A radio amateur’s helper: three thermal adhesives and one thermal paste

Not long ago, I reviewed a thermal adhesive for attaching all sorts of heatsinks to all sorts of electronic components, and one of my readers asked me to test a few more similar products.
So today I’m reviewing—or, more precisely, testing—three types of adhesive and one thermal paste. And at the end of the review, there’ll be a table and a chart bringing together all the thermal (and not-so-thermal) compounds and materials I’ve tested.


Here’s everything I received for testing today.
Three adhesives:
1. Tian Mu
2. Kafuter K-5202
3. Kafuter K-5204K

And one thermal paste: GD9000

I can already hear the question: what’s thermal paste doing alongside three tubes of adhesive? Simple: I have two test heatsinks, and it’s easier for me to run the tests in pairs. That way, I’ll have an even number of “test subjects” in the review.


I’ll start with the thermal paste. I think it’s quite well known among radio amateurs and computer enthusiasts, but I’d never used it before. I was used to the basic KPT-8 and KPT-19 pastes. I actually prefer the former, because the latter, supposedly non-drying, has already started to dry out...


The packaging is just a plastic syringe. It weighs 7 grams; I won’t weigh it myself, since I don’t see much point.
It costs 50 UAH (a little under two dollars), link.


The paste is more fluid than the familiar KPT-8. Interestingly, at first it squeezed out as a uniform mass, but when I took the cap off a month later, I saw that around the edge it had formed something like a little tube of firmer material.
This hasn’t affected its performance at all; it’s just something I noticed.
The stated thermal conductivity is 4.8 W/mK, substantially higher than KPT-8, which is rated at around 0.7 W/mK.


Next up is the adhesive.
I’d heard about an interesting adhesive called Tian Mu, which offers a good balance of price, performance, and volume. Similar adhesives are sold under various numbers—704, 705, 706, and so on. Mine had no number and was simply marked RTV. Also, 704, 705, 706, etc. are sealants or general-purpose adhesives, not thermal adhesives, although they’re often sold on AliExpress as thermal adhesive.
It costs 47.5 UAH (about $1.70), link.


It comes in a tube containing about 60 grams. It has a distinctive “feature”: the metal is so dented that it looks like someone chewed on it. A nozzle is included to make the adhesive easier to apply.
Operating temperature: -50 to +280 °C; thermal conductivity: 1.5 W/mK.


The compound is fairly runny and a deep gray color. It has a slight smell—not the familiar smell of silicone sealant, just a chemical odor.


These adhesives are made by Kafuter. I have two samples: K-5202 and K-5204K.
K-5202 costs 99.75 UAH (about $3.50), link
K-5204K costs 128.25 UAH (about $4.50), link.


Specifications:
K-5202: thermal conductivity 0.8 W/mK; initial curing time (at 25 °C) ≤30 min; operating temperature -60 to 280 °C; gray.
K-5204K: thermal conductivity 1.6 W/mK; initial curing time (at 25 °C) ≤10 min; operating temperature -60 to 280 °C; white.

As you can see, the main difference is thermal conductivity: the 5204 has a higher rating. There are also differences in the secondary specs—the 5204 cures faster—and, of course, in color.


The packaging is also a tube, weighing 80 grams. For some reason, only one tube came with a nozzle; the other one probably got lost “along the way.”


As you can imagine, this compound is also entirely Chinese, so most consumers—at least in our countries—will find the description hard to understand.


Broadly speaking, it seems to be the same kind of mixture as Tian Mu, but there are a few differences.
1. The mixture is less runny. With Tian Mu, I easily squeezed some out and it started to “drip”; this one tried to retract back into the tube.
2. There’s a slight chemical smell, especially with the 5204.



And so, dear readers, here are the four “test subjects.” They will be tested in the following pairs:
Paste + Tian Mu
The two Kafuters

The test setup is the same as in the previous reviews: two heatsinks, two high-power resistors, a power supply, a small board, and a 1 kg weight. The procedure is unchanged too: a 1-hour test, measuring the temperature and calculating the difference every 10 minutes.


And the first thing I noticed: at first, glue came out of the tube, but when I tried to apply it to the heatsink, some kind of liquid started leaking out. I wiped it off and tried again. There was less liquid, but it was still there, and the glue wouldn't wet the surface, instead rolling into strange little clumps. In the end, I had to squeeze some glue onto a piece of paper first, and only then apply it.


I set everything up, let it sit for about an hour and a half, and started the test. The power was 20 W, 10 W per resistor, and the heatsinks were identical. Unlike the Stars 922 glue test, this time I didn't remove the clamp, and each sample was held down with a 500 g weight.


Even without comparing the results, you can see that the glue's thermal conductivity is roughly on par with, or even better than, KPT-8, which is pretty good. The paste clearly performed better, though it was advertised as having better specs to begin with.


I removed the resistors about 6 hours later. At first, the layer of glue might look enormous, but it's quite runny, so most of it simply squeezed out beyond the edges of the resistor casing, leaving only a thin layer underneath. The glue holds moderately well and has hardened noticeably, both on the outside and inside. In some places it was quite difficult to scrape off the heatsink; in others, it had hardened.


The Kafuter glue behaved much like Tian Mu: liquid came out first, followed by the glue. With the 5202, it was almost exactly the same as with Tian Mu; with the 5204, the glue itself came out first, and only then did the liquid start leaking. Here too, I had to squeeze some glue onto a piece of paper first, and only then apply it to the heatsink.

I also noticed another difference between the 5202 and 5204: when pressing and rubbing the latter into place, it felt as if the compound contained sand. There was a certain "grittiness" as it moved parallel to the heatsink surface.


The test "setup" and the "test subjects." The test was run 4–5 hours after gluing.


The result is strange: the 5202 performed better, even though its stated thermal conductivity is noticeably lower.


When I pulled them off, it turned out that the glue held about as well as Tian Mu. But while the compound had hardened on the outside, it was still almost liquid inside (I smeared it around a little with tweezers in the top photo), even though I removed the resistors after the same 6 hours. The glue's texture also felt rubbery; you can see in the photo how the resistor is "hanging on" to the heatsink.
The layer of 5204 glue (bottom photo) felt thicker, even though the clamping force during application was about the same.



The previous test results surprised me a little. I even started wondering whether I hadn't pressed them down firmly enough when gluing, or whether the heatsink had ended up rotated 90 degrees. So I decided to repeat the test.
I removed the resistors, wiped down the heatsinks, positioned them correctly, and glued the resistors back on, carefully checking the pressure. I felt the "sandy" component in the 5204 glue again, then ran the test once more.


Basically, nothing much changed.


But I didn't stop there. I ran the test in "bare" form, without the clamp or the bar I use to hold the resistors down.


The results got a little worse because some of the heat was no longer dissipated through the bar that holds the resistors down, but the difference wasn't dramatic. By the way, I tested the Stars 922 paste without clamping, which means the results with the bar would have been better. I should probably either repeat that test or adjust the results; the latter seems easier.


About a day later, I removed the resistors. But after pulling off the first one, I realized I should have tested them in that state first.
It turned out that neither glue had cured inside yet. There was only a strip about 1.5 mm wide around the perimeter; the rest was still liquid.
Even so, they were very difficult to pull off. For the K-5202, I even had to use pliers, since I couldn't remove it by hand despite applying quite a lot of force. Well, you can pull it off if you really put your back into it, but it takes some effort. I think once the adhesive has fully cured, it will hold much more firmly, which is a big plus for some applications.


Since K-5204K is supposed to have better specs, I ran one more experiment so nobody could accuse me of using warped heatsinks, not pressing them down properly, or something like that.
For this test, I swapped the samples around and also made sure once again that the amount of adhesive and the pressure were the same.


And guess what? There was almost no difference. The K-5204K results were actually even worse, while the 5202 results stayed the same.


Here’s a comparison table; lower is better. Unfortunately, I don’t have much experience putting together charts like these yet, but I’ll figure it out eventually.
The table also includes results from other reviews, some of which are... rather strange :)
Thermally Conductive Rubber
K5 Pro, or What to Do If You Don’t Have Thermal Paste on Hand
Stars-922 Thermal Adhesive


Now for the conclusions. There’s not much to say about the paste: it conducts heat reasonably well, noticeably better than KPT-8, and can easily shave off a few degrees. But if the contact area is large and the power being dissipated is low, KPT-8 will still do the job.
As for the adhesive, I’ll discuss that separately.
First, Tian Mu. In my opinion, it’s simply a great adhesive. It has decent thermal conductivity, dries fairly quickly, and holds moderately well. So I wouldn’t use it to attach heavy heatsinks or components in devices that might get knocked around. But as they say, it’s all relative, and in practice, removing a heatsink attached with it still won’t be easy.
And of course, the Kafuter adhesives. The results are very strange. According to the descriptions, 5204 should perform better and be roughly on par with the Tian Mu shown above, but in reality it’s the other way around: 5202 has thermal conductivity comparable to Tian Mu, while 5204 is noticeably worse.
But aside from its thermal conductivity, I liked the 5202 because it dries faster and holds better than the 5204. Then again, the 5202 is supposed to reach initial cure in 30 minutes, while the 5204 is listed at 10 minutes. It feels like the labels on the tubes may have been mixed up by mistake, although the colors match: 5202 is gray and 5204 is white.

If you asked me which one I’d choose, I’d say without hesitation either Tian Mu or Kafuter K-5202. The 5204 adhesive shown in this review doesn’t make much sense to me at all, since it costs more and performs worse than 5202. Maybe I just got lucky with my samples; maybe they really are this odd. I can’t say for sure, but all the test results are above.


That’s all for today. I hope you found this information useful.