Heatsink Staining

[AsRock]

I was using 91% alcohol. After the surface is clear it still isn't clean. I soaked it for a couple minutes and then rubbed at it with a lens cloth for a couple more. This is buried in the small surface valleys a cloth won't reach into.

Naturally occurring acid, interesting.

Sounds like you need a better cloth. and my bad i should not of said rubbing alcohol either as they has so much crap in it and will typically had all kinds of crap in it.

ArctiClean is really good stuff.

You could check in to trying salt baking powder and white vinegar all though i have onlly tried this on copper coins and totally works.

 

[Bill_Bright]

Toothpaste is acidic.

Not really. Most are neutral and the rest are only very slightly acidic. The reason why is our saliva is an acid as it is the first stage to start breaking down food in the digestive process. So the toothpaste neutralizes that acid. Regardless, you don't leave it on for days and it is much less acidic than the tomato - the base for ketchup.

 

[nomdeplume]

Sounds like you need a better cloth.

I may not have spelled out how severely I stressed these and how little regard was given to their internal well being. This was back in the days when as soon as the register rang it lost half the value and was six months from being outdated. So you used it up and got another. This was a semi-serious attempt to see how reversible the ravages of time and heat were on outmoded equipment. Learning on them had very little downside.

When I pulled the heatsink, on five out of six, it had a bare spot where the processor sat and the processor itself was devoid of any TIM. A better cloth would not have made any difference. A dremel with polishing pad attachment might make a difference.

@Bill_Bright This was quietly heading towards using power and hand tools for kicks on a dumb project. As much as I appreciate the suggestions I don't see myself spending hours scrubbing away with a mild household substance. If ketchup or toothpaste find their way into the path of my dremel I'll take video.

 

[MrGenius]

I've done A LOT of lapping in my day. If there's anything I've learned it's if you want a super shiny mirror-like finish after lapping NEVER use water in the process. Or at least not during the final stages. The super smooth shine you can see yourself in requires extra fine grit used dry. Water wrecks it every time. Not that it matters a whole lot. You want a perfectly flat/straight surface more than a shiny one. I mean ideally you want both. You want to be able to slap a straight edge on it and see no gaps, and have a mirror polish on the surface. With the former being more important. The latter is pretty much what TIM is for(or should be for).

 

[Nuckles56]

If you want a true mirror polish, look for some 1/4 micron diamond paste, it expensive but it will give you an amazing finish so long as you managed to get all the large scratches out with some 4000 grip wet and dry paper

 

[nomdeplume]

Since we have diverted away from using chemicals to release the 'stain'. I've noticed a lot of cpu cooler reviews mention heatsink bases that are slightly convex and smooth to the touch, but nowhere near a mirror finish as faint machining marks are visible (TPU Score: 9.7 & Editors Choice Award). Were this present on lower end and budget version coolers, or significantly fewer top performers, It'd be easy to dismiss.

Having done a fair amount of hands on work refinishing to as flat and smooth of a surface as possible. Relatively, as opposed to exactingly, flat and smooth is possible with a steady hand and care. Mirror polishing on a surface as level as possible was the best choice for some time. Every one of the stained heatsinks I encountered were exactly that. I think if any lapping is attempted by me it will involve a fair amount of higher end knife sharpening techniques and equipment to recreate in small part the complex machined effect on current coolers.

 

[Bill_Bright]

Back in the day (15+ years ago) lapping was often done by enthusiasts to ensure better transfer of heat because the CPU and heatsink mating surfaces were (1) not perfectly flat and (2) pockmarked (microscopically speaking) with pits and valleys and impurities in the metals. Lapping made those mating surfaces as smooth and flat as precision machining allowed. This allowed those users to eke out a few more clock cycles when overclocking while keeping temps below thermal thresholds.

It is important to remember the most efficient transfer of heat occurs with direct metal-to-metal contact of the mating surfaces. So with properly lapped surfaces, no TIM (thermal interface material) was needed, and actually was in the way and counterproductive to the heat transfer process - just as too much TIM is in the way today.

But CPU and heatsink manufacturing techniques have improved significantly since those days - as have the quality and efficiency of today's TIMs (including OEM TIM). Lapping is just no longer needed. It is only good for bragging rights.

If you "need" the couple extra degrees lapping "might" provide, you have other cooling deficiencies that need to be addressed first, like better case cooling and/or a better CPU cooler or a "proper" application of a good TIM.

And remember, lapping metals is unhealthy! It involves sending metal filings into the air that might be inhaled or deposited on the CPU pads/pins, or into the socket or other components if you carelessly lap near your motherboard. You risk torching a Grand Canyon sized trench (microscopically speaking) through your CPU with a ESD, if necessary ESD precautions are not used. Or just dropping the CPU. And clearly it will void your CPU warranty as well.

I am not saying "don't lap". I am saying "know the risks" and "be prepared to accept the consequences" should things go south.

 

[John Naylor]

The staining is thee result of chemical reactions where by something is reacting with the copper. These compounds have less heat transfer capability than pure Liquid Metal TIMs. "Galinstan" is a gallium, indium and tin alloy of with a melting point of -19 C. Gallium is prone to alloy with other metals to varying degrees but this happens much faster because of the electrochemical potential of the two metals. Much like ina CLC with aluminum rad and copper block where ions will flow from the aluminum to the copper, the aluminum is essentially corrodong which results in a build up of oxides on the copper block. This happens on a heat sink also... the stuff ya can rub off is the oxide, the stuff ya can't is "alloy plating". The rate of electrochemical action is dependent on the electrical potential of the two metals. Gallium's electrical potential is -0.53 volts and copper's is +0.35 volts. The the greater the electrical potential difference between the two metal( 0.88 volts), the more powerful the galvanic corrosion cell..... essentially this is a "battery". Good reading here:

https://martinsliquidlab.wordpress.com/2012/01/24/corrosion-explored/

So yes, while the remaining alloy will reduce thermal performance, the effect will me minimal. Still if you are a purist, or if the aesthetics of 'knowing it's there" bothers you, it can be removed. I can not recommend lapping. Well if it is a high end heat sink, the surface will already be smooth but more importantly it will also be slightly curved (convex) to mate more evenly with the slightly concave IHS. By lapping the IHS, you decrease the level of contact and thereby cooling performance. You can remove the metal plating by polishing with a polishing compound but must take pains to do it evenly. Some heat sink manufacturers void the warranty if you lap the base. (See

http://www.legitreviews.com/prolimatech-armageddon-6-heatpipe-cpu-cooler-review_1303

As for paper... anything that is fibrous is a bad idea ... I use a chamois for polishing (AllClad kitchen pots) and a foam swab for cleaning (Indigo Extreme Cleaner) CPUs. If your in a pinch, a coffee filter is a gppd option used with 90+% isoprophyl alcohol.

 

[Totally]

Lapping was commonplace before IHS became the norm and CPU were just bare dies. The reason heatsinks are slightly concave is because IHS are slightly convex (don't know if this still holds true with Intel chips using TIM) letting the two surfaces mate better. So if you are lapping the base you are essentially creating a pocket in between the CPU and heatsink that may or may not be filled with interface material.

 

[Bill_Bright]

Lapping was commonplace

Commonplace? Not even. Most people never heard of it. It was only popular among enthusiasts and since they are only a tiny minority among all users, it was not commonplace.

Never were CPUs bare dies.

 

[Totally]

Commonplace? Not even. Most people never heard of it. It was only popular among enthusiasts and since they are only a tiny minority among all users, it was not commonplace.

Never were CPUs bare dies.

I was referring to forum-goers then, and they consisted of nothing but enthusiasts. The average Joe was not visiting these site back then.

Didn't Pentium ii's and early iii's not have IHS?

 

[Athlonite]

I was referring to forum-goers then, and they consisted of nothing but enthusiasts. The average Joe was not visiting these site back then.

Didn't Pentium ii's and early iii's not have IHS?

Yes as did Athlon and duron and celerons but that doesn't make Lapping and HS a common place thing as Bill said it was really only done by some enthusiasts wanting the absolute max from their HSF's and I have seen people do it to both the IHS (common problem was that a soldered IHS dipped down in the middle because of being soldered) and HSF base

 

[Totally]

Yes as did Athlon and duron and celerons but that doesn't make Lapping and HS a common place thing as Bill said it was really only done by some enthusiasts wanting the absolute max from their HSF's and I have seen people do it to both the IHS (common problem was that a soldered IHS dipped down in the middle because of being soldered) and HSF base

well, if it wasn't commonplace then let it be so. I was under the impression it was a common thing then because when I first joined bit-tech before joining here long ago just about every other thread was someone documenting their lapping process. The IHS dipping wasn't really a problem because solved by making the bases of the heatsinks slightly concave, just a new normal people weren't used to.

 

[Bones]

I've been using Brasso to clean the surfaces of my coolers if needed and it does work as long as they are copper or the surface to be cleaned is.
Gives it a nice shiny surface.

Can't say if it would work well with your exact problem or not but certainly won't hurt anything either - Definitely not abrasive at least.

 

[dorsetknob]

Never were CPUs bare dies.

WTF =

Didn't Pentium ii's and early iii's not have IHS?

Picture proof bill and did you not build any P3/Athlon/duron systems
Early PGA versions have an exposed die,

AMD and exposed Die

Explain these Retail pic's then bill
and those old threads " i think i over tightened my Heat sink and i think i cracked my Die"

Oh and by the way Bill "do you know what those 4 pads surrounding the AMD Die were for???"

 

[Caring1]

WTF =
Early PGA versions have an exposed die,

AMD and exposed Die

Explain these Retail pic's then bill
and those old threads " i think i over tightened my Heat sink and i think i cracked my Die"

I have a few in a drawer still.

 

[nomdeplume]

Incredible how much I'm learning about how exacting this aspect of building a computer is. Did a little research into how processors and heatsinks fit together. Getting a flat surface even height within a given tolerance is easier than something more complex like a curved surface. I had wondered if current Intel chips top cover deflected slightly to help evenly distribute heat. All of 30 seconds later I was running to grab a screwdriver and loosen my cooler screws. Hand tight + a twist with the screwdriver was news to me.

Picture proof bill and did you not build any P3/Athlon/duron systems
Early PGA versions have an exposed die,

By far the worst heatsink I refreshed was covering an exposed die Intel Celeron M 370. Had forgotten all about them. So shiny! So covered in bubblegum from the factory everywhere but the die!

I've been using Brasso to clean the surfaces of my coolers if needed and it does work as long as they are copper or the surface to be cleaned is.
Gives it a nice shiny surface.

Can't say if it would work well with your exact problem or not but certainly won't hurt anything either - Definitely not abrasive at least.

I'll have to see what the current chemical composition is and how to neutralize it when it inevitably gets on aluminum body the copper heatsink is buried in. This could potentially be the go to solution.

 

[Bill_Bright]

WTF =

Is right. It seems some folks don't know what they are even looking at. Come on guys! Those images CLEARLY show the dies encased in epoxy resin.

This is what "bare dies" look like:

and

 

[Beertintedgoggles]

Those P3 shown are most definitely the die and not some easily scratched epoxy resin. On the P3s, the die was bonded directly to the substrate beneath it without the use of micro wires. The P1 shown in the above pics had a bonding layer that used micro wires to connect to the substrate. That's why if you look on the underside of the original pentiums, there was this "square" that isn't included in the P3s, Athlons, etc.

The reason why the P3 doesn't look like the die is there are a few layers of protective silicone over top the actual circuitry (the same layers that make up the actual chip, just no hardware there to protect the actual electronics within from any scratches). However, it for certain is the bare die you are staring at when assembling those "old" systems.

 

[Bill_Bright]

The reason why the P3 doesn't look like the die is there are a few layers of protective silicone over top the actual circuitry

We seem to have a different definition of "bare".

"A few layers of protective silicone over top the actual circuitry" does not suggest the same thing as "bare" to me.

I'm sitting here "bare" naked - except for the layers of clothing I am also wearing.

By your definition, virtually every stick of RAM is populated with "bare" memory modules like this one, right?

 

[dorsetknob]

yup bill "our and industry" definition of bare Die differ from yours

 

[Beertintedgoggles]

That RAM is packaged bare dies.... the P3 is actually part of the silicone that was cut from the wafer.

 

[dorsetknob]

they say a picture is worth a 1000 words

 

[Bill_Bright]

yup bill "our and industry" definition of bare Die differ from yours

I disagree with that being the accepted "industry" definition, but I've only been an electronics technician since 1971 so what do I know? I tend to go by this: What is Bare Die? where it notes "bare die" as being "unpackaged" and further states in part,

Die Handling
Bare die must be handled always in a class 1000 (ISO 6) clean room environment: unpacking and inspection, die bonding, wire bonding, molding, sealing. Handling must be reduced to the absolute minimum, unnecessary inspections or repacking tasks have to be avoided. Use of complete packing units (tray, FFC, tape and reel) is recommended and remaining quantities have to be repacked immediately after any process (e.g. picking) step. To avoid contamination and damages (scratches, cracks)

• Bare Die or wafers must never be handled by bare fingers
• Work only in ESD safe clean room environments

Again, the image shown here would be a "bare" die to me. But fair enough - as long as we are all on the same page now, that's fine.

 

[Totally]

WTF =

Picture proof bill and did you not build any P3/Athlon/duron systems
Early PGA versions have an exposed die,

AMD and exposed Die

Explain these Retail pic's then bill
and those old threads " i think i over tightened my Heat sink and i think i cracked my Die"

Oh and by the way Bill "do you know what those 4 pads surrounding the AMD Die were for???"

Oh screw off Dorset, it was a rhetorical question. I literally laughed for a good minute at bills comment.