Precision LEDs

Most Leds are pretty narrow band, but not 'band pass' they are emitters not filters.
The wavelength is normally in the manufactures data...suggest you look there.

Del

From data sheet, it is seen that LEDs are having a band pass ranging from 40 to 20. Can I have an LED with 1 or 2 as its tolerance?

They don't have a band pass!
40-20 what?
Are you talking about viewing angle? Wavelength or what???
...maybe give us a manuactures part number...at the moment your post doesn't make much sense...
Del

Dear Sir, I was talking about the wave length. We want LEDs with reproduceable wavelength. Anil Jose

If you are asking about beam angles, like reducing 20-40 to 1-2 [degrees], try laser diodes for lowest beam divergence angle.

If you are asking about frequency bandwidth (color purity?), like reducing 20-40 to 1-2 [nanometers], try laser diodes for narrowest emission bandwidth.

If you specify UNITS in your questions, you are more likely to get the answers you need.

Dear Sir, I want to use LED to find out transmittance from a solution. So Laser is not practical. Anil Jose

If you are talking about viewing angle (which I think you are), then you want a 'laser diode'

http://en.wikipedia.org/wiki/Laser_diode

If this is for an indication then use a narrow viewing angle LED (they go down to about 4 degrees) and a lens, rather than a laser diode (which will cause eye irritation and potential damage if you stare directly at it).

Dear Sir, I want a narrow wavelength LED. Then only even same batch LEDs will give reproduceable results. Anil Jose

A wavelength variation that small between parts is going to be very difficult. Mass production batch tolerance and variation is not that tight and will almost certainly exceed your close tolerance requirements.

Have a look at the different LED manufacturer websites as some may offer special close tolerance batch services (where they test the wavelength of many LEDs from their stock and match a small quantity of LEDs to your required tolerance). This is a common practice but will likely be expensive.

Additionally try specialised lighting web resources such as

http://www.photonicsonline.com

Or Globalspec.

If you have the proper test facilities and can accurately measure the wavelength then you best (and cheapest option) will be to buy a lot of LEDs and test them. Say from a batch of 20 or 50 you should find 1 or 2 LEDs that meet your tolerance requirements (remember tolerance indicates worst case variation and the variation will actually vary across, say a bell curve distribution).

Yes it is a bit like the lottery, but with much better odds (and this is the same as the service a LED manufacturer would offer that I have mentioned above).

Dear Sir, Thanks for your guidance. Anil Jose 0 94474 98430

It sounds like you are trying to build a Clarity Monitor or Turbidity Monitor.

What is the solution you are measuring ? All the detail you can provide us will help you toward a proper solution, otherwise it is a guessing competition for us just trying to work out the question. Any/all details please?

Sir, A colored solution from a chemical reaction will be measured for its transmittance. The current technology uses a halogen lamp with a filter to find out the transmittance. Anil Jose 0 94474 98430