When selecting the right Phase Change Material Pad thickness for your application, thinner isn’t always better
As any thermal design engineer knows, all other things being equal, Thermal Interface Materials with higher thermal conductivity and lower thermal impedance are better than those with a lower conductivity and higher thermal impedance (Rth). But is the same true when it comes to material thickness?
A Thermal Design Experiment for CPU and GPU
Our engineers set out to test a range of Phase Change Materials to discover when selecting pad thickness if better thermal properties and thinner bond lines translated to better thermal cooling of CPUs and GPUs.
We started with a selection of Phase Change Materials and ultimately set
| Product (Unit) | Thermal Impedance (°C-cm²/W) | Thermal conductivity (W/mK) | Density (g/cc) |
|---|---|---|---|
| Test Method | ASTM D5470 | ASTM D5470 | |
| PTM5000 | 0.086 | 4.4 | 2.3 |
| PTM7000 | 0.058 | 6.5 | 2.7 |
| PTM7900 | 0.046 | 8 | 2.8 |
| PTM 7950 | 0.04 | 8.5 | 2.8 |
GPU Manufacturers matter
There are many Graphic Processing Unit (GPU) manufacturers, including AMD, Asus, and Intel, but by far the largest GPU manufacturer is Nvidia with the lion’s share of the GPU market so it made sense that we would carry out our test on the Nvidia GPU.
Phase Change Pad Thicknesses
Phase Change Materials can be supplied either as stencil-print versions which require the thermal interface material to be applied to a stencil of a set thickness and then printed onto the substrate.
These stencil thicknesses typically range from 0.2mm to 0.3mm and this is the thickness that is then applied. An important note is that Phase Change Materials provided in stencil-print format need a drying step (either 20 mins @ 80°C or 24hrs at room temperature) in order to have the solvent evaporate and dry to leave behind only the phase change material itself.
Using stencil printing, it is important to note that the height reduction (and therefore the ultimate thickness) is reduced on average about 20%, so it is important to take this into consideration when determining your stencil thickness.
Another option is to but “pre-cured and dried” Phase Change Pads which are the exact same material, but with the solvent removed, leaving only the Phase Change Pad. In these cases, of course, there is no drying step and the initial thickness is also the final thickness.
Phase Change Pad Candidates
After considering the options, we settled on three products: PTM7000, PTM7900 and PTM7950, and three thicknesses: 0.2mm, 0.25mm, and 0.3mm.
Result: Thinner isn’t always better
The results we got were surprising. What we expected was that we would get the best thermal performance (ie. the most cooling & thus the lowest temperature rise) by using the product with the best thermal properties (PTM7950) with the thinnest possible bond line (0.2mm).
The numbers below were not exact but representative of the data we observed to indicate the trend in temperature change
| Product | Thickness (mm) | Temperature Change (°C) |
| PTM7000 | 0.2mm | +2°C |
| PTM7000 | 0.25mm | +1°C |
| PTM7000 | 0.3mm | 0°C (baseline temperature) |
| PTM7900 | 0.2mm | -4°C |
| PTM7900 | 0.25mm | -5°C |
| PTM7900 | 0.3mm | -6°C |
| PTM7950 | 0.2mm | -5°C |
| PTM7950 | 0.25mm | -6°C |
| PTM7950 | 0.3mm | Not tested (was not available) |
Discussion
What we noticed is that generally there was an improvement as we moved from phase change materials with a higher thermal impedance and lower thermal conductivity to ones with lower thermal impedance and higher thermal conductivity (as expected), but surprisingly, there was no improvement as we moved from 0.3mm to 0.2mm.
Surprisingly, the PTM7950 at 0.3mm offered similar thermal performance to the PTM7900 at 0.25mm, but because the material was not available at the time of testing, we do not know the result of increasing the thickness to 0.3mm on the PTM7950.
Analysis of the parts and the data suggested that part of the problem lay in the Nvidia GPUs themselves. The GPUs tested were quite large and it is suspected that these larger GPUs were not as flat as smaller GPUs meaning that closer and more intimate contact was not possible with the larger GPUs.
Many customers have performed similar tests and we see similar results emerging from the their data.
Conclusion
We can conclude with some level of confidence that for larger GPU’s, that thicknesses of 0.3mm are better suited than thicknesses of 0.2mm or 0.25mm. We can also conclude that the PTM7900 is generally superior to the PTM7000 and the PTM7950 also seems to offer better thermal performance than the PTM7900. Unfortunately, we cannot (yet) conclude that the PTM7950 at 0.3mm would not have given better results. Stay tuned for these results when they become available.
As such, for large GPU’s we would recommend the 0.3mm version of PTM7900
Please visit our website to view our entire range of thermal interface materials including our phase change products including the three materials described in this article: PTM7000, PTM7900, and PTM7950. Here you will also find all relevant technical information including technical data sheets and safety data sheets. You can also contact us if you have any questions about any other Thermal Interface Materials.
