Perhaps you can use some diffuse reflector:

https://en.symphotony.com/products/laseview-lineup/?gclid=CjwKCAiA25v_BRBNEiwAZb4-ZZ1-FnGZ_iE8oi11auRe4J-H8WHDEDAvD237FM7l8wvMZMmKYIkMhRoCN54QAvD_BwE

https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=7112

I’m surprised you had difficulty with ND filters. That is the usual approach. Multiple reflections shouldn’t be an issue as they are highly attenuated. If, for example, you attenuate 100X with an OD 2 filter, then the strongest internal reflection has to pass through the filter two additional times and is 1000X weaker than the first pass. This should not be a problem. It is important that the filters be optically flat, but quality ND filters usually are, so that shouldn’t be a problem. ND filters are absorptive and can heat resulting in thermal lensing that might distort a beam, but seeing as you only want 100-1000X attenuation before a CMOS camera it doesn’t seem like your laser is too powerful. Are you sure the ND filters are distorting your beam? Are yours damaged or very low optical quality or something?

In any case, you are certainly welcome to use crossed polarized. However, why wouldn’t those have the same issues? Unlike ND filters, there is nothing to attenuate multiple reflections, but also unlike ND filters you are more likely to find them with AR coatings. Most technologies (except for the ubiquitous polymer film) are no less likely to be optically flat, so they are just as likely to work well. If they aren’t absorptive, many polarizer scan handle more fluence than ND filters, but others like polymer films can’t.

Another typical approach, particularly with high intensity beams, is to use a weak first surface Fresnel reflection, for example from an uncoated window at 45 degrees to the beam. To avoid multiple reflections the window is usually intentionally wedged. You can buy inexpensive wedged windows from the usual sources such as Edmunds or Thorlabs. The Fresnel reflection can be highly polarizing, particularly near Brewster’s angle. This can be a problem as you aren’t truly sampling the whole beam. If you would like to avoid that, you can use two Fresnel reflections, one in the plane of the table and one out of the plane. You can return to the plane of the table with a mirror. Each Fresnel reflection attenuated the beam by about a factor of 25, so two reflections will attenuate by 625X

I agree with Mike, and I implemented this with diode lasers. The interference can come from the glass plate covering the CIMOS sensor. You can tilt the camera a little bit to see how the interference pattern changes. The ND plates can be set at an angle too to prove that they are not affecting the pattern. Pay attention to specification of the ND filters and how much power density they can handle. The ND filters can be of two types: absorptive and reflective. The absorptive are generally better regarding interference artifacts. If you see rings on the pattern, try to adjust sensor position along the axis: you might be not precisely in the waist of the laser beam (the specified focal length of the lens can differ from real). There can be other artificial rings on the pattern coming from small scratches or pieces of dust. You can turn around the beam axis any filter or the lens to see the result. For higher power laser (like hundreds of mW) I recommend to use a beam shutter between the laser and the optical setup, so you can warn up the laser and once it is stable, open the shutter and take a picture, so the lens and the filters are not distorted by heating in the center. I should mention here that sensors without glass cover are also possible to find on the market, but those are more expensive, as they are third parties product with the cover glass removed from a standard sensor.@

Dear Elena,

Try this. Regards, Shigeo.

In the schematic provided by Shigeo Kubota, care should be given to the plates flatness quality to minimize wavefront distortion. The reflectance from each plate is polarization sensitive, so it will work correctly for linear polarized beam.

In the case of linear polarized beam you can calculate an optimal angle closed to the Brewster angle at which the beam will be reflected only a portion of a percent, theoretically linear polarized beam can be 100% transmitted through the plate with right orientation and incident angle. So you can use only one plate and reflected beam can be of much lower power than the transmitted beam. Look up “Brewster angle”.. The unknown is: if the polarization perfectly linear and stable, as well as the beam is perfectly collimated, so the central and peripheral parts of the beam have the same reflectance. Good luck!

I should point out an obvious thing just in case you use reflective attenuation: the reflectors can be wedges or of absorptive glass (thermal issues, use shutter) to eliminate contribution from the back side.

Mike Albert 'The Fresnel reflection can be highly polarizing, particularly near Brewster’s angle. This can be a problem as you aren’t truly sampling the whole beam. '

Do you mean that by selecting only a narrow range of polarization orientations I may miss some modes that contribute to the original laser beam?

Yes, and this can be a problem even with supposedly polarized lasers. Suppose you have a nominally linearly polarized laser, but it does have some small amount of the other linear polarization, say 1%. Often that other polarization has a very different mode (particularly in pulsed lasers). Now suppose you use two Fresnel reflections at 45 degrees to attenuate the beam. At each reflection the P component reflects less than 1%, but the S component reflects almost 10%. You think you are looking at your original beam reduced by 10^4, but the S component has only been reduced by 10^2. It started out as a negligible 1% of the beam, but now it is half of what makes it to the camera. Also, if you are measuring power, you believe the power is 10^4 X the measurement, but that now overestimates by a factor of 2.

