Metallurgist? Copper annealing discussion.

[Les]

Possibly some relevancy here:
http://properties.copper.org/
For influence of Oxygen the link to "Metallurgy of Copper Wire" is worth a look - http://innovations.copper.org/metall...etallurgy.html.

[redleader]

I spent all last night reading through copper.org. I don't think oxygen will be a problem.

According to their information, a tiny bit of oxygen (~.01% iirc) is actually helpful. Above that oxygenated material tends to come out of solution forming fairly large gaps (though widely spaced since the material collects). The graphs they show suggust that the effect is pretty minor.

As for the rate of difussion, I didn't think much of anything will enter a metal crystal so far below melting. I assumed that anneling would require 400C+.

[murray13]

Thx 8-ball for digging into this!!! Very informative.

One question, all this talk about 'cold working' copper decreasing thermal conductivity does it really apply to the extent suggested at the start of the thread? What I mean. Most of what we do to a piece of copper is to a surface. We are not bending, skieving, or compressing to an extent that would cause significant deformation. Yes the surfaces will be deformed to some depth depending on the method used. But the bulk of the mass will not have sufficient stress applied to cause crystal deformation. Or am I out in left field?

Since most copper bar stock people buy is of the cold drawn type I think annealing would be good at some point during the making of a WB. I guess my question is do you have to wait until it's complete (except for final lapping) before you anneal and to what extent machining and lapping will actually have on thermal conductivity?

[8-Ball]

I agree, the deformation introduced by machining is not on the same level as the hard drawn copper. This will have undergone something like a 70% change in area as it is drawn through the dies and so on.

I think the original idea was starting with this as it is a lot harder, making it ieasier to machine, and then performing recovery and recrystallisation to reduce the defect density.

However, that resulting microstructure will be finer than if you had started with soft copper.

8-ball

[Alchemy]

Originally posted by redleader
I spent all last night reading through copper.org. I don't think oxygen will be a problem.

(off-topic)

This reminds me of one of my team members on our capstone design project. We didn't realize until after our report was nearly complete (that is, too late to make major changes) that she got all her environmental information on a particular heavy metal toxin on the website for the lobby group of producers of that particular metal. They, of course, stated quite seriously that the toxin was of no danger to any living creature and made an excellent seasoning on salads.

Okay, maybe not the last part. But the site had a bias that was obvious, and even if it hadn't, someone 2 months away from a B.S. in chemical engineering should realize that the best way to research a possibly controversial topic is *not* entering www.(controversial topic).com in a web browser.

Alchemy was not pleased.

Anyway, thank God I didn't have to pay for this degree. Hooray for South Carolina public universities, where having a third brain cell to rub against the other two guarantees a four-year free ride.

Alchemy

[8-Ball]

Originally posted by murray13
Thx 8-ball for digging into this!!! Very informative.

One question, all this talk about 'cold working' copper decreasing thermal conductivity does it really apply to the extent suggested at the start of the thread? What I mean. Most of what we do to a piece of copper is to a surface. We are not bending, skieving, or compressing to an extent that would cause significant deformation. Yes the surfaces will be deformed to some depth depending on the method used. But the bulk of the mass will not have sufficient stress applied to cause crystal deformation. Or am I out in left field?

Since most copper bar stock people buy is of the cold drawn type I think annealing would be good at some point during the making of a WB. I guess my question is do you have to wait until it's complete (except for final lapping) before you anneal and to what extent machining and lapping will actually have on thermal conductivity?

Sorry, I didn't really read your post thoroughly.

If you are using cold drawn copper, it will have an incredibly high defect density. This is the material compared with soft copper which had only 90% of the conductivity. The machining and lapping will have little effect on the overall defect concentration of cold drawn copper, as it is already high. (The local concentrationaround the milling surfaces may increase)

However, milling a soft copper will raise the defect density several orders of magnitude above the normal. So you will have to recrystallise again.

In summary, machine your cold drawn copper bar stock and then lap the base. Then recrystallise, either 140 degrees C for 15 minutes in the oven or 2 hours in boiling water. Once you've finished the recrystallisation, you could give it another quick lap on 2000 grit wet and dry paper, though I don't think it would be necessary.

8-ball

PS, I'm thinking of writing this up properly when I get the time, along with anything else I can find out relating to materials science in extreme cooling and overclocking. Any suggestions are welcome. (This won't be till the summer)

[pelle76]

Great. This has been a good diskusion and I think we have came to the conclusion that annealing the copper will help us get more efficient blocks.

8Ball: Uve really dug deep in this shit and U seem to know Ure metalls... Thanx.

Now I guess the final thing to do is to make some similair looking blocks and anneal some of them and then run some IRL testing...

I'm about to build my first watercoolingsystem and If I get the time I will make two WW-looking blocks. One anneald and one that isnt.
Since this is my first attempt at watercooling I guess a lot of U guys could make these tests alot easier than me though.

I guess some people will soon try this so I'm really looking forward to see the results.

[BillA]

give us details on the test setup to measure this
- gonna be SOME rig

[8-Ball]

All it would need really is any block machined from cold drawn copper bar stock to be tested then treated for 15 mins at 140 degrees, or 2 hours in boiling water, and then retested.

If there is going to be any noticible change, that's how you'd pick it up.

Though, If you do gain an extra 10% in conductivity, I don't see how much of a difference this would make on performance.

8-ball

[BillA]

hmmm
a 10% gain is about twice the difference between Cu and Ag

what is the basis of the 10% value ?
(I would have guessed FAR lower)
but if the mounting induced variation is ±0.2°C (by an 'expert', more typical is 2 or 3 times this);
I think the difference will be incredibly difficult to pickup

[bigben2k]

Using Cathar's block, assuming that 0.19 C/W is improved by 10%, that would equate to a 1.3 deg C improvement, assuming also the same conditions for the 0.19 measurement (70 W source).

I think that 10% is awfully optimistic: I'll have to see it to believe it.

[edit: miscalculated, see below]

[8-Ball]

My basis for this is the data posted by since87 in the first post whereby Hard copper had an electrical conductivity 89.5% of that of soft copper.

For the majority of materials which use elecron motion in thermal conduction, there is a linear relationship between electricla conductivity and thermal conductivity.

I'm not 100% sure about this but I will be revising "electronic structure of materials" over the next few days, and there is mention of the relationship between thermal conductivity and elctrical conductivity.

Actually, I suppose, I could just dig it out later on today. Will post my findings later.

8-ball

[8-Ball]

Also be aware, that it depends largely on the initial level of cold working and the level to which the recrystallisation goes to completion.

8-ball

[BillA]

the relationship between thermal and electrical conductivity is well understood
(no need to cite the refs)

Null-A Ben
yet again, think harder
it is ONLY the thermal conductivity that is being changed,
not ALL those factors that together form the basis of a wb's "C/W"
jeez, you gotta start checking what you say before you post it
where we at now ? 4500 posts ?
goddam Null-A

[bigben2k]

Ah yes, minus the 0.05 for the TIM, leaving 0.14 for the block... 0.98 degrees.

What else did I miss?

[Since87]

Originally posted by bigben2k
Ah yes, minus the 0.05 for the TIM, leaving 0.14 for the block... 0.98 degrees.

What else did I miss?

Differences in temperature gradient across the "furniture" of the block and changes in heat spreading.

[8-Ball]

I think what BillA is trying to get at here is that thermal conductivity of the material is not simply a constant which is, say, multiplied by a factor based upon the design of the internals to provide an efficiency.

Changing the thermal conductivity WILL change the temperature difference between the water and the cpu, but, for the same reason that this happens, the way the heat gets from the block to the water will also change.

(Note that I am not educated in the area of heat flow, so my choice of words may be incorrect at times, but so long as the general idea gets across, I'm happy)

As you know, the heat flux is the thermal conductivity of the material multipied by the temperature gradient. So, if we have increased the thermal conductivity, and the net heat flux is to remain the same, the thermal gradient must have become shallower.

Lets take White Water as an example and lets suppose it is now made of Aluminium

The thermal conductivity is lower, so the temperature gradient must be higher for the same heat flux. This means that there will be a higher the baseplate will be at a higher temperature than the fins (steeper gradient) as compared with the situation in the same block made of copper, where the temp gradient will be shallower.

This will affect the way the block transfers thermal energy.

I guess what I'm trying to say is that the relationship between thermal conductivity of the block and it's C/W is not a linear one.

Also note, this is grossly over simplified, there are other factors to take into account. This was merely an explanation of a point to be made.

8-ball

[BillA]

Null-A Ben
the above post is an example of why 8-Ball's posts will be read for their content
and in contrast shows how yours confuse, mislead, and obscure every single aspect you purport to comment upon
200 vs. 4400

sad, eh ?

[8-Ball]

Didn't realise I'd broken 200

It was about thirty a few weeks ago, and then I decided that I should start revising for my finals.

Strange isn't it, that the start of my "revising" and the start of my mad posting marathon coincided like that.

Could have something to do with me being 2 weeks behind on my revision schedule.

8-ball

[EDIT: I suppose I did have to do "some" revision to be able to post everything in this thread ]

[bigben2k]

Originally posted by unregistered
Null-A Ben
the above post is an example of why 8-Ball's posts will be read for their content
and in contrast shows how yours confuse, mislead, and obscure every single aspect you purport to comment upon
200 vs. 4400

sad, eh ?

Actually, I'd rather think that it poses questions, hence the question marks, but to each his own.

[BillA]

ok 8-ball
thats it,
you're banned 'till after finals

Regards

[myv65]

@bigben,

Ben, you often seem like a programmer that's trying to create a massive program but has only a list of programming terms to use and none of the understanding of what those terms actually do. You're able to type in the terms, but don't seem to understand the syntax. Bill's point here is that the "C/W" of a block is more than simple conductivity. In the loosest sense, C/W implies a temperature change based on heat load. In the specific case of Bill's testing and waterblocks, it means the temperature between the base of the block and the water flowing through it.

This means that a block's C/W is really two-fold (ignoring minor stuff like convection to air off its surface and virtually zero radiation energy exchange). The first part is conduction of the heat through the block structure. The second part is the convective resistance getting that heat into the water. This second part is purely a surface phenomenon and other than how the heat is distributed over the surface, conductivity has nothing to do with it.

This is why dropping a block's conductivity by 10% via cold-working is not so bad. Conduction through the block is probably (just an off-the-cuff guess here) about 50% (ideal would be 50%) of the C/W of a block. So improving conductivity by 10% would mean a delta-T reduction of about 5% of the delta-T that occurs from block base to fluid.

A change of this magnitude is piss in the wind when compared to the variation that Bill gets from one block mounting to the next. The only way this is measurable is a test geared specifically toward a measurement of conductivity, which not even Bill's fancy get-up could manage.

I'm all for folks exploring the unknown and seeing what's possible, but I prefer to be realistic about the potential benefit. The potential of this one is not so big, especially when compared against using "good" copper versus a lower conductivity alloy or aluminum.

[theetruscan]

Originally posted by myv65
In the specific case of Bill's testing and waterblocks, it means the temperature between the base of the block and the water flowing through it.

I'm probably just confused, but it was my impression (I looked over a few of his tests, but I'm kind of stupid) that BillA's C/W numbers meant the difference between the die and the water flowing through the block, and that would include the thermal interface material layer. Am I wrong? Thanks.

[myv65]

Your question is best answered by Bill, but I'll say that he has made an attempt at quantifying the resistance of the TIM joint so that the actual differences in block performance may be better approximated. The main point I was illustrating above is that the block's conductivity is only a portion of its total thermal resistance. The overall block performance is dictated by both its internal resistance (conductivity, material, and "internal" geometry) and its convective resistance (surface geometry). A good block design balances these two.

[BillA]

theetruscan
several different terms used

"C/W" in quotes is the die to coolant inlet temp difference ÷ Watts
it is in quotes as it includes the TIM joint offset which is not actually part of the wb, and also because the "C/W" is related to the flow rate

T/W has been used to denote the thermal resistance of the wb itself and was experimentally obtained with a TC or RTD in a hole in the bp 0.52" off the bp face
- this had some problems with the hole placement verification
but this data was used to try to quantify the TIM joint thermal resistance (a 'correct' C/W as it is a constant, for a given mounting)

C/W will appear again shortly in a writeup on ThermalDesign rads,
and in another article on rad and fan selection
- this C/W will use the difference between the air and the mean coolant temp (heat exchangers normally use the log mean temp difference but with the small deltaTs it is not worth the effort)

it is NOT correct to say that the TIM joint has a C/W of 0.05
my data suggests a value of perhaps twice that (the 0.05 value was in a 'what if' discussion)
(perhaps Les recalls the threads on OCAU where this was thrashed out - to no definitive conclusion)
Aavid has some good papers describing the calorimeter setup to make such measurements, mine is not really suitable