Poll: What do you use to apply thermal paste?

Back in the day, applying thermal compound was a fine art. I used to spend entirely too much time making sure the surface of the CPU was coated by a thin, perfectly even layer of thermal goop. These days, cooler makers tend to recommend depositing a blob of thermal compound on the CPU and then putting the heatsink directly on top. The force generated by the cooler’s retention bracket is supposed to spread the compound evenly between the heatsink and the CPU, and the method seems to work pretty well—as long as you’re not trying to set overclocking records.

Old habits die hard, and that got us thinking. What to you use to apply thermal paste? Do you take the lazy-but-increasingly-recommended approach, or do you painstakingly spread compound with a special tool, a plastic-wrapped finger, or even a bare digit? Perhaps you eschew tubes of thermal paste completely and stick with the pre-fab TIMs common among stock CPU coolers. We’ve put together a handful of options in a new poll, which you can vote in below or in the middle column on the front page. If you use an alternate method than the ones we’ve presented, feel free to detail your approach in the comments.

Our last poll asked for your opinion on Apple’s patent victory over Samsung. Unsurprisingly, the clear majority deemed the ruling a travesty. 77% of voters disapproved of the verdict, and only 12% thought justice was done. 11% couldn’t decide either way.

Comments closed
    • gerbilspy
    • 7 years ago

    paper match stick

    • clone
    • 7 years ago

    squeeze some out then stir it around with the end of the tube, place the heatsink on it then rotate a little left & right while applying some pressure (not much) to level it off and fill in gaps…… I have used a wide screwdriver in the past being careful not to scratch but why bother?

    when using a stock heatsink it’s already spread across so I just do the partial spin L & R with pressure to make certain before clamping it down.

    • shaq_mobile
    • 7 years ago

    there needs to be a “deflector array” option. thats what people in the 24th century use to solve their problems.

    • sweatshopking
    • 7 years ago

    HOLY CRAP NERDS. MATH? MATH!?!??!!! WHAT THE F!?!?!!? THIS IS A WEBSITE, NOT A HOW TO BE SINGLE FOREVER SITE.

    • aim18
    • 7 years ago
    • xxxSakurachanxxx
    • 7 years ago
    • jamsbong
    • 7 years ago

    I spread it like a putting a butter spread on bread. I would either use the thermal paste container’s mouth or the heatsink itself.
    It does not matter what methods you use, as long as you get an even spread and your technique results in no bubbles within the paste.
    Of course, it should be as thin as possible but thick enough to account for the minor imperfection on the two contacting surfaces. Otherwise you’ll get bubbles, which defeats the thermo paste’s purpose.

    • ColeLT1
    • 7 years ago

    Spreading it (thin) around is the correct method for exposed dies. With an IHS, I have had best results with AS clean, then GELID extreme, spread it on the IHS and the cooler then with a microfiber cloth, spread it all over, leaving only a haze of TIM, then a drop in the center and bolt it up.

    • Delphis
    • 7 years ago

    That little plastic bag the motherboard screws came in – awesome for getting the heatsink installed.

    Edit: Arctic Silver 5 is my paste of choice too, by the way. Just the right viscosity and stickiness to get the job done.

    • siberx
    • 7 years ago

    Depends on the type of chip I’m trying to cool. If we’re talking bare silicon (increasingly uncommon these days) then I’ll use a very high-quality thermal paste and I’ll spread a thin layer over the whole chip. If I’m working with an IHS I’ll put a blob in the middle and let pressure spread it. For a very large IHS (like a modern GPU with an heatspreader) I will often use a more complex pattern (a line, or multiple lines, or a cross) so that the paste doesn’t have to get squashed/spread so far to reach close to the edges.

    The reasoning behind this is that silicon’s thermal conductivity is nothing particularly fantastic, but it can be assumed to be quite flat. Getting contact across the whole die is important to avoid hotspots and is most easily achieved (assuming your heatsink is pretty flat) by applying a thin, even coat.

    For IHS chips, I’ve found it to be pretty uncommon to get a flat surface that mates well on both the heatspreader and the heatsink. Addiitonally, since the IHS takes care of getting good contact to all parts of the silicon chip, getting full coverage is a lot less important than getting good contact at the *center* of the heatspreader where the heat is mostly coming from. To achieve this, a reasonably sized blob near the center will assure good thermal contact where it matters and will avoid any unexpected air gaps that can occur if you spread a layer too thin over a bowed IHS.

    • dashbarron
    • 7 years ago

    Personally? I squeeze out the entire tube on the CPU and then bolt the heatsink on. Then I know I’ll have great heat transfer because the paste is oozing out the sides and onto the motherboard; afterwards, I just use a babywipe to clean the motherboard and I’m good to go.

      • willmore
      • 7 years ago

      The whole TIM industry would like to thank you for your support. Keep up the good work.

    • HisDivineOrder
    • 7 years ago

    I use a case knife.

    That’s how I apply mayo to everything else, so why not use that for mayo between my Heatsink/CPU sandwich? Mmmm… mayo…

    • Chrispy_
    • 7 years ago

    I usually empty the whole syringe over the processor to make sure it’s thoroughly coated. Then I apply the heatsink and bash it down with a mallet to make sure it’s on really firm. If any of the clips break off I found that a small trowel of cement is better than glue because it doesn’t melt when it gets hot.

