5

I have several sheets of a plastic like material that I recovered after stripping down industrial led light banks.

One type is a clear plastic like sheet, the other type is a semi opaque sheet used as a light diffuser. Neither sheet has any marking of any kind. The light banks are generic and unbranded.

I plan to paint them and use them in a craft project, the primer I use says not for use on acrylic or polystyrene.

Without having access to any material data sheets is the a good way to ID the above, for example burn color or hardness, or reaction to certain chemicals?

I'm looking for an ID beyond the normal "paint a sample and see what happens". Which I will be doing anyway.

Aaargh Zombies
  • 2,058
  • 5
  • 21

1 Answers1

7

If your primer tells you what solvent it's based on, you can test a sample of material with the same solvent. One reason for not using a particular solvent with a particular plastic is that it dissolves the plastic. But some surface etching may be desirable to improve adhesion. Acetone, for example, will attack most of these, dissolving polystyrene pretty easily.

Instead, or before testing with solvents, a combination of density and the softening temperature should help. With reference to this table (sadly only in °F), we can see that thermoplastics differ in their density (a.k.a. specific gravity) and the temperature at which they can be formed hot.

For materials that sink, we can use the water displacement method to work out the volume of a sample, and weigh it, to calculate the density. Water has a density of 0.997g/cm³ (to all intents and purpose, 1g/cm³). You need reasonable precision for this sort of thing - curiosity like this is one reason why I have a cheap set of 0.1g kitchen scales. Acrylic is denser than polystyrene and ABS, but polycarbonate is denser than all 3. HDPE will float - just about.

The same table has a column for lower processing temperature. At this temperature, heated in an oven, it should be possible to bend a sheet of the material without breaking, and have it hold its shape. You'll need a good oven thermometer or infrared thermometer, as oven thermostats aren't accurate enough. Start cool and work hotter.

Unfortunately both ABS and PS soften at around 260°F (130°C). And they have the same density. Sheet polystyrene is rather brittle at a cool room temperature (CD jewel cases are made of PS); ABS much less so.

Thin polycarbonate bends in a springy way even down to very low temperatures. So does acrylic, but less so. Another table this time in °C, shows temperatures below which plastic become brittle. These are rather soft boundaries.

Acrylic is a bit harder to scratch than polycarbonate, but without comparing to known pieces it's hard to judge.

If you were going to compare to known materials, they all tend to have distinctive smells when you saw them or heat them (not to burning, but higher than the minimum forming temperature) - but you don't want to inhale great big lungfuls of the fumes anyway.

These are my guesses as to the most likely materials. All are hydrocarbons materials, and will burn with a yellow flame and give off a fair bit of soot. They can also have additives to make them less flammable.

Chris H
  • 6,789
  • 12
  • 26
  • TL;DR - it's not easy. I have in the past had access to lab equipment that was very good at distinguishing polymers, but it's not something you can replicate at home. – Chris H Jul 18 '23 at 12:41
  • 2
    Would a 5MHz benchtop NMR do the trick? There are a few models for sale, and somewhere in my personal archive I've got the schematics for building one... – Perkins Jul 18 '23 at 19:23
  • 1
    @Perkins I've never done any NMR. I was thinking of Raman spectroscopy – Chris H Jul 18 '23 at 19:29
  • Interesting. I'd never looked into Raman spectroscopy before. Seems like it would be even simpler to build though. Big question is if there's open-source software for processing the data like there is with NMR. If there is, then building one for personal use likely wouldn't be too horribly difficult given how cheap lasers and lenses have gotten these days. – Perkins Jul 18 '23 at 19:40
  • @Perkins lasers and lenses can be cheap but the spectral purity of cheap lasers is really poor, and the filters to reject the laser light are expensive. Even with a strong signal you need 8+ orders of magnitude transmission change in a few tens of nm. Then you need a decent spectrometer to resolve the spectrum. I could probably build a really basic system from scratch, but it wouldn't be cheap. From a software point of view it's quite easy, though there's no major established open source tool. What we'd need here are libraries of reference spectra, and there are now published resources – Chris H Jul 18 '23 at 19:51
  • @Perkins There's a Python library for basically everything. Like RamPy, "a Python library that aims at helping processing spectroscopic data, such as Raman, Infrared or XAS spectra." – Mast Jul 19 '23 at 08:50
  • 2
    There's lots of people tinkering with home-workshop NMR, Raman spectroscopy and so on, but the bottom line is that signals are extremely elusive and there's no "easy way" yet. More promising, there's been various attempts to identify plastic for recycling by shining different wavelengths of IR into it and examining the response: the last I saw it relied heavily on "machine learning" i.e. is likely to be snake oil. For the "small shop", there's likely to be no better way than testing with solvents: which after all mirrors paint selection which is OP's question. – Mark Morgan Lloyd Jul 19 '23 at 08:58
  • @MarkMorganLloyd IR has some promise, as does sufficiently constrained AI in spectroscopy. The state of the art a few years ago used AI pre-screening of Raman data for expert review, which is reasonable in some contexts, but not a workshop or a recycling plant. There's also a big problem with dyes in coloured plastic for recycling. – Chris H Jul 19 '23 at 09:30
  • 1
    @Mast-onstrike RamPy would have been handy for me during my PhD, but I still would have had to write the trickier bits of my own code - I had to locate peaks on the shoulders of much stronger peaks which even sequential curve-fitting with subtraction failed to do – Chris H Jul 19 '23 at 09:33
  • 1
    @ChrisH OTOH, there's probably a limited number of dyes added to recyclable plastics (i.e. thermoplastics) heavily weighted towards getting RAL colours, and these would be expected to have predictable IR etc. characteristics. However this is your speciality, not mine so I'd best stop holding forth :-) I think the bottom line in this question is that (a) it's the solvents that matter and (b) OP might be looking at some weird copolymer, so there's no better answer than "suck it and see". – Mark Morgan Lloyd Jul 19 '23 at 09:38
  • @Perkins Presumably an NMR trying to characterise a specimen (rather than verify the presence of a single compound) would have to generate a substantial number of "pings" in order to capture a signal over a fairly wide frequency range. Have you seen the NanoVNA and related TinySA designs? Seems to me like they'd be a good starting point... – Mark Morgan Lloyd Jul 19 '23 at 21:56
  • @MarkMorganLloyd I expect you're right. You'd have to hit the whole range of potential frequencies that might be there, and then compare their proportions to get an idea of what it is. I've never actually tried it though, partly because I'm just an amateur, and partly because what study I've done is from the side of using the resonance to do interesting things to the sample, rather than gleaning information about the sample. – Perkins Jul 20 '23 at 20:37
  • @Perkins this https://www.sciencedirect.com/science/article/abs/pii/S109078072030118X looks interesting, suggesting a price in the low $100s. – Mark Morgan Lloyd Jul 20 '23 at 21:35