Which Thermal Paste to buy and apply (Traditional and Liquid Metal)

This is for the moderators... If this thread is ever made a sticky, can you also merge or move this thread along side it - Thermal Paste Roundup - Performance, Price, Recommendations (Update: Thermal Pads!)

 

Keep in mind he's not talking about a failure with LM paste, but with MX-4.
Also many users have complained about Kryonaut shooting up in temps after awhile also due to pumpout or extreme heat from direct die contact. IC7 or Phobya Nanogrease or Noctua NT-H1 are all good options for thick pastes.

 

I was using Arctic Ceramique 2 & was gonna buy mx4 but opted for maker gel nano.

 

The problem with most TIM's is being liquid they will pump out under high heat. This is true even of LM's. Now if you can keep the heat at say 60c as a rule of thumb the TIM does not get to the point of pump out failure at least for a long time.

LM on my laptop pumped out over time. I had to go back to ICD to prevent the issue. Of course as in all cases YMMV.

 

Is this what you meant? https://www.pccasegear.com/products...Lx4eR9U5zWavkMvKXdhoCB0sQAvD_BwE&gclsrc=aw.ds

​

Phobya isn't available where you live. You Americans have it easy, you have all the good stores where you can ACTUALLY find things to buy lol.

IC Diamond and Noctua NT-H1 and that MasterGel Maker Nano are my best options then? A repasting every 2-3 months is acceptable, but I'd really prefer something that lasted for a year. I'm not worried about electrical conductivity, since there are no resistors on the CPU of my laptop itself. I'll still probably use MX-4 on my dGPU since it works fine on that.

I think the thermal compound really is being squeezed out. Here is a close up of my heatsink for the CPU


Notice the triangle thing on the side with the screw? The two pieces supporting the screw are made of a metal that I can easy bend.

P.S That is the stock thermal paste. Would you believe that? Somehow that is better than MX-4.

 

If the tabs easily bend then you have another issue. That is not much pressure can be between the HS and chip before the supports give way. Also the issue will be keeping the HS level to the chip as I imagine there is quite a bit of flex. Also with that much flex a thin TIM like MX-4 will quickly fail.

 

What LM did you use that pumped out?

 

That was CoolLaboratory original. This was in my P75xx with a Q9200m. It took 4 months but taking it apart I could see the tell tale star pattern from the center of the chip radiating out. This was streaks of the TIM and it accumulated in mass at the outside edge of the silicon. That chip though ran HOT all the time as I used to benchmark the thing to no end.

 

With ICD, Noctua, CM Maker Gel Nano you needn't repaste every 2 weeks but every 2 years would be ideal.
The stock paste looks better than mx4. Grab new thermal pads as well.
I could find Maker Gel Nano which was really priced lower than kryonaut. Now, its priced twice as much as kryonaut.

 

Depending on your load I've seen ICD last even longer than that.

 

I do agree on coolermaster mastergel nano, ive been in the space a long time & its my favourite paste to date

I made a post about it awhile back

http://www.overclock.net/forum/63-o...hermal-paste-coolermaster-mastergel-nano.html

 

Are you Asus11?

 

yeah that's me

 

My friend just bought a cube of TG Conductonaut LM, and I'm thinking of borrowing it and jumping straight to LM with my laptop. Does anybody know if it's ok to use Blu Tack to seal off the CPU with the rest of the motherboard? It's this sticky thing that has the texture of dough (flour and water). Anybody has played with Blu Tack as a child (I have) they'll know that it softens when it gets hot. I put the Blutack in boiling water (95C +/-5C according to my NTC thermometer) and it definitely softened but still kept its shape. Only thing is, it faded to a white and it's now not as stretchy as it used to be, but that shouldn't be a problem when it's down i a laptop shouldn't it?

Here's my idea:

I believe the chances of the LM rolling off is already low, but it's always good to have extra insurance. If a ball rolls off, shouldn't the Blu Tack keep the balls contained within the PCB of the CPU? It can do whatever it likes in there since there are no SMD parts on the CPU PCB
The Core and iGPU die is 1mm up from the CPU PCB, so the BluTack height would have to be at most 2 or 3mm so as to not reduce the already low mounting pressure of laptops. In the past I've used BluTack as a solution to clog things like holes in the window and years later that blob of BluTack is still there, suffering 0C winter nights and 40C summer heatwaves. What are everybody's opinions on this? I imagine BluTack isn't the best idea and something more... heat-resistant... should be used. I've heard of foam dams but I don't think I'll be able to get them fitted in my laptop without reducing mounting pressure. There are no SMD parts on my 8250U.

