Silver plating, an option to lapping?

[bigben2k]

You read it right!

I'm proposing that electroplating (with silver, while were at it) might be a way to achieve excellent flatness on our dear blocks.

The silver won't improve the thermal properties of the block, since it's still copper.

From what I've read, silver can easily be electro-plated to copper, up to a thickness of 300 microns, or 0.3 mm, with a minimum of 0.5 microns (at least according to this website.

From another site, it seems that in order to achieve a very smooth finish, a low current should be applied, but will take longer.

From
this site, it seems that all we need is a 12V battery (or a PSU, maybe?)

(I have yet to find a DIY guide)

Comments?

[Acroam]

Definitely sounds like it would be a great alternative to lapping

[myv65]

It still comes down to the flatness of the surface you are plating. All electroplating will do is apply a relatively uniform thickness of material "X" over material "Y". If material "X" isn't flat, the finished product will not be flat either (although the Ra roughness may be better in the finished product).

Electroplating is also not something for the DIYer. Possible sure, recommended, nah. Some of the chemicals required (for a really nice job) are regulated and require hazardous waste disposal procedures. This is truly one area best left to the professionals.

[bigben2k]

That's probably true.

I did find that there are many, many shops that will do this, in a variety of processes.

If I got a block, I'd consider sending it off for silver plating of the bottom surface.

[gmat]

IMHO, this would be great for a totally different reason.
Firstly, even after electoplating (or anodizing) i'd still lap the surface. Oh and i'd lap it BEFORE so i'm sure that the geometry is alright. The 2nd pass would only be with 1200 grit minimum, to avoid getting through the plating itself, and only to get rid of small imperfections which grew up with the plating process.

Next, i think it's a very good idea for thermal reasons. The thermal diffusivity of copper is not as great as silver, this means the core 'hot spot' is very localized. The silver plating would help "spreading" that hot spot further through the block baseplate, with obviously benefic results.
Note the conditional, i'm just guessing. And one would need to determine the ideal plating thickness for optimal thermal diffusion.

[myv65]

Hey gmat,

The conductivity differential between silver and copper hardly makes for much improvement in spreading of the heat. At 300K, silver's conductivity is 429 W/m-K while copper's is 401. Electroplating also isn't real effective for depositing significant material thickness. Based on this, I wouldn't see a sizeable gain to be had in heat spreading via silver electroplating.

If you want to go the silver route for spreading, you'd have better results from a slug (few mm thickness) similar to the copper insert currently used on some air heat sinks. Even then, with less than a 10% gain in conductivity it would be a pretty minor improvement.

No question that it would be an improvement, just a very, very small one.

I think you'd see far more (temperature) variation simply in installation technique from person to person and in mounting style from block to block.

[bigben2k]

Originally posted by myv65
Hey gmat,

The conductivity differential between silver and copper hardly makes for much improvement in spreading of the heat. At 300K, silver's conductivity is 429 W/m-K while copper's is 401. Electroplating also isn't real effective for depositing significant material thickness. Based on this, I wouldn't see a sizeable gain to be had in heat spreading via silver electroplating.

If you want to go the silver route for spreading, you'd have better results from a slug (few mm thickness) similar to the copper insert currently used on some air heat sinks. Even then, with less than a 10% gain in conductivity it would be a pretty minor improvement.

No question that it would be an improvement, just a very, very small one.

I think you'd see far more (temperature) variation simply in installation technique from person to person and in mounting style from block to block.

Well, that's still a 7% improvement, but you're right, it probably wouldn't do anything that couldn't be compensated with a higher flow rate.

Still, although I believe that the silver plating has a few advantages:
1-although the original surface would still need to be lapped, the plated surface might have a finish that Arctic Silver can fill up better. I was hoping that the plated surface wouldn't need to be lapped, and if I send it out, I'd have to specify that they should not attempt to polish it, because I don't want "buffing compound" as the interface material between the block and the CPU core.

2-I thought it might help to transfer heat from the core to the copper part of the block, so I guess I'm saying that it might offer less resistance, but as gmat pointed out, it might have to be calculated, but I was thinking more simply in terms of using silver to act as a transition surface, as the material to be filled by the Arctic Silver, so I don't think that it needs to be calculated, it just needs to cover 100% of the bottom surface, with enough thickness to make sure that any irregularities are made entirely of silver.