    The excess paste usually leaks down onto the graphics card, so I tend to scrape that off with a knife. If you use a plastic knife you won’t shock yourself (hurts, lol) and I find fewer of those little square lumps on the graphics card come off at the same time.

      • ronch
      • 7 years ago

      Epic.

    • FireGryphon
    • 7 years ago

    When I used thermal paste I’d use a razor to spread it.

    • Krogoth
    • 7 years ago

    Flat, precision screwdriver works.

    • travbrad
    • 7 years ago

    I just squeeze out a short line oriented to match the actual CPU (since Sandy Bridge is a long rectangle), and let the pressure of the HSF do the spreading. I’m talking about a very short line though, maybe 1-1.5cm, not making a line across the whole CPU like I’ve seen pictures of.

    I have used the plastic bag and credit card methods in the past though on CPUs that had no IHS, and instead had the CPU directly touching the heatsink (AthlonXP 1900+ for example)

    • kmieciu
    • 7 years ago

    Toothpick. Or a match if toothpick is not available. Or a toothpick if match is not available. Or … eee… did I mention toothpick?

    • Johnny5
    • 7 years ago

    Apparently the credit card technique can create air pockets that have no way of escaping at the moment you press the sink to the CPU. When I was searching on how to do it, the recommended technique I found involved an x-shape that will spread well across the square surface without creating air bubbles if done correctly. So that, via squeezing out of the tube.

    They tested with pieces of glass/clear plastic so the effects could be seen. I’d post the link if I could remember where I found it.

      • nexxcat
      • 7 years ago

      Is this it?

      [url<]http://www.youtube.com/watch?v=ffK7L0Qj13Q[/url<] Turn off the sounds. The music is really annoying.

        • Johnny5
        • 7 years ago

        It is. Now that I see it the difference between cross and dot method with the size of the cross used is insignificant. It does however nicely demonstrate air bubbles in the spread technique, especially in part 1..

    • DrCR
    • 7 years ago

    Razer blade. (I used to use the finger-inside-a-plastic-bag route, but that seems to only work well for low viscosity thermal paste.)

    • OsakaJ
    • 7 years ago

    I use an old fashion single edged razor blade that I’ve had since the original Pentium days. A pea-sized dollop of grease in the middle of the heatsink and a little blade work is all it takes for a nice even, thin layer.

      • mutarasector
      • 7 years ago

      One of my single edge blades is missing since then, too! Now I know who has it…

    • Ari Atari
    • 7 years ago

    At first I used my bare finger to spread the paste. However, touching the arctic silver always made my finger feel funny, so I began to use a duplicate business card. Any other bare finger spreaders out there get this weird “pain” like feeling?

    • FuturePastNow
    • 7 years ago

    I used to use a dead credit card to spread a very thin layer on the processor. Now I just put a small grain in the center if the IHS and smush the heatsink down. Slide it around a little.

    It’s much faster and there’s no discernible difference.

    • JosiahBradley
    • 7 years ago

    My preferred method is to follow the instructions provided by the thermal compound producer. They have tested the stuff in a lab and know, backed by hard science, what works best for their product. Seeing as the consensus from most manufactures is the, ‘rice grain and apply heatsink method’, I normally end up doing just that.

    Before there were heatspreaders though I was using a razor blade and lots and lots of time. Even with a shim I was afraid to crush a core.

    • Geonerd
    • 7 years ago

    Depending on where the OCD-O-Meter is pointing….

    Fair sized drop in the middle.
    Apply heatsink and clips.
    Remove clips.
    Gently rotate/translate heatsink while pressing down.
    Remove heatsink.
    Inspect smeared goop.
    If there are any bare areas, I’ll apply tiny drops to those surfaces.
    If goop is anything more than a thin film, scrape some away from the edges and goto line 2.
    Quite small drop in the middle, just to ‘refill’ the contact area. (Yea, probably unnecessary.)
    Re-install heatsink and clips.
    Gently rotate/translate heatsink a few mm. as much as clips will allow.
    Done.

    • JohnC
    • 7 years ago

    I use a very old SIM card to spread it. Perhaps it’s not as convenient as other tools, but I’ve used it a few times long time ago (before I had any kind of credit cards or other plastic cards in my wallet) and it kinda became a “tradition” to use it ever since 😉

    • cynan
    • 7 years ago

    TIM is for wimps. Real men use solder.

      • Airmantharp
      • 7 years ago

      Terrifying as that may be, I have to wonder if the benefit might be worth it. Also, just how much of a benefit might you get?

      • internetsandman
      • 7 years ago

      On sale: NH-D14 with free 3770K pre-attached!

        • cynan
        • 7 years ago

        I wonder why no one’s tried to market something like this to particularly gullible enthusiasts…

        With the right re-flow oven and low enough temp solder, I wonder if it would actually be possible without damaging the CPU. Anyone know the TJ Max for Sandy or Ivy while not operational?

      • Squeazle
      • 7 years ago

      Pfft. REAL MEN WELD.

      Actually…
      REAL MEN SMELT. Just pour the heatsink onto the processor. Ultimate custom build.

        • Airmantharp
        • 7 years ago

        For those that are big into the copper cup and liquid nitrogen scene, doing this reliably might prove beneficial to their craft!

      • rrr
      • 7 years ago

      Intel = wimps, and they proved it with Ivy Bridge release.

      Well spotted, sir.