I think the heatsink mounting is really good, since under a full prime 95 load with a raised TDP of 40W my 8250U gets up to 90C and core temperatures are off by only 5C and I haven't had any significant thermal paste degradation on this laptop, even with the MX4 I'm currently using.

I'm aware that LM really starts degrading quickly when temperatures rise above 80C, and it shouldn't be an issue on my laptop since even under a combined dGPU+CPU AIDA64 Load the CPU never pushes above 75C on MX-4

 

I recommend that you use K5 pro to set up that barrier and put a small opening / gap in the barrier to let air pressure equalize.

In my experience, Grizzly Conductonaut are repelled by K5.

 

Oh wait, I could also just use a really thick thermal paste that will stop the LM. Thx for the idea

 

ICD comes to mind

Garbage feces tier thermal paste but is actually useful for non thermally critical application like this.

Note that ICD will scratch the heatsink if you remove it.

 

K5 is a good idea if you already have access to it. But may not be suitable on all applications. Not sure how well it would work with LGA CPU's (assuming you are using LM on the *copper* heatsink and not just under the delid).

But why not use FOAM DAMS like me, @Mr. Fox and @tijgert use? And no mess to worry about cleaning up either (besides the annoying job of cleaning up old LM). I would hate to have to deal with K5 Pro around an IHS, for example.

Like this:


Cut out a square one for the GPU's.
Use nail polish (3 coats) over exposed SMD resistors over CPU and GPU. Use foam dams to stop LM runoff. (Foam dams are NOT to be used in place of nail polish over SMD resistors, but WITH combination with it. If you do not use nail polish, please use Super 33+ electrical tape or Kapton (high quality) tape. Nail polish gave lower temps than tape, according to @judal57

Here is how Mr Fox's foam dams looked like. I don't think you would want K5 Pro to be used in place of those on SOCKETED hardware:

 

I just repasted my laptop with TG Conductonaut Liquid metal!. I originally started with the BluTack method but that just wasn't going to happen. Then I started with the foam dam and wasted an entire hour trying to get the fit right. Unlike the HQ, HK and K chips, the 8250U's die is barely a millimetre raised above the actual PCB, and the mounting pressure of my laptop simply wouldn't press down enough. When I booted the laptop, the CPU would idle at 38C, but shoot up to 98C and thermal throttle when opening edge. Even opening windows explorer caused the CPU to jump to 70C.

The actual application of the liquid metal took about 10 seconds. The LM just stuck onto the CPU. The metal was so easy to spread, yet funnily my friend took 30 minutes to spread it on his GPU die. I also spread the remnants on the cotton bud onto the copper heatsink.

I ended up surrounding the CPU die with a wall of thermal paste. . I sealed off the contacts on the CPU with silicone adhesive. I don't have any pics unfortunately because I am at school right now.

As far as I can tell, idle temps haven't changed but max load temperatures are lower by about 10C. I'll do proper benchmarks when I get home.

 

All of the chips are barely a mm off the actual PCB. Not specific to the 8250U.

If the foam dam caused problems then the foam was far too thick and caused resistance. As I said multiple times, the foam must be VERY porous and light and should compress with your fingers down to the width of a human hair. That's why its best to buy foam locally or find it locally so you can test it the old fashioned way

Anyway glad you got the temps all set!

 

You can as well use K5 Pro for foam dam on BGA.

 

Yes. Even very old caveman era super thick silicon pastes that you don't usually see anymore will work as dams! Like Arctic Ceramique. Have a 22 gram tube of the stuff from over 10 years ago that stuff is thick as all living hell. Not sure if its age or what, but that could also work as a dam as well. (just spread it around the BGA chip housing and it shouldn't go anywhere, especially since it wont be in contact with the main heat source directly!)

Great for "caveman foam dams".

You would be just shocked as to how thick Ceramique is.
Anyone remember the old Radio Shack thermal paste of the Dinosaur era? That's how thick Ceramique is.

 

I've done my testing now; all results below were tested on an 8250U running at a -100mV undervolt.

Before: MX-4
After: Thermal Grizzly Conductonaut

Ambient and idle temperatures are no different.

Prime95 on CPU - Default 15W TDP
Before: 68-70C
After: 60C

But nobody runs their CPU at stock powers do they?

Prime95 on CPU - TDP unlocked to 45W. My 15W 8250U runs at a full 3.4GHz turbo with a power consumption somewhere between 40-41W
Before: Temps ramping up to 80C within 10-20seconds and plateauing at 95-98C 10 minutes later with intermittent throttling.
After: Temps going up to 60C before slowly rising to about 80-85C

Cities Skylines - 40K population - CPU powers ranging from 15-20W with the MX150 fully loaded and power limit throttling (About 25-30W)
Before: 80C
After: 65-70C

Mining - Neoscrypt (or was it Cryptonight?) on CPU and C11 on GPU
Before: 75C
After: 60-62C

Roblox Phantom Forces (Don't kill me for playing it pl0x)
Before: 57C
After: 52C

As you can see, the biggest changes come when the CPU spews out more heat than the heatsink was designed for.

AIDA64 stress test coming in a moment.

 

As expected. +rep. Some as you know still belive that the difference between very good thermal paste as Grizzly Kryonaut vs. Liquid metal is minimal They should take time to reflect on the results.

Please, run Aida64 Stability Stress Tests as the Guide.

 

AIDA64 Results are in! Done as per your wishes @Papusan Stressed until temperatures stopped rising. My laptop's heatsink has quite a large heatspreader on the motherboard so it takes some time for everything to heat up.

What the numbers mean:

1. The temperature drops slightly as the fans spin up to 100%
2. I suddenly realise the integrated GPU is hogging about 4W of the package power causing the CPU to throttle. Then I use Intel Power Balance to prioritise the CPU. The Cores begin to run at their expected 1.8GHz
3. Stress test stopped momentarily and started again now with the correct settings (CPU, FPU, Cache, system memory, and GPUs all stressed)
4. SSD temps spike as Windows thinks its hilarious to update my audio drivers during this test. I doubt it affected temps. Plus the temperatures would have steadied out for the next 8 or so minutes.


Dual 1080p monitors FTW! For those poor mobile users who can't view images:

65C Maximum temperature on CPU Core. Before - 76C
59C Maximum temperature on GPU. Before 54C
CPU limited to 15W for consistency with previous AIDA65 stress test before Liquid Metal and partially also because any higher limits causes throttling with VR-Current limitations.
GPU overclocked +205MHz on the Core + 655MHz to keep in consistency with older test. It power limits anyway to about 25-30W

ARGH! Windows decides to render a little slower than snipping tool. Some boxes are white in HWInfo, but it doesn't really matter because max temperatures tell the majority of the story.

The PCH temperature kept rising throughout the test to until 55.5C, where it stopped rising. I take it as an indicator the whole motherboard warmed up enough.

Tests were run with the default fan curves, which spun the fans to 100% at a threshold of 63C. The ambient temperature was 22C

The CPU temperatures were rock solid for the last 5 or so minutes of the stress test, with each core varying only about 3C. Compared to the MX-4, which had Core temperatures varying by up to 7C. I take it the liquid metal was properly applied and the heatsink was not uneven?

The GPU temperatures were actually higher, and I'm glad. It's an indication that more of the CPU heat is being transferred from the CPU to the heatsink, which in turn is increasing the GPU temp more. Xiaomi has set the GPU to thermal throttle at 80C. As if it's ever going that high.

I'd consider my liquid metal application a success. I believe people with more powerful CPUs (45W HQ and HK CPUs) will see a much higher temperature drop. I 'only' saw a 10C drop across the board

 

Amazing work!

Thank you for doing this!

 

Does anybody have or know a guide to replacing the thermal pads on the 1080m in a P870TM1-G? id like to lower the temps in the entire machine. Put some extreme ultra fujipoly on it or if you have one for the entire machine even better. I have delidded my cpu already and put liquid metal on i stay in cool 40 and 65 on full load and have already lowered the voltage a bit. I might lower it more. Yet my sli setup his 90c sometimes so they have gelid extreme on them. I was scared to put liquid metal. Any suggestions please are welcomed

 

Try some chinese pads or arctic ones. You can ask @Mr. Fox

 

Those work really nice. The blue pads are very soft, gummy and wet feeling. The Chinese pads on eBay are dirt cheap and they work great.

https://www.ebay.com/itm/1pcs-100mm...ction-Silicone-Paste-Thermal-Pad/251735856827

 

Nice pads, but I can't deal with the shipping. Hah. Ordering 30g of Galinstan from Riga for 2 week shipping is one thing, but pads....um... hmm. Yeah.

If you have amazon prime you can try these:
(0.5mm) https://www.amazon.com/gp/product/B00UYTTLI4/ref=oh_aui_detailpage_o08_s00?ie=UTF8&psc=1

1mm: https://www.amazon.com/gp/product/B00UYTTXSM/ref=oh_aui_detailpage_o08_s00?ie=UTF8&psc=1

I would not use 0.5mm on a GPU as most are always 1mm unless otherwise measured by knowledgeable users first. But having some 0.5mm pads sitting around can be quite handy.

 

You must be referring to the wait time on the slow boat from China. The cost of the pads on eBay is far less than Amazon and the shipping is $1.00 LOL. Best to order stuff like this before you need it and then it does not matter if it takes a month of Sundays to arrive. If you need it right away to fix a problem, Amazon is definitely better. I keep stocked up on things like paste, pads, BIOS chips, etc. Cheap stuff that you cannot generally buy from a local store for double or triple the cost of what it is actually worth, I try to keep some stock on hand. When I need those things I need them like yesterday... pronto, with an order of fries.

 

@Mr. Fox Yeah I was referring to the slow boat, sorry!

 

Arctic pads can be had from within the USA for 6$ and double the thermal conductivity.

As for conventional thermal compound, this is my new favorite:
http://www.coolermaster.com/cooling/thermal-compound/mastergel-maker/

I refuse to pay the ridiculous price per gram that thermalgrizzly charges. This is within 1w conductivity and offers 4x more product for the same price as kryonaut.

 

http://forum.notebookreview.com/threads/official-alienware-13-r3-owners-lounge.797884/page-709

Please check post 7083, it might help.

 

MasterGel Maker Nano is really good stuff. Similar performance as Kryosnaut, but a lot more affordable. If you haven't already, give Phobya NanoGrease Extreme a try. It has higher thermal conductance (16W/mK) rating than Kryosnaut or the MasterGel Maker and it is thicker than either of them. Excluding liquid metal, which has no match, this is my new favorite thermal paste now, and it is super-affordable. Kryosnaut is 12.5W/mK and MasterGel Maker is 11W/mK. You can get 3.5g of Phobya for the price of 1.5g of Kryosnaut.

I mentioned this previously in multiple locations. Here is an example of the results I got from Phobya NanoGrease Extreme. I am using it on my desktop now.

It is $12.95 for a 3.5g tube at Performance-PCs.com [ LINK]

Technical specifications:
Content: 3.5g net
Colour: Silver/grey
Thermal conductivity: 16W/mK
Density: 2,7 g/cm³
Viscosity: 85000cps TF
Conforms with ROHS
Extent of delivery:
1x Syringe Phobya NanoGrease Extreme 3.5g
1x Plastic spatula

 

That's very nice, I will have to try this on my GPUs! I'm a CLU guy as far as CPU goes

 

I hope you like it. Another change I made about a year ago, is I no longer apply an X, line or a dot and let the heat sink squish it. I use the spatula and spread a thin layer across the CPU/GPU die/IHS. This gets me more consistent results and uses less thermal paste. There is little or no waste from the excess pushing out the sides and there are no air pockets or dry spots like you can sometimes have using a line, X or dot method.

 

I’m going to disagree with you on this one here, there is a little different system I use to completely eliminate air bubbles. I use an extremely thin layer (with gaps in places) with the spatula to coat the the die, then a small dot in the middle and let the heat sink spread it out over the existing thin/gappy layer. I learned this from UltraNSC when he did several test methods when repairing the PlayStation 3

 

I approve of this. If I didn't have LM on everything, I'd give this a shot.

