I think simple ND-filter should do it. If some further mirrors are used in the beamline one can think about using mirrors with an adapted reflectivity to reduce the intensity.

I agree with Alex: ND filters are simple and effective. You can go with the metal-coated variety (which you can also by continuously variable varieties) or even the absorptive type. As the beam profile is important, it would depend on the spot size as to whether the continuously-variable ND filter would be appropriate.

Another common method is using a high quality polarizer after a half-wave plate. The half-wave plate will allow continuous adjustment of the incident beam polarization, but the polarizer will only allow propagation of the polarization state of interest.

Agree with all above! ND would do the job and preserve the polarization. However, speaking of distortions one can introduce aberrations when inserting say, a glass plate in converging&diverging beams. Make sure that the filter is placed in the collimated beam.

There are several ways to control the output power of the CW laser. As you mentioned using polarisation controller (PC) is one way. But you can use optical attenuator BEFORE PC to vary the CW laser output. The output of the PC will have the same polarisation profile.

I agree with all the previous comments.

Nevertheless, it would be useful to know by how much do you want to attenuate your signal and at what speed if you want to modulate it. Under very thigh constrain, polarizing effect of the optics may become an issue too. If high power is involved non linearity might mess up thing.

In some circumstances, only manipulation of the geometry of the beam might worth.

Appreciated promptly for all valuable comments but:

Dear Prof. Yvan Dutil!

I am working on one nonlinear medium and thus I would like the laser output only to experience the same nonlinearity. Even I don't want other optical components to change the beam profile as a distortion or beam polarization. Thus the attenuator should not remain nonlinear effect on laser output.

A convenient way could be to use a manual filter wheel mounted with neutral density filters. You let your laser beam go through the filters which have a certain optical density. For instance, if you choose an optical density of 2 then you will reduce your laser intensity by a factor of ten to the power of "your optical density" (in this case 2) thus reducing the intensity by 100. Then by rotating the wheel, you will be able to choose the filters you want, so the optical density you want and control the way you reduce your laser intensity.

You can type "manual filter wheel mounted with neutral density filters" on Google and have an example of such a device.

You can use a concave mirror followed by a convex one of the same focus length that reflects a part of the beam.

If you point with your laser to a steel needle you will see the effect.

I had a similar problem than you when measuring pulsotron multimegawat ultraviolet pulses, we used needle filters that catched a little part of the emission without saturating a photo sensor.

It was written a paper with the results of such filter

The use of reflectors instead of refractors improve a little the quality of the filter, but there is still some effect due the limited size of the optics

Data Z-PINCH HIGH POWER NEEDLE FILTER

as I know transmitter filter have nonlinear response to circular polarized light so you should use linearly polarized light with them

The best way in my opinion is to change the optical launch power (OLP) with the aid of EDFA. Please make sure you measure your link budget and photodetector sensitivity in order to understand what is the minimum power you need. You can put aside all the filtering concepts all together. Please make sure you use polarisation controller before and after EDFA. You will be able to maintain polarisation of light that way and control the OLP before fibre spans.

From your statement above, that you want to work with nonlinear media, I would assume, you have a powerful laser (~0.1 Watt or even bigger).

1. All absorptive neutral filters will influence on beam characteristics (both spatial properties and polarisation) because of the heating and can not be used.

2. Depending on a laser power and a beam size, every simple peace of glass, placed in the beam, can also be heated up and influence on the properties.

Based on the stated above, your only choice (if you don't want to work with polarisers) will be reflective neutral filters. With a remark, that the beam size will be also big (1cm or so). The bigger - the better. In order to manipulate the power without changing beam spot position and so on, not a set of filters, but round reflective filters should be used. A wheel set of filters, recommended by Thibault Cremel, would be nice, but I'm pretty sure, the filters will be tilted. Also they could have different thickness.

Round reflective filter is a one peace of glass with a coating, so one will avoid the tilt and thickness difference. Continuously variable filters would be nice, but the power attenuation will differ across the spot and will be even more pronounced with a rise of the beam diameter.

So, I would recommend to use "Round, Step-Variable, Metallic Neutral Density Filters" (NDC-100S-4M from Thorlabs, for example). It is not AD, I just like Thorlabs :).

1. If your laser is polarized, then you only need one half waveplate (mounted on a rotation stage), followed by a polarizer (Glan polarizer is better than a polarizing beamsplitter if you need to reach a good extinction). Then rotating the half waveplate allows to change continuously the power, while the polarization output is set by the polarizer.

