Best Thermal Compound

Hi Guys,

I'm planning on repasting my CPU and I'm looking for the best thermal paste for the job. I've used IC Diamond before my G73JH for 4 years (had the reseller apply it) and it even worked after such a long time. Now, I have some IC Diamond which I was planning on using but I heard horror stories about how it damages the CPU's surface, voiding the warranty. Hence, I'm looking for a similar alternative? I've been looking at the Arctic Cooling MX - 4 and it looks pretty good. I can get it quite cheap however if there is something better I am open to suggestions.

Thanks

 

hi i m also looking for a good paste and i m deciyding or artic cooling mx4 that have a 8,5W/mk ,,,,or collaboratory ultra ,,,

 

A proper application of any decent paste will do wonders vs thermal conductivity, get anything in the 5-10 W/mK range and apply it properly, that's all there is to it. MX-4 is a pretty decent choice.

This discussion creeped up in another thread recently, relevant post: http://forum.notebookreview.com/har...paste-scratch-your-cpu-die-4.html#post9662623 Just follow the link in that post for the math behind it.

 

Get the Coollaboratory Liquid Ultra. I used MX-4 for years and just tried Ultra the other week. Temps dropped by about 7-9 degrees. Just make sure Liquid Ultra does not touch any aluminium as it will destroy it. For the first time I removed the CPU to apply the thermal paste for this reason alone. Application proved to be a lot easier than I expected though.

 

I have an m17x R4 so you probably know best about repasting it.. Can you just explain by what aluminium this paste can touch? Also how do u apply it (spread it etc)? I personally am more in favour of getting the MX4 since its half the price of this and doesn't have these issues...

 

It depends on the thermal paste. If you want to use Liquid ultra, it comes with 2 brushes for spreading on the CPU die. The paste itself is literally made from liquid metal - if you are not careful when pushing it out of the tube, drop might roll out from the CPU and onto the mainboard - that would be very unhealthy for the computer. The easiest way to apply it is to take the CPU out the the socket and place it on lint free cloth, then carefully squeeze a small drop of the paste and spread it with the provided brush so that a thin layer of the paste would cover CPU die completely. Before removing the cpu from the socket, you should clean the old paste with lint free cloth and some rubbing alcohol from the CPU and the heatsink. Any thermal paste only needs to be applied to the CPU die, leaving the heatsink clean. When heatsink is reseated, it will push out any excess paste.
In R4 paste does not have any chance to touch aluminium if you are careful as the heatsink has a copper plate in the middle, which touches the die itself. So as long as you remove the CPU before applying fresh paste and only apply it to the cpu die, you are safe.
In MX-4 case I would leave the CPU in the socket, clean the cpu and heatsink and then squeeze out a rice size MX-4 along the cpu die. Then just reseat the heatsink back and off you go. I stopped spreading MX-4 after I saw a youtube video, where some guy tried different techniques with the cpu and glass checking for air bubbles. In that test, spreading method would leave some air bubbles, which is not what you want between your hot cpu and heatsink.

 

As much as I dislike IC Diamond, it does seem to be the best non-conductive thermal paste on the market right now. MX-4 is also great after an initial repaste but durability is questionable, I've personally experienced temp gains of +5C after a month, but thankfully it leveled off after that.

Liquid Ultra I would not recommend unless you're into hardcore benching and don't mind the risks. Also keep in mind that below a thickness of 0.1mm (if memory serves), thermal conductivity of the paste no longer becomes a limiting factor.

 

I have used IC Diamond 7, switched to Arctic MX-4. Both work well, ICD7 performs slightly better (~1C), but MX-4 is easier to clean and apply. I have read some good opinions about Phobya Nanogrease Extreme, I guess this will be the next TIM for me to try.

 

I've gone ahead and bought Arctic Cooling MX4... So how do you guys usually apply it? Pea - sized drop at centre of CPU and screw the heatsink down?

 

CPU core dies are a bit elongated; I use a line method to apply the TIM. About "half of a grain of rice" amount of paste is enough.

