Sign in with
Sign up | Sign in
Your question

How pure should isopropyl alcohol be for cleaning the HS/CPU

Last response: in CPUs
Share
September 30, 2006 9:57:20 PM

Im having trouble finding pure isopropyl alcohol (can only find 70%), does anyone think I will notice much of a difference in performance vs something in the 90-99% range? I can order off the internet, but it would be nice to find something local instead. Thanks,

Scribs
October 1, 2006 1:05:40 AM

Quote:
Im having trouble finding pure isopropyl alcohol (can only find 70%), does anyone think I will notice much of a difference in performance vs something in the 90-99% range?


Well, as long as you don't drink it! :wink:

http://en.wikipedia.org/wiki/Rubbing_alcohol

Sure! "Wikied"! :lol: 

Whatever concentration above 70% is good enough, for the purpose; however, since it contains water, I'd go for specific cleaning products, such as "Tim Clean", from Akasa.

(This was a very sui generis thread, indeed!)


Cheers!
October 1, 2006 1:32:19 AM

Scribbs,
I'm a Pharmacy Tech where I live, so I know about alcohol. There is NO difference which type of alcohol you use, because either one will do the job. It's the alcohol BASE content that removes the residual thermal paste..........the 99% alcohol will just evaporate quicker. If I were you, I would use the 70 % isopropyl stuff. That way, if you have to rub a bit to remove the leftover GUNK from the heat sink and cpu cover, it won't dry up right away.
Hope this helps

Antec P180 Performance Case
SeaSonic S12 600w PSU
Asus A8N32 SLI S-939 (bios 1103 V02.58 )
RealTek 97 Onboard
ATHLON 64 X2 4800+ Toledo Core (AMD v 1.3.1.0/MS hotfix/ Dual Opt)
2 G's of Corsair 3500LL Pro @ 400Mhz 2-3-2-6-1T
2- BFG 7900 GT OC in SLI (NV 91.31)
WD RAPTOR 74.3 G's / XP Home / SP2 & Apps
Maxtor SATA II 250 G's /Gaming / Movies / MP3's
Maxtor SATA II 250 G's /backup (unplugged)
Sony CDrom 52X
Plextor 708-A DVD/CD rom
Razer DiamondBack Optical mouse
Logitech Z-5500 Dolby Digital 5.1 THX 500w
Related resources
October 1, 2006 2:23:27 AM

Most all major drug stores in the USA carry 91% pure isopropyl alcohol. Use it, it's cheap and works like a dream. I've seen numerous concoctions recommended on this forum, many of them petroleum based. They should be avoided as they will leave a filmy residue that reduces the effectiveness of the thermal paste/compound.
October 1, 2006 3:30:01 AM

Quote:
If you do not have AS5, then some vinegar or something that can remove metalic oxide.


Have you used vinegar as a final cleaning agent before putting on the thermal grease? Vinegar contains a reasonably high concentration of polar impurities, some that are non-volatile. So I'd expect it to leave residuals. Who knows, maybe a thin film of polar gunk is a good surface prep for a HS grease?
October 1, 2006 4:48:22 AM

Wow. I haven't built my first system yet but I am going to before the end of the year if not the end of October. After reading this though why would the CPU need to be cleaned this efficently and what would be the best method to do so? I would think an out of the box CPU wouldn't need anything other then a light wipedown or are we talking about reusing a CPU or possibly changing a heatsink? If someone could help explain please let me know. I was just going to wipe the new CPU down with alcohol at best and that is only if the directions said so.
October 1, 2006 12:22:50 PM

Quote:
If you do not have AS5, then some vinegar or something that can remove metalic oxide.


Have you used vinegar as a final cleaning agent before putting on the thermal grease? Vinegar contains a reasonably high concentration of polar impurities, some that are non-volatile. So I'd expect it to leave residuals. Who knows, maybe a thin film of polar gunk is a good surface prep for a HS grease?

No I haven't actually, I suggest it because acetic acid is good at removing oxidation from metal surfaces. Better than that is ascorbic acid, but I would not recommend orange juice on your CPU :)  ....

For sure, unless your CPU has put in an order for a screwdriver. Regardless, an acid like acetic will react with an oxidized metal surface and in fact will cause some oxidation itself. But again it's possible that the residuals in vinegar could form a sort of passivating layer on the IHS. There are many different kinds/grades of vinegar that are made from various processes so one would expect different levels and chemistries of residuals. Grocery store vinegar is about 95% water. Check this out, particularly this:

"WHAT IS VINEGAR?
Vinegar comes from the bacterial oxidation of alcoholic drinks such as wine and cider, and the formation of ethanoic (acetic) acid gives vinegar its characteristic smell. However, ethanoic acid is by no means the only acid in vinegar. It is not even the most predominant acid - it is simply the one, which, owing to its volatility, makes its presence known! Many other acids are also present, such as tartaric, malic, lactic, citric and succinic acids. There are probably many more besides, but since these latter acids are non-volatile they have no smell, yet they still contribute to the overall acidity of the vinegar. You will need to look up the structures of these non-volatile acids."