        • cynan
        • 7 years ago

        Are you referring to the fact that the heat spreaders in Ivy Bridge use some sort of TIM, whereas Sandy Bridge had them soldered to the cores? (Because, I’ll admit, I wasn’t. Well spotted right back atcha 😉

        Has this actually been proven to result in worse thermal dissipation? Is this part of the reason why Ivy gets so hot when OC’d?

          • Chrispy_
          • 7 years ago

          I was never quite sure about this either; If people have proved that Ivy uses less power, and the system obeys the first law of thermodynamics, the whole rig ought to run cooler than Sandy.

          In practice, I suspect the weird resistive properties of a semiconductor like silicon mean that it’s a positive feedback loop:

          [list<][*<]Ivy's TIM means less effective cooling of the silicon [/*<][*<]Hotter silicon changes the resistance of the circuit [/*<][*<]Higher resistance requires more voltage [/*<][*<]More voltage draws more power at the same current [/*<][*<]More power means more heat to dissipate [/*<][*<]I'vy's TIM means less effective cooling of the silicon [/*<][*<]repeat until the [i<]magic smoke[/i<] escapes.... ;)[/*<][/list<] But I dunno, I'm not really edumacated in the ways of semiconductors; More of a hammer and spanners kind of engineer, myself. ([i<]one that sucks at closing bbcode tags, at that....[/i<])

            • internetsandman
            • 7 years ago

            That sounds like something that should have been caught while Ivy was still in testing though, unless it was decided that because of the lower power draw they could save money by using TIM instead of solder, but I’m not too aware of the specific cost of materials in either case

          • Draphius
          • 7 years ago

          read a review about this, and the solder used on sandybridge is something like 10x more conductive then the tim used on ivy. seems like they were just trying to save money. although its now easier to remove the ihs and do direct die cooling, if u have the guts

    • Bensam123
    • 7 years ago

    I squiggle a bit on one edge and use the edge of a soap box (the laminated glossy side) to spread it out evenly. Then I put on the heatsink and give it a couple rotations back and forth to make sure it’s evenly distributed.

    • Alexko
    • 7 years ago

    Last time I put thermal paste it was some kind of Artic Silver, and there was a special brush sold with the bottle, so I used that.

    Before that, in the good old days where nothing was convenient, I would use Paris subway tickets: [url<]http://uploads.neatorama.com/wp-content/uploads/2010/02/Lartigue.jpg[/url<] Small, rigid enough, but not too much. Pretty ideal, actually.

    • Rand
    • 7 years ago

    Squeeze out a drop in the middle, push heatsink down and twist once or twice to spread it out. Secure heatsink. Done.

    • Meadows
    • 7 years ago

    My option wasn’t in the list (I buy disposable household latex gloves, and use it to spread the paste extremely thinly and reasonably evenly), but “finger inside a plastic bag” is mechanically the closest, so I voted that.

      • Airmantharp
      • 7 years ago

      That’s been my standby- put on the ‘dot’ of paste in the center, spread it out (not to the edges, maybe 2/3rd of the way to each edge from the center), then let HSF clamping pressure do the rest. That’s assuming I’m not lazy and just using whatever the HSF came with- I hit 4.8GHz with my 2500k with the stock paste on my H60.

    • Zarf
    • 7 years ago

    I use an old Starbucks giftcard, but voted Credit Card. It has transparent parts and looks pretty neat. It allows me to get a very, very thin layer of thermal paste on there.

    On a side note, being blunt and telling people that I don’t like giftcards as gifts, nor do I accept them as payment, has been quite nice. Now I can spend that money WHEREVER I WANT! HAH! Take that, terrible retailers!

    • videobits
    • 7 years ago

    Squeeze out a big drop of grease.
    Smash down the heat sink with twist or two and secure in place.
    Wipe up squeeze-out from edges.

    <Look around forums. Enjoy OCD heat sink installers having minor heart attacks now>

    • DancinJack
    • 7 years ago

    Maybe we should just follow the real instructions…

    [url<]http://www.arcticsilver.com/intel_application_method.html#[/url<] [url<]http://www.arcticsilver.com/amd_application_method.html#[/url<] Yes, I know it's "only" for AS. I can't imagine the method being different for others that a similar in chemical make-up though.

    • fellix
    • 7 years ago

    The precise application of the thermal paste made sense back in the years of Athlon XP processors, where the cooling of the tiny and unprotected die heavily depended on the good and even contact with the heat sink, since the amount of safe pressure was very limited. Today, with all the CPUs resting securely under the hood and cooler manufacturers actually shaping the base of the heat sink to match the curvature of the metal caps, a single measured drop of paste in the center is more than enough — the pressure and heat will do the rest of the job.

    • Prototyped
    • 7 years ago

    I apply five smaller-than-rice-grain sized globes of paste (one in the center, four near the corners), and have in the past spread them out using a train ticket (piece of card stock). Most recently I didn’t bother spreading the paste out with a flat card — all it did was to create a mess.

    Then I place the heatsink surface over it and turn the heatsink counterclockwise then clockwise 90 degrees to cause the layer to even out. Then I mount the heatsink using its retention brackets.

      • DancinJack
      • 7 years ago

      That’s probably too much paste.

        • Airmantharp
        • 7 years ago

        WAY too much.

        You need one of those dots, probably less.

      • Waco
      • 7 years ago

      This is what I do too. Never fails.

    • indeego
    • 7 years ago

    Where’s the “Edge of an OCZ SSD” option?

    OH SNAP.

      • Xylker
      • 7 years ago

      Too bad I only have one thumb up for that one. Funny.

      • internetsandman
      • 7 years ago

      This. This times a thousand

    • albundy
    • 7 years ago

    “asked for your opinion on Apple’s patent victory over Samsung. Unsurprisingly, the clear majority deemed the ruling a travesty. 77% of voters disapproved of the verdict, and only 12% thought justice was done”

    12% sounds a little low for apple fanboys.

      • trackerben
      • 7 years ago

      At least that’s something compared to the poll on Samsung for using the legal system to block competition from Apple on the basis of patent violation

      • derFunkenstein
      • 7 years ago

      12% are Apple fanboys AND they think Apple was reasonably compensated. You can call it a travesty against Apple, if you’re that bat shit crazy.

    • odizzido
    • 7 years ago

    I am the 3%. Whatever comes on the heatsink is good enough for me.

    • ludi
    • 7 years ago

    Finger.

    But usually I leave the blob just slightly thicker in the middle so the cooler can do the final spreading.

      • DancinJack
      • 7 years ago

      Exactly what I do. But with my finger inside a Ziploc bag.

    • derFunkenstein
    • 7 years ago

    You’re supposed to spread it?

    I put a grain of rice or so down on my CPU and mash it down with the cooler. I was surprised, however, to learn that so do 45% of the other respondents as well. Way higher than I expected.

      • cmrcmk
      • 7 years ago

      I got an all-in-one water cooler last christmas and tried to spread the paste several times with an old credit card (each time resulting in very high temps and a couple of auto shut offs). I finally broke down and looked up the paste’s instructions which said thin line + mashing the heat sink down = win. Haven’t had a problem since.

      • Krogoth
      • 7 years ago

      It doesn’t really matter that much with modern CPUs.

      You can just apply a small dab where the die is located and let gravity/physical pressure do the rest of the work for you. 😉

        • cynan
        • 7 years ago

        I think it depends a lot on the retention mechs. Some of those cheap plastic spring-loaded push-pin setups don’t look like they offer much pressure (let alone even pressure) at all. Then at the other end of the spectrum you have custom bolt mounted blocks where you can just keep turning until the motherboard cracks. The one that came with mine had no indication of how tight to screw the bolts so I just ended up going by feel…

        Also, the less viscous and gritty the TIM, the easier to apply (the more thinly it spreads and flows into pits).

        • Meadows
        • 7 years ago

        What exactly does gravity do for it?

      • Creech
      • 7 years ago

      I’ve never heard of anyone using rice as a TIM before. How’s that working out for you? 😛

    • not@home
    • 7 years ago

    I used to do just the small drop and put the heatsink on with the idea that it is the best way to eliminate air bubbles. Then I got a Direct touch heat pipe style cooler. That presented some problems so I did some tests. I applied the TIM to the heatsink’s base and set it on a small sheet of glass. I set it on the floor and placed an appropriate amount of weight onto of the heatsink. Then I would wait an hour and, holding it carefully, pick it up and look through the glass. I tried every way I could think of applying the TIM. The best for that heatsink was to use a straightedge to force the TIM into the cracks of the base (alongside each heatpipe), leaving the rest of the base as clear of TIM as possible. Then put a very thin line of TIM along each pair of cracks (there were two side by side on each side of the heatpipes). That TIM had to be very thin. Then put the tiniest little dab of TIM in the middle of each heatpipe lengthwise. This produced the least amount of excess TIM oozing out the sides while still eliminating all the air bubbles. Make sure you use a long lasting TIM.

    • Wildchild
    • 7 years ago

    Spreading creates air bubbles, which isn’t good for transferring heat. I just put a rice or pea sized dot depending on how big the heat spreader is

    • spigzone
    • 7 years ago

    I use common sense.

      • indeego
      • 7 years ago

      Streetsmart spigzone doesn’t #$%^ around, yo.

        • spigzone
        • 7 years ago

        Meet my leetle friend, safety razor …

          • indeego
          • 7 years ago

          So you use a dual-edge razor alone to apply thermal paste?

    • Arclight
    • 7 years ago

    Idk if i’m doing it wrong, i usually put a tiny bit in the midle of the chip, spread it with my freshly washed, throughly dried finger, then i do the same on the heatsink. Only afterwards i put a rice sized beed of TIM, again on the middle of the chip, and i proceed to put the heatsink but this time without spreading it.

    PS: i use Arctic MX-4…..i wonder how much better new TIMs are.

    • eitje
    • 7 years ago

    edge of an envelope or sheet of paper.

      • willmore
      • 7 years ago

      Not a good idea unless you want to embed a bunch of very hard strands of cellulose in there.

    • `oz
    • 7 years ago

    I mostly use Arctic Silver 5 when I apply thermal paste. Their website has differing methods for each CPU. That said, I use a finger in a bag followed by an old credit card to smooth it out if I need to apply it to the whole heatspreader.

    [url<]http://www.arcticsilver.com/methods.html[/url<]

    • GTVic
    • 7 years ago

    The Banana Jr. 6000 comes with thermal paste already installed.

    [url<]http://www.jarnot.com/images/BananaJunior6000.jpg[/url<]

    • flip-mode
    • 7 years ago

    Tines of a fork.

    • mark625
    • 7 years ago

    Like the other professionals here, I only use a razor blade. It’s straight, smooth, rigid, and easy to handle. (Yeah, I know, “that’s what she said”.)

    A credit card might seem to be a decent substitute if you can’t afford a razor blade, but it’s a bit too flexible to reliably create a smooth, even layer.

    The other non-straightedge applicators would leave too much compound. Nothing like a solid ring of goop seeping out around your CPU to make a big mess.

      • Fieryphoenix
      • 7 years ago

      I voted other because I use single bladed safety razors.

    • Deanjo
    • 7 years ago

    Pneumatic caulking gun is the only way to apply paste. 😛

    • bthylafh
    • 7 years ago

    I’m usually happy to use the TIM that comes with a CPU’s stock HSF. I don’t OC so it doesn’t matter.

    If I have to apply, either my finger (my fingers are not oily) or the end of the paste tube.

    • cegras
    • 7 years ago

    I don’t understand why people think that having a thin film of paste serve as a buffer between two metal plates is good. The less paste there is the better, actual contact between the two metal surfaces will transfer heat way faster than any silicone goop with metal or ceramic filler will allow.

    (i.e. the less paste there is, the better)

      • ChronoReverse
      • 7 years ago

      Because your two plates aren’t perfectly flat. With applied pressure and a thin layer of paste, the metal is touching each other at the peaks while the valleys are filled with the paste (which is better than air).

      The paste itself isn’t as important as displacing the air (as it’s been shown that even processed cheese is better than bare metal). I choose my pastes for stability rather than silver content for that reason.

        • chuckula
        • 7 years ago

        Indeed, the thermal conductivity of even the “high end” thermal pastes is lousy compared to a simple piece of metal. However, the thermal pastes do have a much better thermal conductivity than air does. Thus, the trick is to use just enough thermal paste to establish good contact between the spreader & heatsink without air bubbles while also not overdosing on thermal compound and lowering the effective conductivity through the junction.

        • cegras
        • 7 years ago

        Yeah, that’s pretty obvious. But the volume of air in between the two metal plates is miniscule. You really don’t need a lot of paste to fill in the space, I think you’d be quite surprised if you ever bothered to make a back of the envelope calculation.

          • Airmantharp
          • 7 years ago

          I think you should test this for us. 🙂

            • cegras
            • 7 years ago

            2000 grit sandpaper has particle sizes 7 microns. Let’s fill the entire surface with spheres and calculate the volume of 3 x 3 cm.

            So (3 cm / 7 microns)^2 = 32 million spheres * 1.4 E-15 meters^3

            = 5 E-8 m^3

            50 uL. 50 microliters.

            That’s the worst case scenario. But your contact area will be less if the surface is concave. You might need even less.

            • Chrispy_
            • 7 years ago

            Yes, but you’re assuming a viscosity of zero in your model which allows a small amount of paste to spread evenly between the flat surfaces. Even between two polished slabs of glass, something with a very low viscosity (like ethanol) still requires an almost infinite force to spread out for the 3 x 3 cm you mention.

            In reality, thermal paste is like syrup and the surfaces are significantly less smooth than polished ceramics. You’d need exponentially more force to squeeze the paste layer thinner, and some quick mental calculations based on a syrup with a viscosity of 50,000cP (water has a value of one and peanut-butter has a value of around half a million) means you’d need around 1KN of force to get the fluid moving between two plates that are 50 microns apart.

            In other words, to get even a 50 micron layer with thermal paste, you’d need the same force as a large or fat person standing on top of your heatsink.

            [i<]In fact, if it were possible to do so without destroying the two plates, I am fairly confident that a paste gap of 7 microns could support the force of a locomotive running over it, assuming a locomotive has at least eight wheels and weighs less than 800 metric tons. [/i<]

            • cegras
            • 7 years ago

            In turn, you forgot about wetting properties, surface energy, contact angle, etc. Im not squishing out a liquid that would rebound into a ball otherwise.

            Im also not moving fluid between plates, but squishing down.

            • Chrispy_
            • 7 years ago

            Not forgot, just removed from the model to simplify it. Adding things like contact angle, and wetting properties (which would mean the paste has to be calculated as a slurry rather than a Newtonian fluid) just makes the required forces higher, and doesn’t favour your argument in any way.

            The same simplification applies to fluid movement. I’m just modelling for [i<]any[/i<] movement. Specific movement such as that induced by pressure over an area towards edge boundaries is both a nightmare to calculate, and will of course also require more force than that of a simple lateral movement. Yeah, my model is waaaaay too simple to be accurate, because I can't be arsed to work it out properly, but by increasing the accuracy of the calculations, it only gets worse for your argument because I'm modelling a best-case scenario for you with my oversimplification.

            • cegras
            • 7 years ago

            [quote<]Specific movement such as that induced by pressure over an area towards edge boundaries is both a nightmare to calculate, and will of course also require more force than that of a simple lateral movement.[/quote<] I don't think squeezing force is similar to sheer force at all. [quote<]Adding things like contact angle, and wetting properties [/quote<] Why? At least for ethanol, a drop of it on a hydrophilic surface will self wet, obviously not the case for paste. I don't see how increasing wetting properties will make it more difficult to squeeze it over two parallel plates. I'd be interested in seeing your calculations. [url<]http://innovationcooling.com/Indepandentuserpressuretest.htm[/url<] [url<]http://www.wolframalpha.com/input/?i=50+N+%2F+9+cm%5E2[/url<] The higher the mounting pressure, the more metal contact you have.

