High power laser diodes typically have large facet sizes and large beam divergence angles and hence do not collimate well with standard techniques (e.g. aspheric lenses, GRIN lenses, etc..).
I am quite new to this field so I was wondering if there was a technique for better collimating these diodes using combinations of standard optics (e.g. spherical/cylindrical/aspheric lenses) to get better collimation than with the single aspheric optic. The techniques I find in literature all seem to use exotic custom made stuff like microlens arrays and gratings.
Any thoughts on this would be most appreciated.
An example datasheet for one of these lasers can be found here:
http://www.roithner-laser.com/datasheets/ld_div/rlco-808-2000-to3.pdf
The problem of LD collimation is the consequence of their emitter sizes. Low power LDs have 20-40x1 micron size. Medium power LDs have typically 100-200x1 micron, high power LD might have several such emitters stacked on top of each other. Because of such high aspect ratio of emitter it is not easy to collimate the light into circular shape parallel beam. Typically either cylindrical lenses or anamorphic prism pairs are suggested. I have heard about integration spheres being used to homogenize the light. By the way one can use homogenization rods (look at Edmunds).
If you do not care about circular shape and just want to collect the light, then look for microscopy objective - with corresponding field of view (5x-10x) , working distance (couple of mm's) and numerical aperture - they are quite cheap and abundant.
One can remove the TO can and put the ball lens almost in direct contact with emitter for light collimation - this is how it is done in fiber-pigtailed LDs. However, it is almost impossible to do it without experience and precise equipment.
I would personally tend to buy multimode-fiber coupled LD - it saves time an money (compare the price of optical mounts and elements - at Roithner they sell the mm-coupled 2W 808nm LD for 500 euro). Besides, it is less susceptible to mechanical vibrations (given that you fixed the fiber too, e.g. to surfaces using adhesive tape) and much more convenient (you can put it anywhere). The disadvantage is not defined polarisation - the light will have some polarisation ratio (circa 10-20:1), but the plane of polarisation might float if the position of fiber changes (even if the FC connector itself is fixed, this is the reason why fiber has to be fixed).
Please, explain what exactly you want to achieve by colimation, what is important and what is not.
Hi. By saying collimating, I suppose you mean, getting a very small divergence angle, even at the expense of large diameter spot. This can generally be obtained if you use large lenses with long focal length. The [diameter]-[angular opening] remains constant in passive systems, therefore such trick should bring the beam to be very paralllel. Fine details, like divergence with different directions should be well designed, like by two q multi focal lenses etc.
Regarging hi-power, the lens of your should be made from materials and coatings that are compatible with high power at the specific wavelength. Usually Fused-Silice is a good lens material.
The problem of LD collimation is the consequence of their emitter sizes. Low power LDs have 20-40x1 micron size. Medium power LDs have typically 100-200x1 micron, high power LD might have several such emitters stacked on top of each other. Because of such high aspect ratio of emitter it is not easy to collimate the light into circular shape parallel beam. Typically either cylindrical lenses or anamorphic prism pairs are suggested. I have heard about integration spheres being used to homogenize the light. By the way one can use homogenization rods (look at Edmunds).
If you do not care about circular shape and just want to collect the light, then look for microscopy objective - with corresponding field of view (5x-10x) , working distance (couple of mm's) and numerical aperture - they are quite cheap and abundant.
One can remove the TO can and put the ball lens almost in direct contact with emitter for light collimation - this is how it is done in fiber-pigtailed LDs. However, it is almost impossible to do it without experience and precise equipment.
I would personally tend to buy multimode-fiber coupled LD - it saves time an money (compare the price of optical mounts and elements - at Roithner they sell the mm-coupled 2W 808nm LD for 500 euro). Besides, it is less susceptible to mechanical vibrations (given that you fixed the fiber too, e.g. to surfaces using adhesive tape) and much more convenient (you can put it anywhere). The disadvantage is not defined polarisation - the light will have some polarisation ratio (circa 10-20:1), but the plane of polarisation might float if the position of fiber changes (even if the FC connector itself is fixed, this is the reason why fiber has to be fixed).
