Cleaning liquid metal off pure copper

[cleric_warlock]

After running my computer for a month with my first ever application of liquid metal between the cooler and IHS and enjoying some pretty nice temps, I decided to take off the cooler to see how the application held up so I could spot any potential issues that I might not have been seeing. Thankfully, there was no indication that the liquid metal had gone anywhere it wasn't supposed to be, but now it seems like I have a new problem. I had to use gentle yet continuous force to pry the cooler off the CPU since the liquid metal seems to have bonded the pure copper IHS I have to the pure copper cold plate of my cooler. From what I've read, this seems to be normal since the conductonaut liquid metal compound that I used will alloy with copper over time, both absorbing itself into the copper surface and creating some adhesion between the cold plate and IHS.

 

My real question here is what level of cleaning is sufficient before I can re-apply compound and re-mount my cooler? It seems like a lot of what is left on the surface of my IHS and cold plate may be bonded enough that I won't be able to clean it off, but I know that doesn't necessarily pose a problem to heat transfer. I only included photos from my cold plate below since it's in pretty much the exact same condition as my IHS. Do I need to clean it off more than the after pic below or clean it with something other than the 99 proof isopropyl and coffee filters I've been using or just spend a lot of time scrubbing it down? Please help.

 

Before:

 

After:

 

What came off:

[DoctorNick]

Try with a microfiber cloth and more isopropyl. It's normal that LM bonds to pure copper. Shouldn't affect temps though. From a Google search, this looks like many other posts. Haven't tried LM after the heatsink was used with normal paste, so in my experience it just looks like a darker spot, not the greyish area.

[LIGISTX]

59 minutes ago, cleric_warlock said:

After running my computer for a month with my first ever application of liquid metal between the cooler and IHS and enjoying some pretty nice temps, I decided to take off the cooler to see how the application held up so I could spot any potential issues that I might not have been seeing. Thankfully, there was no indication that the liquid metal had gone anywhere it wasn't supposed to be, but now it seems like I have a new problem. I had to use gentle yet continuous force to pry the cooler off the CPU since the liquid metal seems to have bonded the pure copper IHS I have to the pure copper cold plate of my cooler. From what I've read, this seems to be normal since the conductonaut liquid metal compound that I used will alloy with copper over time, both absorbing itself into the copper surface and creating some adhesion between the cold plate and IHS.

 

My real question here is what level of cleaning is sufficient before I can re-apply compound and re-mount my cooler? It seems like a lot of what is left on the surface of my IHS and cold plate may be bonded enough that I won't be able to clean it off, but I know that doesn't necessarily pose a problem to heat transfer. I only included photos from my cold plate below since it's in pretty much the exact same condition as my IHS. Do I need to clean it off more than the after pic below or clean it with something other than the 99 proof isopropyl and coffee filters I've been using or just spend a lot of time scrubbing it down? Please help.

 

Before:

 

After:

 

What came off:

Don’t use liquid metal directly on copper… it’s eating the copper. 
 

Liquid metal also will not gain you anything here, a good paste will be fine. Liquid metal is intended for extremely tight applications where there is not much surface area to transfer the heat out. A very flat surface of a CPU die and the bottom of the IHS, which has a small area to transfer the heat and is extremely flat. Top of IHS and bottom of cold plate have a much, much larger surface area, and are not as flat. The larger surface area means a paste with less thermal transfer will do fine since it has more area to spread the transfer out over, and being less flat it fills in the areas better.

 

But, again, most importantly, the LM is literally eating the copper. You can’t clean it, and it will only get worse with time especially if you apply more. Don’t use LM on copper…. 
 

If you really want to clean it, you need to sand or machine the surface. 

[AnonymousGuy]

31 minutes ago, LIGISTX said:

Don’t use liquid metal directly on copper… it’s eating the copper. 

It's diffusing into the copper, not dissolving it.  aka it's making a stain.

[cleric_warlock]

36 minutes ago, LIGISTX said:

Don’t use liquid metal directly on copper… it’s eating the copper. 
 

Liquid metal also will not gain you anything here, a good paste will be fine. Liquid metal is intended for extremely tight applications where there is not much surface area to transfer the heat out. A very flat surface of a CPU die and the bottom of the IHS, which has a small area to transfer the heat and is extremely flat. Top of IHS and bottom of cold plate have a much, much larger surface area, and are not as flat. The larger surface area means a paste with less thermal transfer will do fine since it has more area to spread the transfer out over, and being less flat it fills in the areas better.

 

But, again, most importantly, the LM is literally eating the copper. You can’t clean it, and it will only get worse with time especially if you apply more. Don’t use LM on copper…. 
 

If you really want to clean it, you need to sand or machine the surface. 

It's not eating the copper though, it's plating the copper in a way that only really causes a lot of cosmetic damage and doesn't affect heat transfer. This is pretty well known and confirmed by many reliable sources such as Gamers Nexus. Machining or sanding to improve the look of the surface doesn't make sense if there's no functional benefit due to the potential that you could be altering the curvature of the contact plates and worsening heat transfer as a result.

 

I'm not used to assessing the difference between normal cosmetic damage and a thermal contact surface that actually still needs cleaning in this particular situation, so that's why I made my original post.

[wildgg]

4 hours ago, cleric_warlock said:

My real question here is what level of cleaning is sufficient before I can re-apply compound and re-mount my cooler? It seems like a lot of what is left on the surface of my IHS and cold plate may be bonded enough that I won't be able to clean it off, but I know that doesn't necessarily pose a problem to heat transfer. I only included photos from my cold plate below since it's in pretty much the exact same condition as my IHS. Do I need to clean it off more than the after pic below or clean it with something other than the 99 proof isopropyl and coffee filters I've been using or just spend a lot of time scrubbing it down? Please help.

 

 

 

I always got a high grit sandpaper and sanded it down a tiny bit but even doing so you mostlikely won't get everything back to new.

[LIGISTX]

8 hours ago, AnonymousGuy said:

It's diffusing into the copper, not dissolving it.  aka it's making a stain.

I was always under the impression it was a corrosive affect, more or less a form of galvanic corrosion. Is this not the case? LM typically warns against use directly on copper for this reason. Pretty sure my tube specifically said do not use on bare copper. 

 

8 hours ago, cleric_warlock said:

It's not eating the copper though, it's plating the copper in a way that only really causes a lot of cosmetic damage and doesn't affect heat transfer. This is pretty well known and confirmed by many reliable sources such as Gamers Nexus. Machining or sanding to improve the look of the surface doesn't make sense if there's no functional benefit due to the potential that you could be altering the curvature of the contact plates and worsening heat transfer as a result.

 

I'm not used to assessing the difference between normal cosmetic damage and a thermal contact surface that actually still needs cleaning in this particular situation, so that's why I made my original post.

See above. 
 

 

Looks like GN does say it’s a stain though, which is interesting. I am fairly certain it’s recommended to not be used directly on copper. 
 

But, either way I suppose, there is no reason to use it in the way you are, a paste will work just as well and will not risk potentially running and causing a short. 

[cleric_warlock]

On 8/10/2022 at 7:30 AM, LIGISTX said:

I was always under the impression it was a corrosive affect, more or less a form of galvanic corrosion. Is this not the case? LM typically warns against use directly on copper for this reason. Pretty sure my tube specifically said do not use on bare copper. 

 

See above. 
 

 

Looks like GN does say it’s a stain though, which is interesting. I am fairly certain it’s recommended to not be used directly on copper. 
 

But, either way I suppose, there is no reason to use it in the way you are, a paste will work just as well and will not risk potentially running and causing a short. 

You must be using a different LM thermal paste than me since conductonaut specifically markets itself as being safe to use on pure copper. If you section off the paste application area properly with masking tape and paper towels, you can easily eliminate the possibility that the liquid metal will be anywhere other than exactly where it's supposed to be. LM spreads very thin so you don't really need much of a buffer to ensure that the mounting pressure of your thermal interface won't squeeze out LM onto sensitive components. There is typically a 2-3C thermal benefit to using LM instead of conventional thermal paste, and my philosophy with this PC build was to leave absolutely no performance on the table. Doesn't really matter much to me if the thermal contact plates look bad as long as they continue to perform optimally, which I have no reason to believe they won't. 

