Official ICD7 Thread

Note

Thermal compounds are are one on one equivalent to lapping compounds avoid moving sink while mounted under pressure or overworking compound on application or removal,



From MSDS Product Identification
Arctic Alumina Premium Thermal Compound
Product Code: AA
SECTION 2: COMPOSITION/INFORMATION ON INGREDIENTS
Product Ingredient Information CAS No.
Aluminum Oxide 1344-28-1
Boron Nitride 10043-11-5
Polyol Ester Non-hazardous
Polyether Glycol Non-hazardous

 

Alright, so its more or less the natural property of the paste, I don't have a bad batch or anything right?

 

Thats why diamond powder is used in knife sharpeners, diamonds are harder than steel(and nickel), therefore they scratch the surface.

 

Most pastes on the market are abrasive as they contain abrasive particles (Harder than the contact material)

Abrasives need to move in order to be abrasive.

Users should be conscious of the fact and that actions like rotating the sink under pressure with a compound in place is called lapping.

Concentrated rubbing with a cloth is called polishing.

My daughter won first place in the school science fair this year, her demonstration was how household polishing compounds could be used as thermal compounds and how thermal compounds could be used as a polishing compound.

This is basic stuff, I have used toothpaste to polish copper and my daughter tested it as a thermal compound with results similar to a stock compound.

Wear when observed is more an observation of the users handling habits of a compound rather than the compound itself.

 

Same as aluminum oxide.

As a general rule gray pastes are composed of aluminum oxide are highly abrasive as diamond.

Zinc oxide (Stock, white compounds) are less abrasive but have less thermal performance. If you are constantly taking the sink on and off a white paste might be a better choice.

To be fair thermal compounds are designed to be installed then left alone, so works fine, and those that change their paste multiple times should anticipate wear and tear as a matter of course.

 

I was in no way speaking against the compound, I'm using it now and love it. I was just saying that common sense dictates that diamond would easily scratch a nickel plated cpu. I personally don't care if it does scratch it as I don't plan on reselling it ever and the cooling properties are superb. Thanks for a great product!

 

Chill out guys haha. I posted to see if anyone else had any experience with this, and if it had happened to others.

I have only applied ICD-7 once, and that was when I first bought it back in March. Before that I had only changed the TIM in my laptop when I first got it, I took off the stock and put on some Shin-Etsu X-23. In any case, the ICD-7 is terrific stuff, as I said before, I've seen a good 4-5C reduction in temps under load, running LinX overnight.

Anyway, chill out. All I was asking is if anyone else had any problems with the scuffing as I did. The CPU was not damaged and it worked flawlessly, so I sold it. I know that diamonds are the hardest material known to man, and I know how abrasives work, as you yourself said there needs to be friction for it to work. I can't see how I could've generated that much friction when there are 8 screws holding my heat-sink in place. Anyway, I'm still using the stuff, I just wanted to know if anyone had a similar experience.

 

We understand that but sometimes the message becomes misconstrued and we like to make sure some processes are clearly understood before they become embedded as a urban myth of the internet.

 

The mere act of fastening the HS will cause the friction. As you squeeze out the excess TIM it has to move out across the surfaces.................

 

I'm with you.

What was that old saying -

"Diamonds are a core's best friend"?

(Well, maybe not.)

One could say that the difference between "scratching" and "lapping" is one of intent.


Seriously, if one feels that they need to replace the TIM yearly and they do it properly, the "scratching" by any TIM probably wouldn't make any appreciable difference for hundreds or even thousands of applications.

 

The compound has been on the market for 3 years with not one instance of scratching, polishing, scuffing till this spring.

We investigated it by comparing 7 different compounds on their ability polish an oxidized copperplate timed with 20 seconds of constant polishing with a cloth. They were all about the same with the exception of the zinc oxide or stock paste which still polished but to a lessor amount.

I believe the issue arises when people compare out of context, causally comparing the compound against itself rather than against the background of the array of the compounds available.

wear is a problem for the overclockers, some who change their heat sink mount more often then they change their underwear. While high profile they really compose only about 1% of the market. Most users replace their mount then forget about it.

An aspect that many overlook is reliability, probably more important than performance. I have a guy on xtremesystems with a 20 system farm, 2 years on the installed mounts with no change in thermal performance and has not had to reduce his overclock in the period. And from the whitebox and SI's I have been getting reports of fewer returns due to thermal failures. Absolute performance is great but if it only lasts a couple of months what's the value in it?

 

i been using ICD 7 like 2 years already and i can say is the best thermal paste

but they should care more about quality control because one time i brake the corner of my i7 920xm due to bigger diamond particles on the paste

i confirm this the second time i have a problem with my i7 980x because the particles make a big fissure on my core and my lapped water-block

i keep using the stuff because the great results on temps but they need to implement better quality control on the product

 

After I have been singing praises about this paste I have finally been bummed out. When I got the tube in the mail I though, by looking at the plunger, that I would easily get ~10 applications. I was wrong. I only got 3 applications out of a full tube of ICD7. Wow, that plunger is very deceptive and leaves me feeling a little cheated. Oh well, next time I guess I will go with a larger tube. I love the paste, but it sure didn't go very far.

