Long-term Liquid Ultra results

Many of you that frequent here know that I am a big proponent of liquid metal in laptops. I have applied Coolaboratory Liquid Ultra 2 times to my beloved behemoth, the M18x R2. First application back in 2013 lasted only a couple weeks, as I foolishly used too thick a thermal pad on the CPU phases and the heatsink no longer fit properly. No matter! Pulled, removed the 1.0mm thick thermal pad, replaced with a 0.5mm pad, applied some more liquid metal, and everything was hunky dory years and years. Fast forward to 2020, I brought her out of hibernation and played some ARK. I couldn't understand why my frames were randomly dropping and my fans were randomly ramping up and down. Silly me, didn't even bother to check what the CPU temp was doing. Opened up Throttlestop, found that my poor 3920XM was bouncing off T-Junction (105C!!!!!). Clearly, it was time for a check up. Many of you veterans with the M18x know tear down is an absolute chore to get to the CPU, but it had to be done. I figured I'd snap some photos of what I found in there for what is effectively 6 years of constant liquid metal application. Take a look!

As you can see, the liquid metal has effectively permanently bonded to the die. I could not remove it without risking damage to the core. See the discoloration on the PCB around the die? Thats where liquid metal had previously slightly spilled out and has permanently discolored the PCB. This was an unexpected result for me.



Liquid ultra has also bonded to the copper in the heatsink. This was an expected result, as it is known galinstan alloys with copper. It has become quite uneven however, and not even the included scuff pad with liquid ultra could really do a good job sanding it down. I left it alone.




After wiping the die down with 91% IPA, i re-applied more liquid ultra without difficulty. It spreads much easier this time due to previous liquid metal bonding to the die.



Cleaned the heatsink mating surface as best I could with the scuff pad and then cleaned with IPA, dabbed the excess liquid ultra from the brush on the mating surface. Again, spread very easily and attached without problem as the galinstan has already made a nice alloy with copper.




I've heard people postulate the alloyed copper + galinstan and lack of a smooth mating surface would severely impact temperatures in the future or other theories regarding degraded performance in the future after liquid metal usage. I'm happy to say, it is not so. I am quite far from the alarming 105C I had seen earlier in the day.



I'd wager this is one of the longest applications of liquid metal I've ever seen or read about in a laptop, and I can report that it works better and longer than any other compound. No need to freak out over liquid metal my friends. I can manage a re-application every 6 years.

 

Thanks for sharing. An honest report on the long term effects of LM supporting my strategy to avoid it unless absolutely necessary.

 

I thought my post proved the opposite. 6 years on the same application is longer than any conventional paste can deliver and the alloying has negligible impact on performance. Unless you care about how your CPU and heatsink are cosmetically, there is no reason not to use liquid metal

 

Well, the pictures speak for themselves. Not saying it's an insurmountable issue, but it doesn't look pretty and would prefer not to have to deal with this myself.

 

Cosmetic != Performance and Performance > Cosmetic

I could care less what my copper or die looks like, but I do care how it performs.

 

Performance under LM may or may not be improved. It's not just a cosmetic issue either: there are permanent physical changes, and some crude lapping etc is required to restore the die & heatsink to usable state. Hopefully the resulting alloy is not detrimental to performance should someone wish to revert to non-conductive paste, for whatever reason.

 

"Hopefully" it's not determinantal? The original post PROVES it is NOT after more than 6 years of use. The "crude lapping" is done with the materials that are shipped with the LM, intentional tools for a standard job. Moreover, the LM life span is YEARS longer than traditional paste. Despite what may appear as a rough surface, the performance/cooling is still better than any pristine surface with traditional paste. If your take away from this is the die/heatsink is ruined you are ignoring the performance data and simply prioritizing your cosmetic preference. That's your right and opinion, even if it's based on nothing more than your personal preference.

 

If you read carefully, I asked whether prior LM application affects performance using subsequent application of a non-LM compound. As a matter of fact, this question is not addressed here. This might be relevant from resale value perspective, for instance.

