CORRECTION: The GPU die contact is silicon, not nickel plated. it was my initial understanding that there was a Nickel layer to interface with the heatsink.
The major problem is not forming a battery, here is is the formation of an intermetallic alloy that is dangerous. Google "purple plague" to understand what I am talking about. The same can be observed with a lot of metals (e.g. copper and tin, mercury and gold, mercury and alumina, gallium and alumina, etc.). I would be more afraid here to have some metal leaking from the surface to the capacitors around the CPU making a short circuit. Most of the capacitors you see around a chip are connected to both VSS and VCC, it can be a major problem if you connect VSS and VCC together. Using an electrical conductor as cooling system without any seal is a stupid and dangerous idea. It is as stupid as using liquid N2 as a coolant on a CPU. Manufacturers know their job, you do not need to add some fancy stuff to a board that have been engineered by a team of hardware/electronic engineers.
Gallium is aluminium or issues because gallium will form a amalgamate with aluminium this amalgamate is very brittle. It also will then react with any moisture and form hydrogen peroxide. So yes there are a few chemical reactions going on depending on the metals, oxidation is a chemical reaction though and all transition metals have oxidation states. As far as removing gallium from the chip, a warm damp rag works fine. :P With nickel I am not sure. I want to say for an amalgamate with nickel to occur you would need to heat the nickle up to around 400c but I am too lazy to look it up so take that with a grain of salt. Also many metals will undergo changes with other metals do to dffusion, aka the Kirkendall effect.
They don't oxidize, they alloy. AlGa alloys are extremely brittle and have much poorer thermal conductivity, which is where the problem comes in. The fact that it happens fast doesn't help things. LM will alloy with nickel and copper, but it does so much slower and the resulting alloy has similar thermal properties as the initial metals. So while you might eventually* have to re-apply, it's not actually hurting the copper or nickel components. *In 1 to 2 years.
Galium forms an amalgamate with aluminium, not so much an alloy, which is why it's so brittle, it seeps into the grain structure of said aluminium and breaks it down almost, it turns it into junk basically.
Yes, other than the termination resistors and capacitors around the die being shorted, Liquid Metal compounds are more effective on bare dies anyways, it's super popular for CPUs, not so much on GPUs.
I did apply Conductonaut on a Gigabyte GTX1080 with a 3-fan open-air cooler, I lost 15°C in almost every scenario (OCCT/gaming) while the fans got slower by a fair margin. It made the cooler so efficient and silent that since I build this machine I don't recommend to switch to an Accelero/Prolimatech so often (it depends on the card you have, of course). The open-air coolers got really good at their job in the past years, and with only a little boost from liquid metal compound they are definitely killing after-market coolers
I need to address something that was not mentioned in the video yet IS mentioned in the liquid metal MANUAL... Your GPU die shouldn't have looked smooth at ALL after cleaning it. I've used Coollaboratory liquid ultra on my own rig for years with a delid'd 4770k.. and among one of the first instructions is to use the abrasive pad inside the kit, which is what helps the liquid metal actually hold and create a true bond between both surfaces. Failure to do so will result in little to any gains. You've proven A) You didn't read the manual & B) You quite literally applied it yet again without having prepped the die (using the abrasive pad lightly as directed).... There's an old saying, if you do things right the first time- well- you wouldn't need to go back and do it again. You've done it twice, a true achievement I'll say /s. Please upvote this for those that are misinformed by the video.
Yup what people should really research what gallium does to other metals before preaching how great Liquid Metal is. The stuff ends up everywhere and is more trouble then it’s worth.
This needs to be revisited after 12 months. That discolouration on the GPU surface isn't great, but I think we'd need a look under a microscope to make any determinations about the liquid metal etching the surface.
heard many people saying that after about a year the liquid metal just hardens and then the thermal conductivity is just bad they had it on a cpu die but I think it would be the same with gpu die, even if it hardens you can remove carefully it wont kill you gpu die but it wont conduct heat well after some time.
I have been using CLP on my 4670K die for 4 years at this point and it is going strong ever since. The temps are the same as the day it was delided and reassembled, and hasn't been taken apart ever since. I cut 20C+ of my load temperatures, and i have used a Hyper 212+, Silverarrow IB-E, Phobya UC-2LT water block and now a XSPC Raystorm water block with the same results to a few degrees. So there is no need to be worried if it will corrode, as the Galium in the compound will not bind or alloy with the nickle and copper in the IHS or cold plate of the water block. On the other hand, using it with a aluminium heatsink is a big no no, as it will eat it away and the heatsink will crumble to dust.
magottyk I would argue that what was left could be removed with an eraser or a fingernail or further cleaning with stronger iso, what was left is a pretty typical sight in the tons of applications i have used it on and has been able to removed on every single die it certainly looks like it has sandblasted the die or etched it but it always comes off. There is way to many variables here, but at a minimum removing all of will take 98% percent isopropyl alcohol or stronger and i can almost guarantee that little presoaked pad wasn't even close, hit that die with some qtips and some strong iso and it will come up shining
What you are seeing on the gpu/heatsink is most likely fretting, caused by the two metals griding against each other on a microscopic scale due to thermal expansion and retraction from it heating up/cooling down with normal use.
