We recently switched to a 405 nm laser diode which has a terrible beam profile -- it is a well-defined rectangle with an aspect ratio of about 10:1. The image below shows the collimated beam which has been expanded slightly for easier viewing. We would like to convert this to a ~ 5 mm Gaussian or tophat beam profile with minimal power losses. What options do we have? A few things we have investigated include

  • Using a anamorphic prism pair. Previously we had a 678 nm laser with a nice elliptical profile which we easily corrected using a 2.5:1 prism pair, but this doesn't work very well for a rectangular beam.
  • Using an engineered diffuser. We picked up a 1" 20 degree tophat diffuser from Thor, which is constructed from dark optical magics. We are able to get a nice ~12 mm collimated tophat from this, but further expanding or condensing the beam screws it up because, uh, magic.
  • Using a normal diffuser (similar to half of the configuration in the photo here or with one of these.) I don't know of any easy way to estimate the losses in this case but may be less than ~60%.
  • Running the beam through a pinhole to get a Gaussian profile, but this would mean ~90% lost power.
  • Over-expanding and cropping the beam, but this would mean ~80% lost power.

Is there any other way to do this that we missed, or is there any sort of standard approach to emission patterns like this?

Bean profile

  • 1
    $\begingroup$ It still seems like the anamorphic prism option would be an improvement over what you have. After compressing the profile in the long direction, you can crop with less power loss. Is the diode performing within spec? Warranty? $\endgroup$ – lionelbrits Feb 10 '15 at 20:28
  • $\begingroup$ This is true, we can't use our current anamorphic prism pair due to the wrong AR coating, but ~$360 seems like a lot for only a partial solution to our problem, although it may still be the best option. The specs unfortunately don't specify emitter shape, only the divergences: thorlabs.com/thorcat/22000/DL5146-101S-MFGSpec.pdf $\endgroup$ – ARM Feb 10 '15 at 21:54
  • $\begingroup$ Contact your rep maybe? Seems like a pretty useless laser $\endgroup$ – lionelbrits Feb 10 '15 at 22:11
  • $\begingroup$ What application do you plan to use it for? If you need a truly Gaussian beam, say for interferometry, then you shouldn't have purchased a laser diode ;). $\endgroup$ – Chris Mueller Feb 11 '15 at 12:52
  • $\begingroup$ @lionelbrits -- I've heard this is actually pretty common with near-UV diode lasers, although maybe it was just because everyone buys the same crappy one. $\endgroup$ – ARM Feb 11 '15 at 16:39

There are some other options I can add. By using two cylindrical lens you can build a beam expander that only expands in one direction, thus correcting for the elongation of the beam.

Once you have corrected for this you can then focus the beam on to a pinhole spatial filter. This would only partially work since the laser needs to have a Gaussian profile to start with before going into the spatial filter. To get a better beam quality you would instead focus this into a single mode fibre optic. This solution has the drawback of wasting a lot of light but it should be better than the 80-90% quoted above.

The other method you can use is to make a kinoform or diffractive optical element. To do this use the Gerchberg-Saxton algorithm to convert your measured beam profile into any desired intensity profile. This produces a phase only hologram which needs a spatial light modulator or a custom built diffractive optical element. It works but unless you have a phase only spatial light modulator kicking about it's a rather pricey solution.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.