Liquid metal is terrible and I hate it

[RejZoR]

I guess liquid metal is exotic enough thing to be posted here... It's gonna be hugely unpopular experience, but I've tried it enough times in different ways and I still hate it because it's just garbage. I was using Conductonaut and I have 1 full syringe of this junk and other one was used twice.

 

I've tried applying it to die only like thermal paste and initially, temps were awesome. After several days, temperatures got horrible to a point system was entirely unstable at same clocks as before using regular thermal paste. Then I read you're suppose to apply it to both, die and heatsink otherwise it doesn't stick well to the other non "greased" side. Ok, did that and behold, after few days, system became unstable again and temperatures were absolutely horrible.

And worst thing, this shit diffuses into EVERYTHING. Copper plate on graphic card was stained to a point I had to sand it. For the second time now. AiO cold plate, same fate. Sanded twice to get rid of the liquid metal shit. You say it doesn't diffuse into nickel plated CPU IHS? Wrong. IHS looked like an alien barfed on it. It was all smudged and ugly and no cleaning with alcohol removed this nasty ass liquid metal crap.

 

I've re-applied CoolerMaster Master Maker Nano thermal paste and temperatures are amazing again. And guess what, I know for a fact they'll remain this way. Unlike with liquid metal. Ugh, NEVER again. NEVER. It's just such a hassle because it takes forever to apply, impossible to remove and its weird diffusion into everything (not even any aluminium involved) which makes it seem like it just dried up on the IHS and cold plate. Still unsure whether I should just throw away both syringes. I'm staying with current CoolerMaster Nano paste or consider Kryonaut sometime in the future.

 

Everyone doing tests with liquid metal getting amazed by temps right after they apply it, but no one seems to test it long term. Maybe people will see it's not all that great. And I was aware of all the things like diffusion into copper, corrosive against aluminium and stuff like that, but I didn't expect it to be this short term thing and so god damn useless in the long run. And long run isn't even years. it's few weeks at best.

 

Just don't bother with liquid metal and just go for the best traditional thermal paste money can buy. It's just NOT worth all the hassle, limitations, problems and impossibility to clean it.

[Tan3l6]

12 minutes ago, RejZoR said:

I've tried applying it to die only like thermal paste

Ay Carramba!

Liquid metal has absolutely different applying method and the dosage is just a fraction of a conventional paste. 

Something like that is a good thing to watch before applying it:

 

I would not recommend using it over the IHS, only delidded CPU's ... so it's becoming obsolete method anyway.

[Light-Yagami]

I've watched many documented test from 3months of use up to a year, and everyone who's done an extensive testing project can confirm that:

 

1: LM continues to provide good thermal conductivity after a good while

2: LM does stain copper heatsinks, because copper DOES react with it - but it's not damaging to the alloy.

3: LM will "dehydrate" as it absorbs into the porous copper and you will have to reapply it down the road. For some people, that's 3 months. For others, it's a year.

4: LM does not react with nickel at all and it best used with it. After that comes the copper, which stains but keeps operating normally. Alluminum isn't compatible because LM form a new alloy with it, completely disintegrating it in the process.

 

Poor performance of LM occurs when: 

 

1: There's too much of it and it pools

2: There's to little of it and it doesn't make good contact

3: When there isn't enough pressure applied on the heatsink

4: When it's only applied on on side of the materials touching - meaning either just the CPU/GPU die or just the cooler. 

EDIT 5: When it's applied in the same manner as regular thermal paste. It's applied completely differently as your normal thermal paste.

 

You might've had a bad experience with it, but there's clear objective evidence what LM is a good cooling method and works very well when applied properly, with the right materials involved.

[Bombastinator]

My understanding is liquid metal is indium and gallium.  Indium is basically good, but gallium causes all kinds of problems. It’s corrosive for one thing.   Indium alone can be used as a TIM but requires really high pressures to make it work.

 

I avoid the stuff myself and like diamond fill thermal paste.  It’s not as conductive but lasts much much better.

[Falkentyne]

17 minutes ago, RejZoR said:

I guess liquid metal is exotic enough thing to be posted here... It's gonna be hugely unpopular experience, but I've tried it enough times in different ways and I still hate it because it's just garbage. I was using Conductonaut and I have 1 full syringe of this junk and other one was used twice.

