Below are the questions asked during the presentation, along with their respective answers.
Q: You said before that if we are unable to develop thermal management technology further than today’s level, that would mean that the development of electronics as a whole would also have peaked. But isn’t it true that as we develop more efficient electronics, the need for thermal management decreased?
A: That is true – but we’ve also seen that any ceiling gained through improvements in efficiency, rather than being utilised towards reducing thermal requirements, have instantly been claimed by making applications more powerful, thus re-establishing the status quo ante.
From “peak thermal”, the only growth can come from improving the efficiency of electronics. This is of course an ongoing process, but to date, it has been a much slower road than improving our ability to dissipate heat from applications.
So, while there would eventually be an “iPhone XI” … it wouldn’t be as exciting. It would hardly be the same kind of leap in capability as previous generations.
Q: Please expand on why 1p Filler is not recommended for automotive.
A: This is simply because the 1p filler doesn’t cure – it stays in a viscous or semi-viscous state indefinitely, and being a gap filler, it won’t be fixed by the pressure between the interface surfaces, but will potentially be free to migrate away from the interface.
By contrast, a 2p filler when pressed out into its final shape will then cure into that shape. Being thus fixed by its own geometry, it won’t be as prone to go anywhere.
This means that in using a 1p filler, one needs to take extra care in designing the geometries and testing the materials in proper conditions. I’m not saying it won’t work – I’ve seen it done successfully – but I am saying it creates a bit more homework, because I’ve also seen it done catastrophically wrong.
Q: What are the highest conductivity TIM materials currently available on the market?
A: This is a very tricky question. As I mentioned, the differences between different methods of measurement are significant. In the spectrum of the highest thermal conductivities – the nominal 25+W/(m·K) range – there are also further complexities. The thermal performance of these materials tends to be have a strong inverse pressure dependency – the thermal conductivity itself actually seems to decrease (and in some instances I’ve seen, “collapse” might be a more appropriate word) with increased deflection.
So, speaking of polymer-based TIM’s and disregarding e.g. solders, if we take the manufacturers’ data at face value, the highest thermal conductivities generally available at this time would be somewhere around 35W/(m·K). However, as always – and especially with these materials – YMMV. Testing and verification in your application’s conditions is strongly recommended.
Q: Is there an IBM or Intel standard method to test thermal resistance?
A: I am not familiar with what test methods these two companies employ, but variations of ASTM D5470 is used by many companies.
Q: What’s the relative cost difference between pads and gap filler materials?
A: As these material types follow very different logics in terms of production, logistics, and design rules, the cost on the bottom line is very difficult to compare mutatis mutandis.
What is the impact of being able to design for narrower bond lines, thus thinner material layers? …of having just one item in the BOM as opposed to one for each unique geometry? …of dispensing with an automated system as opposed to manually? …of not being able to optimise the material to the performance of each individual heat source, but having to select the material for all instances according to the requirements of the most challenging one? And so on.
Ultimately, I cannot say that one in general terms is, everything else being equal, more or less expensive than the other. It will need to be calculated in each individual case.
Q: Some time ago I remember seeing a paper referring to a standard thermal resistance testing method.
A: There are several standard methods out there; they tend to measure slightly different things. Specifically for thermal resistance, ASTM D5470 is by far the best known one.
While on this subject – the difficulty with measuring thermal conductivity is that there is no way to just apply a meter stick to it. It’s not like measuring dimensions, or temperature, or weight, or time – you can’t measure it directly and just read it off.
Instead, you need to create a set of circumstances that allows you to infer the conductivity from something else, such as a temperature change over time, or a temperature drop over distance.
And so, ASTM D5470 infers conductivity out of resistance. HotDisk infers it out of monitoring a temperature change over time. Laser Flash infers it out of thermal diffusivity, and so on.
Each have their own strengths, and their own weaknesses.
Personally, I swear by HotDisk.
Q: What will be the future growth of heat transfer tech?
A: The growth of the requirement for heat transfer tech is, as we have seen, exponential. It will be up to the industry to answer that requirement. And in places like cloud data centres, it seems we are fast approaching the limit of what physics even allows us to do – I’d be surprised if we can grow our ability to handle heat flux from data centre IC’s very significantly beyond the next five or ten years.
This means that the bulk of all the added intelligence that IoT and AI solutions will necessitate will have to be distributed further down the data infrastructure – into the “fog”, “mist”, and “edge”. This in turn will mean that even higher levels of ingenuity will be required from those of us who supply heat transfer solutions into the “moderate to high” (as opposed to “very high to extreme”) heat flux ranges.
So, as Niels Bohr said: “Prediction is very difficult, especially about the future” … but endeavouring to do so anyway, I’d predict that like the requirement of it, heat transfer technology – both as a business and in pure terms of watts dissipated – will also grow exponentially.
Q: What is the better solution for using graphene sheets. One thicker or more thinner sheets?
A: Graphene is still and emerging technology, as far as TIM’s are concerned. I am not even convinced that graphene in sheet form will see much use as TIM’s, they way graphite sheets have been used – they will probably be primarily used as heat spreaders.
But trying to anticipate the gist of the question anyway – thinner sheets will lower thermal resistance by shortening the distance through the material. Thicker sheets will leverage more of the material’s heat spreading ability, thus potentially lowering the thermal resistance by increasing the area.
Which will win out?
That is something that I suspect will greatly depend upon the properties of the graphene sheet being considered and the heat load and other boundary conditions surrounding it. Just off the top of my head, I couldn’t say which is better.
Q: What is the typical compression percentage for a silicone gap pad where you can assume it will behave elastically vs. being permanently deformed?
A: That will depend (a) on the material and (b) on the time elapsed between deformation and relaxation.
TIM pads in general have fairly limited rebound, but there are significant differences. Some pads will permanently deform instantly after just 15% of deflection; others can take over 50% of deflection over weeks and still recover to better than 80% of original thickness.
The range being this wide, I couldn’t say what would constitute typical.
Q: Most thermal gap filler k-values peak in the 7-10 w/m*k range. What do you see as the next step in creating silicone gap filler materials that exceed this range? Is Nolato exploring these new technologies?
A: We certainly are!
Our highest released thermal conductivity at the time of this writing is 14W/(m·K), and more products with even higher conductivities are in development.
As they say “keep an eye on this space!”
Q: How much influence do contact resistance have on the total resistance for the filler? Is conductivity or contact resistance the main resistance?
A: The influence is significant. In very general terms, the thinner the material is, the larger the proportion of the contact resistance. The thicker the material is, the more the bulk resistance of the material itself takes over.
At which point the break-even is, depends largely on the material. Materials with low Shore hardnesses and/or tacky or “wet” surfaces generally show low contact resistances, while dry/high-Shore surfaces have higher resistances.
And BTW – that is the only useful information you get out of the durometer hardness. It will not tell you anything about how stiff the material is against your components.
Just so you know.