Since vinegar is distilled, many of these other compounds are reduced or removed entirely but like most commodity natural products, it's far from pure.

Quote:
[AS5 works well becuase it is a colloidal suspension of silver oxide particles, which has a good thermal conductivity but poor electrical conductivity. However, most metal oxides (paricularly CuO or CuO2) are bad thermal conductors so it is best to have a well polished, oxide free surface.


I agree completely with that logic and have long thought that lapping the IHS can help lower CPU temps because of this. My last step before applying the bead of AS5 is to take a small amount of AS5 on a lint-free cloth, dilute it with ethanol and polish the IHS and HS surfaces. I might have to try a vinegar step right before the alcohol/AS5 step next time.

Edit: typo and grammar
October 1, 2006 12:26:24 PM

Quote:
Wow. I haven't built my first system yet but I am going to before the end of the year if not the end of October. After reading this though why would the CPU need to be cleaned this efficently and what would be the best method to do so? I would think an out of the box CPU wouldn't need anything other then a light wipedown or are we talking about reusing a CPU or possibly changing a heatsink? If someone could help explain please let me know. I was just going to wipe the new CPU down with alcohol at best and that is only if the directions said so.


The recommendations in this thread for sure apply to removing existing thermal grease, but also for a new CPU. Look at Jack's procedure above and the discussion of what is accomplished by the cleaning. You also might want to search out the benefits of polishing the IHS. I've seen it discussed on threads here on Toms and on other enthusiast sites but do not have the links stored on this laptop, only on my desktop back home.
October 1, 2006 1:13:34 PM

Quote:
Wow. I haven't built my first system yet but I am going to before the end of the year if not the end of October. After reading this though why would the CPU need to be cleaned this efficently and what would be the best method to do so?


http://www.arcticsilver.com/arctic_silver_instructions....

Well you could follow the advice and instructions posted by the "experts" here ( many of which sound like they are making a salad or preparing a vaginal douche) or you could simply follow the manufacturers directions posted above.

Call me crazy, but I actually think Arctic Silver knows what cleaning and application methods are the most effective. Unless you tried all these suggestions, with all combinations of CPU's and HSFs, and tested before and after results, how would you know which recommendation to use? BTW, note the consistent reference to isopropyl alcohol in the AS5 cleaning instructions.

Let the flaming begin!
October 1, 2006 1:52:03 PM

Thanks for all the replies guys, Im glad to see this spawned such a fruitful discussion heh. I was figuring that 70% would maybe take a little more elbow grease, and possibly a little airflow to dry off fully, but that it wouldnt be too much worse.

On the topic of preparing a new HS/cpu, I think it also tends to depend on the HS too. Some HS come in a much smoother state than others. There are some that come with a real mirror finish, in which polishing probably would only muck it up (as they are better than you would probably do), while others come in a fairly smooth state (which might benefit from polishing). Unless you are super hardcore, I think maybe just a slight rub down with isopropyl alcohol for both the HS and CPU would do.

Also note however that some HS come with thermal paste pre-applied. In that case, if you dont have AS5 or something to replace it you should probably not remove it, and if you do have AS5 you definitly need to replace it.
October 1, 2006 1:52:33 PM

Quote:
You also might want to search out the benefits of polishing the IHS. I've seen it discussed on threads here on Toms and on other enthusiast sites but do not have the links stored on this laptop, only on my desktop back home.

Well, there are two ways to take the word 'polishing'.
One is lapping to a high degree of flatness & smoothness then polishing it to make it look like a mirror, voiding the warranty in the process.
The virtually eliminate the need for thermal paste providing the two surfaces have been treated and requires the smoothest thermal paste on market (AS Ceramique) at minimal amount to achieve maximum conduction.

That's the plan I like. Having been a professional metallographer, flatness and mirror finish are my mantra. Voiding warranties are part and parcel in our search for overclocking perfection. Om mani padme hom.

Quote:
[The other polishing is like wax polishing... an layer of insulation... :lol: 


I have not used that one before! Thou shalt not wax thy heatsink (except HSs made by Maserati)!!!
October 1, 2006 3:47:39 PM

Quote:
That's the plan I like. Having been a professional metallographer, flatness and mirror finish are my mantra. Voiding warranties are part and parcel in our search for overclocking perfection. Om mani padme hom.


Well then, here you have it, from Akasa to Cooler Master, from tooth paste to 'vegemite'. :lol: 

http://www.dansdata.com/goop.htm

By the way, nice buddist mantra you've got there! :wink:


Cheers!
October 1, 2006 4:05:25 PM

Quote:
Hehe, crazy Dan. He writes for AtomicMPC and lives in the land of Aussie. :D 
One of the people on my list, haven't mentioned him here for a long arse time... (kinda forgot about him... :oops:  )


He's done a nice & funny job, that's for sure! :lol: 


Cheers!
October 1, 2006 6:39:16 PM

You guys are making such a big deal about this, its silly.

Water works, with some elbow grease, any non-polar (organic) solvent should work out ok. 70% IPA is fine, ethyl alcohol should be good, i wouldnt go much crazier than that. I think acetone is just overkill, im sorry, it doesnt take much to remove a TIM with simple ipa.

About removing oxides, this is kind of stupid, im sorry, because its just going to oxidize again. So unless you plan on doing this over and over again, it seems useless.
October 1, 2006 7:34:34 PM

Quote:
You guys are making such a big deal about this, its silly.

Water works, with some elbow grease, any non-polar (organic) solvent should work out ok. 70% IPA is fine, ethyl alcohol should be good, i wouldnt go much crazier than that. I think acetone is just overkill, im sorry, it doesnt take much to remove a TIM with simple ipa.

About removing oxides, this is kind of stupid, im sorry, because its just going to oxidize again. So unless you plan on doing this over and over again, it seems useless.


It takes a few days to oxidize more than a few monolayers of oxide, so unless it takes you 48 hours to seat an HSF, then yeah it makes a big deal.

Unless your talking about aluminum
October 1, 2006 7:59:45 PM

Quote:
You guys are making such a big deal about this, its silly.

Water works, with some elbow grease, any non-polar (organic) solvent should work out ok. 70% IPA is fine, ethyl alcohol should be good, i wouldnt go much crazier than that. I think acetone is just overkill, im sorry, it doesnt take much to remove a TIM with simple ipa.

About removing oxides, this is kind of stupid, im sorry, because its just going to oxidize again. So unless you plan on doing this over and over again, it seems useless.


This actually brings up an interesting little point. I read through a thread in another column (awhile back) a very interesting discussion on aluminum and copper. The poster, while very knowledgeable, never addressed the problem of aluminum oxide.

My little library is woefully thin in the material sciences department, and I’ve often wondered if the "standard" thermal properties generally listed for Al was for oxidized AL or pure Al. The reason this is interesting is while pure aluminum is one of the more ductile metals, with an advantageous heat transfer coefficient, Aluminum Oxide is one of the harder substances known to man. I wouldn’t guess that Al and Al2O3 have the same heat transfer coefficient.

For welding aluminum, Al2O3 plays a very significant role. To successfully weld aluminum, the oxide layer must first be removed (usually by solvent) prior to welding. Failure to due so will result in failure to penetrate the metal thus a failed weld. This being the case, I would guess AL2O3 also impacts heat transfer properties of Al. As such, removing the oxide layer from Al may not be so "stupid"

Why? Simple. In order to oxidize, a material must be exposed to oxygen. No exposure = no oxidation. With thermal grease fillingt eh gap between the HSF and CPU heat spreader, there is no way for the oxygen to reach or react with wither the HSF face or CPU heat spreader, so its not going to "oxidize again".

Additionally, certain thermal greases I have used (I believe the compound packaged with Cooler Master was one of these) actually seemed to have some from of mild solvent mixed in it.

Clearly, this could have an impact on the efficiency of a either a pure Al or AlCu HSF.
October 1, 2006 8:13:18 PM

Quote:


Even aluminum self oxidized to Al2O3, which is highly passivating so initially the oxide layer is very thin. It takes several days for oxidation to build up thickness on solid surfaces.

Si is a good example, if you strip the oxide with an HF etch, then measure oxidation over time it takes about 6 hours before any detectable amount of oxygen appears, and takes a few days before the oxide growth 'saturates'. Al2O3 is similar behavior --- Alumina though has decent thermal characteristics compard to CuO, Cu2O, and SiO2.

It's the copper oxide that is the most annoying, sometimes it is so thick I cannot get rid of it all without baking it off.

Jack


I wrote the response below before reading your response. For the purposes I am more familiar with in Aluminum (structural) the oxide layer may be more critical than when it is used as a heat sink. For welding, the only book I have that addresses aluminum in depth gives a time of "about" 15 minutes to the formation of the first layer of oxide. Additionally, it states because the oxide is so tough, subsequent layer formation adds little protection to the original layer, and is not of significant consequence.

So it may be that for the purposes of heat transfer Al2O3 is far less critical than it is in welding.

In welding, what happens is the Al2O3 actually forms a "skin" protecting the Al underneath. The base Al will actually melt while the skin will form an impermeable layer blocking the fusion of the electrode wire with the base metal.

The question is, why does the Al2O3 "skin". Is it due to a higher melting point, or a different coefficient of heat transfer. In the case of the latter, this would certainly seem to have an impact on the optimal use of Al (or 6061 T6 Al in most cases) for HSF
October 1, 2006 11:10:48 PM

Quote:
The question is, why does the Al2O3 "skin". Is it due to a higher melting point, or a different coefficient of heat transfer. In the case of the latter, this would certainly seem to have an impact on the optimal use of Al (or 6061 T6 Al in most cases) for HSF


Although this will not answer both of your questions, it's a good starting point for the search of structural occurrences at the interface boundary, nevertheless:

Quote:
High-resolution transmission electron microscopy observations on the structure of the interphase boundary between crystalline Al and amorphous Al2O3 coating reveal that an interfacial melting transition of Al occurs at 833 K, which is distinctly lower than the bulk melting point of Al. The crystalline lattice planes of Al near the interface bend or small segments of crystalline Al deviated from the matrix Al grains are formed. Stand-off dislocations formed at the interphase boundary are also observed. The amorphous Al2O3 coating plays an important role in retaining the evidence for structural transition at high temperature to room temperature, which makes it possible to make experimental observations.
[/b]

(http://www.iop.org/EJ/abstract/0953-8984/14/8/316/; just the abstract, since I'm not a subscriber).


Cheers!
October 1, 2006 11:52:04 PM

Quote:
The question is, why does the Al2O3 "skin". Is it due to a higher melting point, or a different coefficient of heat transfer. In the case of the latter, this would certainly seem to have an impact on the optimal use of Al (or 6061 T6 Al in most cases) for HSF


Although this will not answer both of your questions, it's a good starting point for the search of structural occurrences at the interface boundary, nevertheless:

Quote:
High-resolution transmission electron microscopy observations on the structure of the interphase boundary between crystalline Al and amorphous Al2O3 coating reveal that an interfacial melting transition of Al occurs at 833 K, which is distinctly lower than the bulk melting point of Al. The crystalline lattice planes of Al near the interface bend or small segments of crystalline Al deviated from the matrix Al grains are formed. Stand-off dislocations formed at the interphase boundary are also observed. The amorphous Al2O3 coating plays an important role in retaining the evidence for structural transition at high temperature to room temperature, which makes it possible to make experimental observations.
[/b]

(http://www.iop.org/EJ/abstract/0953-8984/14/8/316/; just the abstract, since I'm not a subscriber).


Cheers!

Interesting, but if I am reading that correctly, that is dealing specifically with the al-al2o3 boundry layer, and not the entire crossectional area of the al2o3 layer. Though not specifically stated, as I interpret that, (by "ommision") due to the lower melting point at the transitional layer, both the Al and al2o3 are remaining solid after the transitional layer material has melted.
October 2, 2006 12:18:10 AM

What began with a simple question by the OP, has degenerated into a gross display of pseudo intellectual masturbation. Give it up guys, nobody is impressed. :roll:
October 2, 2006 12:21:32 AM

Quote:
What began with a simple question by the OP, has degenerated into a gross display of pseudo intellectual masturbation. Give it up guys, nobody is impressed. :roll:


Not really. Imagine if you could increase you HSFs efficiency signifcantly simply by doing a quick phosphoric acid etch to remove a layer of oxidation.
October 2, 2006 12:50:46 AM

Quote:
What began with a simple question by the OP, has degenerated into a gross display of pseudo intellectual masturbation. Give it up guys, nobody is impressed. :roll:


Not really. Imagine if you could increase you HSFs efficiency signifcantly simply by doing a quick phosphoric acid etch to remove a layer of oxidation.

That's not the question he asked. If you have some revolutionary method for improving a HSF's efficiency, wouldn’t that be a topic for another thread?
October 2, 2006 1:07:29 AM

Quote:
What began with a simple question by the OP, has degenerated into a gross display of pseudo intellectual masturbation. Give it up guys, nobody is impressed. :roll:


Not really. Imagine if you could increase you HSFs efficiency signifcantly simply by doing a quick phosphoric acid etch to remove a layer of oxidation.

That's not the question he asked. If you have some revolutionary method for improving a HSF's efficiency, wouldn’t that be a topic for another thread?

Well, thats the turn the thread has taken. As to the OPs original question, that could have been answered with a simple yes/no. So, if the thread has gone on as long as it has, it would be indicative that it has taken said turn. As always, no ones forcing you to watch the thread, and if you feel there is no useful info left here for you, or your not inetersted in the subject, then simply dont read the thread. Adding a "this is useless" post is --------usless. This post itself, stating that a "this is usless" post is usless is-----useless.

You and I have now added 4 usless posts to this thread. Are you going to make it 5?
October 2, 2006 1:28:11 AM

Quote:
Interesting, but if I am reading that correctly, that is dealing specifically with the al-al2o3 boundry layer, and not the entire crossectional area of the al2o3 layer. Though not specifically stated, as I interpret that, (by "ommision") due to the lower melting point at the transitional layer, both the Al and al2o3 are remaining solid after the transitional layer material has melted.


I'm no expert on these matters (I'm stressing this point); anyway, since Al bulk melting point is about 933.47 K, that would be correct (although I don't know the Al2O3 melting point...); as for a cross-sectional area characteristics of the AL2O3... I only hope you can bring up some interesting data on this, since you seem pretty knowledgeable on the subject (pertinent to the topic, of course! :wink: ).


Cheers!
October 2, 2006 1:49:28 AM

Quote:
Nice pull, however -- for cleaning up a surface distilled vinegard should work without any major headaches.


Yea, you're talking short term exposure, so pitting shouldn't be a problem at all.
October 2, 2006 1:49:31 AM

Quote:
Interesting, but if I am reading that correctly, that is dealing specifically with the al-al2o3 boundry layer, and not the entire crossectional area of the al2o3 layer. Though not specifically stated, as I interpret that, (by "ommision") due to the lower melting point at the transitional layer, both the Al and al2o3 are remaining solid after the transitional layer material has melted.


I'm no expert on these matters (I'm stressing this point); anyway, since Al bulk melting point is about 933.47 K, that would be correct (although I don't know the Al2O3 melting point...); as for a cross-sectional area characteristics of the AL2O3... I only hope you can bring up some interesting data on this, since you seem pretty knowledgeable on the subject (pertinent to the topic, of course! :wink: ).


Cheers!

Im not even remotely an expert. Im not even a novice. My area of study in my major (MechE) concentrated on machine design/kinematics.

This falls clearly into the area of material science.

Why I got interested was what the effects (if any) al203 has on heat transfer in CPU HSFs. Working with aluminum (i have a small milling machine) for my own personal projects, I have many time wanted to weld aluminum. I have always been very wary though, due to the problems I have read about, my education and a mishap investigation I did once involving failed aluminum welds. The welds were (from visual examintation) perfect. Continuous, consistantly even flow, no gapping, pitting and absolute minimal spall. Unfortunately, failure of the welds also reveal absolutely no penetration of the base metal. Esentially the weld was nothing more than a "solder" or brazed joint. This was caused by failure to remove the oxide layer prior to welding. Because the welds easily passed visual inspection, the only way to have known the welds were faulty prior to placing the component in service would have been destructive testing.

Another interesting little tidbit that adds to my curiosity about aluminum oxide is that as one of the harder sustances known to man it is also commonly used as an abrasive media. "Sand" paper is actualy often aluminum oxide paper.

The fact that al2o3 is so hard, forms so fast and resists the high temps of welding has often made me wonder about its effects on heat transfer as aluminum is such a popular material for radiators, thermocouples and heat sinks.


So, im am curious, but have no answers.

In fact, with Jacks backround in chemistry, I was hoping he would have a quick answer.
October 2, 2006 2:18:41 AM

Quote:
Just quickly, I have had a need in the past to heat and cool silicon directly. This is tricky as silicon is not a great conductor of electricity and I could not use thermal paste when making the thermal contact. Just from past experience getting a clean, oxide free surface was very critical. Thus I advocate a clean surface with any HSF installation.

Metals are good thermal conductors, metal oxides and organic solids are not. The steps I put above should help remove most of that gunk that builds up by sitting around, handling, and such. Some HSF come in polystyrene and I have data that shows that such packaging can find it's way onto the components.... acetone will remove that for example.

Clueless brought up a good point on acetic acid. Acetic acid is a weak organic acid, which can make some metal oxides soluble -- CuO being one, he points out that that run of the mill household vinegar may contain other less desirable contaminants, and he is correct. If I have the resources I will wipe down a Cu surface and run XPS on it to see what is left behind --- if I can get that data I will post it back.


Just a heads up to the non-surface chemists out there... XPS is a technique that is very sensitive to the outer molecular and atomic layers of a surface. Depending on the specifics of the experimental setup, you can localize on the top ~10 to 50 nanometers of the surface. That may sound like a lot of yada-yada BS, but the particulars of surface analysis are not intuitive even to many educated scientists. That tiny bit of surface composition is where alot of the chemical action that matters in areas such as conductivity (thermal, electrical), adhesion, reactivity, etc.

Quote:
Cleaning the CPU surface is just a habitual --- all surfaces clean --- process I use.


I'm a son of a habitual CPU cleaner...

Quote:
Now that this has become a topic, it may be worth looking into other oxide removing solutions. There are several copper or brass cleaning solutions in local HW stores, I will also look at those and see if anything pops out.


That is a very competitive area in the advanced cleaning systems business. Talk about neurotic bosses, I tell ya. Along those lines, you were talking about Si surfaces and cleaning, that's another area with an extensive literature. It's relatively easy to study the surface chemistry of Si so surface analysts use wafers as a substrate often.
October 2, 2006 2:43:42 AM

Quote:
I'm no expert on these matters (I'm stressing this point); anyway, since Al bulk melting point is about 933.47 K, that would be correct (although I don't know the Al2O3 melting point...)


It's about 2000C but varies with crystalline form.

Quote:
Why I got interested was what the effects (if any) al203 has on heat transfer in CPU HSFs.


There are studies out there that I've read years ago but I don't even know if I still have the hard copies. I worked on a team that developed AlN ceramics for high thermal conductivity apps so we had to deal with aluminum oxide layer effects. Suffice to say that the thermal conductivity of aluminum oxide (~18 W/mK) is way below that of Al (~237W/mK). But the aluminum oxide layer is very thin on a HS and although it is a hard substance, it is thin and I'd expect that the typical grease install thins it a bit.

Quote:
Because the welds easily passed visual inspection, the only way to have known the welds were faulty prior to placing the component in service would have been destructive testing.


Are you sure that there are no NDE methods that would work?

Quote:
The fact that al2o3 is so hard, forms so fast and resists the high temps of welding has often made me wonder about its effects on heat transfer as aluminum is such a popular material for radiators, thermocouples and heat sinks.