I delided my Intel Core i7 4790k cpu (4.4ghz) & installed a NZXT Kraken g10 braket + corsair h75 AIO liquid cooler to my 4GB Nvidia GTX 980. Both the cpu die(Under IHS) & the GTX 980 bare die have been applied with Cool labs liquid metal pro. I Noticed a 15 degree drop on both CPU & GPU Temps. Moral of the story from 2014 too 2017 now I have noticed Zero problems with gpu tempature/ no corrosion & Liquid metal pro paste has NOT dried up. I Highly recommend liquid metal pro- just take ur time with application, also like in this video I recommend apply LM to gpu die & gpu heatsink!
I just applied an evga hybrid cooler to a 1080ti using liquid metal and can certainly say that that temp decrease was jaw dropping. I also applied liquid metal to my Cpu (5930k) which under an aida64 stress test would hit 90c, now it sits at around 75-80c. Liquid metal is truly an amazing compound but certainly one I wouldn't recommend to those who want to retain their warranty
I put liquid pro on my 2x MSI 980 Ti Gaming 6G cards when they launched, ran it for about 8 months with no issues at all. Had those cards for 3yrs before I sold them and they ran 24/7, no issues at all. First thing I do when getting any new GPU is to replace the paste, you don't know what crap they may have used and whether they applied to much or too little.
are you really this stupid? it is literally in nvidia and the actual card manufacturers best interest to use a product that meets the requirement of their product , otherwise they will have customer issues and go out of business, what you mean to say is that these things are needed if you are planning to run the product out of spec. No company is going to be like "' we sell top tier products in the performance market, where people upgrade every 1-2 years, lets save 50 cents on each device and risk losing millions in future sales"
@readyrepairs - Intel CPUs that see 15C drops after replacing the toothpaste they ship with with a REAL heat transfer material, if you want an example about how wrong you are
Gallium creates an amalgam with aluminium similar to mercury. Gallium and copper will slowly create a layer of gacu2 over the copper surface but it shouldn't be harmful and it should subside once the layer is between the two metals. Nickel coatinv is recommended to prevent this.
0:27 this is technically wrong Gallium (element in the liquid metal) forms an amalgamate with aluminum it really infuses itsself in the heatsink and weakens the metal to the point where it crumples off in chunks
This leaves me with a question for you. Would you do this to a future graphics card or is this the one and only card you’ll be applying liquid metal to?
It's safe if the cooler is nickel plated and the SMDs on the chip are conformal coated. With copper, the liquid metal reacts and makes an alloy. It does not destroy the cooler as it would with aluminium. You will have to reapply the liquid metal after a few months, since it gets sucked into the copper.
I had Liquid metal on my CPU for 3+years and i removed the heatsink for the first time last week. The temperature was absolutely the same as 3 years ago (60C under load). After removing it i observed the following: - The base of my cooler is a mirror finish, nickel plated and no liquid metal was stuck to it. - The cap on my cpu is lapped, so it is bare copper and the liquid metal is firmly stuck to it. Also the liquid metal is not a liquid anymore, but it also has a mirror-like finnish. It cannot be removed from the cpu with solvents, so i guess i have to sand it down. I think it's a great solution for bringing your temps 1-2c for extreme overclocking, but not really useful for moderate overclocking. DO NOT USE LIQUID METAL ON ALUMINIUM! Use it only on copper and nickel!
in my experience, if you don't apply Liquid metal on both sides, over time the compound seems to migrate to section of the die after 3 months or so ( verified on the imbalance of temps on all cores from one side to the other on my laptop and desktop) . If you apply it on both the die and heat sink it seems to stick better, probably due to surface tension. Since the compound is a bit difficult to apply to begin with, you can imagine upon installation it can make contact but could resist and not fully adhere and migrate leaving less compound on the heatsink side if not preapplied. using a very thin amount on both sides seems to be the best method for me and it's been almost a year installed on 2 cpus, 5 graphics cards, for both my sager laptop and 3d rendering workstation (1200 watts load) liquid metal is serious help for high wattage users, brought my temps down on average by 15 degrees on full load from IC Diamond. By the way, in my experience you don't have to remove the compound if you disassemble, you only need to even it out and add a bit more if needed, and then reinstall. When i need to completely remove it i use metal polish on both sides, works like a charm and totally removes corrosion on the copper.