 

GPU pea sized drop at center; CPU line across the die.

 

Liquid Ultra is the best on the market today. Previously I used gelid extreme. That's one of the best thermal grease on the market today. Certainly in the top 3

 

The very first thing you have to establish is whether your contact is good or poor between your heatsink and die.
Thermal transfer efficiency is linearly inverse proportional to interface thickness and reciprocal inverse proportional to conductivity (i.e. Conductivity ceases to matter at a certain thickness).

This is really easy to determine because when you remove your HS, good contact leaves an area on the die where the paste is really really thin while an even residue layer implies moderate-poor contact. With IC Diamond, this is difficult to establish since the paste is thick and semi-solid by nature.

Poor Contact
If poor contact, then conductivity and chemical stability is king. The reason is if the contact is poor and stability is bad, some excellently conductive pastes will leach out of the TI zone (e.g. HeGrease Normal).
I recommend the ones I've used, HeGrease Extreme or IC Diamond. Both are electrically non-conductive and have excellent stability.

IC Diamond has the advantage in brute stability and if you contact is really really bad, it can bridge gaps as big as 1mm with a minimal loss in efficiency. However, its higher filler loading (which gives it the bulk conductive properties) is a liability when forced to thinly spread. Thus if your contact is good then it is unable to take advantage of it.
Indigo Xtreme | Skinnee Labs

HeGrease Extreme trades stability and bulk conductivity for enhanced spread-ability and sheer conductivity when thin. This is recommended when your contact is good-moderate.
2011 Thermal Compound Roundup – Results Compilation | Skinnee Labs

Good Contact
If your contact is good or excellent, then your TIM brand doesn't really matter. The key properties you need are spreadability and thermal stability. You want something that can go extremely thin but not leach out over several thermal loading cycles.

If you want a brand recommendation, I can probably vouch for TIM Consultants Quantum. This paste has next to no cure (indicating thermal stability) and very good spreadability.
EVGA Frostbite, Phobya HeGrease, and TIM Consultants Quantum Review | Skinnee Labs


The Coollaboratory stuff is without a doubt the very best you can use. However, it has serious problems with removal, aluminium corrosion and electrical conductivity. Basically, if you want to use it, less is more.

 

Make a bead along the center of the die. For laptop CPU's and GPU's since there is no IHS, you will need to make sure there is enough that it will cover the ENTIRE die when the heatsink compresses the TIM, so if anything a little extra is not a bad idea (assuming it's non-conductive and non-capacitive).

 

Ok so let's say I put a pea sized drop and compress it so it spreads out using a hard piece of plastic... Will that be a better idea to ensure that it spreads? The thing is I don't understand what you mean by bead.. do you mean the pea sized drop method?

P.S: Sorry if I'm asking noob questions since this is my 1st time repasting myself..

 

A bead is basically an elongated drop (a short line with the width of the pea-sized drop).


Sent from my iPhone using Tapatalk

 

Laptops cooling systems are typically poorly designed or machined. The ones I've worked on have poor contacts or uneven pressure.

 

Best way to check heatsink-die contact is via a pressure indicationg film. They're pretty good at giving out free samples, and if nothing's changed, the sample is big enough to check CPU + 2 GPUs and still have enough left over for more.

 

Actually in the dental industry, we use something called articulating paper to measure contact pressure. Go to your nearest Dental Supply shop and see if they can provide you a roll. Pressure indicating spray is also effective but a tad difficult to remove.

 

That youtube video you are referring to was very poorly done and staged to show spreading as being horrible. In the spreading method demonstration, he pushed down on the glass as hard as he could, squeezing out the thermal paste. Then he released pressure from it and the glass popped back up, showing not bubbles, which he claimed, but a giant air gap. He didn't do any of the other "tests" in the same way.

With a heatspreader I'm not sure what method is best, because the few times I had a desktop CPU with a heatspreader that I didn't remove I just did the spread method by habit. But with exposed dies, I have always gotten cooler temperatures with spreading (however, the difference is minimal). The goal is to have as thin of a layer of paste between the CPU die and the heatsink as possible. On a laptop especially, that isn't putting mega amounts of force down on the heatsink, it would be awfully hard to squeeze the paste down to a very thin film. I've noticed this when removing heatsinks.

 

Does anyone still use Arctic Silver 5?

I have been using it for years with excellent results. As a matter of fact I just re-pasted my Thinkpad T410 with it and saw a decrease in temps of almost 20 degrees celcius. Mainly because the old TIM was dried and there was a massive amount of it. Here's my post from the Lenovo section of these forums:

 

I also used Arctic Silver 5 a while ago, but it didn't respond well to poor contact and uneven pressure.

Initially the max load temperature was about 20 C lower than the replaced stock TIM.

Then the temperature started climbing back up over a the course of a year, finally prompting me to open up the cooling assembly when there was less than 5 C temperature difference.

 

Just used some last night on an old machine.


Sent from my iPhone using Tapatalk

 

Doesn't AS5 have a curing time of 200 hours?

 

Does it? Doesn't matter - it's looking like the mobo is shot, since replacing the heat sink and fan didn't seem to do anything.

 

I have tested both with the thin line and the spread method. Used heated up gelid extreme during the application . Are 100% sure of the application of the thin line is the best. The temperature on my overclocked i7 4930mx was about 3-4 degrees cooler under various tests with that application. But nothing beats liquid ultra

 

This ^

Liquid Ultra is nothing short of amazing.

 

I'm not too fussed with how the thermal paste is applied as long as 3 conditions are fulfilled:
-not too much (Thermal paste be expensive yo)
-maximum die coverage
-minimal air bubbles

Check out this site Thermal Paste Application Techniques - Puget Custom Computers, fairly comprehensive.
I've been trying out the X method recently and been getting pretty consistent results.

 

Thanks for linking that article, it was very informative. I had been using the full cover and spread thin method for years, but it looks like I will be using the X pattern from now on. Bookmarked and e-mailed to co-workers and rep given. Thanks again.

 

What works best for desktop processors doesn't necessarily work best for laptop processors. Read the comments to the article.

The heatspreader's surface area is much, much larger than the exposed die of a mobile processor, the heatspreader is not flat like an exposed die is, and laptop heatsinks have much lower mounting pressures than their desktop counterparts.

 

Remember that a laptop cpu has a core without heat spreader. The best is the thin line on the processor core without a doubt. Use minimal thermal grease and heat the paste before application.

 

Exactly... as I and many others have preached time and time again, this is just the CPU or GPU die exposed. The *entire* surface of the die needs to be covered. Better to have a slightly thicker section than have no paste at all.

 

I've got everything I need for the repasting.. just need to get some compressed air and I should be sorted... I'm probably going to use the rice grain method and put the heatsink down and let the paste spread that way..

 

So then I will stick to what I have been doing all these years for laptop CPU's and forget the X method. I don't do desktop machine's anymore. All my family is laptop people and all my clients have next day support from the hardware vendor. So the only times I really re-paste anything is for a friend or family member. Thanks for the clarification guys.

 

Just remember to heat up the thermal grease before applying it. Especially important with a thick thermal grease "-)

 

I did some testings on GPU temperatures and I published them here

http://forum.notebookreview.com/har...us-thermal-paste-thermal-pad-replacement.html

I am concentrated on thermal pads mostly and how they should be replaced with thermal paste or copper shim but still it might be usefull to some people.

 

I was told to heat up ICD but does MX-4 need to be heated up too? If I have to warm it up, is it good enough to heat up a cup of water in the microwave for 1 min and leave the tube in the water then or does it need to be heated up more?

