hahaha! it *may* help, but it will not make a miracle if the silicon or the heatsink was bulged. this 7950 is just a phase-change low melting point thermal pad with a thermal conductivity of about 8 w/mk, wich is good, but lower than any thermal paste like: thermal grizlly, mx-4, artic... i would try, but just by my own curiosity (i even bought one), but deffinitely wouldnt put my hope on that thing... i suggest to change the test to "watt vs temperature", as the score fluctuate as the software changes. comparing watt vs temperature you have the same constant wich is the silicon itself, so it could be a more suitable test.
liquid metal isnt evaporate. just sipped into copper. you only need to reapply once more and you are set for years a tip for someone wanna try LM. apply LM to heatsink a generous amount and let it there for about 3 days. the gallium will form a layer of GaCu2 on the surface of the copper heatsink. then wip clean and apply LM normally -> put back to the laptop. this layer of GaCu2 will prevent Ga sipping into Cu again. making the performace last nearly forever
Also i noticed that you had better performance in feburary and worse in July... this could be because the ambient temperature increased - if your room is 5 degrees warmer then the CPU and GPU will also run 5 degrees warmer because the delta T between the CPU and the ambient temperature needs to be the same to conduct the same amount of heat energy.
You should probably try nickel plating the copper heatsinks for the laptop cause I heard it helps with the liquid metal from fusing as much to the bare copper on the heatsink
As another commenter said, the liquid metal seeped into the copper heatsink. This is expected behavior when applied to a bare copper heatsink. When this happens, apply more liquid metal and do *not* polish or try to remove what has already mixed into the copper. This second application will hold for much much longer (possibly forever). Nickel plated copper heatsinks don't have this problem and the liquid metal lasts effectively forever on those. If it "evaporated" then every PS5 sold would be an overheating mess.
No you didn't get scammed, gallium which is part of the LM alloy will eventually diffuse into cooper leaving the indium behind which becomes solid; That's why it's better to use it with nickel plated heatsink in order for it to work long term. But even then electroless nickel that is used by most nickel plated PC parts is too thin and again it will diffuse through. I electroplated my 6900xt oem heatsink to 0.1mm nickel plating thickness and 2 years later no issues with "drying". Eventually it will diffuse though this one too. I started using LM more than 10 years ago in my laptop that's why i know all these now. In fact i make my own LM compound, which is much cheaper if you use it often. In fact it will last me more than a lifetime. edit: Regarding to PTM7950 i haven't use it personally yet, but no it will not last forever and it won't perform as good as LM. In your laptop heatsink it will be ~5oC difference is my guess because your heatsink is very "weak" so you are not limited so much by the TIM. What makes PTM7950 so good is that it doesn't pump out like regular paste and is also very good thermally conductive compared to other good pastes.
How do you make your own compound? Do you make it in small batches or do you make it in one big batch and store it? Do you mind sharing your recipe or advised working conditions? Are environmental contaminants like dust or moisture a concern? Sorry for all the questions, but this seems like an interesting project!
What I learned, after you apply it, you must not disassemble it, the temperatures will get shitty even when you clean it as best as you can. Only solution is to buff out the heatsink, so that it's not "corroded". Reapplied LM with buffing the heatsink last year, and the temperatures are basically the same.
