Collimating Light from a Hg Arc Lamp

[LDF1010]

Hi,
I am trying to collimate light from a mercury arc lamp. I know this question has been asked before here https://www.physicsforums.com/showthread.php?t=413437&highlight=collimation but it was first posted last year and since my situation is a bit different it didn't quite answer all of my questions.

I am using a 100W Hg lamp housed in a OBB PowerArc lamp housing. The lamp has an ellipsoidal reflector with an f number of 2.5. The specs on the lamp (http://www.obb1.com/LightSources/Compact-High-Intensity/PowerArc.html) lead you to believe you get 10W in a 1.2mm spot about 15 cm from the front of the housing. What we didn't expect was all of the diffuse/surrounding light. We put a iris to select only the most intense portion of light but that dropped our power significantly and not to mention burning through irises faster than we can buy them. We have tried a variety of different lens configurations to collimate the light but we are losing up to three orders of magnitude to get a nice collimated beam. Some things we have tried: a bestform lens, two plano-convex lens, an aspheric lens with a variety of plano-convex, biconvex lens. At this point we have taken the iris out to try to retain the power but now collimating has become impossible. We would like to have a beam diameter of 10 mm or less with at least 6 watts of broadband power. Any advice or suggestions? Thank you for taking time to read this.

 

[ZapperZ]

Did you try adjusting the back reflector? Typically they suggest adjusting it until you can image it somewhere.

Unfortunately, I haven't seen any Hg lamp where you get as much power as you expect, which is why it is always advisable to buy a bigger power than you think you need.

Also, how do you burn through the irises? From the heat? You don't have water-cooled irises, or at least have the light from the lamp pass through a condenser first to absorb most of the IR?

Zz.

 

[LDF1010]

We have adjusted the back reflector to obtain a good image. And we have a good image. The problem is now collimating the light without losing a good bit of power. If we use several irises with 1-2 mm apertures we can collimate nicely but we go from >10 W to less than 0.5 W. And that is total broadband power! We will be decreasing that power even more when we filter to obtain a single wavelength.

So far we are able to focus so we have 6.6W going through a 3 mm iris aperture. However, it diverges quickly from that point and have had mininmal luck in collimation. The irises we use are not water cooled. We were looking for a high temperature iris since active water cooling complicates the setup. We finally added a small fan to help cool the first optics. We also have a rather large cold mirror to help cut down on IR but of course that cuts down on the power output and at this point we don't want that because we are just trying to get anything, something to work.

Optics are not my area of expertise so maybe that is why I thought collimating non-coherent light wouldn't be that difficult. If anyone has good links or articles on optics/collimation please feel free to share.

 

[LDF1010]

Also, how do you burn through the irises? From the heat? You don't have water-cooled irises, or at least have the light from the lamp pass through a condenser first to absorb most of the IR?

Zz

Where can I find a water cooled iris? A Google search doesn't provide much help. Thanks.
L

 

[Andy Resnick]

<snip>We would like to have a beam diameter of 10 mm or less with at least 6 watts of broadband power.

What is your application? This is *extreme* high power optics, and I'm concerned that (apparently) nobody on your team has any optics experience. You have to realize that in order to get a low divergence beam, you have to start with an extremely tightly focused spot and I don't know what material can withstand MW/cm^2 intensities for any length of time. And, since you have broadband light, you may not be able to use refractive elements because of dispersion.

Edit: the total optical power output of a 100W Hg arc is, IIRC, around 1W- how do you propose to get 6W?

 

[LDF1010]

Andy,

I do not have any optics experience but the lab I work in is equipped with a Nd:YAG laser so my PI and other co-workers do have some experience in optics. The lamp we are using has an ellipsoidal reflector and at the focal point we measure ~18 W of broadband power. Of course to get perfectly collimated light you need a point source but we were hoping that starting with a 10 mm beam we could get a decently collimated beam while retaining a large portion of the power. Once we received the lamp we realized we had misunderstood the specs and assumed too much. Now we are trying to find a way to collimate as best as possible and retain as much power as possible. We upgraded to a 2" plano-convex and 2" condenser lens and that seems to be the best arrangement so far. We realize that having ~15 W hitting an iris is extreme but we must iris the light to assist with collimation. The lenses we are using are BK7 and seem to be holding up well to the power. We are also using bandpass filters to select certain wavelengths but they are never in contact with the full power of the lamp and tend to only see about 1 W. By the way, we are doing photoacoustic spectroscopy and phase shift cavity ring down.

 

[Andy Resnick]

Some rough estimates can help guide you here- you are starting with a 1mm-diameter arc and a f/2.5 ellipsoidal reflector. f/2.5 corresponds to a half-angle of 0.2 radians. The product of size of the source and the divergence angle is a conserved quantity (the etendue), so call that 0.2 mm*rad.

Now you want to end up with a collimated beam 10mm in diameter- let's say the divergence angle is 1 milliradian, so the target etendue is 0.01 mm*rad. In order to match the two results, the pinhole at the image of the arc must be about 0.01/0.2 = .05mm (50 microns), which would result in a major loss of efficiency of the system- which is your experience.

What you need to give up is either the low divergence or the small diameter. In my system, the arc lamp is coupled to either a fiber or a light pipe- but I don't require the optical power to be so high.