 

All depends on your heatsink setup. Same application don't work equal for all... See also Thermal Paste Application Techniques

 

Sounds more like you are agreeing with a small added twist rather than disagreeing. I have actually done that before and it worked perfectly fine. But, it ended up being more paste than I needed and it just squirted the extra paste out the sides and it ultimately made no difference in my temps. They were identical without adding the dot in the middle. If the heat sink and die/IHS fit together correctly, all you need is a thin layer. If they do not, I am sure the dot on top of the thin layer helps fill up all of the gaps. Like @Papusan said... all depends on the individual setup. All I was saying is I get better coverage and less wasted thermal paste using a thin layer versus an X, line or dot. Might not work as well for me on a different product without adding the dot in the middle. Same basic principle by spreading it, which is potentially better than using an X, a line or a dot. There's nothing left to chance in terms of bare spot and not nearly as much waste.

I think I posted that I have NanoGrease Extreme on my desktop earlier today. That's not correct. I did have, but I forgot that I have Indigo Extreme on it now. That's working really awesome. It turned out super good this time, so when I am done tinkering with trying to get 8700K working on the P870DM-G, whatever CPU ultimately stays in it, I will see if I can get the Indigo to work on it. (It never worked worth a darn on laptops when I tried to make it work before.)

 

Interesting, somebody on NBR was working with indigo extreme heat spring thermal setup circa 2010 but it didn’t seem to really take off.

 

Those are two different products. Heat spring is kind of like a metal pad. There are some similarities between them and CLU Liquid Metal Pad to the extent they all go through a phase change process.

Indigo Extreme XS is slightly different. It is engineered for specific CPUs and takes a different kit for each socket type. It has a built-in liquid metal capture zone to avoid leakage, melts, then re-solidifies after allowing the system to cool off for 30 minutes.

I am getting max core load temps on my 8700K of around 72-75°C with the cores clocked between 5.3 and 5.5GHz, so it is totally amazing results.

http://www.performance-pcs.com/indi...d-thermal-interface-eti-kit-for-lga-115x.html

 

LOL I think I might be the only one around these parts using Prolimatech PK-3. I got a big 5 gram syringe for only $15 on Newegg a few years ago. I lost track of how many repastes I've done, but there's still a lot left.

 

I am sharing here professional opinion as I received from several experts while facing same problem. Some of geeks suggested
Noctua NT-H1 which is good for pc. Few responses were for Thermal Grizzly Conductonaut for huge performance boost. But, according to few guides available on Google, Arctic Silver 5 is the best choice for all kind of CPUs. Here, I found a guide that might be helpful to every newbie. And, the guide link is here for the best thermal paste.

 

That link is just an advertisement. Arctic Silver 5 has been surpassed by several other TIMs in the last several years. Expect several others to chime in here to confirm that Arctic Silver 5 had its day, but that was long ago.

 

nearly any nano-diamond based compound will outlive and outperform AS5

 

It's not really the best any more, but in many comparisons it's in that huge pack that's within 10ths of a degree of each other, so still an OK pick. I think I've still got it in one of my systems, likely to be replaced with Phobya when the time comes to repaste.

 

pretty standard tools in my case: ICD for pasting gf's and/or relatives/friends computers for fire and forget long-term applications. kryonaut for non-conductive high-perf and gallinstan (CLU raw ingredient at 10% the price) for between ihs and die

Sent from my Xiaomi Mi Max 2 (Oxygen) using Tapatalk

 

Prime95 after 5 minutes (35W TDP limited on purpose)

Time for a repaste of liquid metal? The laptop has had the CPU+GPU fully loaded for at least 2 weeks 400+hrs at over 70C. Once I mined (for fun) for 1 week constantly with an average temp of 72-75C. Other times I play Cities Skylines which pushes the temperatures into the mid 70s. I've been actively trying to avoid just this by limiting my CPU temperature to 80C but I guess it happens to all laptops. Just before I repaste I'll raise the TDP to 50W and run the Prime95 benchmark again and see how temperatures jump. I've read on how the gallium can seep into the copper and require a repaste, especially on laptops with elevated temperatures. I know a second repaste of liquid metal would see it lasting much longer.

my guess on how this works:

Some single-threaded tasks like to hog Core 0, thus causing that core (and Core 2, the one next to it) to be hotter and cause the liquid metal on those cores to degrade faster, and a cycle continues.

Your opinions on this? I mean before the temperatures varied by at most 2-3C. The 6+ difference is enough to make the fan spin one 'step' faster when in reality the CPU is about 5C warmer than it should be. I'll have to borrow the liquid metal off my friend. And I'm curious as to how that wall of MX-4 around the LM is faring.