 

I've tried applying it to die only like thermal paste and initially, temps were awesome. After several days, temperatures got horrible to a point system was entirely unstable at same clocks as before using regular thermal paste. Then I read you're suppose to apply it to both, die and heatsink otherwise it doesn't stick well to the other non "greased" side. Ok, did that and behold, after few days, system became unstable again and temperatures were absolutely horrible.

And worst thing, this shit diffuses into EVERYTHING. Copper plate on graphic card was stained to a point I had to sand it. For the second time now. AiO cold plate, same fate. Sanded twice to get rid of the liquid metal shit. You say it doesn't diffuse into nickel plated CPU IHS? Wrong. IHS looked like an alien barfed on it. It was all smudged and ugly and no cleaning with alcohol removed this nasty ass liquid metal crap.

 

I've re-applied CoolerMaster Master Maker Nano thermal paste and temperatures are amazing again. And guess what, I know for a fact they'll remain this way. Unlike with liquid metal. Ugh, NEVER again. NEVER. It's just such a hassle because it takes forever to apply, impossible to remove and its weird diffusion into everything (not even any aluminium involved) which makes it seem like it just dried up on the IHS and cold plate. Still unsure whether I should just throw away both syringes. I'm staying with current CoolerMaster Nano paste or consider Kryonaut sometime in the future.

 

Everyone doing tests with liquid metal getting amazed by temps right after they apply it, but no one seems to test it long term. Maybe people will see it's not all that great. And I was aware of all the things like diffusion into copper, corrosive against aluminium and stuff like that, but I didn't expect it to be this short term thing and so god damn useless in the long run. And long run isn't even years. it's few weeks at best.

 

Just don't bother with liquid metal and just go for the best traditional thermal paste money can buy. It's just NOT worth all the hassle, limitations, problems and impossibility to clean it.

I have LM on my laptop and Vega64 and temps are better than the best thermal paste I have (Thermalright TFX).

LM takes work to apply.  You have to make sure the surface is PRISTINELY CLEAN, you also have to make sure you didnt manage to scrape off any tiny particles (metal flakes, fibers of the applicator, dandruff, dust, anything) when applying it.  You also need to *NOT* have a fully polished surface!!!  Polishing a heatsink or IHS or CPU die before applying LM is the #1 biggest mistake newbies make.  It's literally the absolute WORST thing you can do, which is completely opposite of what you would normally do with thermal paste (regular pastes love mirror finishes).  Not only does it prevent wetting (the LM will simply refuse to stick to the surface and will remain in ball form for next to forever), but even when you do manage to break the LM tension to itself and make it spread, it will simply migrate to itself when it heats up and you compress the surface.

 

I've written some very detailed explanations (here and on reddit) on how to apply LM and how to sand both surfaces with 1500 grit sandpaper to give LM something to adhere to.  Please search as I am not going to write up everything all over again.  I'll say a tl;dr:

 

1) sand both surfaces with 1500 grit sandpaper to roughen them up (this increases LM wetting massively and longterm adherence massively)

2) spend 10-15 minutes wiping the LM that you applied on the surface around and around repeatedly, without applying any extra downwards pressure, to accelerate gallium absorption into the micro crevices the sandpaper made.  Do this to all surfaces.

3) apply a new layer afterwards on all surfaces, fast wipe and spread should then take 30 seconds.

4) Profit.

[Light-Yagami]

@RejZoR Aja če rabiš kešno pomoč mi piši v DM. Tukaj ti ne bo noben pomagal

[Falkentyne]

19 minutes ago, Light-Yagami said:

I've watched many documented test from 3months of use up to a year, and everyone who's done an extensive testing project can confirm that:

 

1: LM continues to provide good thermal conductivity after a good while

2: LM does stain copper heatsinks, because copper DOES react with it - but it's not damaging to the alloy.

3: LM will "dehydrate" as it absorbs into the porous copper and you will have to reapply it down the road. For some people, that's 3 months. For others, it's a year.

4: LM does not react with nickel at all and it best used with it. After that comes the copper, which stains but keeps operating normally. Alluminum isn't compatible because LM form a new alloy with it, completely disintegrating it in the process.