If you ever put liquid metal on aluminium you will be met with an interesting situation of liquid metal bubbling, and actually dissolving aluminium in front of you, if you try to wipe it off with rubbing alcohol or water after it's been sitting on the aluminium, it will actually reactivate the liquid metal dissolving aluminium, essentially locking you in catch-22 cycle of not being able to remove liquid metal. I tried that on what I thought was copper (gpu heatsink), but it was aluminium, anyway, it ate through it like water through sugar cube, and the more I wiped, the more it dissolved the gpu cooler.
Shorting-out the capacitors around the GPU-die is a big risk. I recommend covering the capacitors with tape to prevent this. Even when applying very cleanly there is still a risk that some liquid metal drips down when the cooler is mounted...
Nearly 20 degree C drop from changing my GTX1080ti fans to Noctua 1x120mm and 140mm fan, my GPU now sits around 60c. It is a really, really good mod, i just use Corsair link to now control my GPU fans, works a treat, no need to liqud cool GPU's now. If I put the fans onto 3000rpm (too loud) I can get around 52c temps. Forget buying aftermarket heat sinks etc, juts upgrade your fans.
I wouldn't be comfortable leaving those little capacitor thingies next to the die without protection, like, at least apply nail polish onto them just to be safe. Especially on such an expensive graphics card.
GAnime88 Yeah, regular nail polish should do the job. You can also put isolirband (I have no idea whats the english word for that) over the nail polish. But you need to be careful that it doesn't restrain the contact between cooper and CPU/GPU plate because then your laptop will overheat even more. I just did this an hour ago, 12 degrees celsius drop on both CPU and GPU and no worries about liquid metal leaking and fucking up everything.
You are correct. Use nail polish (Acetone cleans the LM, and nailpolish should you want to clean it up after, and is PCB safe, but careful, those SMCs can get knocked off easily, i use a paint brush to clean the SMC areas, dipping it in acetone, brushing, then dabbing it away slow and sure) OR use Kapton tape, a heat resistant tape you have likely seen in electronics before, and tape over the SMC components to keep them safe,
Super late to a zombie comment section, but I put liquid metal and new arctic thermal pads for the memory modules on my gigabyte aorus waterforce 1080ti extreme. It's already a watercooled card so it runs really cool, but figured I'd do it anyway because why not. It didn't drop much, but it went from high of mid 50s to never touching 50c looping heaven for 30 minutes. It was a small drop, and I probably could have reached the same overclocked without it (managing 2087mhz core), but it's a good experience. I delided and put it directly on the die of my i9 7940x, and even with a very mild boost of 4.2ghz on all 14 cores and only bumping vcore to 1.2, I'm sitting at 30c idle on water as well. Before, the same clocks got me around low/mid 40s. It's totally worth it, but what I'd recommend is you put electrical tape or clear nail polish over the capacitors around the die on the pcb of the board to insulate them in case you put too much liquid metal on and it runs off the board. This will save you shorting out the cpu or gpu and bricking the system. Totally worth it to me, and makes for great practice too! It's not that hard if you're a computer dork too.
The metals that the liquid metal thermal compound is made of does not "dry". The etching that you are seeing is where you aren't pulling all the compound up. In order to pull it all, it requires more than alcohol. I did the same thing with a mini gtx1060 and it dropped the temps between 25 and 30C depending on the load. At full load now, I run 56C and never thermal throttle with a short video card. I never even have to run the fans more than 10% to get the max oc the card's allowed voltage will do. Liquid metal on a video card with any form of afermarket cooling is going to make a huge difference.
Great video! Quick clarification that gallium does not "oxidise" aluminum. The aluminum literally dissolves into the gallium and forms a very brittle alloy. (This is from memory and I'm not chemistry expert, so please don't quote me!)
Is it possible that your GPU die and your heatsink are expanding and contracting at different rates causing the wearing of the surface of your GPU die? That stain on your heatsink looked suspiciously familiar to the wear pattern on your GPU die :( I would guess the manufacturer uses TIM compounds also for their lubricating properties so as to avoid this type of wafer wearing from expansion and contraction of different materials mated together?
Hmm, it could be possible. I assumed it was due to me reinstalling the heatsink a couple times, sort of when you remount a CPU cooler into a CPU multiple times and you start getting some wear.
Nice follow-up. I saw a good 15C drop on my Titan X Pascal with an EK waterblock just by replacing the EK TIM with liquid ultra and boy was I surprised. I have a 360 Predator setup with those QDC connectors that make install a breeze but have been criticized for limiting flow-rate and driving up temps. Quite possible as my GPU would surpass 65C under sustained load which is really stupid for a liquid loop. Now I get better clocks with less fan speed. Awesome stuff.
The so called drying up is from the gallium in the compound being absorbed into the copper or nickel. They turn into a metal alloy. The roughness you felt is the other metals inside the liquid metal that got left behind.
There is an important difference between Thermal Grizzly Conductonaut and Coollaboratory liquid ultra... CLU, over time, will solidify and it will be extremly hard to remove, Conductonaut is the best way to go as it will not solidify over time.
well, my own experience I can say... I went with CLU for 6 months on a delided 6700k. Both CLU between CPU -> IHS and IHS -> Cooler. After 6 months temperatures were starting to rise, just a bit, 1-2 degrees and I checked the Liquid Metal... I've found out few clots, both on the CPU and on the exterior of the IHS, between it and the cooler. Conductonaut is another story, stayed the same for 3-4 months.
hmm, i wonder if that's what's starting to happen to my 6700k. i noticed my temps are starting to climb a bit. i've been cleaning my filter every couple days thinking that's what's affecting my temps. i used the "gel super glue" on the corners of my IHS, i wonder if it's even worth the risk of switching the TIM out...
I second that! About CLU! I have used Coollaboratory's Liquid ultra on several builds and usually after a year or so I'll have to really pry to get the cpu and heatsink apart (not delidded just using it as TIM between CPU headspreader and heatsink/waterblock) and have always had to use abrasive materials to clean them. I have not had the temperatures climb though despite hardening. But if Thermal Grizzly does not harden then it will definitely be a better option. Cleaning sensitive and expensive components with abrasives and polishing compounds is fiddly work.
The 1070 cooler is garbage. I swapped out the stock 1070 heatsink for a 1080 heatsink, the vapor chamber from the 1080 is substantially better than the copper block on the 1070. I didn't have the courage to try liquid metal though. I'd highly suggest giving a 1080FE cooler a try on the 1070 with liquid metal.
Hey man, I'm new to the channel, and I just want to say... The way you talk is nice to listen to (Trying and failing not to sound creepy, lol) you don't studder or misspeak and there's no awkward pauses or beating around the bush. I've watched a couple videos and they're great so far. I think I found another great tech channel :D
Wow thanks so much man, that means a lot. Out of all of the tasks that go into making a video, speaking has been my biggest battle. I try hard not to mumble or misspeak words. It's not perfect, but I do believe I've improved since my first videos. Thanks for the support!
My 7700K is still going strong several months later after having applied CLU. And my STRIX 1080 Ti hasn't gone over 55°C since the application process. All-in-all, with how frequently I upgrade components; this is worth it.
Nice video. You inspired me!.. So, I just put Arctic Silver 5 between my GPU and heat-sink on an EVGA 1070 (Black edition) and gained a 14 degree Celsius drop in temperature! - Overclocked at 2100Mhz, temps went from 68 down to 54! (Running Heaven at 1080P full screen) ... Seriously, I didn't do anything special. Took maybe half an hour to complete... Totally amazed... Unfortunately I was only able to gain an extra 20Mhz increase in GPU speed, but who cares, I can now have quieter fans and the card will probably last longer... And mainly it was fun and rewarding. :) P.S. I have Liquid Metal on order- Lol
Arctic silver 5 just saved my pride of selling a nearly defect product which i bought as used and could not check the problem when i bought it with a light testing. this is a long useful story about what happened with my workstation laptop. If u want read , do it to the last of the story :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: I bought a used DELL PRECISION 3510 , good performance but the laptop runs quite hot especially GPU until it throttles , I felt like i was deceived while i got the laptop quite cheap but this problem make it problematic to me that even if i resell the laptop hiding this problem will make me feel guilty even if the buyer knew the problem or not , cuz DELL built the laptop like this and it is not old model just 2 years so you get what u pay for . but i did not Give up , I asked myself what if DELL did not used the best method to cool down the laptop ? is it from the fan or tight place state , is it a combo of a bad silicon lottery for both CPU and especially GPU !!! I thought more about problem , people do buy a thermal paste when they build their own PCs , and i read about that a semi "miracle" product which could help their CPUs to cool down better , and the idea of GPUs are simply rly similar to CPUs so we could use the thermal pound to it , but hello !!! Dell precision 3510 is rly hard to get disassembled , and DELL have a good reputation of building and assembling their own products !!! And it is a Precision laptop with a firepro GPU not a cheap consumer series !!! and maybe i can not get the laptop back to its original form after i disassemble the laptop cuz maybe there is a trick which help DELL to be the exclusive one which have assemble and disassemble the laptops but not the laptop owner or a tech savvy !!! Ohhhhh my heaaaaad staap it , screw my time and screw my money , it is my hobby long time ago to know all of what is happening inside the tech products , so lets do it man !!! Googled what was that semi "miracle" product which helped people to cool down their silicon products, found it and also i googled which store got it in stock , went and bought it after i finished my language course "i am a refugee" who got his hobby of stationary desktops in homeland but due to not being settled , i was forced to buy laptops I disassembled the laptop with the help of a youtube video , not ez , finally got to the CPU and GPU , what is the problem ? ah i remembered , got a tissue with the help of my relatively expensive perfume i removed old thermal pastte on both of the CPU and GPU , then it took me 2 times to get the laptop to its original "working" form , in the first time i knew i will do mistakes so i did not 100% tighten back all of the screws , did try to boot and check the laptop , several issues including wifi card and keyboard , did check again were i missed , fixed my mistakes then i tightened all o the screws and turned the laptop on , good it is working , lets test if the problem still exists , turned on the game , played for continuous hours , the laptop is not designed for continuous gaming , still there is heat , i am waiting until it throttles , Hello dude what is your problem , why do you not slow down from 50fps to 15 fps !!! Do i did it right !!! wow I am genius , yet awaiting the chance to use my brain for better things , and the laptop did the job whithout insane temps and throttling , managed to reduce the GPU temps by 5-15 and removing the throttling issue , Now i can keep it to myself or sell it with peace of mind , and actually i did a little bit danerous business !!! I sold the laptop to an another immigrant whom their reputation is a little bit damaged in other european countires "madafacka people leave nothing to simple good people" i thought about not generalizing cuz it is quite stupid and extremely dangerous. I did the business , he did not give me cash nor he used the most known way to transfer the money but he used the traditional method of transferring the money , a bank transfer using application in mobile phone , we have different accounts belong to two different banks which it takes 1 working day to get transferred , was a little bit hesitate , he could give me cash but refused at first cuz cash is harder and more time consuming to put it again in my bank account and i can not say if the cash notes are fake or not , So he transferred the money , he showed me everything and told me to wait a 1 working day ,and gave me a pic of his id card , I did w8 till the next day , and checked the bank account , wow i got the money !!!! HE DID NOT DECEIVE ME !!! That is the story behind acrtic silver 5. took me half hour to write , thanks for reading.
Your card was probably performing better since it got better heat dissipation so it was running faster which could be why you only dropped 3 degrees Celsius
The fade on the GPU is mostly residue. You could etch it away with a 5% hydrochloric acid , but you have to be really careful. Especially nice when you applied LM on IHS on CPU and the CPU died and you have to RMA it (No SN readable) after that treatment you can read it again. Performance wise it doesn't matter at all.
The heatspreader on a CPU is actually nickel-plated copper. They warn you not to use Aluminum coldplate heatsinks due to pretty fast corrosion that will lead to complete structural failure in days. Gallium-based TIM on a CPU will do very little harm, it's everything else that may get damaged.
Seriously who is saying liquid metal dries out over time? It's a metal. It's liquid because that's it's natural state as long as it is withing certain temperature range. It's not going to dry out just as water isn't going to dry out...
Very limited sources on this stuff, while the die package itself is Silicon, it was my initial understanding that the top layer of the contact area was nickel plated.
xDevs_com depends on the manufacturing - sometimes the die is coated with multiple metals, especially if a soldered IHS is in use, since you cant solder silicon to copper Usually the copper is coated with nickel and the silicon is coated with titanium, Vanadium and Nickel Then they are soldered with Indium.
Liquid metal reacts to copper, while not a significant reaction they do react and long story short one is positively charged while the other is negatively charged causing them to naturally pull each other and since it's easier for liquid metal to move over to the copper it does. The TL;DR is they attract each other and both conveniently cancel out the others charge completely (well symbolically) the gallium in liquid metal will cancel out the coppers negative charge and create a neutral charged zone making it appear as if the liquid metal has dried out (as the gallium a liquid metal will separate from the rest of the liquid metal and other additives in it) and for the application this is a good thing as it will prevent further "drying out" that neutrally charged zone breaks the pull copper has on the gallium in liquid metal. If you remove it ... it will happen again .. if you leave it than it will prevent it ONCE the neutrally charged zone has been sufficiently built up.
Gallium isn't going to evaporate, the vapour pressure is stupidly low and the difference between the melting and boiling point very large compared to most metals.
By the way, if by any chance, you'll open the thing again and wish to clean the metal off, I suggest you do this one thing: Blow it with a hairdryer! Liquid metal may have a low melting point compared to most other metals, but it is still quite high. Gallium itself melts a little above room temperature, but not at room temperature as others may say. Also, due to the nature of alcohol, using alcohol to clean the metal will only encourage it to freeze, as alcohol absorbs the heat and evaporates quickly. I think using dry cloth is the most optimal way to clean off the metal (along with the use of a hairdryer). That way, it will be less likely that the thermal compound will leave a mark on the CPU.
I used Liquid Metal, specifically CoolLabratory Liquid Ultra, on my AMD FX 8350 with a Full Copper Air Cooler and I am here to say after a year(ish) it about ruined the top of my CPU. It took several hours to clean and get back to something that did not look fried to death. The Copper Air Cooler was fine, typical 3 min clean up job, but the AMD IHS was wrecked. After cleaning it off I used regular TIM and nothing seemed to be wrong with it. The IHS was not pitted or anything so I am not sure if the Liquid Ultra was baked on or something. I did not overclock the CPU or anything and it did not run much cooler than before so needless to say I did not ever give CoolLabratory a second chance... Cheers...
it's probably thermal contact resistance - thermal resistance between two solids, not thermal conductivity in given solid. It's hard to overcome that but getting rid of oxides on solids should help, because it will help the liquid metal to wet the surfaces, the surfaces sould be flat and care should be taken while applying liquid metal.
In terms of Putting LM on copper will cause it to galvanically react with the copper and gallium will leach from the LM to the copper effecting performance of the TIM. it will leave a metal coloured spot and does need to be reapplied a few times after the first application on fresh copper as it DOES dry out. After one or two applications the galvanic process has ran its course on the effected area leaving the metal colouring. I can confirm this, having applied it only once to the copper HS for over 1 years and the liquid metal was hard and brittle, and not in any way like how it was applied fresh. This experience was with cool laboratories liquid ultra.
How did you account for GPU boost in your testing methodology? How do you know that your GPU wasn't boosting higher because of the increased thermal headroom?
Liquid metal has Galium in it and after a while it does dry up, it doesn’t oxidize the metal it destroys them over time slowly look up how gallium affects metals and you find out why you can’t use it on aluminum
I don't have a Computer Engineering Degree, but Mechanical Engineering is close enough. The only way to provide more heat distribution is the bring the atoms in contact with each other. This, in essence, does make the molecules interact with each other, causing them to possibly oxidize under intense heat. We could see the thermal breakdown happen for sure at temperatures much higher than 80C. I do think that liquid metal is a dumb idea though. For serious cooling, the best option would be evaporative cooling. I think Ethanol boils at 78C, which if used instead of water, would cap your max heat at 80C. Everything I would've thought of going into a good cooling system has already been installed. Ethanol might already be used in some cooling systems. What I'm seeing here is legit "miracle" cream. Anybody who says otherwise either doesn't know what they're talking about, or doesn't have a degree in Engineering. If you're serious, you should try an evaporative methanol system. The Heat sink appears to have liquid in it. If you could calculate the volume, and possibly the stress of the metal, you could, in theory, get your cooling to 70 degrees max. You would need the exact amount of methanol, enough so that the vapor pressure doesn't blow up and cause a devastating fire. Certainly a huge risk, but definetly high reward when looking for major decreases of 12% in heat.
This could be even better with ammonia based evaporative cooling. I can't understand why no one have made a closed cooling system with ammonia. It's in use for refrigerators in camping vans and is completely self sustained and close to soundless. Using this for a CPU cooler could be expensive, but would be the ultimate silent solution.
If I had a heatpipe that contained 80C evaporating material and attached it to some fins and shoved it on a hot chip it would not magically cap itself to 80C. For someone with a mechanical engineering degree you sound like a complete fucking idiot. What do you want him to do, pour methanol over the chip? Go back to school moron.
True, but it wasn't the point of this. Now we know that for blower cards, no matter the thermal conductivity at that point, it's the whole rad that is a limiting factor! Also, if you are being more agressive on the fan curve (such as me and my 980Ti) you can drop temps to the mid 60's without touching anything else.
the faded surface is because of your cleaning method alcohol does almost nothing to liquid metal it would be the same if you tried cleaning it with water or with nothing at all.. use a mild acid like white vinegar the next time.
applying to both surfaces is theoretically supposed to get around the high surface tension of the Liquid gallium, or whatever metal it is. similar to when you are gluing a surface to another surface, it is a good idea to apply glue to Both surfaces, to assist in adhesion or use an abrasive to scar one or both surfaces , but with a GPU diode, and a Heat-sink cold plate, you cant use an abrasive to scar the surfaces, as that would damage either surface, so applying the LM to both surfaces is a way around this.
Why does anyone expect someting not so good to happen when using Liquid metal? Liquid metal is better than normal thermal compound in any way except that it`s electrically conductive.
LM does interact with copper and dries out during the period of about 9-12 months with hot chips like GTX1070. Copper absorbs liquid gallium in small amounts over time, high temps help this process. The question is what happens if you reapply it over the years, how the situation will be. One point of view is that copper, once absorbed enough gallium becomes plated with it and further every new application will last longer as the direct chip + heatsink contact will become Gallium + Gallium, as opposed to Copper + Gallium.
Liquid metal will "dry up" over time in contact with copper. Gallium will move out of LM and embed itself in copper. Gold plate copper heatsink to prevent that.
A bit of advice - this paste will not oxidize any aluminum, as aluminum is so darn reactive it will oxidize itself on the surface in a few hours after exposing to oxygen, creating a very thin but mostly inactive shell of corundum. So all you aluminum elements are already oxidized, apart from those you've scratched recently. But then again. aluminum oxidizes only on a very thin outside layer, which prevents further rust. Problem with aluminum in this case is that this liquid metal paste, or alloy, can't moisten the aluminum surface too well, as it is oxidized. This paste needs a metalic surface to spread on neatly, and on the oxidized aluminum it will tend to form droplets instead. That's why using it on aluminum plates is not recommended, it's safe but it may actually perform worse than your average thermal paste. You could try to sand such surface a bit before applying this alloy (so it would touch bare aluminum), but considering how expensive things below the cover might be - would you? It's worth trying when you have aluminum heatsink, but you can expect cooling performance to drop as the surface will still oxidize a bit every time some oxygen somehow goes between the surfaces and things will go as mentioned above eventually.
It works fine with copper. The problem with using it on aluminium is that the liquid metal creates an allow with alumiunium, and they have high solubility. The same happens with mercury and indium for example. It will completely penetrate and destroy aluminium, once aluminium is scratched (removing aluminium oxide layer).
The people that say that liquid metal is drying up,its probably only in a solid state,as Liquid metal is mostly gallium,wich destroys Aluminium,and melts at about 25c,Liquid Metal does Not dry up,and there is no need to Re-apply it after you took the cooler off for example for cleaning,hope i helped some of you.
I use Liquid metal on both of my 2080ti's both watercooled and it greatly improved the temps the top card now sits at around 35-47C° instead of 70-80C° with an light oc (measuring under full load in gears of war 4 benchmark and heaven benchmark)
The Aluminium problem comes from the Gallium content of the liquid metal. It diffuses into the Aluminium and weakens the crystal bonds of the metal, resulting in complete destruction. There are plenty of videos on this.
Good video! I used it 5 years ago for the first i7 EVGA Classified full liquid build that was notorious for high heats. Applied it to the processor only, 'correctly' (not brush but plastic wrap)!EDIT! *The brush is better with the new formula*, and saw a 5* drop from artic silver. Overclocked to 4.2 for 5 years with no issues to date. Rig just got retired in perfect condition. Never saw the need to use it on a GPU thats replaced about yearly to bi-yearly.
On polished-metal or GPU-die surfaces, it is absolutely possible to clean without abrasive methods and get residue-free and shiny surfaces again. Use acetone first, then alcohol. Acetone dissolves the residue of that LM completely.
You missed one huge part... GPU Boost 2 will boost up as high as it can go to it's set temp and power limit. By using the liquid metal, you basically made the cooling better, allowing the gpu core to boost higher. So I'm sure you now have higher boost clock compared to how it was at stock. You should have recorded the benchmarks from before and after because THAT is the thing that is going to change, not so much your temps (unless you were using water cooling)
Liquid metal, ie gallium, is not wet because of moisture but is a metal that is liquid at room temperature, just like mercury is liquid at room temperature.
Nearly bricked my 5700xt but using liquid metal. Didnt know about its issue when I bought it and it was causing shorts and red pixels all over the screen indicating a dying card. Clean and put regular paste on and back to normal. And i even blew a resistor on the gpu! That was my fault because it spread when i put the water block on. But either way just stick to thermal paste guys, not worth the risk imo.
I think you are correct that the blower style cooler is the bottleneck, not the thermal interface between the dye and coldplate. LM is probably best reserved for CPU delidding in high efficiency water cooling loops. The staining is probably just permeation of the LM into the pores of the substrate when it is heated, which may actually improve thermal transfer efficiency. BTW, removing (or applying) LM without masking off the dye first is probably not a good idea. At the very least, you may create more cleanup work for yourself, and at worst you may end up with a short circuit.
The only thing thats negative about Liquid Metal is that it REALLY sticks to every nook and cranny. While the Silicon of the Die is pretty flat, the copper sink has grooves that comes from the production. The Liquid Metal really sinks into them, thats what also boost the cooling, but make cleaning it without sanding off the Copper impossible. So far I only heard 2 major things that can happen over time. First the LM could dry out. The Liquid Ultra seems to be more prone to that. But it'll only dry out on the edges if you paste it too thin, like only on the Die and not on the cooler. The second is that LM wanders off.. Thats more a CPU problem if you have it vertical. Funny thing, there seems to be reports that the Liquid Metal defies Gravity in that regard XD since it sometimes wanders upwards. But thats only in VERY Rare cases. As long as you apply it correctly nothing happens over time. There are 3 major rules. 1. Apply thin on DIE and Cooler. 2. Don't apply on Aluminium EVER EVER EVER.... 3. Avoid any spillings, especially on metal contacts... FYI I delidded my i7-7700k and used TG Conductonaut. Even with Stock Fan its now cooler then a Hyper EVO 212 without Delid. Using an AIO it's at 50-55°C on 4.5 GHz and 65-70°C on 5GHz. Without delid the CPU would Overheat (95+°C) when overclocked. (bad Silicon Lottery draw)
Wouldn't slight corrosion help with thermal conduction? Lets say you used the liquid metal compound on an aluminium heat sink. After removing the compound, if you put normal thermal compound on it the surface area of the aluminium heat sink would be greater, theoretically resulting in slightly better thermal conduction.