Hey thanks for sharing your experiment, we can do some simple math and find that it really shouldnt matter what kind of thermal paste or liquid metal you apply as long as you apply some.... thats because the thermal paste is so thin and conductive enough, that the temperature difference across the paste is never more than ~2 degree °C even at max load when youre conducting 200Watts. So even if you welded the cold plate of the heatsink together with the IHS of the cpu you would only see a temperature improvement of 2 degree °C at best. You know the unit for heat conductivity "Watts per meter Kelvin" or W/mK...? What it means is if you have a cube of 1x1x1 Meter and a heatconductivity of 1 W/mK and a 1 Kelvin temperature difference across two opposing faces of the cube there will be 1 Watt of energy flowing through the cube from one face to the other. And that energy flow is liniearly proportional to Area, thickness, Heatconductivity and Temperature difference in Kelvin. (Kelvin is basicly the same as °C). So Really the unit is Watts per square meter area divided by meter thickness times Kelvin. - the Square meter and meter cancel out to leave only meter which can be confusing... Knowing this we can do a simple calculation for the heatpaste. Heatconductivity = ~ 8 Watt/mK Thickness = ~ 0.0001 Meter Area = ~ 0.001 Square Meter (10 square cm) Kelvin = ~ 2 degrees Watts = ( 8 W/mk * 0.001m^2 * 2K ) / 0.0001m = 160 Watts So even at only 2 degree added temperature across the thermal paste you will already conduct 160 Watts which is way more than a laptop needs - so realisticly the temperature across the thermal paste will only be 1 degree... What this means is that you cant improve temperatures of the computer much using different thermal paste... only when its dried and bad will it make sense to change out. Undervolting might help you alot more... if you undervolt your CPU and GPU just 10% you will see a delta T decrease of 10% also... so if youre running at 90 degree and 20 degree ambient with a 10% undervolt you will run at 83 degree... or 7 degree less.
This is a very interesting video. Thank you for posting it. I've experienced the exact same problem as you described. I've been using liquid metal with an Intel i7-8700k to help improve cooling for overclocking. After applying liquid metal for the first time in January 2023, the CPU temperature dropped significantly by about 15 degrees Celsius, and I was able to easily and safely overclock the CPU to 5 GHz. After about 6 months or so, I noticed a slight increase in the CPU temperature. It continued to get worse, to the point where I had to lower the CPU clock speed. After a couple more months, around October/November time, I had to lower the clock speed again to 4.7 GHz. At that point, I decided to delid the CPU, and I found that the liquid metal had completely dried out. I cleaned all of the old liquid metal from the CPU die and the heat spreader and then re-applied the new liquid metal. Cooling performance was restored, and I was again able to increase CPU speed to 5 GHz. I have since upgraded to an AMD 7800X3D, so I'm not using the i7-8700k as much these these days. However, I suspect I would be getting close to the point where I would probably need to consider reapplying the liquid metal again. From my experience liquid metal is a short term solution that needs to be repeated every few months. I haven't experienced any long term damage caused by the liquid metal. However, my heat spreader is nickel plated. As far as I remember, I think copper does get damaged to some extent by liquid metal. If you replace the heat spreader with a copper version, I think Nickel plating is the way to go. If the heat spreader is plated everything should be fine. However, the liquid metal will still dry out and you will need to periodically clean the CPU and heat spreader and reapply fresh liquid metal. The same is true for most types of thermal paste, it's just that the liquid metal dries out a lot faster.
If applying liquid metal to your cpu or gpu droppes temperatures that significantly, then that must be because there is something wrong with the mounting of your heatsink and the gap between heatsink and gpu is too large. With proper mounting there would only be 0.1mm of heatpaste between IHS and heatsink and the Temperature difference across the heatpaste would only be like 1 degree even at 100W of power.
this is uneducated comment. the temp rised not because of liquid metal are bad long term. but because gallium reacts with copper. just apply one more time without sanding the copper gallium alloy on the heatsink, liquid metal will last forever with performce as day 1
Try ptm7950 it make good contact, will not damage your device it and you don't need to change it out occasionally like paste. I used it on my laptop and it make 10C improve from artic mx-4. Hope that helps.
for a quick Benchmark, Thermal Grizzly is absolutely unmatched. But for long term usage, Arctic eventually beat Thermal Grizzly. All of their products be it Thermal Paste, Thermal Pad, Thermal Sheet, Liquid Metal. So, if you're aiming to achieve world record benchmark within 1 day, use Thermal Grizzly if you're aiming for long term usage and stability, use Arctic
I think the "1 day" comment is a bit harsh, lol. I've been using liquid metal for quite some time. From my experience, liquid metal degrades a lot quicker than regular thermal paste. However, if applied correctly using the correct materials, it should last at least 6 months. If the liquid metal is replaced at regular intervals (every 6 months or so for a CPU that is heavily used), it will be fine. The main problem seen in this video is the use of a copper heat spreader that has been damaged by the liquid metal. Having said that, if you want a longer-term solution that lasts for several years, regular thermal paste is definitely the way to go. Liquid metal is a high-performance and high-maintenance solution ;).
