Clemens J.M. Lasance
Editor-in-Chief, Fall 2010 Issue
Some time ago I devoted my editorial to the problems caused by the veeerrryyy slow adherence to the use of SI units to which the U.S. committed itself in 1872. When talking globalization we should speak the same scientific language, and when there are conflicts of interest, logical reasoning should prevail to decide upon the best solution, and when this turns out to be difficult some standardization body should produce the verdict. One example is the incorrect use of thermal impedance by some vendors to describe the thermal resistance of a Thermal Interface Material (TIM). The logical reasoning to get rid of this habit is the following,
“Thermal impedance,” with unit m2K/W is not the right word to indicate unit area thermal resistance, because it violates the electrothermal analogy commonly in use. First, the term is historically reserved to describe time-dependent thermal resistance with unit K/W. In limiting cases, for frequency zero or large enough times approaching steady state, the impedance becomes equal to the resistance. Second, in the electrical world ‘electrical resistance’ and ‘electrical impedance’ have the same unit, namely Ohm. Consequently, ‘thermal impedance’ should have the dimension K/W, not m2K/W. But the real problem is that sticking to the current definition of “thermal impedance” will cause a lot of confusion in the future, because the use of dynamic test methods is the obvious choice for application-specific tests, one output of which is a thermal impedance. It should be added that regarding TIMs, in situ testing is the only way for a designer to achieve a value that makes sense in real life. When quoting the performance of a TIM per area, we propose to use ‘unit area thermal resistance’ or ‘unit thermal resistance’ or ‘area thermal resistance’. The final word is to a standardization body.
Let us summarize the pros and cons.
Pro: The use of the word is generally accepted by some vendors and users in the TIM branch of industry.
Cons: (assumption: we agree on the usefulness of the electrothermal analogy)
• Thermal impedance (m2K/W) is not the right word to describe the output of steady state TIM testers, for reasons outlined above.
• An increasing use of dynamic test methods is expected, one output of which is a thermal impedance (K/W), leading to a lot of confusion.
• Conflict with the established use of dynamic thermal impedance (Zth) in the power device industry.
Alternatively, we could decide to refrain from using m2K/W. However, using m2K/W has its merits:, because the unit area Rth is directly comparable to the inverse of the heat transfer coefficient, or alternatively, equal to the ratio of thickness over effective thermal conductivity, a metric to compare the thermal resistance of various TIMs with different areas. As an example, take the generally accepted feeling that future TIMs should exhibit an Rth < 5 mm2K/W. Let’s assume the Rth of the cooling solution should be less than 20% of the Rth of the TIM, otherwise improving the cooling would not be effective. What does this mean for heff? heff> 1.000.000 W/m2K ! (Recall that we talk effective h, including the area enlargement by a heat sink.)
How to improve upon this situation? Customers unite! Refuse to buy from steady state thermal impedance believers!
Following an example of a letter you could write:
Dear TIM vendor,
I hope you agree with your valued customer who is considered king by your sales department that your choice of continuing to use ‘thermal impedance’ to describe what is essentially a unit area thermal resistance, with dimensions m2K/W, is not in my interest. It leads to unacceptable confusion because thermal impedance is a historically defined parameter to describe the thermal transient response with units of K/W.
Thanks for your understanding.
