, and use the
steam for cooking, showers, laundry, steam cleaning carpets, etc.

Unfortunately google doesn't want to give up any pages that talk
about watts/cm^3 right now:)

Regards,

Mike Monett

Antiviral, Antibacterial Silver Solution:
http://silversol.freewebpage.org/index.htm
SPICE Analysis of Crystal Oscillators:
http://silversol.freewebpage.org/spice/xtal/clapp.htm
Noise-Rejecting Wideband Sampler:
http://www3.sympatico.ca/add.automation/sampler/intro.htm
 
D

D from BC

Jan 1, 1970
0
Some questions:

1. How much percentage are synthetic diamonds better at
transferring heat
than metals like copper and other metals in todays cooling
devices for
computerchips ? A rough percentage estimate is good enough for
me.

2. Does it make sense to use synthetic diamonds for cpu coolers
or is there
another limiting factor, like maybe the ammount of heat that can
be
transfered to the air ?

Bye,
Skybuck.


------------------
Reminds me of an email I got long ago..and it's still on my
computer and I've pasted below...

..
[Received Email]
Hi D

The atoms of any crystal are in a state of vibration, their
average kinetic energy being measured by the absolute
temperature of the crystal. In certain phenomena it becomes
evident that this energy is divided into discrete bundles; the
energy bundles behave like particles in some respects and are
termed phonons.

With diamond particles in a thermal adhesive, when these phonons
reach the outer edge of a diamond particle it seems they tend to
be reflected back into the particle rather than conveying the
energy (heat) to the resin or an adjacent diamond particle.

We are working on solutions to this phenomenon, but simply
adding diamond powder to an existing thermal adhesive or making
a thermal adhesive with pure diamond filler is not nearly as
effective nor as simple as it would seem to be.

This being said, our OEM customers consistently tell use that
our adhesives are the best performing ones available. You may
be interested in our Medium Cure Ceramic Epoxy and our Slow Cure
Ceramic Epoxy. The Slow Cure Ceramic Epoxy performs on par
with out silver adhesives.

[My Email I sent]
Hello

I know that diamond is an excellent heat conductor and I'm
curious as to why Artic Silver does not have an "ultimate
thermally conductive" adhesive? I realize price might be higher
but some people don't mind.
 
B

BFoelsch

Jan 1, 1970
0
Won't work. To suck up 10KW from few cc the water would have to flow so
fast that it would likely heat up the heatsink. Of course I didn't do the
math :)
Well let's see.

10kw = 34,130 BTU/hr

That works out to a Flow-temp rise constant of about 68, in gals/min
and degrees F.

In other words, 1 gpm will move 34,000 btu.hr at a temperature rise of
68 degrees F, 2gpm will do it at a 34 degree rise, etc.

I'll let someone else put a number on the surface area of heat
exchanger required.
 
R

Robert Baer

Jan 1, 1970
0
Boris said:
Won't work. To suck up 10KW from few cc the water would have to flow so
fast that it would likely heat up the heatsink. Of course I didn't do the
math :)
Perhaps the use of a low boiling point liquid that has been
previously cooled to almost freezing may do; laminar flow on the
heatsink surfaces will be a severe impediment.
 
R

Robert Baer

Jan 1, 1970
0
John said:
Pure diamond is roughly twice as good a heat conductor as pure copper.
Isotopically pure diamond is several times better again.

The problem with diamond is that it's not available in big chunks. A
thin film of diamond doesn't spread heat laterally, so interposing it
between a chip and a heat sink makes the thermal situation worse, not
better. Diamond would only be useful if you could make the whole
heatsink out of it, or mount the chip on a big and *thick* hunk of
diamond embedded in a copper heatsink. I could tell you more except
for that damned NDA.

There are diamond-filled greases and epoxies. Their thermal resistance
isn't much better than other types, since the thermal resistance is
dominated by the grease or epoxy between the fill grains. In that the
diamond fill is usually big grains, the diamond stuff may well be
worse.

Nowadays, probably the best heatsink is a water-cooled copper block.

John
Better yet, a water-cooled block of silver.
 
J

jasen

Jan 1, 1970
0
Won't work. To suck up 10KW from few cc the water would have to flow so
fast that it would likely heat up the heatsink. Of course I didn't do the
math :)

allowing for a 20K rise in the temperature of the coolant
that'd require about 2.48l per second

possibly a 1 or 2 KW pump would be enough :)
 
mike said:
Someday we will be able to make large diamonds cheaply. Then we will
make solid diamond heat sinks with internal water channels and
dissipate 10 kw in a few cubic centimeters. Pentium 4's will come
back in stile and run at 10GHz!

Pyrolytic graphite looks good too, 1900 W/mC at 30C, vs 386 for pure copper
at 20C, according to page 338 of the 1998 Schaum's Outline on Heat Transfer,
but manufacturers only spec 300 or so. I wonder why.

Nick
 
N

Nick Maclaren

Jan 1, 1970
0
|>
|> The problem with diamond is that it's not available in big chunks. A
|> thin film of diamond doesn't spread heat laterally, so interposing it
|> between a chip and a heat sink makes the thermal situation worse, not
|> better. Diamond would only be useful if you could make the whole
|> heatsink out of it, or mount the chip on a big and *thick* hunk of
|> diamond embedded in a copper heatsink. I could tell you more except
|> for that damned NDA.

Standing for Not Deliverable, Actually? :)


Regards,
Nick Maclaren.
 
N

Nick Maclaren

Jan 1, 1970
0
|>
|> Someday we will be able to make large diamonds cheaply. Then we will
|> make solid diamond heat sinks with internal water channels and
|> dissipate 10 kw in a few cubic centimeters. Pentium 4's will come
|> back in stile and run at 10GHz!

Illiad has said it all:

http://ars.userfriendly.org/cartoons/?id=20061114


Regards,
Nick Maclaren.
 
A

Ancient_Hacker

Jan 1, 1970
0
It all depends on how fast you need to push the IC's and your weight
constraints. Right now the best bet might be to run a plain old freon
refrigerant past the back of the CPU chip-- plain old mechanical
compressor and all-- IF you have the space and can tolerate the power
and weight. Much cheaper than using diamonds and you can get much
lower chip temperature and therefore more speed.
 
J

joseph2k

Jan 1, 1970
0
Robert said:
Better yet, a water-cooled block of silver.

Not even cost effective, but probably the best that can be done in an
ordinary office or household setting. In an industrial setting using LN2
in a copper plate wins easily.
 
F

Fred Bartoli

Jan 1, 1970
0
BFoelsch a écrit :
Well let's see.

10kw = 34,130 BTU/hr
Arghhh...


That works out to a Flow-temp rise constant of about 68, in gals/min
and degrees F.

Arrrgggghhhhhh...

In other words, 1 gpm will move 34,000 btu.hr at a temperature rise of
68 degrees F, 2gpm will do it at a 34 degree rise, etc.

AAAArrrgggghhhhhh...

I'll let someone else put a number on the surface area of heat
exchanger required.

Not me, you just killed me...
 
R

Richard Henry

Jan 1, 1970
0
Lorne said:
Synthetic diamond generally around 250% better than copper, pure diamond
around 500% better


In an ideal world it would be the best solution, however I would swop my
diamond heatsink for a fanned copper heatpipe plus a brace of 8800GTX's any
day!!!

Synthetic diamonds can be made at any level of "purity" desired.

What do you mean by "pure diamond"?

Where do you get your figures for your thermal conductivity comparisons?
 

Similar threads

Top