Author Topic: really proper way to apply thermal compound?  (Read 6900 times)

0 Members and 1 Guest are viewing this topic.

Online coppercone2Topic starter

  • Super Contributor
  • ***
  • Posts: 9234
  • Country: us
  • $
really proper way to apply thermal compound?
« on: December 09, 2018, 11:03:42 pm »
If you really wanted to do this, is there a better method then the 'baboon fishing for ants' way we normally do it by smearing it around with a fine brush?

If both surfaces are super clean, would it not be better to do something like place a frozen sphere of the compound (or hemisphere) in the center of the die, warm it up so it flows, then compact it warm and use a precise jig to clamp it down to some known seperation distance?

What is the crazy semiconductor manufacturers way of doing this? Maybe spinning a layer on it like a photomask?

also can vacuum degassing possibly get rid of air bubbles if combined with warming?  or possibly replace voids that are normally air filled with voids filled with more thermally conductive solvent present in the thermal compound?
« Last Edit: December 09, 2018, 11:14:45 pm by coppercone2 »
 

Offline IanB

  • Super Contributor
  • ***
  • Posts: 11790
  • Country: us
Re: really proper way to apply thermal compound?
« Reply #1 on: December 09, 2018, 11:13:38 pm »
The purpose of thermal compound is to displace any air between the mating surfaces (air is a better insulator than the thermal compound, which we hope is a thermal conductor). Therefore the best way to apply thermal compound is as a small blob in the center of the mating surfaces and then compress the surfaces together so that the blob spreads out sideways displacing air as it goes. (If the surfaces are large then maybe one or more beads might be better than a single blob.) It also wouldn't hurt to smear the surfaces first with a very thin layer of compound to remove any pockets of air in the surface pores and assist the spreading process.

Overall, the thinnest layer of compound possible would be the ideal as thermal compound does not conduct heat as well as metal does.

(Note, it is not a good idea to spread the compound before mating the surfaces as this will cause pockets of air to be trapped and will reduce the heat transfer efficiency.)
 

Online coppercone2Topic starter

  • Super Contributor
  • ***
  • Posts: 9234
  • Country: us
  • $
Re: really proper way to apply thermal compound?
« Reply #2 on: December 09, 2018, 11:17:21 pm »
i wonder if the lapped surface should be measured first with a surface texture gauge, then a thermally conductive compound of grit size appropriate to fill the voids is used (non appropriate for the larger connection), lapped off, then the proper thermal compound applied that possibly has different properties since it works to fill a different bulk geometric profile rather then micro cavities.
 

Online coppercone2Topic starter

  • Super Contributor
  • ***
  • Posts: 9234
  • Country: us
  • $
Re: really proper way to apply thermal compound?
« Reply #3 on: December 09, 2018, 11:19:16 pm »
The purpose of thermal compound is to displace any air between the mating surfaces (air is a better insulator than the thermal compound, which we hope is a thermal conductor). Therefore the best way to apply thermal compound is as a small blob in the center of the mating surfaces and then compress the surfaces together so that the blob spreads out sideways displacing air as it goes. (If the surfaces are large then maybe one or more beads might be better than a single blob.) It also wouldn't hurt to smear the surfaces first with a very thin layer of compound to remove any pockets of air in the surface pores and assist the spreading process.

Overall, the thinnest layer of compound possible would be the ideal as thermal compound does not conduct heat as well as metal does.

(Note, it is not a good idea to spread the compound before mating the surfaces as this will cause pockets of air to be trapped and will reduce the heat transfer efficiency.)

I thought this was the best method but I thought I was advised against it some where before. I did not understand why, I thought maybe its because if you have a giant blob rather then a thin layer you get some kind of point stresses that end up causing the compound to separate out under the increased pressure that would not be present if spread out more. I wonder if a domed mold used with chilled compound could spread it out more from the beginning with less peak pressures.
 

