A new heatsink compound (thermal grease)

Archived from groups: alt.comp.hardware.overclocking (More info?)

Hello All

Henkel Loctite have released a new heatsink compound which comes in the form
of a bar of material, deployed rather like a lip-stick.

You rub it on the heatsink/component interface and it applies an appropriate
layer of solid material. under high clamping pressures, the solid undergoes
a change to a liquid at 60 deg C (approx. 150 deg F) and is forced into all
the tiny gaps (present due to imperfections, machining tolerances etc.).

The thermal conductivity is 20% lower than the usual thermal grease - it
says it's 0.03 degC/W at 20psi clamping pressure, dropping to 0.02 degC/W at
100psi - and gives 100% surface wetting with easy application, it says.

My Prescott, o/c from 3 to 3.3 GHz, is running at stock voltages with the
stock Intel heastink. CPU temps reported by MBM5 (Abit AI7) are 60-62 deg C.
I plan to re-mount the heatsink having applied some of this thermstrate, and
I'll let you know if I see a drop in temperature. The current heatsink
compound is the usual white alumina-loaded stuff.

Here's some info on thermstrate:

Does anybody know what the thermal conductivity is that's quoted for the
silver-loaded materials?


RMC, England
2 answers Last reply
More about heatsink compound thermal grease
  1. Archived from groups: alt.comp.hardware.overclocking (More info?)

    On Sat, 25 Sep 2004 23:51:52 +0100, "RMC" <RMC@home.com> wrote:

    >Hello All
    >Henkel Loctite have released a new heatsink compound which comes in the form
    >of a bar of material, deployed rather like a lip-stick.

    Seems this kind of stuff has been around for awhile, they sell it to
    system builders.

    Not very good.

    I'd use radio shack white stuff instead.
  2. Archived from groups: alt.comp.hardware.overclocking (More info?)

    Sounds like it is more aimed at integrators/production:
    o The critical factor there is skill & cycle time

    GM can turn out perfect all plastic cars - but the cycle time per panel
    is still that bit more than steel, and so the economics aren't there yet.

    The difference between (modern) TIM, silver-loaded & plain heatsink
    compound is relatively small. Not using too much is as big a factor in
    its performance as the particular material chosen - it's 2-3oC at most.

    So give it a try, but don't expect something particularly radical:
    o Heatsink & CPU die/spreader are two metal surfaces with microscopic voids
    o Voids can be filled with air, or with some form of heatsink compound
    o Heatsink compound's heat transfer is significantly inferior to metal
    o However, heatsink compound's heat transfer is significantly better than air

    So it doesn't make your heatsink bigger. A better compound may minimise small
    localised hotspots on the die/spreader, and may help some overclocking situations.

    The heatspreader itself can be improved - although not by end-users:
    o Heatspreader undersides are a plate & glob of glue to the chip die
    ---- thus the heatspreader presents a single flat plate to the die
    ---- thus the glue/thermal-compound is thicker than it need be
    o Alternative heatspreader undersides are a plate of 4 (2x2) or 9 (3x3) posts
    ---- the posts contact the die more closely than a normal heatspreader
    ---- so less thermal compound exists in the interface, better transfer, fewer hotspots

    Physical mockups & flotherm (3D FEA) have been used to prove the concept,
    but I think the chip makers are still using the older type - "it's cheap, why fix it".

    Playing around with a duct from the CPU heatsink outlet to a nearby exhaust fan
    can have a bigger impact on your temperatures. Foamex (sign makers) is a foamed
    PVC between a thin skin of PVC - a thermoplastic to easily hand moulded into ducts.
    Dorothy Bradbury
    www.stores.ebay.co.uk/panaflofan for quiet Panaflo fans & other items
    www.dorothybradbury.co.uk (free delivery)
Ask a new question

Read More

Overclocking Heatsinks