[LIGISTX]

1 hour ago, cleric_warlock said:

You must be using a different LM thermal paste than me since conductonaut specifically markets itself as being safe to use on pure copper. If you section off the paste application area properly with masking tape and paper towels, you can easily eliminate the possibility that the liquid metal will be anywhere other than exactly where it's supposed to be. LM spreads very thin so you don't really need much of a buffer to ensure that the mounting pressure of your thermal interface won't squeeze out LM onto sensitive components. There is typically a 2-3C thermal benefit to using LM instead of conventional thermal paste, and my philosophy with this PC build was to leave absolutely no performance on the table. Doesn't really matter much to me if the thermal contact plates look bad as long as they continue to perform optimally, which I have no reason to believe they won't. 

I have the same stuff, it’s what I used when I delid my 4770k years and years ago. 
 

As far as sectioning off the area, LM can run once it gets warm, and if it runs down the IHS… you will have a really magical smoke type of day. This is why when you delid and use LM, if there are any exposed conductive surfaces around the die, you use something like clear nail polish or a non conductive conformal coating to cover them in case any of the LM runs… much harder to do outside the IHS as you have no real way to do this. 
 

Yes, LM transfers heat better, but it is a much thinner material then normal TIM. TIM is thicker specifically so it can fill in all the microscopic hills and valleys between the IHS and cold plate; LM may be to thin in some areas to make perfect contact between both surfaces. Obviously the best option is no thermal interface at all and have two perfectly machined surfaces, but that isn’t happening on a PC. 
 

Also, I’d argue, the potential gain of a few c isn’t going to be a detriment to overclocking. A few c here or there isn’t going to matter, that’s well within the noise of normal day to day usage. 

[cleric_warlock]

On 8/12/2022 at 3:23 PM, LIGISTX said:

I have the same stuff, it’s what I used when I delid my 4770k years and years ago. 
 

As far as sectioning off the area, LM can run once it gets warm, and if it runs down the IHS… you will have a really magical smoke type of day. This is why when you delid and use LM, if there are any exposed conductive surfaces around the die, you use something like clear nail polish or a non conductive conformal coating to cover them in case any of the LM runs… much harder to do outside the IHS as you have no real way to do this. 
 

Yes, LM transfers heat better, but it is a much thinner material then normal TIM. TIM is thicker specifically so it can fill in all the microscopic hills and valleys between the IHS and cold plate; LM may be to thin in some areas to make perfect contact between both surfaces. Obviously the best option is no thermal interface at all and have two perfectly machined surfaces, but that isn’t happening on a PC. 
 

Also, I’d argue, the potential gain of a few c isn’t going to be a detriment to overclocking. A few c here or there isn’t going to matter, that’s well within the noise of normal day to day usage. 

After seeing the results of using liquid metal for the last month, I have to disagree with some of your points. I have stress tested my machine near 90C for extended periods of time and the liquid metal showed no signs of having flowed from where I originally applied it. I was actually quite surprised by how well the LM stayed in the areas where it was originally applied. My guess is that if there are serious problems with liquid metal flow, far too much has been applied to the thermal contacts or the LM formulation itself isn't the best. 

 

Also, as far as ensuring adequate coverage and air pocket closure it is best practice to paint both the cold plate and the IHS since the LM layers are thin as opposed to conventional thermal compound where you only apply it to the IHS. If the contact between both plates is good and you apply LM to both surfaces, you should have no air pockets to worry about. The thermal performance I've been seeing on my machine supports that conclusion.

 

I'd also argue that liquid metal gets the thermal interface as close as it can be to the perfect metal on metal since the compound when applied properly will alloy with both sides of the thermal interface effectively creating an adhesive metallic bond between them and making for highly effective heat transfer through a well formed metal crystal structure.

[hkl2003]

For my next build I would really like to go the extra mile and use liquid metal thermal paste on my AIO cpu cooler, which will be pure copper. Thermal Grizzly warns that liquid metal can be used with pure copper surfaces, but will leave marks. Nickel-plates copper is recommended, but there aren't many AIOs with this type of surface.

 

A google search brought me to this thread.

 

How did you solve the problem, was cleaning with a tissue enough? Did you notice a worse cooling performance after re-installing the cooler? Any info on the results would be much appreciated.

[Neggy-Z]

10 hours ago, hkl2003 said:

How did you solve the problem, was cleaning with a tissue enough?