Offline DaJMasta

  • Super Contributor
  • ***
  • Posts: 2289
  • Country: us
    • medpants.com
Re: really proper way to apply thermal compound?
« Reply #4 on: December 10, 2018, 12:20:03 am »
The recommendations I've seen for lapped surfaces and optimal cooling are generally:
Smear a very small amount of the compound, less than opaque thickness, all over each side, then connect (optionally with a small amount in the center).  This does require very flat surfaces and every even pressure applied, because if either isn't right there will be air gaps with no contact, but since TIM is generally less thermally conductive than the metals it sits between, the idea is to minimize the amount while still getting maximum contact area.

Generally, the surfaces aren't exactly smooth so you need a little leeway in the application, I've seen a fair number of techniques, but some options are:
Smear around and dot in center
Pea sized dot in the center
X over the contact area
X with dots in each quadrant
Dots over each die (if you know where it is) with or without a smear
Even opaque application (if you just want to guarantee complete contact)


Not 100% sure on what would be the "best", but there are some reasonable options depending on the surfaces.  Worth noting that different TIM compounds probably prefer different applications - more viscous compounds will not spread as much under the same pressure, so more frequent smaller dots or similar would be preferable.  Nice thing about the even smear is that it doesn't have to move around much.
 

Online coppercone2Topic starter

  • Super Contributor
  • ***
  • Posts: 9234
  • Country: us
  • $
Re: really proper way to apply thermal compound?
« Reply #5 on: December 10, 2018, 12:25:57 am »
how do you find air gaps? what test is used for it? freeze disassemble and do surface inspection?
 

Offline IanB

  • Super Contributor
  • ***
  • Posts: 11790
  • Country: us
Re: really proper way to apply thermal compound?
« Reply #6 on: December 10, 2018, 01:13:37 am »
Just measure temperatures. The purpose of heat sinks and cooling systems is to cool the device producing heat. So measure the temperature and make sure it is within expected limits. Many devices have an onboard temperature sensor to make this feasible. If the temperature goes too high something has gone wrong.
 

Online coppercone2Topic starter

  • Super Contributor
  • ***
  • Posts: 9234
  • Country: us
  • $
Re: really proper way to apply thermal compound?
« Reply #7 on: December 10, 2018, 01:32:10 am »
that only means your trying to meet a spec not see whats actually going on. how would you actually measure it to ttune a process?
 

Offline IanB

  • Super Contributor
  • ***
  • Posts: 11790
  • Country: us
Re: really proper way to apply thermal compound?
« Reply #8 on: December 10, 2018, 01:40:58 am »
You do science. You have the hot temperature in the device being cooled, you have the cool temperature on the outside of the heat sink, you can do math to find out the heat transfer resistance of the thermal compound and compare that to expectation.
 

Online coppercone2Topic starter

  • Super Contributor
  • ***
  • Posts: 9234
  • Country: us
  • $
Re: really proper way to apply thermal compound?
« Reply #9 on: December 10, 2018, 01:49:47 am »
that won't tell you if something happened like a density change due to lack of proper 'suspension' vs air gap. Or contaminated fluid with some kind of larger particle etc.

Say viscosity change due to pressure change that causes a suspension to break down or something.
 

Online wraper

  • Supporter
  • ****
  • Posts: 16792
  • Country: lv
Re: really proper way to apply thermal compound?
« Reply #10 on: December 10, 2018, 01:50:17 am »
I have some experience with using LOCA adhesive for attaching touchscreen to display where I could actually see the result. I can say that the best way is is to apply it as X. It's the best way to fill all of the space with minimum amount of paste and avoid air bubbles at the same time. If you apply only drop at the center it will spread as circle and space you need to fill is not round. So either drop needs to be huge (a lot of paste will squeeze out) or it won't fill the corners.

 

Online coppercone2Topic starter

  • Super Contributor
  • ***
  • Posts: 9234
  • Country: us
  • $
Re: really proper way to apply thermal compound?
« Reply #11 on: December 10, 2018, 01:54:02 am »
it spreading out does not create microbubbles in it because it kind of tears as it spreads on a microscopic level?

