I'm trying to bond semiconductor samples (Si and Ge) area ~1-2cm^2 to fiberglass printed circuit board (PCB) We use West system marine epoxy for other things. 105 resin 209 hardener (long cure time.) So that's what I used. (mixed at standard ratio)

I wanted a controlled thickness for electrical isolation. So I also tried adding some fillers to the epoxy. Glass beads (9.8 mil .. kinda thick IMO, sprinkled on surface.) Alumina oxide, 240 grit. (~1 part Al2O3 to 2 parts epoxy, by weight.) All samples (but one Ge) were from an old piece of Si wafer. Samples and pcb were cleaned in acetone, and scrubbed with cotton tipped applicator. Epoxy mixed. applied and samples pushed into place.

And allowed to cure for 24 hours.

They where then dunked into liquid nitrogen (LN2). After a few dunks, warming in room air, the samples glued with Al2O3 filler had fallen off.

After more dunking the samples held with raw epoxy and beads fell off. A few more tortures that also included warming more rapidly with a heat gun. And I'd lost everything but the Ge sample.

As a final abuse the Ge sample was taken from LN2 and placed into a cup of warm water, several times. It remained attached.

All the bonds failed at the Si interface, and the epoxy remained attached to the PCB (except for the glass beads, that failed everywhere.)

So what's wrong?

My first thought was about the coefficients of thermal expansion (CTE) Here's a link to some values, Si is very low.
The pcb is ~ 12-14 ppm.

I then thought about cleaning. The old Si samples may have all sorts of hand grease.

The final difference is that the Si samples are polished on both sides, while the Ge was only onthe top... the bottom was rough.

wow, that was a long question, (sorry)
I made up a new batch of samples today to try and answer those questions. They will cure over the weekend.

I also wonder if I need a different epoxy? The West 105 stays somewhat pliable.
I don't know if that's a good or bad thing.

  • $\begingroup$ I don't know much about bonding to surface boards, but West 105 should not stay pliable if it is mixed correctly. Is it possible that one problem is with your metering or mixing system? It is critical that the components be mixed in the correct ratio. I have used a sensitive scale in the past to measure how much of each component is used. $\endgroup$
    – Ethan48
    Jan 24, 2015 at 1:29
  • 1
    $\begingroup$ @Ethan48, Sorry I guess pliable is too strong a word. This epoxy does seem a bit softer than others I've used in the past. I did use a nice scale (0.01 g resolution) to weight the components. $\endgroup$ Jan 24, 2015 at 13:52
  • $\begingroup$ There are epoxies specifically sold as "die attach" epoxies. Some are conductive but some are not. On the other hand, I don't know how robust they'd be once you add filler. Abelstik, Epotek, and Masterbond are names that come to mind for these products. $\endgroup$
    – The Photon
    Jan 24, 2015 at 19:37
  • $\begingroup$ @ThePhoton, I sent an email to Masterbond today. (they must have ~100 different epoxies, a bit intimidating.) I had decent luck with a clean Si wafer. The polished side of the wafer stuck to everything, the unpolished side fell off the aluminum slab. I had no luck making a controlled thickness. (except with Ge) I placed two pieces of kapton tape (2 mil) at each end of the sample with epoxy in between. Samples cracked right at the tape /epoxy line. I need tape with the same CTE as the epoxy... or visa versa. $\endgroup$ Jan 27, 2015 at 4:46
  • $\begingroup$ Do you really need the epoxy to provide electrical isolation? Because most die attach epoxies will be intended for use with a very thin bond line. Could you modify the PCB so that the pad the chip rests on isn't electrically connected to anything? Could you add an alumina (or other insulating material) spacer between the pcb and the chip? $\endgroup$
    – The Photon
    Jan 27, 2015 at 4:48

3 Answers 3


A few things, hopefully helpful:

  1. Differential thermal contraction is almost certainly your enemy. For most engineering materials, the vast majority of thermal contraction occurs between 300 and 77 K, the two temperatures you are working at. Your PCB is almost certainly shrinking much more than the thing attached to it and cracking your epoxy (normal resin epoxies are notorious for cracking in cryogenic environments).

  2. I work with cryogenics for my 9-5 and we use "GE varnish" for just about everything. Also called IM7031 varnish. Thins with ethanol/toluene mixture, and can be baked dry. Tends not to crack in cryogenic environments. It will hold pretty well without a cure, too.

  3. Another, more permanent option is Stycast, which comes in different flavors for different thermal properties. If you want a LESS permanent option, Apiezon N or H grease works well. H grease is thicker (maybe necessary if you have a big sample that weighs ~1g rather than ~10 mg). Both undergo a glass transition at low temperature and hold on tight while giving electrical isolation and thermal contact.

  4. If you are concerned about intermittent electrical contact, cigarette paper can be wetted by just about all of the "goos" I mentioned, and will make sure there is no accidental contact. Put a layer between your two samples.

  5. A good general purpose reference on cryogenic techniques is Jack Ekin's book, Experimental Methods for Low Temperature Measurements.

  • 1
    $\begingroup$ This part of my problem has moved to someone else. But thanks for the reference to Jack Ekin's text. There are few good low temperature books. G.K Whites "Experimental techniques in LT physics", I've had forever (well since grad school.) I hadn't thought about cigarette paper. I did get it to work by placing some teflon tape under the ends of the sample, Epoxying, and then removing the tape... only a thin layer of epoxy that (presumably) took up all the thermal strain. $\endgroup$ Oct 26, 2017 at 16:15

I think for that I'd probably try using MasterBond EP21TCHT-1 for what your looking for, It has excellent performance on difficult to bond materials and is also excellent at cryogenic temperatures from minus 450 degrees F (4 degrees K) right up to +400 degrees F. The surfaces need to be spotless though, absolutely grease free and slightly roughened for best adhesion.


I'd agree that you're looking for an epoxy that stays relatively flexible upon curing, especially if the thermal expansion rates of the epoxy and substrates are different.

If you haven't already done so, I'd recommend searching for "underfill" epoxy. There are a few Loctite and Masterbond products that fit the bill. They typically flow very well during application, have good structural stability (don't expand/contract), survive solder re-flow temperatures (~250C), and remain flexible upon curing. The cryogenic properties you're looking for might be harder to satisfy.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.