 

three to 4 apps on full 30mm IHS die 1/2 cc volume same volume as a tube of AS5 for comparison. ICD is more viscous and has a higher bulk loading per OEM spec for long term reliability.

Basic material costs are high for the diamond, might want to try our new ceramic compound we are releasing in two weeks 4 grams for $5.00 retail if diamond is to pricey for you.

 

Thanks for the response and the information. How many applications roughly would I get out of a larger tube? I go to school right now for game development and when my classmates see these temps I'm sure they will want to get their hands on some for their machines.

Thanks,
Chase

 

ICD24 is about 3.4 times as much as ICD7,

so, if ICD7 gives 3 to 4 applications, then ICD24 would be about 10 to 13+ apps.

At least, that's with my math skills.

 

Maybe I should refine my question. How many applications could I get on a laptop vs desktop with the IDC24? I don't know the different die sizes off the top of my head.

 

You have to adjust/ratio for die/IHS size

The 10 to to 13 number would be based on a full 30mm X 30mm contact area.

if the contact area is half that on a bare core, use half as much and the number of apps would double

 

I have also found if your CPU really heats up well you can use even less. Mine went to 99C and the goop spread out real thin between the HS and CPU. It ended up placing a huge amount to the sides of the die.........

 

I would lean towards the high side in amount as the excess is just squished out and under heat and pressure the compound will resolve down to your ultimate Bond line thickness (BLT) which is your average particles size.

Experienced users get a good feel for amount by doing frequent installs and know the margins

Too little and you may end up redoing your application - Not worth being frugal IMO in this situation. For the pennies saved taking it apart twice and reinstalling for those who do it infrequently a little overkill is best.

Diamond is an electrical insulator so no worry about shorting any components.

 

I am commenting not to tell people to use less but that if the cpu really heats up the BLT is alot thinner.I had used it on the P8400 and P9600 but these never really got that hot. The x9100 @ 3.7 GHz heats up faster than pushing in a cigarette lighter in your car. This extreme heat without a doubt makes the BLT thinner. When I pulled it appart it was as thin as any AS5 application I'd ever used before, with the other cooler CPU's there actually was a fairly thick ICD film over the CPU after quite a long usage period.

You are better off with excess as when the HS and CPU flex from heat the TIM actually can move slightly in and out from between them. If you have an excess amount on the outer edge it draws TIM in and out. Without that excess it will bring AIR in and out, not a good thing. Agreed these are miniscule amounts but even that can affect thermal transfer..........

 

I'm curious -

is the thickness of the BLT of importance or is it more important that the transfer of heat to the heat sink is as effective as possible?

I would think (just thinking) that as long as one has a good, clean transfer and the HS can then be as efficient as possible, then you've got close to the best possible result.

If the CPU and the HS were perfectly smooth and had a clean, complete area where they were touching, wouldn't that be as near a perfect transfer as possible?

Just thinking out loud.

 

The most important is full contact between the two surfaces. area's unmated will not transfer any heat at all. You then of course want to air pockets etc. as this then becomes that much less of a contact area to transfer heat. You also want as thin a BLT as possible while maintaining 100% coverage.

While ICD in a thick BLT will have less of a heat loss transfer but the thinner the less resistance to the heat transfer. Other higher resistance TIM's rely more on a thin BLT though and even with a thick BLT ICD will most times have as good or better heat transfer capabilities than a good silver TIM with a thin BLT.

 

Good explanation

maybe if you lapped sink and IHS together with a lapping compound to a .0000002 tol. you wouldn't need thermal compound


36 original Johansson gauge blocks wrung together, held together solely by the adhesion of their extremely flat surfaces. 1907.

 

Going over the data as I am going to be wrapping this up and posting a final report in a week or two. This is a little long in the tooth the last 20 free tubes Of IC Diamond only 4 posted results. For those that did not post per gentleman's agreement they will posted and immortalized on the "Wall of Shame" for their failure.

In any event below chart is not final as all results are lumped together, GPU/CPU will be broken out on final.

What is interesting is I would categorize any improvements 10 C and over as catastrophic failure of the stock compound. I count 12 out of 42 roughly 25%

That's a real high number vs the standard PC which are more like 1-2%.

This would translate to early component failure, reduced life expectancy, increased returns/costs of owning a laptop etc.

You would think the manufacturers would pay a little more attention to a detail like thermal compound. But then again they would sell fewer laptops

 

Throw me a free tube and I'll test it however you like!

 

that's what 17 other people said.
Time to wrap it up

 

Only 4 result returns, I'd break too many forum rules posting my pressent thoughts. All I can say, and not get banned for life, is "Come on people, at least be good to your word".

Personally I was more than happy to pay for my own and place a testimonial as others have as well. ICD is the best TIM out there right now.

Stock TIMs have never been good. OEMs typically just put into it what they have too only. Since these are equiped with such an under rated CPU to what the system can handle I am sure there was no eal concern here. With the P8400 and even the stock TIM I was no where near the CPUs rated levels and with ICD went well below that.

This is probably why Gateway didn't place more of a cpu in these as they'd have to take more care. Which they probably did with the Q9000 variations. I am sure their thought was if we upgrade the CPUs we would have to manually address the TIM with care which is the norm for the proccess. I'd agree it would be nice for the manufacturer to have cared a bit more but for an economy system it just was not a major enough of an issue.

Being as this is coming to a close let me be the first to thank you for your product and for your work here. Without all the accumulated knowledge myself and others may not be enjoying our ICD along with the cooler running CPU/GPU's today............

 

That sucks that people would do that. Good luck with the posting of all of your results.

 

Actually this was a pretty good group and I actually enjoyed the process here, smart people on the forum and I have no complaints. This stuff runs on a kind of a bell curve so in the end it always peters out, goes with the territory I just like to poke the non responders a bit, sometimes they even jump.

Everybody should be thanking Capper5016 as he initiated the whole project here then disappeared?

This is the end of the giveaway part, final report of all the tests in a week or so. I will update and comment from time to time after the final.

One thing from a field test perspective is I will revisit to check long term reliability. Notebooks, game stations are the most thermally challenging applications going. The overclocking PC guys Change their compounds more often than they do their underwear. You notebook users because it is such a PITA only do it because you have to. It will be a good field test on long term reliability which from my perspective is really more important than actual performance numbers.

 

Well I do have ICD on one of my notebooks and half of my desktop(GTX 465) if you ever need any info from me!

 

I'll take you up on it.

 

Updating my charts this weekend and hit the 500+ user tests milestone. Raw data chart, haven't broken out he individual competition compounds yet, getting a little crowded

 

I just want to say thanks again for the free tube of IC Diamond. I have purchased another 24 carat tube for a desktop I am building . Also applied some to my friends 9800 GTX and temps dropped almost 10c over the stock compound.

 

xxERIKxx, Always appreciate the feedback. especially when it's positive

 

Will IC7 work in laptops that don't use cooling pads and simply thermal paste alone e.g. Dell Studio XPS 16, or would would the additional space create a problem by perhaps touching something else?

 

Yes, the ICD replaces the "simple thermal paste".

Remove (clean off) the "simple thermal paste" and the ICD now fills the space you've created between the heat source and the heat sink.

It should fill the space better, conduct the heat better, and effectively allow the heat sink to cool your system better.

At least, that's what it appears to have done for me.

 

Indeed, it does seem that most of temperature drop comes from the poorly added stock thermal compound as my GPU has run fairly cool from the start, while others have had severe problems with the heat. The temperature drop on my GPU after adding the ICD7 was only around 5 degrees, however this has allowed me to OC the GPU a bit more, which is nice of course..

Thank you for the excellent TIM, it was really easy to add and is working nicely both for the CPU and GPU. However, the one tube was just not enough for adding TIM on CPU, Northbridge and GPU + memories, but luckily ordered 2 tubes, just to be sure..

 

Can anyone give a recommended application size for a core i5 mobile processor? Is there a formula to determine what volume to apply for a given surface area?

 

The common guideline is to use a "pea-sized" amount of ICD, which was at least what I used for X9100 CPU, and that seemed to be working all right for me...

 

is there any other place in UK that i can order some IDC7 except Chiled PC since there is no left in stock.
I should change my laptop thermal paste after 2 years of use and idle temps pf CPU about 45-50C this summer.

 

Check EBAY, I have a guy there in the UK with a load of tubes.

Shipping Chiled PC and they should have some the end of the week also added two new UK resellers so next week you will have more options

 

thanks, i will wail till IDC will arrive at chilled PC i have plenty of time. lets lope IDC24 wont go out of stock very fast planing to get 2 of them to cause i want to do chopper mod to chipset and gpu ram or i will look for a good thermal pad if i wont do ti.

 

what is a chopper mod?

 

he means "copper" mod.

if you still dont know what this is - copper shim is sandwiched between two layers of TIM (AS5/ICD7) and placed between heatsink and northbridge to maximise heat transfer; essentially an upgrade to the stock thermal pad.

 

I know what a copper mod is. I thought had some new sweetness.

 

Does anyone know where I can buy ICD7 in Calgary, Alberta?

 

Most big stores are in Vancouver and not many sell it. The only store that I'm aware that has it in stock in Canada is Hardware Canucks, which is an online store.

 

How do we let the compound cure? Do we leave the machine off, have it constantly on?

 

Here are my results:

Code:

Before
Idle		Max
---------------------------
C0	44		74
C1	39		72
GPU	48		75


After
Idle		Max
---------------------------
C0	42		75
C1	40		71
GPU	48		73

Here's a visual indication of how much paste I used:

I know that I didn't apply 50 PSI, since it was too awkward to do so, but I was expecting a greater drop.

EDIT: Second attempt. Applied ~25% more to each.

Code:

After
Idle		Max
---------------------------
C0	41(-2)		71(-3)
C1	38(-1)		67(-5)
GPU	45(-3)		71(-4)