Light testing for leaks in a thin metallic membrane

[spiri]

TL;DR

Is there automated equipment that can look for cracks in thin metallic membranes

We have a small (15 X 25mm), thin membrane (0.03mm thick) that we currently inspect for pinholes and small cracks manually by backlighting and inspecting under magnification, one at a time. Does anyone know of an instrument that can do this inspection for us, with a high degree of accuracy?

 

[bobob]

You might try a photomultiplier tube.

 

[Dullard]

I don't know of anything 'off the shelf,' but there are a number of folks who could build that for you. Assuming that you want maximum information (magnitude, location) and flexibility, A 'vision system' is a good way to go. Contact National Instruments for a list of capable integrators (I'm sure that there are a lot of other options).

 

[bobob]

A photomultiplier tube is off the shelf and you can count photons with it. There are also avalanche photodiodes available off the shelf. You don't need complicated intruments to look for light leaks. All that's required is the phototube (or avalanche photodiode), a power supply and an oscilloscope.

 

[Dullard]

I agree that a photomultipler could be a good solution. I guess it comes down to what 'a high degree of accuracy' means.

 

[bobob]

Well, since a phototube can count photons at the single photon level. I'm not sure how much more accurate you could get. Those are used in spectrophotometers. They are sensitive enough that if there is enough light for you to see, it's way to bright for a photoutbe. Light through a pinhole ought to cause a phototube to go crazy.

 

[Dullard]

How is a phototube at logging the number and location of defects? This sounds like production equipment - that kind of information can be pretty useful, particularly if someone is trying to figure out where the production defects originate.

 

[jrmichler]

Check out machine vision systems. I once put a request into a machine vision industry web site, and had a salesman call the next day. And he had exactly what we needed.

Think carefully about necessary resolution, because your application might only need VGA resolution. Also, monochrome, not color.

 

[Baluncore]

Crack detection in metals can be done in a number of ways.

1. Pass a current through the metal and monitor the distribution of magnetic field using a magnetic fluid or dust. Cracks cause asymmetric currents, that distort the magnetic field.

2. Spin a metal disk in a magnetic field while detecting fluctuations in the field with a pickup coil. The AC signal detected is proportional to shape or anisotropic resistivity in the sample.

3. Wet the metal with a solvent containing flourescene. Dry the sample by evaporation, then observe under UV light. Cracks will wick fluid that will evaporate along the line and reveal the crack. Bulk treatment of the samples is possible, followed by automated scanning of the individual membranes.

4. Apply a pattern of 4⋅n electrical contacts to the metal membrane, maybe using a vacuum table. Imagine four of those contacts to the membrane as the corners of a balanced Wheatstone bridge, or a strain gauge bridge. Pass a fixed current through two opposite contacts while measuring the voltage difference between the orthogonal opposite pair. Look for voltage imbalance due to crack resistance asymmetry. Circulate the orientation of the measurement.

Automatic feeding and testing will need to know the format, size, shape and the material being tested.

 

[bobob]

How is a phototube at logging the number and location of defects? This sounds like production equipment - that kind of information can be pretty useful, particularly if someone is trying to figure out where the production defects originate.

The op did not ask about locating the holes or cracks. The op only asked about finding out whether or not there was one.

 

[Dullard]

That's not so clear in my reading, but I sense that you have to be right. So: OK.

 

[Baluncore]

The op did not ask about locating the holes or cracks. The op only asked about finding out whether or not there was one.

If existing equipment could reliably identify the location of a hole or a crack then that would confirm the membrane was faulty. Restricting the choice of equipment to methods that detect the presence, but do not detect the position, will make it very hard to find an available existing solution. The vast majority of crack detection systems are designed primarily to reliably locate any crack.