Those are the sort of errors you avoid by reflecting once in plane and once out of plane. Each linear eigen-polarization has been reflected once as S and once as P. The net reflection is something like 1% X 10% = 0.1% if you reflect at 45 deg incidence (the 4% above ignored all of this and is appropriate for near normal reflections, sorry. Also, please calculate your own Fresnel coefficients for your wavelength, material, and angle of incidence). Both polarizations are attenuated the same amount and the sampled beam is a good representation of the full beam.

NOTE: The S linear polarization is always reflected more than the P linear polarization, so this is a problem at all angles. It is just particularly bad near Brewster’s angle.

I agree again Mike! I was using this approach with CO2 20W CW laser, but the small reflected part was needed only for reference, so the beam profile was not measured.

To resume the discussion I should say that any bike reinvention is not a good idea when there is a well tested absorptive ND filters.

Here is an example of two pictures taken with DALSA Genie camera.

The two collimated beams from the same laser diode interfere. The large rings are artifacts from interference within the sensor cover glass, the small rings are due to interference of the main beam with the spherical wave coming from a micro scatterer (a small crack or a piece of dust) located probably on the ND filter. On the second picture the two collimated beams are not completely converged.

The test was done with CW laser.

Profile measurement of a pulsed laser is a much more complicated task bearing in mind possible beam pointing instability etc.@

I agree with Leonid. There is really no reason the ND filters shouldn’t work, and you may actually be seeing a problem from somewhere else (like a window on the camera). If the ND filters will work, that certainly is a nice convenient approach.

Elena Renzhiglova , your question has following unclear details: how powerful is your beam? Is your beam polarized? Do you need to know numerical power distribution or only relative shape of intensity distribution?

If the beam power is high, say several watts or more, you need to think about influence of its thermally absorbed power on parameters of element, which absorb it. Such heat can even destroy your element, for the case of absorbing ND filter. Reflecting ND filters are more safe.

For the case of set ND filters think about situation, when such filter can be not rectangular slab, but kind of the wedge with small angle. Sometimes I met such situation.

I'm proposing you to use as calibrated beam attenuator relatively thick (15..30 mm) transparent glass optical slab. If you will set angle of incidence slightly more than angular divergence of your beam, you will see space separated multiple reflections or transmissions of your beam. Diaphragm in this set of the beams can help you to separate most convenient beam for calibration.

Advantages of such solution are following.

1. It is not difficult to calculate corresponding set of intensities of reflected/transmitted beams, if you know slab refractive index and use Fresnel formulas.

2. You have a possibility to make a choice, what of the consecutive beams is optimal for the dynamical range of your registration system and what is its attenuation value.

3. You don't have thermal absorption of beam power, so no thermal effects for any power value.

4. Such attenuator will practically not depends on polarization of laser beam because angle of incidence on glass slab will be close to normal one, for which Rp~Rs: difference between reflectivity of different polarization components will be weak.

Good luck and happy New Year!

I have so many interesting ideas now, thank you everyone.

Additional comments.

If you use Polarizer/Analyzer ensemble to change intensity of powerful beam, one (or both) will absorb significant power, so this is a good way how to destroy it, even if you use their prism version.

If you like to have precise attenuation value without any influence of polarization, you should measure profile of your beam for two 90° difference azimuth orientation of attenuator, that I proposed, and calculate average distribution value.

To attenuate collimated powerful (high power density) laser radiation, reflection from an optical wedge of 5...10 degrees is well suited.

ND filters absorb laser radiation which causes local heating and formation of thermo-optics. Perhaps this is the cause of distortions.

Leonid Vesselov This will probably sound far-fetched, but has anyone tried to do some sort of 'deconvolution' of the interference pattern knowing the cover glass parameters?

You can hardly know the cover glass parameters such as flatness and thickness deviation. I was using Gaussian best fit for laser beam parameters estimation (in my test the beam was coming out of SM optical fiber). If you have no prior info about the beam you would need to experiment more to find out, which is challenging, especially for a pulsed laser. If you can share more details on your laser, I can possibly be of assistance. contact [email protected]

Leonid Vesselov what is the vertical stripe pattern on your pictures? For a Gaussian beam interacting with it's normal reflection, I would probably expect circles, judging by the phase distribution.

There are two laser beams to be perfectly collimated and aligned to coincide at certain distance for laser Doppler velocimeter. So the vertical strips (of high contrast) - was what we needed to be perfectly straight and evenly spaced. Each beam was Gaussian. We could block one beam and measure the other and finally align them. The low contrast rings are artifacts of parasitic interference.

Use an optical flat with high quality ANYI-REFLECTION coating with the appropriate "reflectivity"! You will be all set.

Chandrasekhar Roychoudhuri Yes, I thought that filter makers have thought of this problem. I am using filters without the specs and I assume there are coated for the visible light. But, I have learned to analyze the patterns in my images thanks to this thread:)

Hi Elena, I don't know how powerful your laser beam is so I can suggest multiple ways:

1. You can just use a 4% reflection from the glass plate to observe the intensity distribution on a CCD camera.

2. If that's still too much then cut it down using a single ND filter, probably with ND =6.0. I always chose reflective ND filters over absorptive ones in high power applications.

3. You can also cut down the beam using beam splitters and then use an ND filter.

I hope this helps and good luck with your experiments.