Additionally, this post doesn't really answer the question how this particular system would perform on comparison using a leading durable non-LM paste, say Phobya NGE

 

In essence you're telling someone that has built a purpose built machine that you could never justify such modifications or use because he/she/someone else may not be able to return the machine to it's original state at a later date. The OP has no intention of using regular paste, as he has re-applied LM with success after 6 years and shared it here. Resale value or potential new owners are all supposition to prove your preference.

 

Essentially, I am not saying that at all - please re-read my comments if in doubt. No question regarding OPs choices and priorities, just a calm analysis of the post, and the anecdotal results.

 

Not only that, I could never get the performance I have now without liquid metal. I had tested several pastes before including AS5, GC Extreme, MasterGel Maker, Antec Nano 7, IC Diamond, etc.
Nothing got even close to liquid metal. You must remember 4.2GHZ all-core full time on a laptop in 2012-2013 was insane. The goals at that time were shooting for maximum performance.

todays goals are different. The thin and light gaming laptop trend has forced users to try and keep things under control from having a meltdown. Example, Razer Blade machines. I’ve had 3 of them now, all bouncing off the thermal limiter or close to it (mid-upper 90s C) and the ridiculous temperatures were only tamed with liquid metal.

 

I havent seen any question marks in any of your posts in this thread.

Am I missing context here?

 

Perhaps I read into things. I can see your point, we just have landed on opposite sides of a decision. Appreciate the calm and candor in your responses.

 

One can pose questions without the use of question marks. Please see the comment you quoted after "Additionally," for an example.

 

Your non-LM compound list is extensive for sure, but misses a couple prominent top performers. Also "nowhere near" could mean a lot of things, 5C difference for example, which would be acceptable in this case.

Anyway, thanks for sharing, interesting post.

 

I just checked my 1.5 year old LM homemade Galinstan application on my GT73 throttlebook and it's as good as ever.
1C core temp deltas running prime95 at 4.7 ghz @ 1.2848v (89/88/89/88)

That's proof my method of buffing the heatsink with 1500 grit, then spending 10 minutes wiping a layer around it (to force early absorption of the gallium, until what is left over starts feeling much thicker than usual), then applying another layer on top of it actually works (of course the heatsink has to be reasonably flat or previously sanded flat or it won't matter).

Papusan's baking method also works (assuming you can bake the block at 100C, you clearly can't do that on AIO's, but for laptop heatsinks and IHS delids its fine).

 

I wouldn't bake a heatsink at 100C. The heatpipes will deform and fall off.

 

Did another test.
86/85/86/86 with lower ambients (maybe like 74F).

 

i doubt 100c would **** it up because some CPUs have have operating deltas and throttles at 90+C. i pushed a noctua to 90+c after the fan started dying. My M17x R4 is known to get pretty toasty too.

 

This is just one of the many examples showing why nickel plating is needed for long-term liquid metal stability.

 

Fast forward to 2020, I brought her out of hibernation
So when did the LM dry out? Not when you brought it out of HIBERNATION!
Are you claiming the record for the longest time before realizing it was no longer working?
Seriously! Im confused..

 

Yeah, LM definitely isn’t lasting 6 years of usage in a laptop with relatively high temps (compared to a desktop) and bare copper heatsinks. From personal experience, on a high wattage DTR (100W+ CPU and 150-200W GPU running in the 70s to 80s), LM lasts about 2 years of heavy usage before losing its efficacy. Still breaks down slower than a traditional paste though, and has lower temps.

 

Ooops!
Our hosts post is a prime example of an epic "FAIL" on the heatsink side application of LM. You can see in the pic the defined outline of the true die contact patch with the LM smeared half again out the sides of the patch.Keep in mind that the heatsink in the pic is in its operational position " facing UP" ( 99.9 of the time) Given a good banging around, where do you think that LM is going to go when stressed. The LM creeps to the smear. Air gets into the die-sink button interface and the reactions begin.Game over for any hope of longevity of the application.

Watch the video from ASUS ROG when they are applying LM to their Production lappys. They use a Stencil. They are also using FOAM BARRIERS.

 

Looksee Here!


Yes! You can get Nickel Plating $olution. Watch on YouTube and "Follow the Yellow Brick Road"

 

.Yes,i saw that too. By looking at the condition of the sink it is obvious to myself and by his own admission, that it was a "Shot in the Dark" proposition. By looking at the contact patch on the sink its obvious he broke the big rule. Everyone is so damn concentrated on that Die that they just slobber LM on the Sink button. You have to have the LM STENCILED on that button as close to and lined with the die dimensions as precisely as you can get it and in the right amount. Reason being that it is surface tension that keeps everything stable and in place. Any smears to the outside of the die dimensions on the heatsink button are going to draw LM like a magnet when shook or jarred and the Dam-"Gallium Oxide Skin" is broken. Break that surface tension between the two parts and your hooped,period. Read up on"Surface Tension" physics.
1: Did he do the proper preparation No cut cleaning time by half/Most important part/epic fail.
That copper has to be absolutely clean and etched.Hydrochloric (Mureatic) acid 10-15% in Distilled water (not for Tube Squeezers)
2: Did he use the correct components in the correct amounts.No just poured it on
The worst thing he did was to put the COPPER Heatsink and ALUMINIUM coldplate in the solution TOGETHER. Ultimate contamination!!
Did YOU watch that video I posted, all the way thru?
N0!. But even on a not so great turnout of a first shot, there is a hell of a lot less mess to clean up than without it.
EDIT: OK. I watched his video of how he plated the heatsink. He used CHEMICAL PLATING!
Thats about as far from the "BRUSH" ELECTROPLATING video I posted as a moon shot. You cant build Squat for a nickel coating that way. Cats axe for plating your grannies teapot or spoons for a re shine. Do ya think they plate car bumpers that way? Polish-Acid etch-Copper plate-"THICK" NICKEL PLATE.
Another shiny video with a ho-hum ending. Forgiven if he does a "Brush" ELECTROPLATING version.
Frankly I cant understand why he even attempted that. All he has done is put that "suggestion" in peoples heads that Nickel Plating doesnt work. Conehead!

You can grow a beard Electro Brushing and the Nickel will just keep building. In the video I posted he was using a Gold premix solution.We would use a Nickel Acetate Mix. As long as that pot is full you can keep on building your Nickel thickness on a properly prepped and etched base.
Just another guy that doesnt realize his motherboard is upside down when he is working on it and that LM does not fall "UP." Still using Nail Polish for an insulator.Whatever happened to 5 minute epoxy? Its cheaper than nail polish! Coneheads!
Wanna Tip?: After coating your lappys LM STENCILED heatsink, give your heatsink a firm "whack" on the edge of the table. Now you got the right amount on it. (or not)
For Coneheads: Whack your Puter on the edge of the table!

 

Though I didn’t take photos, I recently sold off my dad’s Alienware 18 which also had LM since 2015 on the 4930MX and it was working perfect, and Haswell chips have been known to be blast furnace. Only reason I removed it was to clean off the LM in case it was to spill out during shipment. To my surprise, it was mostly solid at this point and semi fused the heatsink to the CPU die. It didn’t damage either when I separated, but there was some LM that had fuses to the CPU die as well as the heatsink that were completely unremovable. I used the included scuff pad on the heatsink but I left the die alone and repasted with AS5 to ship to its new home.


But my point is that machine saw daily use on a very hot chip and the LM-TIM was never compromised. Thermal performance was as great as the day it was applied

 

Good to hear! I wonder how many people are aware that even fingerprint oil on the die or the heatsink button can sabotage the whole ballgame? Read the below a couple of times. Strain out the technical jargon and you will have a better understanding. That Gallium Oxide SKIN and Surface tension is whats holding it all together and keeping air from contaminating the core. Give it the best chance with matching sharp edges to keep everything ":in the bag". No smearing on the Heatsink!! Use Acetone for cleaning. Use care and due caution (read dont splash it around like rain). Unless you have a rare reaction to it, it will not do ya any harm.Simple soap and water cleanup of hands and dont get it in your eyes. NOT recommended for those that ask what is the best way to apply thermal paste, which is like asking "how do I put the toothpaste on my brush"..
Please note the comments in that Paper about Pressure. Many on the threads recommend "HIGHER" heatsink pressure. Not So! STABLE and accuracy in the initial placement of the heatsink.