            • lilbuddhaman
            • 7 years ago

            My head hurts.

            • Chrispy_
            • 7 years ago

            So did mine. That’s why I’m in IT now 😉

            • Chrispy_
            • 7 years ago

            Calculations are almost meaningless because of how inaccurate I’ve been, and bbcode commenting systems aren’t exactly formula-friendly.

            How about a real world example of how, no matter how much force is applied, two surfaces with fluid between them are never going to squeeze out enough fluid for the microscopic ridges to make proper contact and interlock:

            Launching a 50,000 ton aircraft carrier down the dry dock ramp where it was constructed. Half a billion newtons disributed over a few load-bearing edges totalling no more than a few square meters. Yet still there is a fluid film and the thing slides down the ramp. These days it’s synthetic oils but in the old days they just used things like lard and soap. You just can’t squeeze a fluid out of a gap that easily. The smaller the gap, the greater the exponent of the force needed.

            Other instances include fluid bearings that can be found in pretty much every piece of 19th and 20th century engineering, the first example I though of being steam locomotives that pulled a dozen carriages with driven wheels connected to the piston rods on peg/sleeve/oil bearings. They couldn’t make ball-bearings hard enough to cope with the forces back in the early days and even pulling hundreds of tons behind them, oil bearings did the job just fine thanks to that viscous layer of [i<]impossible-to-completely-squeeze-out[/i<] oil.

            • cegras
            • 7 years ago

            I wish you had just read what I wrote in the very beginning then, because I posited that the less paste there is, the better (I did not advocate for squishing them into interlocking at all, although there are services that exist that can ‘mate’ surfaces). In fact, from the way you approach the argument it’s worse to have more paste because you’ll then be in the fluid squeezing, rather than air displacement, regime.

            [url<]http://www.mae.ncsu.edu/ward/publications/ward-phys1.pdf[/url<] I want to say that figure 5 shows it is indeed possible to do this, but I'm not a physicist, just a theorist.

            • Chrispy_
            • 7 years ago

            Heh, I was just defending my back-of-an-envelope calculation that you invited me to do in your reply to ChronoReverse:

            I don’t disagree with “the bare minimum” being the right answer. The important thing is to eliminate air gaps because though you rightly say that the volumes involved are miniscule, the difference in thermal conductivity between air and paste is so vast that even a relatively thick layer of paste can be better.

            • cegras
            • 7 years ago

            But the question is, is it easier to squish a large area of 50 micron thick fluid outwards, or is it easier to expand a smaller region out into air?

            • Chrispy_
            • 7 years ago

            Dunno; That study you linked is a measure of newtonian fluids where the force applied by the surface tension of the miniscus is enough to overcome the pressure exerted between two plates.

            Since most pastes these days have more in common with wet sand than newtonian fluids, I suspect that model breaks down quite fast. My reasoning being that most pastes claim to have particles of solid metal/ceramic in them, but I can’t back that up in any way other than “gut feeling”…

            Short answer, use cheap white grease – whenever I take off a heatsink using that stuff, there’s barely any goop left where the die sat – it’s almost all been squeezed to the edges. When you take off a heatsink applied with the silver/grey particulate compounds, there’s always a visible layer left that ‘breaks up’ leaving some goop on the die and some on the heatsink.

            As you said, less is better, and [url=http://www.dansdata.com/goop.htm<]experiments with toothpaste and vegemite[/url<] would seem to indicate that it doesn't really matter what you use, as long as it doesn't dry out 😉

            • shaq_mobile
            • 7 years ago

            its a good thing my forearms are in good shape!

            • Arclight
            • 7 years ago

            Just a small nitpick, don’t you mean convex? Like the Thermalright coolers, they have convex base in order to establish a better contact with the chip. That said how would your calculation work with DHT (direct heat pipe transfer) coolers?

          • Squeazle
          • 7 years ago

          Well yes. That’s sort of the point. I don’t think anyone else here emptied an entire tube of thermal paste, clamped a heatsink on and thought “Yeah, that looks right.”

          Moderation. And a microscope. Mirror flat is not as flat as one might think.

            • lilbuddhaman
            • 7 years ago

            [quote<]I emptied an entire tube of thermal paste, clamped on the heatsink and thought "Good Enough." [/quote<]

          • ChronoReverse
          • 7 years ago

          Practical experience trumps any “back of the envelope” calculation that makes too many assumptions.

          [url<]http://www.hardwaresecrets.com/article/Thermal-Compound-Roundup-February-2012/1490/5[/url<] Even lipstick does better than bare metal.

        • bwcbiz
        • 7 years ago

        You’re missing the OP’s point. He’s not saying don’t use any paste, he’s saying that most people apply the paste so thick that the metal doesn’t touch metal at all. His ideal is to have the paste ONLY in the valleys and not covering up the peaks-an absolute minimum of paste to fill in the cracks on both surfaces.

        Incidentally, my technique is to spread as thin a layer as possible across the surface of both the heatsink and the cpu, add some drops on the CPU to ensure that major gaps due to curvature are filled in and then smush. Assuming the paste isn’t pre-applied to one or the other.

