Does thermal paste lose its efficiency over time?

Good Question.

I think so, the generic grease that comes with the HSF or CPU I would say would only last and I mean be fully effective maybe in just a year, considering you use the pc on a somewhat daily basis. A good thermal paste like Artic Silver 5 I think would last a year or two.

Also since there's heat involve, the warmer or hotter the cpu the shorter the lifespan of the thermalpaste.

So there are many factors like the quality, heat, exposure and etc.

But for me I think it should be replaced on a yearly basis, but then again the pc would be outdated and needs upgrading. :wink:

yeah I think Arctic Silver is pretty good up to about 2 years. after that it starts getting a wee bit solid/crumbly and should be re-applied.

awesome question I have no clue as I just slap the stuff on and usualy dont worry about it... (I upgrade so often I doubt the stuff ever gets old lol) I should probably redo the TIM on my OCed Northwood as its just the stock thermal pad not even AS but the temps seem fine so I never bother  

I hear AS5 gets better with age. it settles and forms a more uniform thinner layer over time.

I think it would be nice to record the temps you get when you first apply AS5 and then re recored the temps a month later.


The thermal paste on those intel stock coolers is another story, over time that grey waxy stuff turns into a hard, dry kina grit.

good question btw

From Arctic Silver's website:

that's a good point. Also make sure to clean up all the dust from your heatsink/fan and your case fans to make sure they are in the same condition when you first applied the paste.

I say no. It's purpose it to fill in the microscopic gaps in the HS and cpu, and once it's there and has set up, leave it alone.

I guess I'll never know since I upgrade once a year.

Ive had my computer for 6 years and it runs 24/7 Never replaced the thermal paste. And the cpu usually averages 65% during normal use and i play games with it that max it out (older games) and it has never overheated. It only has one fan that is the on the Power supply. I dont think it degrades. Room temp is usually around 75 F

In my experience, yes thermal paste can deteriorate over time. I lost a processor because of it not too long ago..

It was a self built 900mhz amd athlon t-bird (awesomely fast computer when it was 1st built) and probably after being through 4yrs of college, with no a/c in 90 degree weather, it was time.
After the processor died, i pulled it out and all that was left of the grease was a little residue on the sides of core. I think the how much heat thing has a lot to do with it too, though. The thing ran HOT (130F+ at idle, not much more at load though)
I don't know about cheaper / more expensive stuff though...

I doubt you can attribute that directly to any breakdown in the paste.

I would say re-apply every 2 or 3 years.

Theremal paste contains liquid, as with any liquid it will evaporate over time. Evaporation causes the paste to eventually shrink and crack, thus less paste will be contacting the heatshink and the CPU.

Worse, as paste drys out it also cracks and that can result in tiny little air pockets in what is left of the paste. Last I heard, air is really not a good conductor of heat.

On a similar yet totally unrelated note, I read something several years ago (sorry, no sources) that did tests on heat conductions with a single paste, and then after applying one paste, then removing it and applying a different paste.

The bottom line was that you never will get all the old paste off, so pick the one you want from the beginning and stick with it.

Now, back to the topic. I personally think this is sort of like that hype surrounding the power drawn by computers and the opinions on what size of power supply one should have, but that's just my opinion.  

The problem is that it would be difficult to determine whether any benifit was due to aging paste or suboptimal initial application, unless you did some very carefull benchmarking when you first applied the paste.

Not just the temps, but the fan RPM's, room temp ...

It would take a lot of careful setup and patience to fully test this.

Well AS5 has 99,9% pure silver in it, so it is obviously the silver that does the heat transfer and I don’t think that it would even matter that the 0.1% of liquid dried out if it even would. All that 0.1% of liquid is used for is so the consumer can have the silver dust stick to the cpu until they slap on the hsf and so it wont blow away before the hsf is on.

on another note a about cooling paste, I have seen a liquid metal such as mercury used as a conductor between a cpu and a hsf. The stuff I saw wasn’t mercury because that sh!t is poison but it was a liquid metal non the less and I thought that would work rather well

i usually reaplly my thermal paste every time i do a deep clean of my computer, every 6 months or so...bucause i remove the HSF and then i have to clean and reaply the thermal paste.

From my personal experience, thermal paste of good quality should be good enough to maintain their efficiency over time. Well, from my personnal experience at least.

My first P4 (a 1.8 Northwood) had a slight increase of average temperature over the 18 months or so I used it. I realized that it only had it thermal pad when I upgraded it to a P4 3.0-800fsb Northwood in january 2004 if I remember right. For my new computer I used Artic Silver paste. It was, and still is at 38-40C (can't tell in F) at idle and reach max 50C during heavy gaming (Half-Life2 :Episode1 right now, AWESOME game by the way!!!) or video encoding that I do a lot. I especially do a lot of transcoding from different video format with application that use multithreading, which make the cpu heat even more.

So, these number are for sure entirely accurate because I'm fairly scared of cpu overheating and first thing I do when I quit a game (after say 2 hours of continuous gaming) or heavy video encoding is to check the cpu temperature. I also remember freaking out when I saw my new cpu loaded temperature going from 47-48 to 50 compare to my old Northwood 1.8 overclock to 2.4-533, which is nothing I now realize.