http://soft-matter.seas.harvard.edu/index.php/Eutectic_Gallium-Indium_(EGaIn):_A_Liquid_Metal_Alloy_for_the_Formation_of_Stable_Structures_in_Microchannels_at_Room_Temperature

 

https://www.liquidmetalalloy.com/liquid-metal Guess Who!

This is how simple it is. As for the time it takes? The more time you spend cooking the brew, and the longer swabbing it on,the thicker your Nickel Plating. Make sure you understand the term "electrocleaning" and do it before plating or your just spinning ur wheels.You should see bubbles on the surface and possible "whiff" of steam if its working properly. Watch the Gold Knife video and understand the term "Wetting Out". Nickel Anodes ( I used 1x6x.003) are $10-13 on eBay https:// www.youtube.com/watch?v=dt-_7mCargE. . The darker the green, the more ionized Nickel in the solution and the better the process goes. I suppose there will be wizards that challenge such a https://www.aliexpress.com/item/1005002275805773.html?spm=a2g0s.9042311.0.0.8e674c4dTSF5esa simple process thats been around for near 100 years. Your making a protective coating here, NOT making Pretty! You DO NOT need one of those BS plating kits for $$$$$$. I used THIS in the link above, 12v for making solution. 5v for plating and electrocleaning. Rinse everything in water between steps and wear Rubber gloves to keep your greasy little mittens off the goodies. Thats a reflection of the swab on the GPU button!

KISS

 

Plug stripped back to expose brass terminals.2 1/2 " STAINLESS STEEL bolt with head cut off soldered into disposable metal brush. Some wire and alligator clips soldered on.Some heatshrink tube. 400 grit polish on heatsink buttons.Muriatic Acid cut with water for electrowash for 5-10 minutes till water rinse coats the button evenly-"wetted". Hot water rinse and then swab on the nickel in circles. Swab pad cut off top of old socks (clean) and Zip Tied. Stencil mask everything and WEAR RUBBER GLOVES!!
Use FRESH SWABS after each step. DONT reuse. Contamination!

CAUTION: ADD ACID TO WATER TO DILUTE (in that order). AVOID BREATHING FUMES WHEN MIXING - It puffs a small cloud when acid hits the water. Keep the swab moving when cleaning.You will see small bubbles around the swab. Thats Hydrogen originating in and cleaning out the "pores" of the copper. You will see a trace of Steam if you are moving to slowly. Moving to slowly both Cleaning and Plating can burn the copper so use a quick circular motion.
READ ALL INSTRUCTIONS ON THE MURIATIC ACID BOTTLE BEFORE OPENING.

EDIT 8/20/2021: Seems this messy step can be avoided if you take the time for a smooth 400 grit wet and dry acetone soaked finish and Super Cleaning with Acetone and stiff brush. The copper will take on a dull even blush when ready for plating but you must use Acetone for the prep.

 

1. Exact same as when applied after 4 months heavy use.Stenciled contact patch and LM worked into copper over 3 hour period, 6hr sit and lite coat before final assembly. What appears to be what the Coneheads call PUDDLES or PUDDLING, is where Surface Tension "PULLS" the LM to a point when the heatsink is removed. Eg: If you pulled it 100% straight up while removing, the LM "Puddle" would be located dead center of the Die. You can see by the puddles, I lifted the stuck GPU side first, let the sink down and then lifted it all straight up. Note Bubble trying to escape upper right GPU. To much LM. Also ridge line upper left.The LM remained wet after 4 months because I had TEMPERED the copper with a first coat and let it sink in 6-7 hours.
2.LM wiped off clean with acetone - Epoxy on foam is a PITA to clean off. My bad!
3. Smooth area is where there was good contact to Die. Of no importance with LM but Perfect is always better.
4.GPU and CPU smooth areas match original MSI paste good contact areas = HeatSink-Die poor alignment. Note x acto knife lines from tracing original msi paste contact patch for taping a LM stencil later.
5.EVEN Penetration from working the LM into the STENCILED copper and getting that SHARP matched edge. This pic after HEAVY sanding on 4 months of use CLU on copper sink during rebop to Nickel.
6.STENCILED nickel plate shining from wet plating solution.Indicates final finish.
7.Nickel Plating dry and dulled out. New plan, corner 1/2 split, overlapping corners, window seal self stick foam. CPU dam in this pic was redone to match overlapped GPU corners. Any sink button-coldplate gaps sealed with epoxy on the coldplate side. Work in progress.
8. Typical work load encode,264/1920 to 265/1280, factory fan settings. Only thing changed was ThrottleStop- Options "Prochot" offset 3c from stock MSI 95c to 98c and the Thermalright Odysey 12 wmk GPU-CPU pads.
My thoughts: OK to use on straight copper if stenciled and worked in to "Temper" the copper. Use Acetone and CLEAN,clean,clean!!
Idle and full load temps with Nickle plate are roughly 1-3 degrees cooler than LM applied straight to copper.Probable cause, penetration into copper of gallium and lower conductivity of the formed alloy. And Yes, Alice, your 20 or 3080 will run warmer. Not the point! Oh yeah! I have 2 heatsink-fan assemblies if some of the pics are confusing. Prepping the original CLUltra on Copper right now for Nickel.
IF A CONTACT PATCH IS NOT PAINTED ONTO THE HEATSINK BUTTON, EITHER COPPER OR NICKEL, FULL CONTACT WILL NOT HAPPEN. You can check this statement by coating the chip dies with LM and installing the heatsink. Remove the heatsink and observe the Buttons. NADA!! ZIP! ZILCH! Maybe a million little "sweat" bubbles but no direct contact.
How much LM to use? Check your original paste heatsink contact patch right after you pull the sink.. If the paste is thin and even over the whole contact patch, you have good contact so you would use less LM. In my pics the patch was 1/2 and 1/2 so gave a good bit more than the CLU guide recommended "thin" layer.Obviously too much on the GPU side. You can tweak the heatpipes some for alignment, but they will resist bending at a point, and at that point will double over on themselves so quick you cant blink!
Tweaking one side usually puts the other side out.Chasing rabbits! How far down the hole do you want to go? http://forum.notebookreview.com/thr...d-metal-safety-insurance-guide.817207/page-15
See my other ramblings here!

Stumbler View attachment 193984

 

You mask the heatsink buttons when nickle plating so you dont get Aluminum Contamination from the Coldplate. Mask Twice! Once for the Electro-washing (no aluminum onto the copper) where the tape will get ate a bit. And final mask for the plating with all edges on the copper button for both stages. Mureatic Acid is sold in a % reduced solution.Use 10-15 % of that reduced solution,mixed in water for electro washing. If you get a " burn" on the copper it is a PITA to work out. Go on the lesser side till you get on your game.

Edit 8/20/2021: Have found that a good scuffing with 400 grit wet and dry cloth soaked in acetone is sufficient for prepping the copper.Just wash super clean with acetone and stiff brush afterwards. Copper should have a dull, even blush when ready for plating.Superfine paper like 1500 grit is detrimental to adhesion.Equate Shiny with Mirror with Slippery. Nickle is one tuff bullet proof coating.But! You want it applied to a textured surface which the 400 grit will give you.The less you have to scuff, the closer your tolerances will remain to factory specs.If you think your fingers are precision instruments that can provide more accuracy than lasers, your lost!
FYI: Once the thermals are transferred thru the Nickel plate to the copper button and heat pipes, that heat is retained in the copper and is reflected back into the button and heat pipes by the Nickle plate and NOT back out the bottom onto the surrounding CPU/GPU area and motherboard.