I don't think this stuff is very fluid like, like the glue is. It's more like a suspension I think, like chocolate chips in a dough ?  Are you sure there is not some kind of sheer surface forming that can introduce a void? Or kind of like surface evaporation or other effects making a surface thats kind of like a cake thats beginning to bake in the oven where the very skin is stiffer, and if you poke it with a knife you get cracks. Like imagine you just put a dollop of muffin compound on a baking tray, broil it till the surface gets kinda solid, then smush glass down on it. I think you will have voids because of the cracks.

I am also weary of using a transparant surface as a window because the surface tension of the whatever is different so it might behave differently.
« Last Edit: December 10, 2018, 01:59:44 am by coppercone2 »
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21606
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: MK14

Online wraper

  • Supporter
  • ****
  • Posts: 16792
  • Country: lv
Re: really proper way to apply thermal compound?
« Reply #13 on: December 10, 2018, 02:04:24 am »
it spreading out does not create microbubbles in it because it kind of tears as it spreads on a microscopic level?

I don't think this stuff is very fluid like, like the glue is. It's more like a suspension I think, like chocolate chips in a dough ?  Are you sure there is not some kind of sheer surface forming that can introduce a void?
That adhesive is not really that fluid. It's more fluid than thermal paste but not nearly like water. The trick is that air bubbles are not entrapped because touch/spreading of adhesive or paste starts from single point thus pushing all air out.
« Last Edit: December 10, 2018, 02:06:48 am by wraper »
 

Online wraper

  • Supporter
  • ****
  • Posts: 16792
  • Country: lv
Re: really proper way to apply thermal compound?
« Reply #14 on: December 10, 2018, 02:10:53 am »
Relevant: https://www.infineon.com/dgdl/Infineon-AN2006_02_Application_of_screen_print_templates-AN-v1.0-en.pdf?fileId=db3a304412b407950112b40ed3f71297

Tim
This way is good as well. It provides routes for air to escape. But you won't do this well without properly designed stencil.
 

Online coppercone2Topic starter

  • Super Contributor
  • ***
  • Posts: 9234
  • Country: us
  • $
Re: really proper way to apply thermal compound?
« Reply #15 on: December 10, 2018, 03:00:04 am »
I still think the glue is a true solution but the other stuff is a colloid at best.
 

Offline DaJMasta

  • Super Contributor
  • ***
  • Posts: 2289
  • Country: us
    • medpants.com
Re: really proper way to apply thermal compound?
« Reply #16 on: December 10, 2018, 03:47:52 am »
As for checking for air gaps, using regular thermal material will generally show you the outline of the bubble when you remove it to reset it, but I'm not sure how to know before you remove it aside from seeing a temperature that's too high and deciding there was probably a problem with the application.

Most of the reason commercial products or coolers with TIM pre applied come with the thick sheet of heatsink compound (which is often internally reinforced with a mesh) is that the thickness and width ensure a fairly even, full coverage installation.  I think virtually none of the time it's the ideal low thermal resistance connection, but it is consistently reliable at making a decent connection, so if the rest of the thermal design is done around that assumption, automated assembly or end-user assembly and all its variables can still get reliably good results.
 

Offline Nerull

  • Frequent Contributor
  • **
  • Posts: 694
Re: really proper way to apply thermal compound?
« Reply #17 on: December 10, 2018, 04:09:05 am »
These guys tested several application methods and found differences of a fraction of a degree on high-wattage CPU applications.



People obsess over methods of thermal compound application but real word differences seem minimal to non-existent.
 

Offline thermistor-guy

  • Frequent Contributor
  • **
  • Posts: 365
  • Country: au
Re: really proper way to apply thermal compound?
« Reply #18 on: December 10, 2018, 04:32:59 am »
...
People obsess over methods of thermal compound application but real word differences seem minimal to non-existent.

I change the CPU/GPU paste on several computers every few years, either because of an upgrade or just regular preventative maintenance.

I use Arctic Silver, and the 2-part Arctic Clean. My steps are, using gloves throughout:

* clean the surfaces gently but thoroughly with cotton buds, and then lint-free laboratory wipes (Kimwipes or similar;
* let dry for at least 30 minutes;
* apply the paste to the CPU with a glossy business card (new in box, untouched); spread in X, Y, and X-Y directions so it's a thin even coat;
I don't use the edge, I bow the card in half and use the soft bend as the spreading surface;
* install the heatsink; this step usually squishes the paste a little but not much;
* run a thermal benchmark, to test the installation.