this is uneducated comment. the temp rised not because of liquid metal are bad long term. but because gallium reacts with copper. just apply one more time without sanding the copper gallium alloy on the heatsink, liquid metal will last forever with performce as day 1
@@asdfsdfsadfasdfsf you haven't watched the video didn't you? At very first day when he applied the Liquid metal, he didn't even "sanding" the copper surface contact at all. He just cleaned the previous Thermal Paste with cloth and alcohol and that's it. And yet look what happened in the video.
@@niezzayt3809 he just said he reapplied. never mention about clean it or not. also the picture of his heatsink clearly shown that there are ununiform color on the contacting area means not enough gallium. if you understand what happend on that reaction. you will just understand what i said. if you dont. just "believe" also he didnt mention about ambient temp. Fer and Jul shouldnt have the same ambient temp.
i've been using liquid metal for nearly a decade now, and in all my applications, all due to reasons others have detailed below, required a 2nd application of LM after about 6 months from the initial application. now i use kryosheets from thermal grizzly.
@@Azeraph yea there for sure is, the kryosheet is what i would recommend. i dont know what kind of thermal application you're using it for, but im finding a good 3-5 C difference between the two types of thermal grizzly pads, i'm cooling a ryzen 9 5950X with an AMD wraith max air cooler so im after ever C i can get :)
PTM7950. I got it. Please stop mentioning it in EVERY DAMN COMMENT!
hahaha!
it *may* help, but it will not make a miracle if the silicon or the heatsink was bulged. this 7950 is just a phase-change low melting point thermal pad with a thermal conductivity of about 8 w/mk, wich is good, but lower than any thermal paste like: thermal grizlly, mx-4, artic...
i would try, but just by my own curiosity (i even bought one), but deffinitely wouldnt put my hope on that thing...
i suggest to change the test to "watt vs temperature", as the score fluctuate as the software changes. comparing watt vs temperature you have the same constant wich is the silicon itself, so it could be a more suitable test.
@@GrulbGL ptm7950 igor lab google this and check their results.
@@GrulbGL it's not a pad, and in practical it shows better results then any thermal paste.
Lol
liquid metal isnt evaporate. just sipped into copper. you only need to reapply once more and you are set for years
a tip for someone wanna try LM. apply LM to heatsink a generous amount and let it there for about 3 days. the gallium will form a layer of GaCu2 on the surface of the copper heatsink. then wip clean and apply LM normally -> put back to the laptop.
this layer of GaCu2 will prevent Ga sipping into Cu again. making the performace last nearly forever
Do you clean all of the LM on the cpu and heatsinks and then reapply?
Also i noticed that you had better performance in feburary and worse in July... this could be because the ambient temperature increased - if your room is 5 degrees warmer then the CPU and GPU will also run 5 degrees warmer because the delta T between the CPU and the ambient temperature needs to be the same to conduct the same amount of heat energy.
You should probably try nickel plating the copper heatsinks for the laptop cause I heard it helps with the liquid metal from fusing as much to the bare copper on the heatsink
Gallium will chew the surface of regular heat sink with no special coating, so yea the problem is what u said
As another commenter said, the liquid metal seeped into the copper heatsink. This is expected behavior when applied to a bare copper heatsink. When this happens, apply more liquid metal and do *not* polish or try to remove what has already mixed into the copper. This second application will hold for much much longer (possibly forever).
Nickel plated copper heatsinks don't have this problem and the liquid metal lasts effectively forever on those.
If it "evaporated" then every PS5 sold would be an overheating mess.
No you didn't get scammed, gallium which is part of the LM alloy will eventually diffuse into cooper leaving the indium behind which becomes solid; That's why it's better to use it with nickel plated heatsink in order for it to work long term. But even then electroless nickel that is used by most nickel plated PC parts is too thin and again it will diffuse through. I electroplated my 6900xt oem heatsink to 0.1mm nickel plating thickness and 2 years later no issues with "drying". Eventually it will diffuse though this one too. I started using LM more than 10 years ago in my laptop that's why i know all these now. In fact i make my own LM compound, which is much cheaper if you use it often. In fact it will last me more than a lifetime. edit: Regarding to PTM7950 i haven't use it personally yet, but no it will not last forever and it won't perform as good as LM. In your laptop heatsink it will be ~5oC difference is my guess because your heatsink is very "weak" so you are not limited so much by the TIM. What makes PTM7950 so good is that it doesn't pump out like regular paste and is also very good thermally conductive compared to other good pastes.
How do you make your own compound? Do you make it in small batches or do you make it in one big batch and store it? Do you mind sharing your recipe or advised working conditions? Are environmental contaminants like dust or moisture a concern?
Sorry for all the questions, but this seems like an interesting project!
What I learned, after you apply it, you must not disassemble it, the temperatures will get shitty even when you clean it as best as you can.
Only solution is to buff out the heatsink, so that it's not "corroded".
Reapplied LM with buffing the heatsink last year, and the temperatures are basically the same.
wdym by buffing the heatsink?
@@xCheatah polishing
Hey thanks for sharing your experiment, we can do some simple math and find that it really shouldnt matter what kind of thermal paste or liquid metal you apply as long as you apply some.... thats because the thermal paste is so thin and conductive enough, that the temperature difference across the paste is never more than ~2 degree °C even at max load when youre conducting 200Watts.
So even if you welded the cold plate of the heatsink together with the IHS of the cpu you would only see a temperature improvement of 2 degree °C at best.
You know the unit for heat conductivity "Watts per meter Kelvin" or W/mK...?
What it means is if you have a cube of 1x1x1 Meter and a heatconductivity of 1 W/mK and a 1 Kelvin temperature difference across two opposing faces of the cube there will be 1 Watt of energy flowing through the cube from one face to the other. And that energy flow is liniearly proportional to Area, thickness, Heatconductivity and Temperature difference in Kelvin. (Kelvin is basicly the same as °C).
So Really the unit is Watts per square meter area divided by meter thickness times Kelvin. - the Square meter and meter cancel out to leave only meter which can be confusing...
Knowing this we can do a simple calculation for the heatpaste.
Heatconductivity = ~ 8 Watt/mK
Thickness = ~ 0.0001 Meter
Area = ~ 0.001 Square Meter (10 square cm)
Kelvin = ~ 2 degrees
Watts = ( 8 W/mk * 0.001m^2 * 2K ) / 0.0001m = 160 Watts
So even at only 2 degree added temperature across the thermal paste you will already conduct 160 Watts which is way more than a laptop needs - so realisticly the temperature across the thermal paste will only be 1 degree...
What this means is that you cant improve temperatures of the computer much using different thermal paste... only when its dried and bad will it make sense to change out.
Undervolting might help you alot more... if you undervolt your CPU and GPU just 10% you will see a delta T decrease of 10% also... so if youre running at 90 degree and 20 degree ambient with a 10% undervolt you will run at 83 degree... or 7 degree less.
This is a very interesting video. Thank you for posting it. I've experienced the exact same problem as you described. I've been using liquid metal with an Intel i7-8700k to help improve cooling for overclocking. After applying liquid metal for the first time in January 2023, the CPU temperature dropped significantly by about 15 degrees Celsius, and I was able to easily and safely overclock the CPU to 5 GHz. After about 6 months or so, I noticed a slight increase in the CPU temperature. It continued to get worse, to the point where I had to lower the CPU clock speed. After a couple more months, around October/November time, I had to lower the clock speed again to 4.7 GHz. At that point, I decided to delid the CPU, and I found that the liquid metal had completely dried out. I cleaned all of the old liquid metal from the CPU die and the heat spreader and then re-applied the new liquid metal. Cooling performance was restored, and I was again able to increase CPU speed to 5 GHz. I have since upgraded to an AMD 7800X3D, so I'm not using the i7-8700k as much these these days. However, I suspect I would be getting close to the point where I would probably need to consider reapplying the liquid metal again. From my experience liquid metal is a short term solution that needs to be repeated every few months. I haven't experienced any long term damage caused by the liquid metal. However, my heat spreader is nickel plated. As far as I remember, I think copper does get damaged to some extent by liquid metal. If you replace the heat spreader with a copper version, I think Nickel plating is the way to go. If the heat spreader is plated everything should be fine. However, the liquid metal will still dry out and you will need to periodically clean the CPU and heat spreader and reapply fresh liquid metal. The same is true for most types of thermal paste, it's just that the liquid metal dries out a lot faster.
If applying liquid metal to your cpu or gpu droppes temperatures that significantly, then that must be because there is something wrong with the mounting of your heatsink and the gap between heatsink and gpu is too large.
With proper mounting there would only be 0.1mm of heatpaste between IHS and heatsink and the Temperature difference across the heatpaste would only be like 1 degree even at 100W of power.
this is uneducated comment. the temp rised not because of liquid metal are bad long term. but because gallium reacts with copper. just apply one more time without sanding the copper gallium alloy on the heatsink, liquid metal will last forever with performce as day 1
Try ptm7950 it make good contact, will not damage your device it and you don't need to change it out occasionally like paste. I used it on my laptop and it make 10C improve from artic mx-4. Hope that helps.
Your problem is that the heatsink "eating" up your liquid metal since it's surface isn't solid. Do another run and you should see better results
for a quick Benchmark, Thermal Grizzly is absolutely unmatched.
But for long term usage, Arctic eventually beat Thermal Grizzly. All of their products be it Thermal Paste, Thermal Pad, Thermal Sheet, Liquid Metal.
So, if you're aiming to achieve world record benchmark within 1 day, use Thermal Grizzly
if you're aiming for long term usage and stability, use Arctic
I think the "1 day" comment is a bit harsh, lol. I've been using liquid metal for quite some time. From my experience, liquid metal degrades a lot quicker than regular thermal paste. However, if applied correctly using the correct materials, it should last at least 6 months. If the liquid metal is replaced at regular intervals (every 6 months or so for a CPU that is heavily used), it will be fine. The main problem seen in this video is the use of a copper heat spreader that has been damaged by the liquid metal. Having said that, if you want a longer-term solution that lasts for several years, regular thermal paste is definitely the way to go. Liquid metal is a high-performance and high-maintenance solution ;).
i used genuine ptm 7950 for over a year and my laptop temps are the same :)
this is uneducated comment. the temp rised not because of liquid metal are bad long term. but because gallium reacts with copper. just apply one more time without sanding the copper gallium alloy on the heatsink, liquid metal will last forever with performce as day 1
@@asdfsdfsadfasdfsf you haven't watched the video didn't you?
At very first day when he applied the Liquid metal, he didn't even "sanding" the copper surface contact at all. He just cleaned the previous Thermal Paste with cloth and alcohol and that's it.
And yet look what happened in the video.
@@niezzayt3809 he just said he reapplied. never mention about clean it or not. also the picture of his heatsink clearly shown that there are ununiform color on the contacting area means not enough gallium.
if you understand what happend on that reaction. you will just understand what i said. if you dont. just "believe"
also he didnt mention about ambient temp. Fer and Jul shouldnt have the same ambient temp.
yo idk if you know this or not but ptm7950 is a good alternative to liquid metal
i've been using liquid metal for nearly a decade now, and in all my applications, all due to reasons others have detailed below, required a 2nd application of LM after about 6 months from the initial application. now i use kryosheets from thermal grizzly.
Whats the difference with the graphene sheets?
@@Azeraph between the graphene sheet and liquid metal? Or thermal grizzlys 2 types of graphene pads? I have all 3.
@@volkswagenmember Yeah, is there a marked difference for temp between the 2?
@@Azeraph yea there for sure is, the kryosheet is what i would recommend. i dont know what kind of thermal application you're using it for, but im finding a good 3-5 C difference between the two types of thermal grizzly pads, i'm cooling a ryzen 9 5950X with an AMD wraith max air cooler so im after ever C i can get :)