Thermal Compund Shootout

[Since87]

What's this? Could it be?

The Holy Grail of Cooling

Found in the Digi-key catalog. (Digikey part number P11440-ND)
Listed with the low low price of $29.88 for a 60mm X 90mm sheet.
None in stock.

P11440-ND“PGS”
(Pyrolytic Graphite
Sheet) Heat Sink Sheet
Features:
• Excellent thermal conductivity: 600 to 800 W/(m.K),
twice as high as copper, three times as high as
aluminum
• Light weight: the same as water (Specific gravity: 1.0
g/cm3), 1/9 that of copper, 1/3 that of aluminum
• Flexible sheet, easy to be cut or trimmed and
withstands repeated bending
• High heat resistance: Stable up to about 400°C even
in the air
• Low thermal resistance
• The main component is carbon. Minimal impurities
and outgassing
• Excellent heat transferring performance. Heat sink and
even distribution of heat
Application Examples:
• Notebook personal computers, DVDs, DVCs, mobile
phones
• Semiconductor manufacturing equipment (Sputtering,
Dry etching, Steppers)
• Optical
Items Characteristics
Thickness 0.1±0.05mm
Density 1.0 g/cm3
Thermal Conductivity (a-b plane) 600 – 800 W/(m.K)
Thermal Resistance 0.4°C/w(10N/cm2)
Electrical Conductivity (a-b plane) 10000 S/cm
Extensional Strength 19.6 Mpa
Expansion Coefficient (a-b plane) 9.3 x 10-7 1/K
(c axis) 32 x 10-6 1/K
Heat Resistance 400°C
Bending (angle 180, R5) 10000 cycles
Dimensions – mm Digi-Key Price Panasonic
L x W Part No. Each Part No.
125 x 180 P11438-ND 94.62 EYG-S131810
125 x 90 P11439-ND 51.46 EYG-S091310
60 x 90 P11440-ND 29.88 EYG-S060910
General Characteristics of PGS
PGS (Pyrolytic Graphite Sheet) is a heat sink sheet with high
thermal conductivity and high flexibility.
PGS is made of graphite with a structure that is close to a single
crystal.
This is achieved by highly-oriented polymer film sheet, a process
which has never been implemented before.

 

[Since87]

Darn, I requested the datasheet from DigiKey.

If I'm interpreting it correctly, the material is only highly conductive down it's length, and not through its thickness. I'm not sure how it would be useful. I've uploaded the datasheet.

 

[wildfrogman]

Concerning the pyrolytic graphite heat spreaders,

I have been reading into these things for quite a bit. It allows the heat to be spread out evenly sideways first~faster etc but still transfers the heat like its supposed to into the heatsink~waterblock that is attached to it. One thing though 0.1mm~0.05mm thickness i dont think is enough since most everybody uses copper base thicker than 1/13th inch which would be equivalent to 1/26th inch of copper in heat spreading performance. For a small superhigh flow thin base waterblock with copper fins or such should work better but overall i would think thicker than 0.1mm would be optimal for best performance on a highly overclocked cpu. Nice info though as it is one of the first i have seen for sale.

 

[deeppow]

Results after 3 days testing.

At fsb=198, Vcore=1.85, cpu mult=9.5 I get the following:
.........delta-T AS3....................delta-T G751
................9C....................................9C
where the delta-T is Tcpu-Tenvironment as measured by MBM5 and a mobo diode.

I have no statistics related to G751, i.e. this is for a single application.

I have 23 mounts with AS3 at the above settings with an average delta-T of 9.2C and a standard deviaton of 0.21. These are using MBM5 readings which show only integer temperature readings so the average and standard deviation relect the number of samples more strongly. Also the results don't show a Gaussian characteristic which averages and standard deviations imply, however this is most like due to the use of MBM5 integer temperature readings.

I ran a quick study related to Vcore variation using this mount of G751 since the delta-T is dependent on that and cpu frequency. Obtained the following results: table. I had to vary the cpu mulitplier because things weren't stable at the lower Vcore.

My conclusion is that these are very hard numbers to get with any accuracy and sure hope Joe has the equipment and time.

 

[1Time]

I've been using this Shin-Etsu product for about a week now and I've only seen a 1C drop in my load and idle temps. Can't compare it to my previous AS3 temps though since I also modded my case and changed fans. However, I can say I'll be switching back to AS3 as soon as I get the chance since I suspect it will do better than this stuff.