What I find, each maintentance cycle, is that the previous coating was quite even with no obvious voids, and the benchhmark gives reasonable temperatures
on the fully loaded system.

So yes, do it with care - don't scratch the surfaces, and do a proper clean - but don't obsess either.
 

Online TERRA Operative

  • Super Contributor
  • ***
  • Posts: 2870
  • Country: jp
  • Voider of warranties
    • Near Far Media Youtube
Re: really proper way to apply thermal compound?
« Reply #19 on: December 10, 2018, 05:30:27 am »
A good way to ensure even coverage and exclude air is to apply the thermal paste in an 'X' pattern then clamp the heatsink on.
This will make sure the thermal paste reaches the edges but won't capture air.

On the other hand, ESAB welders specify to apply a small amount of heatsink compound then dab it around with the flat end of a small foam paint roller until a thin even coating is achieved. They even supply the paint roller with the tube of thermal paste in the kit.
If that process is good for IGBT modules passing hundreds of amps, I'm sure it'll be ok for a CPU or vreg etc.

On the topic of lapping, apparently 1200 grit emery paper leaves grooves that are approximately the same size as the silver particulates in Arctic Silver 5 thermal paste. Anecdotally, this should mean the silver is able to more effectively fill the gaps and make a close thermal interface.
Where does all this test equipment keep coming from?!?

https://www.youtube.com/NearFarMedia/
 

Offline Mr. Scram

  • Super Contributor
  • ***
  • Posts: 9810
  • Country: 00
  • Display aficionado
Re: really proper way to apply thermal compound?
« Reply #20 on: December 10, 2018, 05:38:25 am »
This discussion seems to reappear over and over and never seems to be conclusive. There are a few popular camps with have valid points on paper, but which in practice don't seem to make any distinguishable difference. Pick what works for you and be happy with it.
« Last Edit: December 10, 2018, 05:42:03 am by Mr. Scram »
 

Offline GregDunn

  • Frequent Contributor
  • **
  • Posts: 725
  • Country: us
Re: really proper way to apply thermal compound?
« Reply #21 on: December 10, 2018, 06:00:59 am »
People obsess over methods of thermal compound application but real word differences seem minimal to non-existent.

This.  If it were really important, they wouldn't give thermal compound, CPUs and heat sinks to gamers and trust them to not screw it up.
 

Offline Mr. Scram

  • Super Contributor
  • ***
  • Posts: 9810
  • Country: 00
  • Display aficionado
Re: really proper way to apply thermal compound?
« Reply #22 on: December 10, 2018, 06:10:24 am »
This.  If it were really important, they wouldn't give thermal compound, CPUs and heat sinks to gamers and trust them to not screw it up.
I suspect the typical demographic of the target group is a large part of the ongoing discussion. Teen men with a point to prove aren't quite known for their nuanced discussions and factual opinions.
« Last Edit: December 10, 2018, 06:13:21 am by Mr. Scram »
 

Online coppercone2Topic starter

  • Super Contributor
  • ***
  • Posts: 9234
  • Country: us
  • $
Re: really proper way to apply thermal compound?
« Reply #23 on: December 10, 2018, 06:27:09 am »
whats interesting about CPU is the fact that you can replace the paste between the die and the lid of the IC in a modern intel chip to improve its performance. surprising for 1000$ ic.

but seriously, what about spin coating? Its used for extremely even distribution of photoresist during transistor manufacture.
 

Online jbb

  • Super Contributor
  • ***
  • Posts: 1128
  • Country: nz
Re: really proper way to apply thermal compound?
« Reply #24 on: December 10, 2018, 06:32:12 am »
In the high power semiconductor line, screen printing a pattern of small diamonds (or similar) on seems to be common.

For Some systems (eg automotive) we are starting to see devices which have fins or pins for water cooling built right into the bottom plate. No heatsink compound anywhere!
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf