This sounds like environmental stress cracking. PC is very susceptible to it and ABS is as well but to a lesser extent. Basically, any trace of oil (from drill bit) or cleaning product makes the plastic crack and fail. There are tests for this such as bending a plastic bar over a set radius then applying detergent like Isopar and timing the failure.

Ways to prevent it include not exposing the part to these liquids, chosing a different grade (ABS with higher acrylonitrile content is more resistant), or chosing another plastic such as polypropylene which does not have this problem.

Chris DeArmitt PhD
President - Phantom Plastics LLC

Consulting, ideas and training on plastic materials
www.phantomplastics.com

I've had similar issues with some telecommunications components where molded PC/ABS parts were cracking after time under assembly loads. There are many possibilities.

Is it a single discrete crack or a lot of cracks in the general area? As mentioned above, ESC is a possibility, and may show up as a bunch of cracks. A single crack could be creep rupture resulting from static overload. In the picture, is that the crack coming up through the bottom of the picture and stopping in the socket? Hard to see from the photo, but that looks like a crack tip, as if the crack initiated below the picture. From this it does not appear to be ESC from an assembly lube. You need to determine the crack initiation point. Fractographic examination can do that.

Is the cracking always is the same location? Could this be a weld line opening up?

What goes into the enclosure, and what environment is it in? Is this a new part, or a sudden occurrence in an existing part? Did you recently change material or molder?

If this was mine, start with the following. I would run melt flow testing on fresh pellets, on a failed part, and on an unfailed part that's been in the field. Properly molded parts would have melt flow numbers 10-25% higher than fresh pellets. Higher than that indicates a process problem, like poor drying, etc. If the values for the finished parts were significantly different, I would send out a sample for FT-IR just to confirm that the molder didn't slip in another material. I would examine the part for weld lines and see if they coincide with the fractures. And I would do a little fractography to see if the crack initiation site can be determined. ESC shows up as a misty, featureless surface. Creep rupture will look completely different. An impact failure will look completely different.

I have to ask: where are the parts molded?

How big is this part? I would be willing to take a look if you sent me a sample.

Rick Fischer
Principal Engineer
Argonne National Laboratory

rickfischer51:

This part is an enclosure of a hand held devise (300mm x 75mm x 35mm) which has been manufactured for the last two years. The cracks started to appear about 2 month ago in the lower two bosses of the part, their geometry and starting point are random but they always appear in the lower two sockets.

After assembly there is no sign of stress/cracks, but after 1-3 month they are beginning to appear to the naked eye.

The device is HHD and it is no exposed not in use and not in our storage to any extremes.

The part is molded in Taiwan. (which claim that he does not changed resin and use only virgin material)

I'm looking for a test (environmental possible) to simulate the cracks appearing?... and i'm wondering is there a way to test the molded parts (yield / break tensile test) to validate the material?

• http://files.engineering.com/getfile.aspx?folder=8db42bfc-d5ad-4e24-8087-34

The sudden appearance after two years may be significant. Is there any kind of date code on the parts, either directly or on the contents, that can be used to track the time to failure? Something appears to have changed,either in your part or the contents.

Taiwan is not bad. My employers have usually had good luck with stuff sourced in Taiwan. Mainland China, not so much.

Heat generally speeds up any process involving polymers, but you have to be careful. It is not selective. It might speed up creep rupture, but it would also speed up oxidation(ABS) and hydrolysis(PC). You could in theory heat the parts in a vacuum desiccator and accelerate the process. I worked on a problem with hydrolysis in a PC/ABS part where we heated to 90C and got 7 years of aging in a week. There is the so-called "10 degree rule" that states the rate of a thermally activated process doubles fro every 10C increase in temp. This is very approximate and gets applied to everything and anything, but if taken literally, a 20C increase would reduce fracture from 3 months (90 days ) to 0.9 days (22 hrs). Dont get greedy and go too high. You could initiate some other failure mode.

Not sure what you mean by validating the material. I suppose a simple tensile test could help. Make sure you use a dog-bone shaped specimen. I've seen tensile test data on straight specimens that was at best misleading and at worst, totally hosed. My reservations here is the best you could get is the same strength as the data sheet, and that tells you only that the material is just as strong as the data sheet material. The test will tell you if the part is brittle, but you can also tell that by cutting out a piece and bending it with your fingers. You could compare values from new, good used and failed parts. This might confirm the presence of material degradation. The best tests for that are melt flow(cheap and easy at any good lab) and impact testing. If you are trying to determine if its the right material, that is what FT-IR (Fourier Transform-Infa Red Spectroscopy) is for. Its cheap (a couple hundred bucks at a good polymer lab) and there is no arguing with it. There are do it yourself methods like the flame test that may help. See Simple Methods for Identification of Plastics by Dietrich Braun.

There is a bigger picture here. Your making product and shipping it, and the longer you mess around, the greater your exposure. You have to get this fixed and quick, or this will get expensive and/or your customers will scamper. Find a good lab (I use MEI in Virgil IL) and get going. A consultant with experience in forensics can do fractography ( I use Engineering Systems. Aurora IL. The Madison Group in Madison WI is supposed to be good.)

Again, I'm willing to take a look, just because I'm curious. I have elementary training in fractography, 21 hrs of graduate level coursework in polymer science and 30 years of plastic part design experience.

Rick Fischer
Principal Engineer
Argonne National Laboratory

Very good comments from Rick apart from the vacuum dessicator idea (that would remove the oxygen needed to oxidize the ABS and the water needed to hydrolyze the PC).

Chris DeArmitt PhD
President - Phantom Plastics LLC

Consulting, ideas and training on plastic materials
www.phantomplastics.com

Chris, that's the idea of the vacuum desiccator. All I wanted to do was accelerate the cracking, not add oxidation or hydrolysis to the equation. The reported failures are relatively short term with no reported extreme environment, so I was going on the premise that this was simply a stress issue. In the hydrolysis study mentioned above, we placed deionized water on the bottom of the desiccator, then ran melt flow on the samples after exposure at one day intervals. In that case, the samples were unstressed and we were trying to drive hydrolysis. If that produced no results, we could have repeated it with a dose of O2 instead of H20. Of course, a good forensic analysis would start with a complete usage history to determine if oxidation or hydrolysis (or ESC, impact, fatigue, UV exposure, etc) could be the problem.

Rick Fischer
Principal Engineer
Argonne National Laboratory

Excellent tip Rick, thank you! it very informing, your help is highly appreciated.

Could you please explain how you calculated that 20 deg increase temp for 22 hours will approx. simulate 3 month real life?

Oh krapski. I did orders of magnitude instead of halving. Spending too much time with those wackey physicists. A twenty degree rise would reduce 90 days to 22.5 days. Thirty degrees C would reduce it to 11 days. Remember, this is a very very rough estimate. Thermally activated processes typically follow an Arrhenius relationship, ie rate = A*exp(B/T). You would run some experiments, plot the natural log of the rate vs the natural log of 1/T, then fit a curve to get A and B. The 10 degree rule is Arrhenius for dummies.

Rick Fischer
Principal Engineer
Argonne National Laboratory

Hi Rick. That makes a lot of sense. Thank you for explaining your intention with that test method.

Chris DeArmitt PhD
President - Phantom Plastics LLC

Consulting, ideas and training on plastic materials
www.phantomplastics.com