 

Poor performance of LM occurs when: 

 

1: There's too much of it and it pools

2: There's to little of it and it doesn't make good contact

3: When there isn't enough pressure applied on the heatsink

4: When it's only applied on on side of the materials touching - meaning either just the CPU/GPU die or just the cooler. 

EDIT 5: When it's applied in the same manner as regular thermal paste. It's applied completely differently as your normal thermal paste.

 

You might've had a bad experience with it, but there's clear objective evidence what LM is a good cooling method and works very well when applied properly, with the right materials involved.

Yagemi: you forgot: LM hates polished surfaces.  Trying to apply LM to a polished surface is asking for bad temps in a matter of days.

@ Everyone:  Here's proof I know what I'm talking about.

Got 150 grams of stuff I made myself.

 

[RejZoR]

I'm just gonna stick with high end thermal paste. Temps are still excellent and it's effect is consistent through its entire age. Be it months or even years. I'm not going to keep re-applying this garbage nonsense every few weeks after system just out of the blue starts shitting itself. Until I finally figured out liquid metal is the last thing I did to the system. Just nope. All the surfaces were mechanically cleaned to remove LM crap and they won't see it ever again. And any system I'll be building in the future will be the same. I will replace generic grease with high end one because that does make a difference. I didn't apply Conductonaut to my laptop because it has aluminium plate between Ryzen 2500U and heatpipes, but giving it CoolerMaster Nano paste dropped temps dramatically and it's so much cooler it runs passive most of the time and fan only kicks in when I really put some load on it.

[Light-Yagami]

1 minute ago, RejZoR said:

I'm just gonna stick with high end thermal paste. Temps are still excellent and it's effect is consistent through its entire age. Be it months or even years. I'm not going to keep re-applying this garbage nonsense every few weeks after system just out of the blue starts shitting itself. Until I finally figured out liquid metal is the last thing I did to the system. Just nope. All the surfaces were mechanically cleaned to remove LM crap and they won't see it ever again. And any system I'll be building in the future will be the same. I will replace generic grease with high end one because that does make a difference. I didn't apply Conductonaut to my laptop because it has aluminium plate between Ryzen 2500U and heatpipes, but giving it CoolerMaster Nano paste dropped temps dramatically and it's so much cooler it runs passive most of the time and fan only kicks in when I really put some load on it.

Vzem Grizzly Kryonaut na mimovrstah, Preverjeno deluje najbolje. Če rabiš vsako stopinjo 

[RejZoR]

1 minute ago, Light-Yagami said:

Vzem Grizzly Kryonaut na mimovrstah, Preverjeno deluje najbolje. Če rabiš vsako stopinjo 

Using CoolerMaster Maker Nano and it's cool (pun intended). I'll try Kryonaut next time to see how it compares. Believe it or not, still have a syringe of MX-4 from like a decade ago lol. Still use it for low end but still demanding cooling. Last time I smeared it on my router as replacement for thermal pads on chipsets and temps were like 20°C lower on processor and antennas lol.

[Light-Yagami]

1 minute ago, RejZoR said:

Using CoolerMaster Maker Nano and it's cool (pun intended). I'll try Kryonaut next time to see how it compares. Believe it or not, still have a syringe of MX-4 from like a decade ago lol. Still use it for low end but still demanding cooling. Last time I smeared it on my router as replacement for thermal pads on chipsets and temps were like 20°C lower on processor and antennas lol.

Večinoma vse termalne paste delujejo enako v okoliščinah, kjer ne gledaš za vsakim procentom ki ga lahko dobiš. Potem ko greš bolj v ekstrem, npr pri laptopih - potem je vredu če imaš res top izdelek.. Saj jaz sem do zdaj tudi uporabljal nekaj od Cooler Masterja in je vredu delal.. Ampak v naslednji laptop gre vrjetno Kryonaut, če ne kar LM.

 

Fajn bodi

[WkdPaul]

* thread cleaned *

[porina]

I'm half and half on LM vs traditional pastes. Right now, I'll consider it only for higher performance desktops that I intend to overclock hard, but not below ambient. I only use it between die and IHS. 

 

On re-application, I've not had to do it on any of mine, yet. Looking at my order history, I first bought Conductonaut in June 2017, so that suggests my oldest application is now 3 years good. I never used up my tube, but I did also home brew a batch which I also use. Can't say I noticed any difference between mine and Conductonaut. I now have more than a lifetime supply of the stuff. So much cheaper to make per volume, but it doesn't scale to tiny amounts.

 

I have to wonder if it played a part in the death of my previous gaming laptop. Temps dropped greatly although it didn't alter performance much, presumably that was capped. One day, it wouldn't boot and I never found a fix for it. I don't know if LM could have migrated where it shouldn't.

 

Maybe I'm getting old and/or lazy, but I've not used it on all the systems I can. If they reach some level of "good enough" why bother? With both Intel and AMD moving to solder on their (at least higher end) CPUs, there's less need for it.

 

1 hour ago, Bombastinator said:

My understanding is liquid metal is indium and gallium.  Indium is basically good, but gallium causes all kinds of problems. It’s corrosive for one thing.   Indium alone can be used as a TIM but requires really high pressures to make it work.

I'm not sure if the commercial offerings like Conductonaut ever gave their exact composition but their website says it uses indium, gallium and tin. That's the same mix I used, where I approximated the gallinstan ratio on wikipedia. Indium I believe is the "solder" used by Intel and AMD. Pressure isn't the problem, it's that to apply it, it's melting point is at 156C. So you have to manage the thermal ramping to not break anything as it heats and cools. When mixed with gallium, you can get the melting point below 20C which makes it easy to work with. Tin also does something, but only a tiny amount is used compared to gallium.

[Falkentyne]

21 minutes ago, porina said:

I'm half and half on LM vs traditional pastes. Right now, I'll consider it only for higher performance desktops that I intend to overclock hard, but not below ambient. I only use it between die and IHS. 

 

On re-application, I've not had to do it on any of mine, yet. Looking at my order history, I first bought Conductonaut in June 2017, so that suggests my oldest application is now 3 years good. I never used up my tube, but I did also home brew a batch which I also use. Can't say I noticed any difference between mine and Conductonaut. I now have more than a lifetime supply of the stuff. So much cheaper to make per volume, but it doesn't scale to tiny amounts.

 

I have to wonder if it played a part in the death of my previous gaming laptop. Temps dropped greatly although it didn't alter performance much, presumably that was capped. One day, it wouldn't boot and I never found a fix for it. I don't know if LM could have migrated where it shouldn't.

 

Maybe I'm getting old and/or lazy, but I've not used it on all the systems I can. If they reach some level of "good enough" why bother? With both Intel and AMD moving to solder on their (at least higher end) CPUs, there's less need for it.

 

I'm not sure if the commercial offerings like Conductonaut ever gave their exact composition but their website says it uses indium, gallium and tin. That's the same mix I used, where I approximated the gallinstan ratio on wikipedia. Indium I believe is the "solder" used by Intel and AMD. Pressure isn't the problem, it's that to apply it, it's melting point is at 156C. So you have to manage the thermal ramping to not break anything as it heats and cools. When mixed with gallium, you can get the melting point below 20C which makes it easy to work with. Tin also does something, but only a tiny amount is used compared to gallium.

This user (from where I learned to make my own galinstan) only saw a 2C difference between conductonaut and self-made galinstan, which he made.  I have the same results he did. 

https://imgur.com/a/T6JlP

 

The grizzly website simply mentions "increased indium content."  But anyone who majors in metallurgy or chemistry knows that their w/mk numbers are baloney.  Any time you have an alloy of two metals, the w/mk is always going to be lower than the lowest metal mixed.

 

One example :

 

https://www.electronics-cooling.com/2006/08/thermal-conductivity-of-solders/

 

Quote

For example, the thermal conductivity of AuSn (80/20) solder is 57 W/mK which is lower than the conductivity of either of the parent metals of gold (315 W/mK) or tin (66 W/mK)

 

[Bombastinator]

38 minutes ago, porina said:

I'm half and half on LM vs traditional pastes. Right now, I'll consider it only for higher performance desktops that I intend to overclock hard, but not below ambient. I only use it between die and IHS. 

 

On re-application, I've not had to do it on any of mine, yet. Looking at my order history, I first bought Conductonaut in June 2017, so that suggests my oldest application is now 3 years good. I never used up my tube, but I did also home brew a batch which I also use. Can't say I noticed any difference between mine and Conductonaut. I now have more than a lifetime supply of the stuff. So much cheaper to make per volume, but it doesn't scale to tiny amounts.

 

I have to wonder if it played a part in the death of my previous gaming laptop. Temps dropped greatly although it didn't alter performance much, presumably that was capped. One day, it wouldn't boot and I never found a fix for it. I don't know if LM could have migrated where it shouldn't.

 

Maybe I'm getting old and/or lazy, but I've not used it on all the systems I can. If they reach some level of "good enough" why bother? With both Intel and AMD moving to solder on their (at least higher end) CPUs, there's less need for it.

 

I'm not sure if the commercial offerings like Conductonaut ever gave their exact composition but their website says it uses indium, gallium and tin. That's the same mix I used, where I approximated the gallinstan ratio on wikipedia. Indium I believe is the "solder" used by Intel and AMD. Pressure isn't the problem, it's that to apply it, it's melting point is at 156C. So you have to manage the thermal ramping to not break anything as it heats and cools. When mixed with gallium, you can get the melting point below 20C which makes it easy to work with. Tin also does something, but only a tiny amount is used compared to gallium.

I understand it can also be done cold but requires high pressure.  More than is good.

[Mick Naughty]

Like most things that have been done a million times successfully, if there’s an issue it’s prolly user error. 
 

No temp issues initially or over a year so far. Don’t do it if if that’s horrible. Replace the bad tim and get a better cooling solution. 

[vong]

My LM application on my 7700K at 4.9 GHz has been doing well and I haven't reapplied in like 2-3 years, maybe I should though

[Guest]

On 6/19/2020 at 2:02 PM, Falkentyne said:

  But anyone who majors in metallurgy or chemistry knows that their w/mk numbers are baloney. 

 

 

This right here. 

 

Never understood the de-lid replacing actual solder with LM. It would only be useful in situations that do not include an IHS plate.

[Falkentyne]

10 minutes ago, ShrimpBrime said:

This right here. 

 

Never understood the de-lid replacing actual solder with LM. It would only be useful in situations that do not include an IHS plate.

The solder is thick and the die is also thick, so replacing solder with LM and shaving the die down a few hundred microns (and the edges of the IHS to match it, if you are relidding) can easily drop temps as much as 13-15C, which is quite substantial, with an actual relid of that shaved down (edges only) IHS.  (That's exactly what I did on my 9900k--200 micron die shave, 300 micron IHS edges shave, to compensate for both the solder z-height removal and lower die height).

 

Delidding a 10900k is completely pointless unless you're going full direct die.

[RejZoR]

3 hours ago, Mick Naughty said:

Like most things that have been done a million times successfully, if there’s an issue it’s prolly user error. 
 

No temp issues initially or over a year so far. Don’t do it if if that’s horrible. Replace the bad tim and get a better cooling solution. 

There wasn't any user error. And the results were identical whether I smeared LM just on die or on die and heatsink (temperature and staining wise). Initially, results were good and then it just out of the blue stopped functioning (I gave LM another shot after watching Der8auer nickel platting the heatsink). Only reason I even spotted it was I fired up game P.A.M.E.L.A. and it turned my PC into furnace, fans spinning like insane (because they were controlled by temperature that wasn't getting removed) and it kept locking up system. Which was odd given it's an indie game and I never had problems with AAA games. After I didn't have stability issues for years. Then I did Cinebench test and CPU was hitting 87°C and more, something I never had even with normal thermal paste. There wasn't any user error, LM just plain sucks ass. Maybe it works if both cooler and CPU are nickel platted in some highly specific way, but given how my 5820K IHS looked like after LM application, I wouldn't even bet on that. It was all smudged from LM and no wiping or amount of alcohol removed it. I had to literally sand away a layer of IHS to remove the LM garbage. And same was on pure copper on heatsink side. Something I was expecting because of LM diffusion into copper, but didn't for the IHS. Coz everyone kept saying how nickel platting prevents that. I guess 5820K doesn't have nickel platted IHS then. I'm not gonna bother with LM anymore. Gave it a shot twice and it fucked up everything. I'm gonna stick with highest end thermal pastes. Don't stain shit, have amazing temperatures and easy to clean any time without having anything ruined.

2 hours ago, ShrimpBrime said:

This right here. 

 

Never understood the de-lid replacing actual solder with LM. It would only be useful in situations that do not include an IHS plate.

LM under IHS only made sense as a replacement for regular white thermal paste Intel smeared on some series of CPU's. Replacing solder with LM is stupid, you're just changing medium that has almost the same characteristics. It did made sense replacing cheap generic thermal paste they came with. People had great results there even just replacing it with Kryonaut or anything higher end for that matter.

[Guest]

12 hours ago, Falkentyne said:

The solder is thick and the die is also thick, so replacing solder with LM and shaving the die down a few hundred microns (and the edges of the IHS to match it, if you are relidding) can easily drop temps as much as 13-15C, which is quite substantial, with an actual relid of that shaved down (edges only) IHS.  (That's exactly what I did on my 9900k--200 micron die shave, 300 micron IHS edges shave, to compensate for both the solder z-height removal and lower die height).

 

Delidding a 10900k is completely pointless unless you're going full direct die.

No, I mean yes..... Ok You did more than simply replace the solder with LM which is beyond where I was headed with that.

 

Can also get into IHS replacement with larger and thicker or smaller and thiner. 

 

But 13c drop is impressive. Nice job with that!

 

Does the cpu socket clamp still hold pretty tight with 500 micron of material missing??

 

[Guest]

9 hours ago, RejZoR said:

There wasn't any user error. And the results were identical whether I smeared LM just on die or on die and heatsink (temperature and staining wise). Initially, results were good and then it just out of the blue stopped functioning (I gave LM another shot after watching Der8auer nickel platting the heatsink). Only reason I even spotted it was I fired up game P.A.M.E.L.A. and it turned my PC into furnace, fans spinning like insane (because they were controlled by temperature that wasn't getting removed) and it kept locking up system. Which was odd given it's an indie game and I never had problems with AAA games. After I didn't have stability issues for years. Then I did Cinebench test and CPU was hitting 87°C and more, something I never had even with normal thermal paste. There wasn't any user error, LM just plain sucks ass. Maybe it works if both cooler and CPU are nickel platted in some highly specific way, but given how my 5820K IHS looked like after LM application, I wouldn't even bet on that. It was all smudged from LM and no wiping or amount of alcohol removed it. I had to literally sand away a layer of IHS to remove the LM garbage. And same was on pure copper on heatsink side. Something I was expecting because of LM diffusion into copper, but didn't for the IHS. Coz everyone kept saying how nickel platting prevents that. I guess 5820K doesn't have nickel platted IHS then. I'm not gonna bother with LM anymore. Gave it a shot twice and it fucked up everything. I'm gonna stick with highest end thermal pastes. Don't stain shit, have amazing temperatures and easy to clean any time without having anything ruined.

LM under IHS only made sense as a replacement for regular white thermal paste Intel smeared on some series of CPU's. Replacing solder with LM is stupid, you're just changing medium that has almost the same characteristics. It did made sense replacing cheap generic thermal paste they came with. People had great results there even just replacing it with Kryonaut or anything higher end for that matter.

I did my 8700K replaced the cheap TIM with IC Diamond. Typically I like diamond nano pastes. My favorite actually Antec Formula 7. 

 

But LM seems very successful for some people. Myself I dont use it. I can be a little more careless tthat way!

[RejZoR]

Shaving off die is entirely unrealistic thing to do at home. Or with advanced machinery. It's something Intel could do with 10th gen at their factory.

[Bombastinator]

1 minute ago, RejZoR said:

Shaving off die is entirely unrealistic thing to do at home. Or with advanced machinery. It's something Intel could do with 10th gen at their factory.

Might be a specificity of language issue where shaving and sanding (lapping?) might mean the same thing.  Not sure in this case.  

[RejZoR]

Just now, Bombastinator said:

Might be a specificity of language issue where shaving and sanding (lapping?) might mean the same thing.  Not sure in this case.  

Till you shave/sand/lap away core's internal connections... because you don't know where they are positioned within the die.