 

MX-4 is plug and play (or more like open and spray )

 

Let me share my experience and findings:

I am a PhD student at a pretty good university with more access to intruments than the average enthusiast. So I decided to.. ahem.. make use of a home built instrument inside a controlled insulated box that measures thermal conductivity of thin films. The instrument creates the film itself by pressing together two plasma cleaned surfaces so there is no variability of mounting here. I tested out 5-6 different TIMs keeping cure times in mind and most turned out to exaggerate their conductivity values. AS-5 in particular was one of the worst offenders (NREL's results which are similar to mine are published here: http://www.nrel.gov/docs/fy08osti/42972.pdf). MX-4 and GC-Extreme were also in the same basket. Prolimatech PK-3 was a pleasant surprise where their numbers actually were pretty spot on. Another surprise was a new comer from a company called Thermene. I had tried out their Gen 1 paste last year and wasn't impressed at all but the new Gen 2 is very impressive. I will say I got 2 batches to be sure and one batch was not as good so I did let them know about it and they got back to me saying that batch was a part of a small custom build order so it isn't very representative of their retail product.

tl, dr: AS-5 <<< MX-4 < GC-Extreme = PK-3 < Thermene

In real world applications, you wouldn't really see a whole lot of difference between the top 3 pastes so just go with what's on sale

I also saw CLU mentioned in here so let me say this- it is great between a de-lidded desktop CPU and the IHS but don't bother using it on direct contact with a nickel plated heatsink. It is better than the older CLP but it still stains nickel enough for an observant technician to deny RMA if ever you need to send it in for repairs.

 

^not if you have a spare heatsink on hand

Also, CLU beats every other paste hands down for laptop CPUs with a bare die.

 

f you overclock your CPU, then the CPU is hotter than standard. What I have discovered at a higher temperature over some time, some of the cool paste is not keeping quality. That is the main reason for choosing a good quality. Tested a so-called good thermal grease (Noctua nt-h1) and tested my overclocked machine in some tests. Approximately one month after, I saw the temperature of my cpu increased , it was consistently 5-6 degrees Celsius warmer during tests. This increase was not as big with Gelid extreme, maybe 1 degree Celsius higher after some use. After application of the Liquid Ultra 3 months ago there is no increase in temperature after many tests. During the tests, it is generally a wattage on my cpu at 85-95watt. At temperatures 80-90 degrees over time, a cheap colingpaste gets worse. By heating up a thick good thermal grease, it is a little softer to applying cpu. It's guaranteed to be one of the better result.Even Gelid extreme working poor by improper Application. The very best thermal grease is often thicker in consistency than the inferior editions.

 

It is very rarely the CPU is broken. If you are sending your computer for repair is just and add on standard thermal grease. Before applying liquid ultra then it is obvious that the CPU is in order before you do it. Remember that liquid ultra saves your CPU from the heat, then it is not destroyed.

 

Well CLU bonds to the heatsink like white on rice, so unless you can (and have the patience to) get it off, much easier to just swap in a new heatsink.

 

No matter how much work it is with Liquid Ultra (clean the CPU and heatsink and apply new) then it is worth it.

 

Only if you know what you're doing and are willing to take the very real risk of frying your CPU and possibly your mobo.

 

QUOTE=n=1;9691109]Only if you know what you're doing and are willing to take the very real risk of frying your CPU and possibly your mobo.[/QUOTE]

Is this an option then? This is the difference between a bad thermal grease, application of Dellienware and my application of liquid ultra on i7-4930mx With 4x42. View attachment 113081 [

 

Oh I'm not disputing CLU gives good results (I'm using it myself), I just don't want people reading this forum to dive in head on without understanding what they're getting themselves into.

 

As I said before, I was very pleased with also gelid extreme. Not so happy with Noctua NT-H1 and Arctic Silver 5. Noctua NT-H1 has been very much praised, but it is too bad after some period of use. I hope someone tries out gelid extreme and that we see some temperature results. It may be what it takes, if they will overclock a bit

 

I've found all the other pastes don't last very long under extreme conditions too. As long as your heatsink is copper and you have a steady hand painting the metal on I consider it safe.
Since moving from IC7 I've gone from 4.3ghz to 4.6ghz and have kept the same temps. Love the stuff.

 

Yes maybe a painting course had been the thing for everyone? :laugh: