IR transparency behaviour of epoxy and

[abbydee75]

Hi, we've got a device with an optical transmission line made as follows:
1) an infrared transmission device containing a led surrounded by either Dow Corning Sylgard or Araldite D and a 0.4mm thick PES disk as "external window"
2) a 0.5mm free air path
3) a receiver with the same architecture of the tx

We must do a thermical cycle to validate the product: 3h@-20°C, then 3h@+70°C, all this for 2 times.
What happens is this:
a) without thermical cycle our electronics has a gain X on the optical amplifier
b) with Sylgard we see that after the first -20°C the gain need to be lowered (i.e. the optical attenuation diminish), after +70°C the effect goes on the same direction but is less important and another -20 does almost nothing so we can say that the first cycle tends to improve and stabilize the path loss
c) with Araldite D the effect is more intense and after every step, both -20 and +70, ther is a substantial improvement without stabilization.

We made the cycle on some specimens without filler and saw no variations at all, so it must be the polymer that makes the difference.

Can anyone please suggest an interpretation of this?

 

[Demon3]

To me this looks like it may be due to post curing of the epoxy making it contract and put pressure on your window. Therefore I would suggest one of the following:

1. Cure the epoxy longer
2. Cure the epoxy at higher temperature
3. Use a catalyst to help the cure
4. Choose a more reactive epoxy

 

[abbydee75]

Hi Demon, I just talked to the guy who made the tests.
He says that in his last try he removed the PES windows on both tx and rx, so there are just the diodes embedded in the polymer and the air gap between.

 

[Demon3]

Hello,

Thanks for the info. I still think that it may well be the post curing of the epoxy stressing the diode and thereby changing the way light passes out.

 

[Demon3]

The reason I suspect that is that the epoxy is curing a bit more every time you cycle from -20°C to 70°C and so the effect lessens with each test cycle. Still just a guess though.

 

[Compositepro]

The fact that you see the most significant change after a -20C cycle I think rules out any postcure effect. You will not get postcure at that temp. You will maximize the mechanical stresses due to thermal effects. I would surmize that the mechanical stress on silicon in the LED is changing its characteristic response to light.

 

[Demon3]

As I understood it the effect lessens with every temperature cvycle and that is what makes me suspect post cure during the 70°C part of the cycle. A straight mechanical effect cannot explain the observed variations.

 

[IRstuff]

It may have nothing to do with the cure, per se. You've got silicon potted in an acrylic potted in something else.

What's the CTE match or mismatch? Even if it were perfectly matched, you could still have problems, since the LED body may be sufficiently small enough to self-relieve the compression/expansion stress, relative to the diode portion.

TTFN

 

[Demon3]

Well let's divide this into two effects.

I agree that the variations may well be due to strain on the LED caused by the epoxy.

The fact that the magnitute of the magnitute of the variation lessens with each heating cycle may indicate that the epoxy is post curing.

One idea that could potentially help is to first coat the LED in a layer of curing silicone rubber. Then coat the silicone (after cure) with epoxy. The silicone (or other rubber) should act as a buffer when the epoxy contracts. So the epoxy will squeeze the silicone and not the LED! Another way to achieve the same effect is to add very small rubber particles to the epoxy or choose a lower moldulus epoxy.

Your idea of adding filler was a good one but it didn't work. I think that's because although adding filler reduces shrinkage on cure it also increases the modulus of the epoxy (relative to the unfilled) and thereby increases the force on the LED.