2. If your laser is not polarized, you could do the same, but then you need first to introduce a polarizer.

Adding to what Patrick has said, I'd recommend a Glan-Laser polarizer, as these are made with the best quality Calcite and are less vulnerable to damage. The three types of Glan-whatever are Glan-Thompson (held together with epoxy, only good for lower power beams), Glan-Taylor (air spaced rather than glued so tougher) and Glan-Laser (like Glan-Taylor but made with superior materials). Just make sure you have a beam block to safely absorb the rejected light.

ND filter!

For measurement of beam parameters we often use the front surface reflection from a wedge (or multiple wedges) to reduce the flux. After your measurements simply remove the wedge from the beam line.

This works with ultrafast lasers, even with very high pulse energy.

Morteza,

From your response to Dr. Dutil and your original problem statement, it is not clear what you are trying to preserve about the polarization, nor what the original polarization is. Nor is it clear whether you are trying to sample the beam, or if the reduction in intensity is an end in itself.

I would suggest that you consider using glass plates (wedges would be good, to ensure that the second-surface reflection is directed out of the beam-of-interest) in the following fashion:

Place the first plate (wedge) so that it is within one degree of perpendicular to the original beam. Behind this wedge, place a graphite absorbing block. Use another to catch the wedge second-surface reflection, which is less useful because its polarization will be more affected by the angle of the second-surface.

Place the second wedge so the first surface is within 1 degree of perpendicular to the reflected beam. Again, place a graphite absorber behind the wedge, and another to catch the second-surface reflection. Since it is unnecessary to pass the beam through the center of the wedges, they can be arranged so that the beam passes the edge of the first wedge without too much trouble. This beam will be reflected without modification to profile and the polarization will be easy to calculate based on the actual angle required to ensure that the beam does not strike the first wedge or the absorber behind it, from brewsters angle equations.

I was doing 50W absorption with a 1/4" extruded graphite-powder-in-glue-matrix blocks without significant heating damage to the blocks. (They were 4" by 6" by 1/4").

Edit: Just a reminder, you get (depending on wavelength and glass type) between 2% and 4% reflection at each glass surface. If your beam is very powerful and your profilometer very sensitive, you can let the light bounce at 1-degree angles many times before analyzing, with a) a .02^n power, n= each reflection and b) increasing polarization to the plane perpendicular to the plane where the beams end up lying. Fresnel's and Brewster's law suggest that, based on intensities (square of the usual amplitude) 7 degrees per reflection causes 1% more parallel reflection than perpendicular reflection. At 5 degrees, it is .5%, at 2 degrees .08%, at 1degree .02% (For glass with an index of refraction of 1.5). In each case, reflection should be less than 4.0004% of the incident beam, and more than 4.00000%.

I hope this helps.

Based on my experience, reduction in the power intensity without changing the polarization state of the laser can be done using either non polarizing beam splitter or an optical isolator as well as neutral density filter NDF individually or all together,absolutely you have to choose these equipments with properly wavelength

Have a look at this. http://www.liquidcrystaltechnologies.com/Tech_Support/OAS_Cell.htm

It's about non polarising liquid crystal shutters.

If you decide on neutral density filters be aware that the plastic ones are not good in the near IR, i.e. they are not as dense as indicated.

To my opinion it exists the long list of methods and optical elements to solve this task. For example, we used tunable attenuator made of color or neutral (gray) glass in form of two rectengular prisms oriented each other by hypotenusa sides to make rectengular profile. When first prism move relative to second optical absorbance can be changed in broad limits for laser beam in insident point where the total thickness of two prism varies.

Other aproach is to apply the very known simple telesope system made of two lenses with different but matched focuses: this case wave front and intensity will be changed without any distorsion and polarizational variation.

Also it is possible to use for this aim the phase Bragg grating transmission which changes with deviation of angular Braag condition from exact Bragg angle for given wavelength.

Wish your any success.

-

Be cautious when using an ND filter, as suggested by some. The absorbed power will heat the filter and lead to thermal lensing, thus changing the beam propagation parameters. I remember being fooled by this effect when doing nonlinear optical spectroscopy during my PhD thesis many decades ago. The ND did not only affect the power, but also the beam cross section (and hence the intensity=power/cross section area) in the focal point of a lens placed downstream of the ND. It took me days to figure this out...

The half-wave plate - polarizer combination suggested by many is a standard work horse that works well for most applications.

ND non-polarizing filter

You can use neutral density filters.

Depending on the power of the CW laser, you may use a chain of ND non polarizing filters (beginning with the NDF of lesser attenuation factor, and ending with the higher one). If the polarization across the wave front is critical, you may have to check it after the attenuation you need has been reached.

