See if you can use a large diameter concave mirror that would reflect the wavelength of light. By slightly tilting the axis of the mirror you can locate the focused beam off-axis and see what the spot size looks like. If the intensity is not large enough you can place a small focal length biconvex lens to intensify the image (focused beam).

Dear Vishnu,

if your array is customizable and depending on the type of LED, you can add some SMA bulkheads (Avantes URL: https://www.avantes.com/products/fiber-optics/item/261-fiber-interconnects) and use a small optical fibre to guide the UV radiation to your desired target.

Since you have an array of 20 LED, you'll need 20 bulkheads and 20 optical fibres (or 10 Y-shaped ones; Ocean Optics URL: https://oceanoptics.com/product/lab-grade-fiber-splitters/).

Afterwards, you just need to build (or 3D-print) a support to hold the optical fibre ends, with a slight angle change so that you have a single focal point.

Regards.

Both CVI and ThorLabs offer high f# UV grade fused silica lenses with much smaller focal lengths.. However, to get the f# you want, you will need to use two or more lenses. Also, be aware that any demagnification of your image will be accompanied by a proportional increase in cone angle (NA). Etendue is conserved.

There are a few tricks you can use. For example, use a collimating lens on each LED, then aim each LED/lens at a set location. Then image that location, with demagnification, to your final position.

Be aware that whatever you do, you will probably end up with your image just a few millimeters from your final optic. How close will depend upon how demagnified you get it.

I will reiterate Damien Gray's suggestion of lensing each LED individually. The 120 degree viewing angle (about 0.87 NA) is problematic (couldn't switch to something a little smaller?) Aspheric lenses specifically designed for collimating these kind of high NA light sources are readily available such as here: https://www.edmundoptics.com/p/15mm-dia-075-na-uv-vis-coated-uv-fused-silica-aspheric-lens/3395/

That is only 0.75 NA, so you will lose some light, but only about 25%. Actually, that is to where the coating is no longer guaranteed, but the lens will sort of function out to 0.83 NA, so it should do a pretty good job. Machine some sort of jig to hold the LEDs and the lenses in a tight pattern and at the right working distance.

With the LEDs collimated and with the beams placed in a pattern as close together as possible, you will have the minimum possible etendue. For example with these lenses a closely spaced hexagonal pattern of 19 LEDs will make a ~50 mm collimated beam at something like only ~ 5 X diffraction limited. You can then focus that down with a 2" lens to a very tight spot. For example this lens:

https://www.edmundoptics.com/p/50mm-dia-063-na-uv-coated-uv-fused-silica-aspheric-lens/21780/

should focus more than, say, ~ 50% of your LED light into as little as a ~ 1 um spot. That is undoubtedly much smaller than you want for your purpose, but that indicates the possibility. For a larger area, simply work away from the focus of the 2" lens. Also, just to make a point about how small the spot can be I picked the fastest 2" lens from Edmund, but since you probably want a larger area, and you will therefore be working away from focus, you should choose a slower lens to give yourself a more convenient working distance behind the 2" condenser.

If you want to go cheap for the collimating lenses on each LED, consider sapphire or UV silica ball lenses. If you have time and some polishing equipment available, you can polish the front surface of your LED's package to a flat surface, polish a flat on the ball lens, and join with an optical epoxy to create an immersion lens. Bonus points if you mount at the aplanatic point to avoid a lot of spherical distortion. More bonus points if you pick the right index of refraction for the optical epoxy. Ball lenses are available in many sizes (focal lengths) and from multiple vendors. Optical epoxies are also available from multiple vendors and many are UV cured, making your use fairly easy (watch out for long term degradation, though).

Note that 275nm LEDs are commercially available with immersion lenses to reduce the output angle. You have not mentioned power requirements, so I don't know if they would work for you.

See if you can use a large diameter concave mirror that would reflect the wavelength of light. By slightly tilting the axis of the mirror you can locate the focused beam off-axis and see what the spot size looks like. If the intensity is not large enough you can place a small focal length biconvex lens to intensify the image (focused beam).

Same answer, two different profiles...

Strange...

To provide effective energy transfer from UV LED (Lambertian source) to sample providing highest irradiation (energy concentration on sample) one need to apply two steps:

(1) to collect and collimate LED radiation - The LED lens based on internal reflection or Parabolic aluminum reflector could be used as simplest solution. Parabolic Al reflector seem as less effective but solution available on the market.

(2) Focus collimated beam onto sample.

Parabolic concentrators could be used for group of LEDs but it is recommended to use it for individual LED, due to dimensional constrains.

Some other solutions are possible, too.

In any case, the problem is not trivial, but require serious optical analysis and calculations.

Good luck

You can use a concave mirror while setting the lEDS on a circle a round the axes of the mirror. If the light of the LEDS is coming out parallel to the axes it will be collected in the focus of the mirror. The mirror can have also parabolic shape.

However, the the use of lenses may be more suitable because as the wavelength decreases it will be very difficult to get smooth surfaces for the mirrors better than a small fraction of lambda.

Best wishes

Best wishes