to detect nano sized variations in surface ... AFM can be used

otherwise ultrasonic waves can be used to detect minor cracks ...even x-rays can be used

Dear Michael,

Try to consider the use of two groups of optical methods offering a relatively big scope of opportunities in detections of small scratches – interferometry (e.g. different varieties of white-light interferometry) as well as methods, which are based on imaging and analysis of scattered light (e.g. laser scatterometry).

Best regards,

Wojciech Kaplonek

An optical profiler which uses interferometry would probably work well in your case.

http://phx.corporate-ir.net/phoenix.zhtml?c=111487&p=irol-newsArticle&ID=1080832

The visibility of scratches can be enhanced by appropriate illumination angles. Glancing incidence illumination works well for inspecting scratches on glass surfaces. If the sample is on a microscope, rather than illuminating from above or below, try illuminating from the side with a powerful desk lamp. The scratch scatters light and appears as a bright stripe against a darker background.

Optical Coherence Tomography will detect features of this size and will allow you to profile them in depth as well as in B-scans (cross-sections) and 3D (volumetric). It's non-contact and the depth of field can be made quite long to allow you to look at curved surfaces.

The companies Sensofar as well as Optocraft may have products suiting your needs.

They have optical profilers as well as white light interferometers.

You can find them at:

http://www.sensofar.com/sensofar/

http://www.optocraft.de/index.en.php

Thanks for all of the great suggestions!

Our budget is rather limited so we will try the simplest things first, non specular scattering to see if we have the dynamic range and then perhaps a verification phase using either a white light interferometer or a commercial optical profiler, if external funding can be obtained. I will also look into Hough transforms, which are new to me, but seem promising especially for non planar surfaces.

Try to observe your samples in near glancing light or, better, arrange polarizing microscope configuration - linear polarizer and quater wave plate or polarizer and analyzer if you have transparent sample. To make proof-of-principle measurement you may use cheap linear polarizing films or even circular polarizing film (e.g. from 3D screen glasses). Propably you dont need $1M to make your measurement :)

If you have some biologists at hand, ask them for their dark-field microscopy.

Interferometry and digital holography both are suitable for scratch detection even using a Lloyd's mirror interferometer one can detect scratches in linear patterns for universal applications digital holography is more suitable. Both these methods are non-contact and performs parallel processing of large area simultaneously.

You can detect the small scratches by monitoring the noncoherent (diffuse) light scttering from laser beam from moving (scanned) surface of your subject. In the point of intersection of focused beam and scratche You can detect some encrease of the scattered light power.

We once tried an experiment at college and I believe it might help. Use a CW laser and aling a pinhole and a lens right in front of that so you can increase the size of the light spot. Then just direct it into the surface transparent or mirror) you want to analyze and all scratches will be highly visible. Good luck on your experiments. Best regards.

Other alternate optical techniques are dark field schlieren (where DC light is blocked by small aperture placed at focal point) and phase contrast imaging depending on nature of substrate.