Again, studies exist but are academic. For real world apps, the oxide is going to be on the metal. You can prepare oxide-free surfaces, laminate them to an oxygen barrier, etc., but within the lifetime of a computer HS, you will get sufficient penetration to compromise the interfacial chemistry. In other words, all of the good oxygen barriers we looked at had other prohibitive problems.

Quote:
In fact, with Jacks backround in chemistry, I was hoping he would have a quick answer.


Let me know if that was useful or if I missed what you are after.
October 2, 2006 2:49:41 AM

Quote:
That's not the question he asked. If you have some revolutionary method for improving a HSF's efficiency, wouldn’t that be a topic for another thread?


Think "incremental" more than "revolutionary".

Quote:
[Adding a "this is useless" post is --------usless. This post itself, stating that a "this is usless" post is usless is-----useless.

You and I have now added 4 usless posts to this thread. Are you going to make it 5?


Welcome to my world, where the useless collides with the sleepless clueless and most observers are thankless.
October 2, 2006 3:23:27 AM

Quote:

It's about 2000C but varies with crystalline form.


Much higher than elemental Als 840'C

Quote:
There are studies out there that I've read years ago but I don't even know if I still have the hard copies. I worked on a team that developed AlN ceramics for high thermal conductivity apps so we had to deal with aluminum oxide layer effects. Suffice to say that the thermal conductivity of aluminum oxide (~18 W/mK) is way below that of Al (~237W/mK). But the aluminum oxide layer is very thin on a HS and although it is a hard substance, it is thin and I'd expect that the typical grease install thins it a bit.


Which leads to a potentially very interesting experiment: CPU HSF efficiency with/without the oxide layer. Even though the layer is so thin, because it has such low thermal conductivity relative to elemental Al (my reference doesn’t have the heat transfer numbers for 6061 T6 either) it must have some impact, even if it is only negligible. This would be a great experiment for the boys at Toms "lab".

Quote:
Are you sure that there are no NDE methods that would work?


Not for that piece at that large (at least at the time in 1999) It was 60 ft long x 6 ft wide and about 1-2 ft thick constructed of minimum 1/2 thick AL (either 6061 or 5053). Again, visual inspection revealed no reason to suspect the welds and they had been signed off by a certified inspector.

Quote:
Again, studies exist but are academic. For real world apps, the oxide is going to be on the metal. You can prepare oxide-free surfaces, laminate them to an oxygen barrier, etc., but within the lifetime of a computer HS, you will get sufficient penetration to compromise the interfacial chemistry. In other words, all of the good oxygen barriers we looked at had other prohibitive problems.


Absolutely. Using any kind of coating adds another layer with different properties. But, with thermal grease acting as a barrier, no al2o3 should form, so if the user removes the oxide layer immediately before installing the heat sink, no layer should form afterwards eliminating any kind of interferance.


Quote:
Let me know if that was useful or if I missed what you are after.


Super info!!!! :D  Thanks very much!!

Now, if we could just get someone with the proper equipment to test the heat transfer of an oxidized al HSF vs a "clean" HSF [/hint]

For both you and Jack, the best aluminum cleaner I’ve found is a mixture of inhibited (phosphoric I believe) acid and fluoride used for cleaning trailers and aircraft, located here:
http://www.baczymes.com/Merchant/merchant.mv?Screen=CTG...
October 2, 2006 4:45:02 AM

Quote:
[...recommend such method to others in this forum, for the sake of a practical ~0.001C temperature decrease.....
.


Well, to acheive a .001'C descrease, I would recomend opening your computer case and blowing on the HSF.

Does this help solve the problem? :wink:

:lol:  :lol:  :lol: 
October 2, 2006 5:16:54 AM

Quote:
Water works, with some elbow grease, any non-polar (organic) solvent should work out ok. 70% IPA is fine, ethyl alcohol should be good, i wouldnt go much crazier than that. I think acetone is just overkill, im sorry, it doesnt take much to remove a TIM with simple ipa.


You're contradicting yourself. You say that a nonpolar organic solvent should work, then you mention water, IPA and ethanol. All of these are considered moderately polar solvents. Nonpolar solvents are those such as hexane or benzene.

Quote:
About removing oxides, this is kind of stupid, im sorry, because its just going to oxidize again. So unless you plan on doing this over and over again, it seems useless.


I'm not certain, really. If you look at the history of HS/ISH prep procedures you'll find quite a range and some of the performance improvement claims are huge. For example, over on XS, I've seen claims that lapping the plating off of the IHS lowered the CPU temp by 10C. That is difficult for me to accept but I have seen measurable improvements by lapping the IHS and also by lapping HS's that have surfaces that aren't smooth. I always bought into the theory that the improved cooling after lapping was due to less volume of air in the IHS/HS gap and by the kind of crush zone you get from two mildly convex surfaces. But since I've never attempted to design an experiment to isolate the mechanism, I've got no physical proof.

Regarding whether or not a surface is going to re-oxidize again, it very well could depend on the specific IHS involved. I've read that many current IHS's are copper plated with nickel or a nickel alloy. A few years ago, I read of aluminum plated IHS's. For sure, an Al-plated IHS will quickly reoxidize, but just to a very shallow depth. And since I've never read an unequivocal thermal grease recipe and haven't tried to determine the chemistry involved, I have no idea if there are any oxidation inhibitors in the mix. If someone like AS has learned to build in an inhibitor that doesn't reduce the thermal conductivity of the grease, then an oxide removal right before grease application could be beneficial.

There are hypotheses of all kinds of phenomena but often no real proofs. That's life. If you believe something strongly and proof is important, go for it. I've gone along pretty closely with the online instructional on the AS site and like Jack am a little bit more into thorough cleaning. Doing ultra-trace analysis, surface analysis and materials characterization for three decades can tend to make you overkill certain situations but tell me this: what harm can it do to get the ISH and HS "too" clean?
October 2, 2006 6:03:17 AM

As you pointed out, alumina is a relatively insoluable crystaline structure. The correct method for removal is via abrasion.
October 2, 2006 6:48:32 AM

Quote:
As you pointed out, alumina is a relatively insoluable crystaline structure. The correct method for removal is via abrasion.


Well, I certainly do not recommend taking sandpaper to your CPU. :) 

The stuff I linked to and a little 0000 steel wool does wonders for al without causing any significant deformation, and without signifcant effort.
October 2, 2006 6:53:54 AM

Just my way off saying wipe it off. That's all it really needs.
October 2, 2006 7:02:53 AM

HCL will that help with native oxide. Planning on lapping my cpu and heatsink, anyone in here know a good procedure, i.e. grit, surface.
October 2, 2006 8:05:33 AM

Quote:
As you pointed out, alumina is a relatively insoluable crystaline structure. The correct method for removal is via abrasion.


Well, I certainly do not recommend taking sandpaper to your CPU. :) 

The stuff I linked to and a little 0000 steel wool does wonders for al without causing any significant deformation, and without signifcant effort.

I did not read that, but if it is extra fine grit I guess no big deal :) 

I should read up some more on this, people are very touchy about this subject obviously :) 

Jack

You dont have to use the steel wool. I started using it way back when. A buddy of mine gave me a 3/4 full gallon of the stuff back in 94/95 to clean the wheels on my truck. The only thing this stuff wouldnt take off all by itself was the baked on brake dust, which is why I started using the steel wool. Otherwise it worked wonders.

The directions are simply (on my gallon dated 10/91) for spray application dilute 50 to one. For diptanks or brush application dilute 60 to one. For vertical sufaces work from the bottom up. Rinse with plenty of fresh water to flood all surfaces.

The stuff really does an amazing job.
October 2, 2006 1:06:11 PM

Quote:
Im not even remotely an expert. Im not even a novice. My area of study in my major (MechE) concentrated on machine design/kinematics.


Comparatively, I'm just a designer. :wink:

Quote:
The fact that al2o3 is so hard, forms so fast and resists the high temps of welding has often made me wonder about its effects on heat transfer as aluminum is such a popular material for radiators, thermocouples and heat sinks.


Maybe this helps a little (it's more about Al2O3 coating techniques than uncoating; there are plenty of good papers on the net; I've got some highly technical & not exclusively related to bulk Al & Al2O3, from MIT, UCL, Aerospace,...; hence, largely off topic):

http://www.nanofilm.com.sg/downloads/case_study_al.pdf


Quote:
So, im am curious, but have no answers.


Same here. :wink:


Cheers!
October 2, 2006 1:36:26 PM

Quote:
the idea of this is to add more surface area to the heat spreader over the bell of the proc. if there is more surface area you can get rid of more heat. one way to add more surface area is by using sand paper on the heat spreader.
if you try this i would recommend that you make sure you have enough thermal grease and rough up the heat sink first and see if it helps. its cheaper to replace that.


I disagree. It's the opposite of adding more surface area. What you'd like are two flat or very slightly convex surfaces and you want a very high degree of smoothness. The idea is to exclude voids that can hold air because air is an excellent insulator. Now, if you could make a perfect fit between rough surfaces, like an interlocking sawtooth, then you'd have more surface area and still exclude air. That might help but I've never heard of anyone trying it.
October 2, 2006 1:57:16 PM

Quote:
It's a virus, a downgrade to this forum.


(So much for exclusive prerogatives.)

*Deleted*


Cheers!
October 2, 2006 8:28:59 PM

Quote:
HCL will that help with native oxide. Planning on lapping my cpu and heatsink, anyone in here know a good procedure, i.e. grit, surface.


There are procedures published all over the place. I've seen posts here on Toms but just now grabbed these with the help of Google:

http://www.computerpoweruser.com/editorial/article.asp?...

http://www.xtremesystems.org/forums/showthread.php?t=11...

http://forum.overclock3d.net/showthread.php?t=4735

OK, so you have some recommendations there... I've been trained as a professional metallographer. Metallographers grind and polish stuff, often to reveal internal microstructure and get light or electron microscopy images. I've taken classes at Metals Park near Cleveland, taken classes at the World Materials Congress, traveled to the development labs at Struers (a world leader in polishing science) etc., etc. My resume regarding metallography is extensive, OK? So trust me when I say this: it takes a VERY skilled person to hand lap something and have it be flat when they are done. So when you learn to lap by the procedures above, pay very close attention to how they tell you to keep the surface flat. What happens is the the outer edge of the HS and IHS will get ground away more than the center. Makes sense, right? Think about it. You end up with a convex surface and your real goal when hand lapping is to reduce deformation uniformly across the surface, then clean it and step down to the next finer grit. If you end up with a slightly convex surface, say out of flat by 25 to 50 microns across the IHS and HS, that's actually OK. It means that the region of the IHS right above the chip itself will get pressed into the HS. A concave prep would be much worse and some are speculating that C2D IHS's are coming out of the factory concave. I'd have to see real data from a profilometer before I'd believe it. You'd think Intel would have a flatness spec since they know all about heat removal, right?
October 2, 2006 8:57:24 PM

Quote:
And that is the reason why this thread is perfectly acceptable to exist in this forum and to talk about the stuff you guys are talking now and I don't mind one bit.

Not pointing my finger at anyone, but I have seen a quite a few threads here starting out as a simple topic of a new product/question then randomly turning into a total nerd-fest because someone can't keep unrelavent, unpractical knowledge to themselves and splur them all onto the threads for half or so pages long. Wasting my time reading and their time typing it out.

Anyway, the point I'm making is with the exception of threads like this which was destined for such talks and I in fact kinda started it.
When it comes to other threads where such extensive theoratical knowledge are not needed, try refrain from posting them and bore people with unrelavent stuff. :wink:

I control my knowledge to know when it's appropriately needed and applies it to the thread and when not need if the thread just wants a simple anwser.

BTW, I'm used to be a 50% scientist, 50% practical guy, but now it's more of a 40%-60% due to the type of work I'm doing now.



Wow. What does it say about someone when they are labled a geek by a computer geek? 8O

:D  :D  :D 
October 2, 2006 9:24:31 PM

Ive never been a fan of recommending lapping to people for the reasons you stated. Ive always believed that for someone who didnt know what they were doing, the potential benifits were far outwieghed by the hazzards, and that people didnt really understand the levels of effort, care and precision actually required.

I know your familiar with this stuff, but for the people who've never seen it, the two samples Ive pictured (brass and 1018 steel @ .5inx.5in) were polished for microscopic grain structure analysis in my metalurgy class about 15 years ago. The brass sample took me about 8 hours and the steel about 16 hours or lapping/polishing. Even after 16 hours of polishing, you can still see (by the corrosion & discoloration) how many pits are left in both the brass and steel.



Where lapping really helps is with poor qaulity surfaces. If you buy a good HSF from a reputable manufacturer, lapping really isnt going to help that much.

Also thanks for more info. I wasnt aware that some manufacturers were now nickel plating the copper on their HSFs. Also your mention was the first Ive heard of rumours that Intels heat spreaders may be concave. That should be easy enough to prove/disprove - take a cast and throw it on an optical comparitor
October 2, 2006 10:36:00 PM

Quote:
I control my knowledge to know when it's appropriately needed and applies it to the thread and when not need if the thread just wants a simple anwser.


Thank you.


Cheers!
October 2, 2006 10:45:26 PM

Nice link and interesting info....I cant belive intel would have such crappy QC to allow that.....unless the concavity is there purposely:
to allow for contrained thermal expansion
or
to insure heat spreader contact with the die face(this would be my guess)


.....oh well, I guess I have to add that to list.

Thanx for sharing the info
October 2, 2006 10:50:26 PM

Now dat's purdee:



Amazing how lapping can do, when you know how to do it properly.
October 2, 2006 11:19:39 PM

Quote:
Now dat's purdee:



Amazing how lapping can do, when you know how to do it properly.


Whatever you used, it's embarrassing! :D 

Edit: Wait! That's from the link wusy provided...


Cheers!
October 2, 2006 11:35:11 PM

Heh, ya it was from that link provided.

Just curious, never looked for articles lapping the IHS.

When I saw that, I was in awe... :lol: 

I dunno, I'm just not brave enough to do something that extreme, not to mention lapping the HS itself. :lol: 
October 2, 2006 11:57:38 PM

Quote:
I dunno, I'm just not brave enough to do something that extreme, not to mention lapping the HS itself. :lol: 


Usually, I don't go that far as lapping/polishing an HS, mostly, out of ignorance and lack of spare time; now that I've solved the former (sort of), I've got to better manage the later. :wink:


Cheers!
October 3, 2006 12:30:13 AM

Quote:
Nice link and interesting info....I cant belive intel would have such crappy QC to allow that.....unless the concavity is there purposely:
to allow for contrained thermal expansion
or
to insure heat spreader contact with the die face(this would be my guess)


Had I not known better from this thread (namely, temp decrease after lapping), I'd say that is a really good point... but hard to control, due to differential thermal expansion of the materials & package instabilities, when heated/cooled...


Cheers!
!