      • JosiahBradley
      • 7 years ago

      Thermal paste cost around 5-10$ USD. Machining a 50$ block of copper to within microns of precision cost a lot and requires a CNC machine or equivalent. Also, how are you going to machine down the heatspereader and guarantee it is perfectly flat. Even minimizing these problems completely leaves you with just enough air to create hot spots, that given the laws of thermodynamics will only lead to a worst case scenario here. One that involves the CPU triggering its kill switch to prevent melting, literally.

      But I implore you to attach a 50$ heatsink with absolutely no thermal interface material to a 300-400$ CPU and see what happens. *Hint*: without a kill switch, lots of smoke and a nice burning smell.

        • cegras
        • 7 years ago

        [quote<] that given the laws of thermodynamics[/quote<] Are you sure you know what you're talking about here or are you throwing buzz words around? Heat transfer (i.e. the heat equation, diffusion, etc) is a kinetics problem. [quote<]But I implore you to attach a 50$ heatsink[/quote<] I should quote myself: "The less paste there is the better"

          • A_Pickle
          • 7 years ago

          [quote<]I should quote myself: "The less paste there is the better"[/quote<] Yes, but that's why CPU heatsinks attach to the CPU firmly. They'll pretty effectively squeeze out what they don't need, and keep what they do.

            • cegras
            • 7 years ago

            I’m not so sure about that. Not all mounting mechanisms are equal.

            • derFunkenstein
            • 7 years ago

            no, but they’re all really fucking tight.

          • glynor
          • 7 years ago

          I think people mostly read that “less paste there is better” and assumed you meant “no paste is best” (which is clearly not what you meant).

          I concur completely with what you wrote: metal-on-metal contact is best, and that’s what you want. The thermal compound is ideally only present to [i<]fill the gaps[/i<] from imperfect machining. Of course, you can't apply it in an absolutely ideal manner. Perhaps a better way of phrasing it would have been: Use the bare minimum of paste that is absolutely required, and no more.

          • rgreen83
          • 7 years ago

          So none is the least less, so the best?

      • NeelyCam
      • 7 years ago

      love the downthumbs

      • clone
      • 7 years ago

      have you tried going no paste?

      I’ve long since assumed a less is more position personally but would be curious to see the results if you had gone without.

    • Squeazle
    • 7 years ago

    The first 4 options could apply to many things.

    The first 3 made me think that I was taking an entirely different survey from what the title was.

      • burntham77
      • 7 years ago

      Now that’s funny.

    • WillBach
    • 7 years ago

    Why isn’t “tongue” an option? J/K 😉

    I voted for “finger inside a plastic bag” but I did use my finger last time because it was a rush job at work. Was washing my hands like Lady Macbeth afterwards, the stuff was supper tacky.

    • Squeazle
    • 7 years ago

    Razorblade.

    • NeelyCam
    • 7 years ago

    I use a credit card, just so I don’t have too much goop in there.

    But overall, I don’t think it matters, if you use good stuff like Arctic Silver. Especially if you’re applying it on an Ivy Bridge..

      • anotherengineer
      • 7 years ago

      Above or below the shield or both? 😉

      • MadManOriginal
      • 7 years ago

      You must be part of the 1% if you can so haphazardly use a CC to spread heatsink paste :p

        • NeelyCam
        • 7 years ago

        Cancelled cards..

      • oldog
      • 7 years ago

      Visa or Mastercard?

        • lilbuddhaman
        • 7 years ago

        Cheap long distance minute cards are the best…or anything that is particularly thin and flexible. It lets you get both a scraping and scooping action going. And if you’re REALLY lucky, the card would have been cut by a low-quality machine, which at first might sound bad, but those ultra-sharp edges it creates make for a great thermal paste application and spreading tool.

          • willmore
          • 7 years ago

          This^^^

    • Chaseme
    • 7 years ago

    Razor blade as well.

    Just happend to have one ready on the bench many years back and it looked suitable. Turns out I can get a nice thin equal coat of paste every time.

    • ronch
    • 7 years ago

    Does using the most difficult/complicated method (say, using a special tool) really improve heat dissipation vs. using the simplest, easiest method (finger)? I mean, as long as you’re careful to use just the right amount of TIM and spread it well enough, it should work just fine. And in my experience the finger method isn’t inferior.

      • Diplomacy42
      • 7 years ago

      isn’t pressing down with the heat-sink technically the “simplest, easiest” method?

      the reason people use a razor blade is to minimize the amount of paste on the chip, and keep it as thin as possible.

      the reason people put a dab and push down is because its fast and easy.

      the reason people use their finger is because they like to copy the finicky obsessives, but aren’t themselves finicky and obsessive. this is generally good, except in this case it leaves air bubbles, which is bad.

        • ronch
        • 7 years ago

        Regardless of whether you put a dab/blob and relied on the pressure to spread out the TIM, or manually spread out the TIM all over the heat spreader using your finger, I’m referring to using your finger instead of using tools like blades, credit cards, pieces of thin plastic with cutouts, etc. Using your finger in the absence of the above-mentioned tools is faster, cheaper and works well enough, IMHO. As for air bubbles, I don’t think it’s really a big deal, and it’s not like your temps will rise noticeably if there are air bubbles, unless your application was an epic fail.

    • Chrispy_
    • 7 years ago

    Credit-card jobs for exposed dies only. Squeeze out a small drop and apply heatsink normally for everything with a heatspreader on it.

    Whatever I do, the rules are “not too much” and “don’t arse about trying to get it perfect”.
    It really doesn’t matter that much, and any excess will squeeze out of the sides over time anwyay.

    • indeego
    • 7 years ago

    Finger inside a plastic bag, finish up with razors that come out of my fingers.

    • Ryhadar
    • 7 years ago

    I voted for the “drop + heatsink pressure” option but if the heatsink manufacturer’s instruction say otherwise that’s what I’ll do.

    • glacius555
    • 7 years ago

    Goat cheese.

      • DPete27
      • 7 years ago

      How about [url=http://www.hardwaresecrets.com/article/Thermal-Compound-Roundup-February-2012/1490/5<]toothpaste, mustard, or lipstick?[/url<]

    • colinstu12
    • 7 years ago

    Razor blade.

    SERIOUSLY. “Finger”?!! Your finger has oils and dead skin cells on it… I wouldn’t want that anywhere near between my IHS and HSF.

    Finger in plastic bag is nearly as bad. Loppy and ugly paste jobs every time I’ve seen with that. Blegh. I’d soon rather do the squeeze methods (pushing hsf onto pea/rice-sized drop of tim) then use the pastic bag.

    • satchmobob
    • 7 years ago

    Another one for squeeze and press. Used to do the whole credit card thing but experimented with both methods and the temp differences were not worth mentioning.

      • DPete27
      • 7 years ago

      Good point. If the difference is only < 2C, why spend 10 minutes making a perfectly even and thin spread when you can just plop some TIM in the middle and tighten down the heatsink.

    • Forge
    • 7 years ago

    I have never hand-spread. It’s far too easy to get something “flat” that isn’t, and then trap an air bubble in your TIM.

    I used to do a blob, and then attach HSF, sans clips, and move it very gently in a small circular motion. It gave a nice even spread. Now I don’t even bother with the circular motion, most clips make that too difficult/dangerous.

      • rrr
      • 7 years ago

      Yeah, it’s better to use your hand to spread different substances than TIM.

    • anotherengineer
    • 7 years ago

    In other news

    thermal past is poisonous, don’t touch it!!!!!!!!!

      • Diplomacy42
      • 7 years ago

      didn’t TR post a whole thing on their front page that essentially said

      “don’t do this?”

        • anotherengineer
        • 7 years ago

        For the heck of it I wanted to see if I could get more negatives than Neely, but I failed horribly.

        • shaq_mobile
        • 7 years ago

        really? ive been doing it for years. oh well. i also work on cars, which offers you lots of opportunities to inhale, touch and soak in some pretty funky stuff. ever used airplane paint stripper?

      • derFunkenstein
      • 7 years ago

      I agree it’s not something you want to eat.

    • chuckula
    • 7 years ago

    I used to be in the spread out a thin layer with razor/credit card camp, but then I saw some online videos done with plexiglass that really showed the drop + pressure method works a whole lot better and avoids producing air bubbles in the TIM.

    [url<]http://www.youtube.com/watch?v=ffK7L0Qj13Q&playnext=1&list=PL5BDF8C3F398A7CB6&feature=results_main[/url<]

      • colinstu12
      • 7 years ago

      there were flaws with that plexiglass demo. I can’t remember exactly what but there was a thread over on [H] about it. The razor method is still fine.

        • MadManOriginal
        • 7 years ago

        I would question the pressure applied during those tests. Don’t standard retention mechanisms apply over 100 pounds (maybe I have the units wrong) of pressure?

          • UltimateImperative
          • 7 years ago

          You may be thinking of the pressure applied on the CPU package by the retention clip on Intel LGA sockets. I doubt the little plastic clips on the standard retail heatsink could withstand anything like 100 pounds.

    • DPete27
    • 7 years ago

    I used to spread it out with a credit card until I had a problem once where the TIM didn’t “bond” well to the heatspreader and the base of the heatsink. After reading [url=http://www.hardwaresecrets.com/article/What-is-the-Best-Way-to-Apply-Thermal-Compound/1303<]this article[/url<] I switched to a dot in the middle and letting the heatsink flatten it out. (BTW, they [url=http://www.hardwaresecrets.com/article/What-is-the-Best-Way-to-Apply-Thermal-Grease-Part-2/1392/1<]tried the same techniques with a heatpipe cooler[/url<]

      • MadManOriginal
      • 7 years ago

      Those articles are funny – they say that their margin of error is 2 degrees, and yet when the results differ by 2 degrees or less they draw a conclusion about which method is better or worse.

        • DPete27
        • 7 years ago

        True, but the chances of getting an air bubble between the heatspreader and heatsink when spreading the TIM is much greater than letting the heatsink spread a dot/rice grain IMO

    • MadManOriginal
    • 7 years ago

    Other: squeeze out a dab then a razor blade.

    *I will add I use that for exposed dies. For heatspreaders, I’ve used center dot, line, and spread methods depending upon how the die is oriented under the heatspreader.

      • eofpi
      • 7 years ago

      Same here. It’s also convenient for cutting inferior TIMs off heatsinks.

      • SoM
      • 7 years ago

      same here aswell

      razor blade

      can’t believe it’s not in the poll

      RAZOR BLADE

      edit;

      plastic wrapped finger ? hmm, perhaps a trojan

        • willmore
        • 7 years ago

        Uhh, that’s not thermal paste…

Pin It on Pinterest

Share This