Making CU -and AL shiny. How?! - Page 2

RoboTech · 10-13-2002, 11:23 AM

Good info on lapping, etc...

(confirms much of what I try to practice)

1. Flat
2. Clean
3. Smooth
4. Shiny is mostly for looks

It seems like most discussions generally center around lapping the heatsink/waterblock surface. But what about the other half of the equation - the CPU die top surface. At times I have attempted a light lapping on the Athlon die, but I'm always afraid to do much. Without a good way to fixture the CPU and hold it FLAT, I'm not sure how much good I am really doing either.

Anyone have any info or experiences to share on lapping the CPU? Thanks,

This is a guide I wrote a while ago Heatsink and Waterblock Lapping Guide
It covers the basics and a little theory for do-it-yourself at home lapping, but not to the extent you guys are talking about here. Any comments or suggestions to help make it better are welcome...

PS Back to the original question _ Making Cu & Al shinny. I use the following procedure to clean copper blocks (NOT aluminum) prior to silver plating.
1. Wash it well in soap and water (pre wash)
2. Quick rinse and brush off in dilute sodium hydroxide solution (degrease)
3. Flush well with lots of water (neutralize)
4. Quick rinse in dilute Nitric acid (20 seconds)
5. Thourough rinse with lots of distilled water (neutralize)
The copper surfaces should be very shinny at this point...

MadDogMe · 10-14-2002, 05:53 AM

Quote:

At times I have attempted a light lapping on the Athlon die, but I'm always afraid to do much.

What did you use?. the actual surface should be ideal should'nt it?. silver writing aside

...

...Someone must have an Idea on how to get that silver writing off?. It leaves an imprint on a block every time! I've got a whole load of then overlapping on my ole SK6

.If it's proud enough to imprint that often, it's too much!!. do you think the heat 'softens' it up?. should I take a blowtorch to it? :| .

BillA surely you've got an idea?, you must have taken it into equation on your TIM quest before?, or is it a constant, so not important?...

what would remove it without scratching/affecting the core substance(which is?), pollish it off?, or as flat as possible?. which pollish?...

BillA · 10-14-2002, 08:08 AM

years ago I was a QC Supervisor at a co. called Signetics when LSI was just starting
the (much smaller then) wafers were FLAT
they are FLAT today
(the etching and subsequent vapor deposition notwithstanding)
the individual (CPU) dies are from scored FLAT wafers
the CPUs are therefore also very FLAT

I believe that anything you could do would make them less flat
I suggest accepting that the die is as it is

BTW, my heat die is copper and refinishing it's surface is a royal PITA
(had to make a 2in. sq. plate to stabilize it during lapping)

Cathar · 10-14-2002, 08:36 AM

I second Bill here.

CPU die are extremely flat.

The only exceptions were the P3's, but it wasn't the die itself that wasn't flat, the FCPGA P3 dies had a protective thermal interface coating on the die that was not flat. Very careful grinding with 1200grit sandpaper took that coating off to reveal the flat silicon die and a small gain as a result.

AMD CPU dies are flatter (with the exception of the etching) than anything you'll achieve through lapping. Lapping to remove the etching will just result in a "crowned" non-flat CPU core that will overall have less direct surface area contact with a copper plate than with the etching left intact.

RoboTech · 10-14-2002, 11:38 AM

Quote:

Originally posted by MadDogMe
What did you use?...

I used wet 1200 grit paper , very lightly. From what I am hearing now, I'm glad I didn't dwell on it very long and go too deep...

So I guess I shouldn't worry about the die surface being flat. Plenty of other "opportunities" to ponder...

Thanks,

MadDogMe · 10-15-2002, 03:36 AM

What's the core is made from?...

Is'nt etching engraved?...

MeltMan · 10-17-2002, 01:22 PM

What's the core is made from?...

Silicon...

Its so obvious it hurts!

MadDogMe · 10-18-2002, 02:27 AM

Sorry dude, obvious to you(and half the world maybe

), but I always equate silicone with tube or sealant and the cores not made out of either of those is it?...

Now can you tell me whats in the core?, is it layered?, or is it one 'lump'?. what is it there for if it contains no 'curcuitry'?.is it there to protect something underneath?. I'm wondering if it can indeed be engraved with channels for direct die cooling?, and how deep if yes?...

bigben2k · 10-18-2002, 08:30 AM

For what it's worth...

I contacted the makers of the ATTA tape (see liquid forum) and they pointed out to me that Intel was a big customer. It seems possible to me that they use the stuff under the P4 heatspreader, but we'd need to hear it from someone who's done it. I'm probably way off base...

I don't know what the core is made of, but I doubt that it's simply silicone. The P4 heatspreader is very thin, less than 1 mm, so I wouldn't cut into it. It was designed as a heat spreader, and it's probably best to leave it at that.

MadDogMe · 10-19-2002, 04:02 AM

I was thinking of a Palamino core, or a T~Bred(can't spell Thoroughbred

) it's about 1~1.5mm thick is'nt it?, some 0.5^mm grooves would double surface area for direct die cooling...

RoboTech · 10-19-2002, 08:54 PM

Hey MadDog,

It's not just you, a lot of us get our silicons and silicones confused...

That little "e" makes a big difference!

Quote:

Originally posted by MadDogMe
Sorry dude, obvious to you(and half the world maybe ), but I always equate silicone with tube or sealant and the cores not made out of either of those is it?...

Silicone refers to silicone tubing (Siloxane polymers and amorphous silica) and silicone RTV sealants, etc. And no, CPU cores are not made out of silicone.

Quote:

Now can you tell me whats in the core?, is it layered?, or is it one 'lump'?. what is it there for if it contains no 'curcuitry'?.is it there to protect something underneath?. I'm wondering if it can indeed be engraved with channels for direct die cooling?, and how deep if yes?...

The CPU (and other integrated circuits) are built up on a silicon substrate in many layers. Photolithography, etching, masking, doping, etc., etc. Newer chips contain various varieties of doped silicon to make all the semiconductors (P-types and N-types) and numerous copper or aluminum traces to connect things together. It is not possible to drill or engrave coolant channels in an existing chip but I beleive it may be possible to design them in, in the not too distant future.

Just think if you built a chip using diamond as the substrate instead of silicon! Fantastic thermal properties and some interesting optical qualities as well for a hybrid electro-optical 3D chip or cubit. Now that could lead to some serious number crunchin...

RoboTech

MadDogMe · 10-20-2002, 04:44 AM

Light speed!, wonder what the Mhz equivilent is

...

I think they're be CPU's designed for DD H20 cooling soon, micros channeled/bored perforated cores for high pressure/velocity mega cooling...

I think if you could make a wafer thin Cu 'plate' with 0.5mm^ channels it'd do direct die cooling some good, it might not technicaly be 'direct die' then, but imagine double the surface area!, TIM layer is the setback though, you'd have to epoxy it as you'd not be able to put any pressure on it to bear...

10-13-2002, 11:23 AM	#26
RoboTech Cooling Savant Join Date: Sep 2002 Location: Cincinnati, OH Posts: 229	Good info on lapping, etc... (confirms much of what I try to practice) 1. Flat 2. Clean 3. Smooth 4. Shiny is mostly for looks It seems like most discussions generally center around lapping the heatsink/waterblock surface. But what about the other half of the equation - the CPU die top surface. At times I have attempted a light lapping on the Athlon die, but I'm always afraid to do much. Without a good way to fixture the CPU and hold it FLAT, I'm not sure how much good I am really doing either. Anyone have any info or experiences to share on lapping the CPU? Thanks, This is a guide I wrote a while ago Heatsink and Waterblock Lapping Guide It covers the basics and a little theory for do-it-yourself at home lapping, but not to the extent you guys are talking about here. Any comments or suggestions to help make it better are welcome... PS Back to the original question _ Making Cu & Al shinny. I use the following procedure to clean copper blocks (NOT aluminum) prior to silver plating. 1. Wash it well in soap and water (pre wash) 2. Quick rinse and brush off in dilute sodium hydroxide solution (degrease) 3. Flush well with lots of water (neutralize) 4. Quick rinse in dilute Nitric acid (20 seconds) 5. Thourough rinse with lots of distilled water (neutralize) The copper surfaces should be very shinny at this point... Last edited by RoboTech; 10-13-2002 at 08:42 PM.

10-17-2002, 01:22 PM	#32
MeltMan Cooling Savant Join Date: Aug 2001 Location: KS Posts: 374	What's the core is made from?... Silicon... Its so obvious it hurts! __________________ MeltMan Lurker Supreme!

10-18-2002, 08:30 AM	#34
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	For what it's worth... I contacted the makers of the ATTA tape (see liquid forum) and they pointed out to me that Intel was a big customer. It seems possible to me that they use the stuff under the P4 heatspreader, but we'd need to hear it from someone who's done it. I'm probably way off base... I don't know what the core is made of, but I doubt that it's simply silicone. The P4 heatspreader is very thin, less than 1 mm, so I wouldn't cut into it. It was designed as a heat spreader, and it's probably best to leave it at that. __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

10-14-2002, 08:08 AM	#28
BillA CoolingWorks Tech Guy Formerly "Unregistered" Join Date: Dec 2000 Location: Posts: 2,371.493,106 Posts: 4,440	years ago I was a QC Supervisor at a co. called Signetics when LSI was just starting the (much smaller then) wafers were FLAT they are FLAT today (the etching and subsequent vapor deposition notwithstanding) the individual (CPU) dies are from scored FLAT wafers the CPUs are therefore also very FLAT I believe that anything you could do would make them less flat I suggest accepting that the die is as it is BTW, my heat die is copper and refinishing it's surface is a royal PITA (had to make a 2in. sq. plate to stabilize it during lapping)

10-14-2002, 08:36 AM	#29
Cathar Thermophile Join Date: Sep 2002 Location: Melbourne, Australia Posts: 2,538	I second Bill here. CPU die are extremely flat. The only exceptions were the P3's, but it wasn't the die itself that wasn't flat, the FCPGA P3 dies had a protective thermal interface coating on the die that was not flat. Very careful grinding with 1200grit sandpaper took that coating off to reveal the flat silicon die and a small gain as a result. AMD CPU dies are flatter (with the exception of the etching) than anything you'll achieve through lapping. Lapping to remove the etching will just result in a "crowned" non-flat CPU core that will overall have less direct surface area contact with a copper plate than with the etching left intact.

10-15-2002, 03:36 AM	#31
MadDogMe Thermophile Join Date: Aug 2002 Location: Just shut up ;) ... Posts: 1,068	What's the core is made from?... Is'nt etching engraved?...

10-18-2002, 02:27 AM	#33
MadDogMe Thermophile Join Date: Aug 2002 Location: Just shut up ;) ... Posts: 1,068	Sorry dude, obvious to you(and half the world maybe ), but I always equate silicone with tube or sealant and the cores not made out of either of those is it?... Now can you tell me whats in the core?, is it layered?, or is it one 'lump'?. what is it there for if it contains no 'curcuitry'?.is it there to protect something underneath?. I'm wondering if it can indeed be engraved with channels for direct die cooling?, and how deep if yes?...

10-19-2002, 04:02 AM	#35
MadDogMe Thermophile Join Date: Aug 2002 Location: Just shut up ;) ... Posts: 1,068	I was thinking of a Palamino core, or a T~Bred(can't spell Thoroughbred ) it's about 1~1.5mm thick is'nt it?, some 0.5^mm grooves would double surface area for direct die cooling...

10-20-2002, 04:44 AM	#37
MadDogMe Thermophile Join Date: Aug 2002 Location: Just shut up ;) ... Posts: 1,068	Light speed!, wonder what the Mhz equivilent is ... I think they're be CPU's designed for DD H20 cooling soon, micros channeled/bored perforated cores for high pressure/velocity mega cooling... I think if you could make a wafer thin Cu 'plate' with 0.5mm^ channels it'd do direct die cooling some good, it might not technicaly be 'direct die' then, but imagine double the surface area!, TIM layer is the setback though, you'd have to epoxy it as you'd not be able to put any pressure on it to bear...

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)