The Notebook Review forums were hosted by TechTarget, who shut down them down on January 31, 2022. This static read-only archive was pulled by NBR forum users between January 20 and January 31, 2022, in an effort to make sure that the valuable technical information that had been posted on the forums is preserved. For current discussions, many NBR forum users moved over to NotebookTalk.net after the shutdown.
My question is, could this be used in a laptop / desktop? And has someone done it? I mean, it has more than double the thermal conductivity of the Kryonaut Extreme and isn't electrically conductive. Seems like a decent paste if you can use it...
The boron nitride particles in our compound are in an
aqueous suspension. Once it’s heated, the liquid portion of the suspension evaporates,
leaving behind a crumbly, clay-like paste. This paste has a tightly packed crystalline microstructure, and is extremely conductive (31.4 W/mK at 100 °C).
Water + Metal / Circuits = Bad
To evaporate the liquid from it you would have to exceed 212F which in the CPU/GPU arena is a bad thing.
Thermal compound only fills microscopic air gaps. Since thermal conductivity of air is less than 1 W/K-m.
Heatsinks are tapered to squeeze out all thermal compound except in those microscopic gaps. Since direct 'semiconductor to heatsink' contact is hundreds of W/K-m. Direct contact conducts most heat.
Thermal compound is only for reducing temperatures by single digit degrees. Other parts of a thermal chain, from silicon die to ambient air (including direct semiconductor contact), are about reducing temperatures by tens of degrees.
Thermal compound is successfully hyped when specification numbers are ignored. When other parts of that thermal chain are not understood or addressed. And when knowledge comes only from a massive disinformation campaign. Then thermally equivalent compounds can market at massively inflated prices.
Same disinformation says thermal compound must be replaced. The stuff remains just as thermally conductive even decades later. But that would also hurt profit margins. So disinformation recommends constant thermal compound replacement.
Only when it's not working any longer / i.e. fails to keep temps under 100C
I has one paste I was testing as a replacement fail after 4 days to keep temps in line under 100C w/ just browsing windows open.
Normally this is true unless you have a defective paste in place then the difference will be more substantial. Also instead of looking at just max temps the consistency of the temps over a period of time are more important to reduce the flex in the chips / boards in which they are applied from heat/cool cycles causing brittle / failed connections.
This is a very interesting thermal paste. I may just try it out.
I've been looking for a non-electrically conductive alternative to liquid metal that is better than regular thermal pastes, and this seems to be the answer. The only downside is the curing time, which I can deal with.
Test is simple. Attach a heatsink without thermal compound. Measure temperatures. Then repeat that with thermal compound. If temperatures drop more than a few degrees, then something else is defective. Maybe a defectively machined heatsink (with surface defects or not tapered), or heatsink is not properly pressing firmly on that semiconductor, or some contamination exists between that heatsink and semiconductor (obstructs direct contact).
Even mayonnaise is a perfectly good thermal compound (if life expectancy is irrelevant). Mayonnaise may not remain just as thermally conductive 20 years later. So we use well proven thermal compounds.
If temperatures are approaching 100 degrees C, then something else is defective. Other items in a thermal chain from silicon die to ambient air must be addressed. Other items do tens of degrees cooling. Thermal compound only does single digit degrees cooling - if those other parts are properly installed and not defective.
For example, one learns about the quality of that heatsink. Using another essential specification numbers: watts per degree C.
Due to education from hearsay and advertising, many foolishly believe thermal compound does everything. It is a least relevant item in a thermal connection from silicon die to ambient air. All other parts of that thermal chain do more.
Knowledge from advertising and subjective 'disinformation' articles are why many want thermal compound to do what it does not.
If temperature reductions are more substantial, then find and fix a defective heatsink or how it is not properly mounted. 100 degrees C says another part of that thermal chain is defective. Most heat transfers in direct contact - hundreds of W/K-m.
Also, boron nitride is extremely abrasive - You can scratch a bare die at a minimum with application, and I can imagine plausible situations where the die could be chipped under normal stress with a heatsink attached (GPUs and heatspreaderless use cases).
Chain of thermal is one thing and can change from something as simple as the drivers you're using.
Protection from overheating and frying your machine is the purpose of the paste and other components within the chain.
For instance when I repasted and put things together I didn't push the fan cable in all the way and when I put things back together and hit the power button things fired up and within about a minute it started beeping due to the lack of transference with everything else in place and the 2nd fan working 100% on the single heat sink.
When using a different thermal substance that was much thinner it worked fine for several days and then failed gradually increasing temps consistently over 100C.
So, taking into all accounts of this or that in the chain still requires a decent compound to be applied to remove any gaps and spread the heat load to the sink. In a
laptop at least there's not much play when it comes to thermals if just one part of the cooling system isn't working or making contact. In a
desktop environment it's a bit different with larger air volume being able to move across the components. In my desktop I don't use a paste but a graphite pad between the die and tower to make contact. While it's sufficient for daily 24/7 use with sporadic spikes when under significant load it probably wouldn't be my choice for consistent load like gaming. For a while I was running the tower/pad with the tower not fully tied down to the board and the temps were a bit higher as noticed by the fans running at higher RPM's than they normally would have been. After tightening down the tower properly on both screws the temps/fans dropped back into the 30C range.
I'll try this paste out. Seems interesting. But if it isn't cured, will it work properly?
Might throw this on some piece of hardware I don't care about and see what happens, like an old video card or CPU.
Apparently you missed the point. With or without thermal compound, direct 'semiconductor to heatsink' contact transfers most heat. That thermal compound only make microscopic air gaps thermally conductive. Therefore only reduces temperatures by single digit degrees.
Direct contact and thermal compound are two parallel heat conducting paths. If any thermal compound remains where direct contact must be, then temperature increases.
Furthermore, all heat only transfers in the center half of that semiconductor. Since the heat generator (silicon die) is only millimeters in that center areas. Thermal compound should be applied so sparingly as to only slightly impeded into the outer half.
Meanwhile, no CPU frys. Even in an 80486, a processor that got too hot simply throttled. Apparently an Intel patent that long ago has expired. Now everyone uses throttling so that a missing heatsink does not fry a CPU.
If that CPU got another 100 degrees hotter, it still would not suffer hardware damage. Keeping it below 100 degrees C is about averting software crashed. Silicon dies are that robust - heat tolerant.
What is more important in that thermal chain? Design (quality of) a heatsink. Air flow. Ambient air temperature. Thermal compound is nothing more than an afterthought. It does that little.
Keeping a CPU below 100°C also keeps power efficiency from going out of whack since cooler temperatures yield less electrical resistance. It'd be nice if Intel would raise their TJMAX value to something like 125°C or 150°C though. As you said, silicon dies are extremely robust. Their lifespan is on the order of decades, so they're extremely unlikely to ever fail during their useful lifetime.
That being said, cooler temperatures do make it easier to achieve higher clockspeeds while using less power, but yeah the temperature concern is blown a bit out of proportion. Good thermal paste is still good to have though since every bit helps.
Resistance has almost no significance. Temperature changes affect timing and threshold values. Temperatures are taken as high as possible while maintain a maximum clocking speed. Only one signal too slow or too fast will corrupt execution of one machine instruction; resulting in a complete system crash.
Even voltages above logic one and below logic zero are put into a statistical calculation to determine how many years a computer will operate without crashing on one failed instruction. Clock speeds are selected so that a crash might happen once every 50 years (again, don't remember that exact number) when a CPU executes 24/7.
People overclock when more frequent failures are acceptable. Reliability is a primary consideration when selecting CPU voltages, clock speeds, and maximum temperatures.
Problems? See this thread at archive.org.