In ideal event you can to use mirrors or lenses for 2-10 X beam expander and by using variable diaphragm get required laser power

Another solution it might be useful by using two or three mirrors with low reflectivity as well as it can be changed the mirrors distance to reduce the intensity

Not to be negative, but some of these responses are ignoring that the beam profile is deemed important. While spreading the beam and vignetting to select a small portion of the beam does reduce power, it also complicates the question of how the beam spreader (whether curved mirror or other optics) affects the beam profile, and complicates the measurement by requiring the vignetting aperture and the profiler to move through the plane of the widened beam, sampling piece-meal. Additional distortion of the profile by interference at the edges of the vignette aperture are added, free of charge.

Mr. Ahmed, changing the distance between mirrors will only have an effect if the beam is not collimated, or if the mirrors are not flat.

Mr. Sharif (originator of the question), I wonder if the purpose of this measurement is to determine the initial polarization and beam profile parameters, or to look for changes? If you understand the initial polarization and profile parameters, spreading the beam and measuring for changes within the profile (sampling) might be sufficient. But if you are looking to set up the initial measurements, it seems to me that a set of first-surface reflections from glass wedges is still your best bet. If modification of polarization at

Yes that is right, because the intensity is dividing the light's power by the effective area, so when the light spot is wide the light intensity will be less

You might consider using a thin half-wave retarder before a polarizer. If the light from your laser is linearly polarized, rotating the retarder will rotate the angle of polarization, which will result in greater or lesser loss in the polarizer (which remains fixed). Because it is relatively thin, rotating the retarder should have less impact on the beam profile and position than rotating a thicker polarizer.

A potential disadvantage is that you may not be able to find adequate optics if you are operating at multiple wavelengths. This may also be more expensive than using a pair of inexpensive sheet polarizers or ND filters.

I think ND-filter is a good option.

If the beam is large enough, you can attenuate it with a free-standing transmission mask with features much smaller than the beam. By design you can control the transmission fraction and preferential polarization. Edge diffraction is an issue, but you really eliminate wavefront error.

Another option is an absorption filter. With AR coating, this eliminates interference effects. Variable absorption could be provided using a fluid cell with control of cell length or dye concentration.

I would go with a waveplate and a thin film polarizer. If the power is not high you could also go with a polarization cube (the polarisation contrast will be better). With that setup you could change your intensity continuously.

I would consider two solutions:

1. You can use a variable laser attenuator kit that is composed of a half wave plate, a polarizer and a beam dump:

http://www.newport.com/Variable-Attenuator-Kit-for-Lasers/502568/1033/info.aspx

2. You could first expand the beam as necessary with a beam expander than use (ND) filters and reduce it with a beam reducer:

http://www.thorlabs.de/newgrouppage9.cfm?objectgroup_id=1580

May it be easy

A non-polarizing beam-splitter;

A reflected (not perfectly) plate, the reflected intensity can be varied by varying the incident angle of the beam.

I am currently using hight power pulsed laser. I need to attenuate the power without altering the shape. In my case, ND filter showed thermal lensing while illuminated.

I have a half wave plate and a glann-taylor prism. The prism stay at fixe position and rotating the half-wave plate allows to attenuate the power without affecting the pulse shape.

Acousto-optical modulators could be used for this, I think. 

You can even use a feedback loop to get rid of intensity noise.

Generally, the variable attenuator comprising a half-wave plate (HWP) and linear polarizer works very well.  Be careful of power level, however.  I have seen polarizing cubes heat up due to finite beam absorption and create thermal blooming of beams.  You may want to look at Glan-Laser polarizers for high power applications.  

Bottom line: know what your attenuator is doing to your laser in the relevant domains (wavelength, polarization, spatial mode).  It is easy to unintentionally change a property of your laser if you are not careful.  

Neutral density (ND) filters have their place, but they add considerable problems for laser beam measurement.  Thermal blooming is a big problem for medium to high power beams.  Wavefront error is another major concern.  These filters generally don't have great flatness.  

Another attenuator I like is to use uncoated wedges, near normal incidence.  This is good when you need to knock down power substantially, i.e. you sample ~4% of the beam using the first surface reflection.  As long as the wedge has good surface figure, and you work near normal incidence, you are not changing any beam properties except the power.  

neutral density filters is one of the components that reduce intensity of laser beam without any change in polarization.

I have used sometimes the beam splitters or diffraction gratings..

If your aim is to use passive components (such as filters) they will induce losses on the optical output of the CW laser. There is also the possibility of using optical attenuator provided that polarisation maintenance is considered. 

ND filters or a mechanical chopper (Although it is nor an optical element)