Also, since the block is still a copper block, the rate of heat transfer would be limited to that of the copper.

So, the theory still stands:
1-get a copper block.
2-lap it as much as possible.

then, depending on how well lapped it is, will determine the required thickness of the plating.

What would be the size (max depth) of the irregularities, if the block was lapped to 600? 800? 1000? 1200? 1500? 2000? (assuming a circular pattern, for lapping)

3-Have the block's bottom plated to the thickness (above)

Alternatively, we could pick a thickness that would cover just about everything.

4-Get it back, and lap it again, with the finest of the finest of sandpapers, using water as a lubricant.

How does that sound?

BTW, the Arctic Silver website kindly points out that typically, only 1% of a surface comes in contact with another one. 99% is contacted through the paste. (true?)

[myv65]

Originally posted by bigben2k
BTW, the Arctic Silver website kindly points out that typically, only 1% of a surface comes in contact with another one. 99% is contacted through the paste. (true?)

Marketing hype, nothing more. You'll also see claims that AS improves your temperatures by up to 5°C or more. Utter hogwash. The only way it could do that is if you didn't use an interface material. Even the thermal tape used on the retail HSFs has a thermal resistance and as-melted thickness that results in a delta-T on the order of 2-3°C for a stock XP chip. Biggest advantage is AS doesn't make a mess like tape. This makes heat sink or block changes much cleaner.

PS: I use only AS on my systems

[MeltMan]

2 words

Ooooh and Aaaah

Thats why I would silver plate

[bigben2k]

Originally posted by MeltMan
2 words

Ooooh and Aaaah

Thats why I would silver plate

but you couldn't see it if you only had the bottom plated...

Of course you can extend my idea, and silver-plate the whole block: there's your "Ooooh and Aaaah" factor!

Just a side note on what gmat said: I don't think that there's really any kind of calculation required, because it would only show that the best thickness for silver to be applied, would be the thickness of the baseplate...

As a fancy block idea, it might be possible to slip a silver slab, the size of the CPU core, into the baseplate of a copper block...

[MeltMan]

You mean like a slug? All the way through to the water surface? Or more like an alpha heatsink has the copper inlay? Either way it would rule. How about this idea, Silver melts fairly easily right? use an Acetylene torch to melt silver into a copper "dished out area" under the waterblock. You could make the surface very rough so that you have maximum silver to copper surface area.

[gmat]

No no, just base yourself on an existing copper WB, for example a TC-4. Take its base thickness (without turbulators) and compute the best silver plate thickness to spread the heat evenly without diminishing returns. Intuitively it would be rather thin, we want maximum radial diffusion on that layer, with minimal transversal diffusion.
Myv65, you said that electroplating could yield a 300µ layer, i think it would be more than enough. And that, as an extra bonus added to surface quality.

(edit) ha meltman posted while i was typing. Hmmm good idea, but not for the faint hearted Acetylene torch you said ? mmhhhhh (/me looks crazy) nyark nyark nyark mmmmhhhhh nevermind.
(edit of the edit) that what i'm holding on the photo.

[bigben2k]

I meant a slug, yes, right up to the water.

Silver has a melting point of 961 degC, so I think it'd be OK, right?

[MeltMan]

Heh, I know one of those torches will melt what ever you throw at it. Steel is nothing in the way of an Acetylene torch. They are quite fun to play with.

As far as having the water touching the silver itself, wouldnt that pose some interesting corrosion conditions? Copper touching silver with water... I think that would be bad.

[gmat]

Yep. Thats what i modded my YY case with. Overkill, but so quick, and so fun.

[bigben2k]

Originally posted by MeltMan
Heh, I know one of those torches will melt what ever you throw at it. Steel is nothing in the way of an Acetylene torch. They are quite fun to play with.

As far as having the water touching the silver itself, wouldnt that pose some interesting corrosion conditions? Copper touching silver with water... I think that would be bad.

Yes, you're right, it would make the copper corrode, but we can't afford to make the whole thing out of silver (nor silver rads), so it's a trade off.

Besides, copper corroding isn't so bad... We'd have to use WW.

[bigben2k]

Originally posted by gmat
No no, just base yourself on an existing copper WB, for example a TC-4. Take its base thickness (without turbulators) and compute the best silver plate thickness to spread the heat evenly without diminishing returns. Intuitively it would be rather thin, we want maximum radial diffusion on that layer, with minimal transversal diffusion.
Myv65, you said that electroplating could yield a 300µ layer, i think it would be more than enough. And that, as an extra bonus added to surface quality.

What do you mean by "without diminishing return"? As far as I can tell, silver will improve (albeit slightly) the heat spreading ability, regardless of how thick it is, and copper still being most of the material, it would still resist the heat transfer at the same rate, regardless of wether or not there's a silver plating.

I'm just having a hard time seeing what you mean. I can tell that you're talking about spreading the heat sideways with the silver, so that the copper can absorb the heat from what would effectively be a larger area for a heat source, but since the heat source is still the same, I don't think that the copper will perform any better.

[airspirit]

I don't know if this has been mentioned yet, but that is a bad idea. Wery bad indeed.

For every different material the heat must pass through it actually SLOWS conductivity, just like how sound and electricity step down in power with each new material they have to pass through. The gain you would get by plating the sucker will NOT make up for the loss in thermal conductivity. What you'll have happen is the heat will be more eager to spread out through the silver plate to all plated areas of the block than to go to the block itself.

I suspect that this would be a waste. If you want to use silver, make the entire bottom plate of your block silver so the silver contacts the water. Then you only have two steps of transfer. Adding the copper as a transient barrier would make it more difficult to transfer heat than the copper alone.

Just lap the sucka. Cheaper and more effective.

[gmat]

Thats what i meant by "diminishing returns". Airspirit, what's adding resistance is *interface* not material itself. In the case of a "perfect" interface (obviously, brazed / melted materials) there's near zero resistance.
Also a thick plate of silver is out of question considering costs, and forget contact with water...

What you'll have happen is the heat will be more eager to spread out through the silver plate to all plated areas of the block than to go to the block itself.

Well
1 - it *would* spread through the plate
2 - then the plate would more evenly tranfer to the block
in a way, think of this as an "extension of the core".

[bigben2k]

Well, now I see what's going on!

I was thinking that plating was as perfect an interface as one could get by melting silver onto copper. Obviously, I might not be right. I just assumed, that silver would be broken down into individual ions, and latch themselves onto the copper.

As for heat transfer, since silver is not much better than copper (about 7%), I don't believe that it would make a measurable difference in the block's performance.

Really guys, I was just thinking about improving the contact surface to the core, and I mean the mechanical construction of the contact surface.

It'd be unusual to apply Arctic Silver to silver, but I'm guessing that it would reduce the heat transfer resistance of a poorly lapped copper surface.

[myv65]

Don't confuse interface resistance with thermal conductivity. gmat is right on the mark. Imagine for a moment that silver was a thermal superconductor, that is a material with infinite thermal conductivity. In this case, it would have the same temperature at every point within itself and would spread the heat from the core area (on the order of 100 mm^2) to the entire base of the block uniformly (on the order of 4000 mm^2, give or take). This would be wonderful.

Then there's the interface resistance that would still exist between our "perfect" conductor and the block material. As noted, bonds that involve a chemical reaction such as electroplating or brazing have a thermal resistance very near zero. This slight thermal resistance would have far fewer detriments than the advantage gained if we really could spread the heat from the core uniformly over the bottom surface of our blocks.

[bigben2k]

hence the diminishing return calculation, as posed by gmat.

So the question becomes: is it worth it to electroplate silver to copper? and if so, what thickness does it need to be in order to compensate for the interface resistance?

Of course we can't compensate for the interface resistance, since it draws from the overall thermal conductivity of the block. So off-hand, silver plating is a bad idea. It'd be better to improve how we lap the copper.

Right?

[MeltMan]

I think two identical blocks must be forged out of the depths of creation, one of which will be dipped in the bath of gods (silver plate), the other is destined to be lapped and tortured to perfection. Both blocks will then duke it out on a level playing field with the victor being thrust into the spotlight.

or somthing like that

[Sanjuro]

Silver tarnish right? It would be nice to electroplate with something that doesn't tarnish and doesn't kill thermaltransfer in the same time. I hate green copper from hand acid

[bigben2k]

MeltMan: I'd like to see it too!

Sanjuro: All you have to do is cover the metal, because it's the oxigen in the air that reacts with the metals.

In this case, Arctic Silver on the contact area. Everything else can be varnished, except where the coolant contacts the copper.