Sciencemadness Discussion Board

Controlled destruction of phenolic adhesive - advice?

Twospoons - 26-11-2023 at 18:47

We have an issue at work where we need to strip the polyimide layers off a flexible printed circuit in order to assess the mechanical failure of the copper traces. To do this I'm using NaOH in an 80/20 mix of isopropanol and water at 60C. This is very effective at hydrolyzing the polyimide.

The problem is the adhesive that is used to bond the polyimide layer to the copper layer. As far as I can tell this is usually a nitrile modified phenolic resin. All I've been able to find so far is that phenolic resin seems to resist pretty much everything, but is not recommended for use with caustic solutions.

The NaOH solution does slowly damage the adhesive over a couple of hours, but I was wondering if anyone has a suggestion for something that might work a bit faster, but wont affect the copper layers. And preferably doesn't require a hazmat suit.

WGTR - 27-11-2023 at 07:38

Quote: Originally posted by Twospoons  
We have an issue at work where we need to strip the polyimide layers off a flexible printed circuit in order to assess the mechanical failure of the copper traces. To do this I'm using NaOH in an 80/20 mix of isopropanol and water at 60C. This is very effective at hydrolyzing the polyimide.


Is your failure in the copper itself, from something like bending fatigue, and you are trying to cleanly isolate the copper from the substrate/adhesive for analysis?
Quote: Originally posted by Twospoons  
The problem is the adhesive that is used to bond the polyimide layer to the copper layer. As far as I can tell this is usually a nitrile modified phenolic resin. All I've been able to find so far is that phenolic resin seems to resist pretty much everything, but is not recommended for use with caustic solutions.

The NaOH solution does slowly damage the adhesive over a couple of hours, but I was wondering if anyone has a suggestion for something that might work a bit faster, but wont affect the copper layers. And preferably doesn't require a hazmat suit.


Oxygen plasma? That even etches polyimide. It’s not extremely rapid, but it is clean.

Rainwater - 27-11-2023 at 11:01

https://www.researchgate.net/post/Does-someone-have-experien...
Elbow grease and an exacto knife.
This may be of interest
https://doi.org/10.1002/pen.760311007

[Edited on 27-11-2023 by Rainwater]

Twospoons - 27-11-2023 at 13:28

Quote: Originally posted by WGTR  

Is your failure in the copper itself, from something like bending fatigue, and you are trying to cleanly isolate the copper from the substrate/adhesive for analysis?


Thats exactly what we're doing.
Quote:

Elbow grease and an exacto knife.

Thats exactly what we're trying to avoid, as it risks causing more damage to the copper we're trying to analyze.

So it looks like I'm already using the best route available. Even if the NaOH doesn't remove the resin, it does seem to weaken it to the point where mild abrasion works.

Rainwater - 27-11-2023 at 16:46

If you're not attempting a repair but just analysis of the copper traces,
hot sulfuric acid that has been degassed might work.
I mentioned degassed because any O2 present can react and destroy the metal.
If trace amounts are present, damage can occur.
I'm not sure how well you want the metal preserved for analysis.

Copper will harden each time it is bent.
Without annealing, it will become brittle and break
Solving this problem will result in great wealth

Twospoons - 27-11-2023 at 18:10

We're aware of the limitations of flex pcb and try to adhere to the IPC standards for bend radius ... but sometimes the mech team gets its a bit wrong and the bending doesn't occur where intended. Which is why we have a test lab set up to do life cycle testing. The flex in question was failing at ~30,000 cycles, and we need to get it to half a million at least. So .. some work to do there :D

I don't think I want my co-workers handling hot sulfuric acid. Not so sure I want to handle it myself in our current workspace.

[Edited on 28-11-2023 by Twospoons]

WGTR - 28-11-2023 at 05:43

Well, you probably already know this, but it is best to not overlap copper on top and bottom sides of the PCB. Failures are more likely to occur in those areas, usually on the outer radius of the bend.

WGTR - 28-11-2023 at 05:53

One thing that may work surprisingly well is to melt part of a sodium hydroxide pellet onto a soldering iron tip. This will likely strip away your substrate in a matter of seconds, with the advantage that the residue is water soluble. Since it’s a tiny amount it’s not particularly dangerous, although gloves and goggles are needed. The catch is that it would have to be a clean soldering iron tip, otherwise it will tin the copper very well.

Twospoons - 28-11-2023 at 17:00

Worth a try. It certainly works for tinning Litz wire - dipping the wire in NaOH soln then heating with solder strips the enamel and tins the strands in one go. I'm sure I can find a manky old tip to try this with.

Rainwater - 28-11-2023 at 18:14

Not really pcb related, but can be applied to your situation.
I deal with this problem on moving parts within a machine. The only real
solution is what we call high strand count lubricated cable,
conductor brittleness is still an issue, but life span is
greatly increased by the construction of the cable.

instead of a few large conductors, its made up of many, many tiny ones, sandwiched with a conductive
grease. They are hard to terminate, requiring tinting and a sealant on the ends. you will not simply
solder this stuff to a board, or stick it under a lug.
the lubricant can leak out of the cable and make a mess of things.

Pros,
106 90 degree bends @ a bend radius equal to the diameter of the cable

25% of that for 160 degrees
sizes from 0.3mm to 500mm2 (thats 1000MCM for us here in the west)

up to 2kv rated with 12, 24, 240, and 600 volt options
currents as low or high as you would like.

Cons,
hard to terminate,
leaks are conductive so shorts are a big problem.
high inductance 150pH per foot, 16-14awg
High self polarization index when working with DC
Large thermal noise
self ringing characteristics.
very expensive.
not really useful for signals above 40kHz
variable propagation delay

but its a very simple design,
if one strand breaks, the grease and close proximity of the other strands prevents an open fault
there are so many fibers, that the likelihood of multiple breaks, in the exact same place, is very low.

One of my monthly task, is to reposition cable accumulators so that the hard moments do not occur in
the same spot. And some of these machines will put over 100k bends in 24 hours.
I use conductive barrings for when I can but some signals like thermocouples , that isn't an option.

[Edited on 29-11-2023 by Rainwater]

Twospoons - 29-11-2023 at 16:38

Ah, yes, well we obviously work in very different industries. My issue involves a magnetic card reader head moving about 2mm, with the flex carrying traces for an anti-tamper grid. Multistrand cable is just not an option.