My 2 cents.  

the 99.9% silver is the purity of the silver, not the silver content of the paste. This is a common misconception. If it were 99.9% silver content it would look like silver and not be a grayish paste.
I figure the paste will last until the computer reaches obselescence, after that point it probably doesn't matter if the paste is working to full capacity anyway.

-mcg

Yep, if AS 5 was actually 99.9% silver it would be solid silver, not a paste.

Tell me if I'm wrong but the paste is so the heat sink dosent get melted onto the cpu so changing it should not be a problem once on.

No, the paste is to fill the microscopic gaps in the HS, mainly. Most of it gets pushed out when you clamp the HS down and what remains is only in the tiny grooves in the HS and the processor. It has nothing to do with glueing anything down, or preventing anything from being melted or whatever. It's purpose is to fill voids and increase thermal conduction.

The thermal paste conducts the heat from the CPU to the heatsink.

If the CPU and the heatsink both had 100% flat 100% smooth surfaces which made perfect contact it probably wouldn't help.

But on a microsopic level the surfaces are far from flat and the trapped air makes an excellent insullator.

No paste and your CPU will burn up, but there won't be any melting.

Aluminum doesn't melt until 660C, Copper is good to 1084C and Silicon to 1414C.

I suspect there is no single answer that works in every case.

A computer that has not been moved much may need reapplication less often than a computer case that is used at LAN parties, for example. Bumping the case or shaking it during transit could easily effect the cpu-heatsink interface.

Additionally, a computer with better cooling and airflow will have a temperature range that is smaller than a computer that has poor airflow and frequently transitions from very cool (when off) to very hot (when on and under load). The difference in magnitudes between the cycles of heating/expanding and cooling/contracting between two computers with different designs could be very large. These cycles, and especially their intensity, could conceivably play a role in the process by which the cpu-heatsink interface goes from air-free to cracked and dry.

I'd almost be willing to bet that the cpu wouldn't burn up, almost....which is why I'm not going to test it myself. :wink: It certainly would run hotter than it should/could with paste.

Thank You for the correction I was thinking that when I posted that but I thought "well thats what the as5 site says, sounds wrong but.."

this is what I read from as5's web site

A CPU can most definately be permanently damaged by overheating.

I know someone who killed a CPU that way because they wanted to test to make sure everything was put together properly before mounting the heatsink. Apparently an Althon XP lasts less than ten seconds without a heatsink.

What would likely happen if used no paste is the on die thermal sensor will trigger the BIOS to shutdown the system before this happens.

However for awhile AMD's systems had bad thermal protection compared to Intel, and you could lose a CPU if your heatsink fan quit spinning.

Early Athlon's lacked an on die thermal sensor and even when they were added, it still took awhile for the motherboard makers to take full advantage of them.

it does loose its efficiency over time.monitor the CPU temperature during the first time you apply a new thermal paste(using the motherboard software which provided on the disc) , record the lowest and highest temperature. after sometime,(1-2months) or untill u notice the temperature is 5 degrees higher than the one u recorded earlier, this is the time u should reapply the thermal paste.

I never said it couldn't be damaged by overheating. I said I wasn't completely convinced not using paste would cause it. I'm sure using paste is a good idea. I'm not sure the CPU is going to burn to dust under the HS without it.

As an addition, I wouldn't hesitate to test a Taulatin processor without paste. Next time I have an opportunity I might do it with a P3 800 (coppermine) I can spare, but I probably would avoid doing the same with a Prescott. I actually have K6/2 500's running without paste right now, and have been for 6 years. All in all, I'm sure it varies from part to part.

That doesn't make much sense, what about dust buildup? And I know on a daily basis my room heats up about 5C from early morning (as im posting) to early evening (about 4pm). Does that mean I should replace my AS5 on a daily basis? Heh, no. This is mainly due to a crappy AC unit.

The conditions when you record temps should be identical or as close as possible to the original conditions, so just saying when its 5 degreees higher its time to replace is too small of a margin.

Reapply every 1-2 months?!?! AS5 takes 200 hours just to cure. Depending on computer usage, that may be 1-2 months just to fully cure.

And Zunaro is right 5 degree C aint crap.

It does go off.

My graphics cards where getting warmer than usual, and I decided to take the fans apart and clean them, but it was no different.
I took the heat sink off to see if there was enough paste, which there was. I then replaced the paste by cleaning it all off and putting on some cheap thermal paste, and the temperature has gone down buy quite a bit.

My old Barton XP is the same. It is now used as a small server and on all the time. A few days ago it turned itself off because it over heated, the fan had been cleaned a few times but made no difference. I suggested new paste, and the temperature dropped to an acceptable level.

I asked Arctic Silver:
There is no need to reapply our thermal compounds unless the thermal interface has been physically disturbed by bumping the heatsink or rough transport.

Feel free to post this if you like. Just don't post my email so the bots won't pick it up and add to my SPAM load.  

--
Colin Thompson
Arctic Silver, Inc.
I guess he means that there is no degredation over time.

That it artic silver, not every one uses it.

How ever it is interesting to know for the future