Please, explain what exactly you want to achieve by colimation, what is important and what is not.
Shavkat is making good points and suggestions. I would add:
- optical tappers can be an option to homogenize the beam as well as reduce the image of the source
- elliptical mirror. Great for collection very large diverging angle (much better than lenses) but does not do well with large sources.
The main problem which you will face from high power diodes is the large extended size of the source
I have collated a laser for imaging purposes. I tried quite a bit of lenses but the best result I got was with using a 2" monochromatic lens having a focal length of 10cm. I ended up with a collated beam with a diameter of almost 1" and almost Gaussian pattern free beam front. I also tried microscope objectives but I couldn't get as big of a diameter and pattern free beam as I got using the mentioned single lens.
Thank you all for the many good ideas, most of these are new to me.
The homogenizing rods I have tried and it works well but unfortunately not suitable for our application. The microscope objectives give very good divergence angles but seem to have high losses. Although, maybe this is due to using objectives in the 600nm wavelength.
The most critical factor for me is to reduce the beam divergence while not loosing too much of the laser's power.
I suppose I should have mentioned the details of the application before. This is really for a silicon pumping experiment at 850nm. The pump beam needs to be patterned through a mask before it reaches the silicon surface. The best way I have found to do this is by using a beam ex pander followed by a mask and then the silicon. The mask and hence the pattern can then be changed quite easily which is what we require.
The smearing of the mask image on the silicon surface seems to be determined by how well the beam is collimated. (There is of course some diffraction but this is not a a problem as the power in the diffracted components is much less than the main beam).
Due to the nature of the experiment the mask cannot be placed directly on the silicon surface. (in fact the silicon surface requires at least 100mm clearance from any other object). We are currently using a TO-Can 2W laser diode and require at least 15W/cm2 intensity at the silicon surface over an area of 2mm x 1mm. Hence we can afford to waste some power but not much. The final shape of the beam is also not critical since it goes through the mask.
The current divergence angle i am getting with a single aspheric lens (f=2.75mm, NA=0.55) is around 2.2deg and this causes excessive merging of the mask elements. I would really like to get this close to about 0.3deg.
If the incidence angle on your silicon is not important, I would personnaly use of a homogezining rod couple to a standard imaging 2F/2F assembly. I would use a rod with the section of 2 mm × 1 mm, place it as close as possible to the diode, place the mask at the output of the rod and image the mask on the silicon with a 2F/2F assembly.
Regarding the value you give, you can't reach a 2mm×0.3° from a 100µm×30°. From the BBP of your diode, the best you can reach is 2mm×1.5° on your silicon.
J.M., thank you for the interesting idea. I will give this a try. Maybe 0.3deg was a bit ambitious but i will settle for anything less than 2.2deg.
Just remember that the Beam Parameter Product (dimension×divergence) is nearly constant. Why not lighting 4mm×0.75° on your silicon with a F20mm lens? Good luck.
If the laser diode is a linear bar array then it can be collimated to a reasonable degree in a simple way. Take a small piece of glass fiber ( standard multimode of 400 micron diameter). Remove the plastic cladding by stripping it or burning it. After cleaning it thoughroughly stick the fiber rod in a XYZ translation stage and position it very close to the diode surface until you get good collimation in the fast diverging direction. Then with the help of a cylindrical lens of (25 or 50 mm focal length) you will be able to collimate the slow diverging direction to a good degree. With this a reasonably collimated square shaped beam can be obtained. In this way I collimated a high power diode bar ( ~30W power) to pump DPSS laser under axial pumping configuration and found the astigmatism in the diode beam was greatly reduced. To get a circular beam the collimated square beam can be focused and coupled to a fiber ( beam quality will degrade due to fiber modes). Also a circular aperture can be used after the square beam and can be re-imaged on the target with proper de-magnification.
Alternative a lens-duct can be used to collimate high power diode array but the design of lens duct is complicated and its installation is cumbersome.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Optics