You'll need more then just a tissue for cleaning if there is any solid buildup, I've always used a piece of scotch brite, the rough side of a cleaning sponge is also good, as this isn't as harsh as using sandpaper but still gets the surface smooth.

 

You shouldn't notice any performance loss from continued use, providing both surfaces are cleaned and prepared properly.

[cleric_warlock]

On 3/7/2023 at 10:31 AM, hkl2003 said:

How did you solve the problem, was cleaning with a tissue enough? Did you notice a worse cooling performance after re-installing the cooler? Any info on the results would be much appreciated.

I reinstalled the cooler with the same liquid metal and everything and it has been working great for months. There has been no significant change in operating temps before and after my remount. Pretty much all of what you see in my photos is cosmetic damage to the copper plating that is inconsequential to heat transfer. All I did before remounting the cooler was scrub both plates down a bit with some microfiber cloth dipped in 100% acetone before I reapplied LM to both plates. The plates were already so smooth before I did any cleaning that they probably would have been fine if I just skipped cleaning and remounted them with a fresh coat of LM. If you run your fingers over either of the thermal plates and feel any major bumps, sand those down with fine grit sand paper. If the plates are as clean as can be and feel fairly smooth to the touch, you are good to reapply. The cosmetic damage to the plates can look worrying, so keep in mind that unless you feel any serious bumps on the plates' surfaces, doing any sanding may worsen the quality of the thermal interface by creating new air pockets. Sanding should be used as a last resort and only in the specific areas that don't pass the smooth feel test.

[KAMPLC GAMES]

On 8/10/2022 at 8:22 AM, AnonymousGuy said:

It's diffusing into the copper, not dissolving it.  aka it's making a stain.

Indeed after 3 years can confrim. It will uses some stain but mostly you can clean out from brass .  Maybe will stay  some little edge from CPU which is not fatal especially if you using water colling kits which means after 3 years is life  expetancy done, i mean water pump from kit colling cooler, should die after 3 years on average. So if you are purchasing water cooling kit you cant expect that will last for 6 years or 7 years. Maybe if you don`t use regulary computer, but hey you build to enjoy it

[KAMPLC GAMES]

On 8/12/2022 at 10:36 PM, cleric_warlock said:

You must be using a different LM thermal paste than me since conductonaut specifically markets itself as being safe to use on pure copper. If you section off the paste application area properly with masking tape and paper towels, you can easily eliminate the possibility that the liquid metal will be anywhere other than exactly where it's supposed to be. LM spreads very thin so you don't really need much of a buffer to ensure that the mounting pressure of your thermal interface won't squeeze out LM onto sensitive components. There is typically a 2-3C thermal benefit to using LM instead of conventional thermal paste, and my philosophy with this PC build was to leave absolutely no performance on the table. Doesn't really matter much to me if the thermal contact plates look bad as long as they continue to perform optimally, which I have no reason to believe they won't. 

After 3 years use of liquid metal only small stain was on brass side of cooler. CPU  with no worries cleaned and donated to a friend, was good as new

 

[hkl2003]

On 3/14/2023 at 11:03 PM, cleric_warlock said:

I reinstalled the cooler with the same liquid metal and everything and it has been working great for months. There has been no significant change in operating temps before and after my remount. Pretty much all of what you see in my photos is cosmetic damage to the copper plating that is inconsequential to heat transfer. All I did before remounting the cooler was scrub both plates down a bit with some microfiber cloth dipped in 100% acetone before I reapplied LM to both plates. The plates were already so smooth before I did any cleaning that they probably would have been fine if I just skipped cleaning and remounted them with a fresh coat of LM. If you run your fingers over either of the thermal plates and feel any major bumps, sand those down with fine grit sand paper. If the plates are as clean as can be and feel fairly smooth to the touch, you are good to reapply. The cosmetic damage to the plates can look worrying, so keep in mind that unless you feel any serious bumps on the plates' surfaces, doing any sanding may worsen the quality of the thermal interface by creating new air pockets. Sanding should be used as a last resort and only in the specific areas that don't pass the smooth feel test.

Thanks for the update. With this issue out of the way I can use liquid metal without having to worry.

[cleric_warlock]

For anyone who stumbles upon this thread, I really felt I should link this more recent post I made about the things you should do to correctly reapply liquid metal now that I have much more experience: