Hey there's an interesting factoid I seem to now remember myself as well. And that stuff (in the lightest weight??) is about the consistency of contact cleaner. It smells weird tho. So yeah, how about that stuff then?
Taking a look at it, it appears the FC-40 would be a nice choice for a single phase system, though should be sealed (Vapor Pressure = 3.0 mmHg @ 25C).
Product Data Sheet
MSDS
It's not environmentally friendly, and I recall on the expensive side.
I've been looking into it for other possiblilties as well, and recall a military spec MIL-PRF-87252C. It contains 12 products, but I'm not willing to spend $30 for a copy. Yes, I'm being cheap.
Fortunately, I did run across one of them, XCELTHERM 500M.
Product Data Sheet
MSDS
It looks good, and both it and the Flourinert are less viscous than Silicon Oil. Flourinert is 1.8 centistokes @ 25C, while the XCELTHERM 500M is a bit higher. (5 cSt @ 40C, and looks to be ~2.2 - 2.4 cSt in the upper 60's/low 70's C range).
Chart. Not quite "contact lens cleaner", but not too bad. It actually can get a lower viscosity than water, but not until the temps are above expected (0.318cSt @ 260C
HOT). Thermal conductivity remains stable across the entire temp range (0.1412 W/m-K @ -62.2C to 0.1232 W/m-K @ 260C).
It's rated for a closed loop system, so should work in existing liquid cooling components, non corrosive (big plus), and is environmentally friendly. I've no idea of cost, but an email should take care of that.
(sizes in 1 Qt & larger)
No, I highly disagree.
The entire liquid cooling "movement" was born out of complete desperation and flew on the wings of corporate neglect. Dell, and Apple and such only added water-cooling as a "give'em what they want" afterthought.
It's older than that to me. I was thinking in terms of military projects, primarily aviationics cooling. Boeing used an oil cooling system back in the B1B bomber (I'm not sure of the details on this one), and it's still being done. Lockheed Martin is using such systems as the F-35/JSF (closed loop, not just oil in a passive box). The first copy of the military spec above (rev. A) was published in 1993 IIRC.
Then there's various bits of research. Not just coolants though, as my familiarity gets into material manipulation (optical gates). IBM was doing research on this in the '80s that I'm aware of (one of my professors was leading a project they funded).
If it was properly engineered the silicon itself would have cooling channels built into the design and not a lame-ass HIS just capped on top with a piece of inefficient thermal tape actually making the contact. Socket and package design would at least include some kind of component-level emersion and an exchange flow apparatus - or something.
As a guess, the reasoning was simple. Cost. They had to compromise somewhere, and liquid was, and still is, a more expensive proposition. It would be possible, but it would mean some sort of "liquid block" package. Then add in the standardization issues (packaging, not the die), it would have been a mess. As they could get away with the direction used, it wasn't considered critical IMO. Necessity may change this, but I'd think other alternatives would be approached first, if at all possible.
So, no, I have to think that we're using water and thick copper block sitting on top of the HIS because the people thinking this stuff up are, well, I'll just say "untrained" to be nice about it.
As mentioned, I think cost is the major reason (technical difficulties do exist, but given enough $$$, they could be overcome). But the idea of "untrained" isn't entirely accurate, but not that far off either. In my case, I was aware of the heat issues (interning & many conversations with professors), but I've never spent much time doing cooling systems to that extent. It's always been via experience on the "back end", as a single Thermodynamics course only gets you so far...
And it may not even be a required course. It was a choice when I went through. I opted to take it instead of Dynamics, as it seemed more applicable to me. But even then, I've had to learn by doing.
As for the copper foil I guess something like this would last at least ten years with something like contact cleaner flowing over it steadily 24/7.
I guess water without a softening agent would eat through it sooner though. Three to five years maybe?
It would have a lifespan. An inhibitor is added to ethylene glycol to help, but it eventually can eat it's way through.
As I think about it, such a thin Cu surface would also be rather delicate to deformation, rendering it's advantage moot, if not detrimental. (Too much TIM to fill in the dents (past scratches), and the chip's cooling isn't as efficient as the thicker blocks on the market). Maybe to the point of DOA. 3rd party installation would be at the users peril (voided warranty), and system vendors wouldn't want to deal with it due to defect rates (real or imagined).
Adding some nickel to the mix would help, but I'm not sure how thin it could go without deforming under installation & mounting bracket forces.
That's why precision die cast might be better (than stamp forged, assuming that was what you were thinking), as you can add material to strategic locations to counteract these issues. But I'm still not certian as to the plausible thickness on the mounting surface.