those liquid metal are an alloy of germanium and something more, which I do not remember, this is quite common alloy and give it enough time to search you will find an information about it German does not oxidize aluminum as you said, but spontaneously create alot with aluminium, which is very brittle and at the same time it removes aluminium oxide layer from aluminium and aluminium is extremely active metal which is so active that it oxides spontaneously on even briefest contact with air and aluminium oxide formed prevents any further oxidation so basically removing the only protection from oxidation you enable aluminium to oxide constantly with air. it is not the other metal which cause it, but air and aluminium itself!
i did this to my evga 1070 ftw2. And i used silicone for temperd glas which you normal use on your own to cover the the area around the gpu die. So this silicone can survive temperatures up 300c. And my gpu temp is 58-59c when playing BF1 on ultra 1080p. I do recommend this!
iv added thermal grizzly conductonaught, and what i have found is that the cooling is very dependent on the bottlenecks of thermal transfer and heat dissapation. the liquid metal takes care of the transfer bottleneck, but is then dependent on the fans ability to dissipate that heat. when added at stock fan curve my card only dropped by 3 degrees celsius as well. but when turning up the fan curve i noticed an additional 7-9 degree drop over the temps with the same fan curve and standard thermal compound. so my bottle neck became the heat dissipation provided by the fans. to get the most out of this compound i recommend any cards with an open shroud dissipation, especially triple fan radiators that minimize the dissipation bottleneck, and not blower style cards.
Hmm, I'm not sure if Nvidia dies are nickel plated, I think these ones are just silicon or possibly doped silicon, I don't feel like taking acid to one to see though, although I have a few dead GPUs from older generations lying around, I might try it on one of those.
The best way to clean those rough surface is the regular thermal paste. Just rub them and it will clean those liquid metal left overs. I learned this yesterday when I was switching from liquid to regular paste. I think it acts like somekind of a polisher.👌
I have seen with laptops that after 3-4 months it need to be redone cause the liquid metal seems to get absorbed in the copper heat sink. After the second time it keeps working fine.
We tried it in our linecards many years ago. After a couple years in the field some of them got oxidation so bad that temp of the chips raised up more than 10 deg C. It was a big mess and we got to scramble to get rid of them and bring back the silicone grease. For gaming it is possibly ok as life of such products is relatively short not 7-10 years required in high end switches and routers.
In this case you don't have to worry about it but -- just in case it hasn't been said already -- probably not the best idea to use liquid metal on a 90deg. mounted GPU (ie motherboard parallel to the floor) since gravity would eventually pull the liquid down and out from under the copper plate.
Don't quote me on this one, and correct me if I'm wrong, but: The liquid metal (TIM) usually is made up of Gallium, Tin, and Indium, in different proportions(can't really find out what those are). Nickel is mostly safe to use with liquid metal because the electric potential difference is quite small. Copper is kind of safe? Copper's electrical potential is closer to Gallium's. "Gallium has a potential of -0.53 volts and copper has a potential of +0.33 volts, the difference between the two metals is going to create a "battery" with a potential of 0.86 volts.Gradually this battery will deplete; as the gallium migrates over to the copper. You may have heard some stories about how gallium "dries up" and evaporates. In many of those cases it is likely that it acted like a battery and migrated over to and became part of the copper heat sink." ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-Q8Xo43sfLgY.html ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-4HKpMYJ-6go.html Thoisoi clips, from what I can gather he's some sort of chemist? ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-nFRlV5mN65g.html forum.notebookreview.com/threads/something-to-think-about-liquid-metal-compatibility-with-copper-heat-sinks.800890/
So if I'm understanding correctly, the "staining" you see on copper is Gallium making an alloy with copper. Gamers Nexus did a video on this. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-fnbovjT4JLQ.html
Russian youtubers seem to have more information about the long term consequences ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-nFRlV5mN65g.html ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-ZuzSx0HmpR0.html
SIlicon is extremely stable material and even with galium and other parts of the liquid metals it won't corrode the die. I don't know what the ruff surface is, but I guess it's some material kind of burnt on top of the die, or it's the gallium that stick to the surface and doesn't come off.
I'm pretty sure the effect on aluminum has nothing to do with oxidation (rusting in common language). The effect is caused by gallium - and gallium has no oxygen in. What happens is that the gallium can seep into some other metals (like aluminum most notably) and start to break apart the bonds that keep it together as it inserts itself into the molecular gaps - making it become extremely brittle. Obviously this would be terrible to have happen on an aluminium heatsink - it would likely start breaking apart from just the mounting pressure. It probably also wreacks havic with it's thermal characteristics.