Confocal Spectroscopy

In today’s post I would like to discuss confocal spectroscopy. The topic can be extended to imaging spectroscopy although I will not cover it explicitly.

In confocal spectroscopy you image a scene on a camera an analyse the spectrum of a small localized region in that scene. This is represented in Figure 1 where we image a street light and analyse the spectrum at the centre of the image, where the light bulb is. Here, an analysis of the spectrum reveals that the street light is actually a high-pressure sodium lamp.

A photography of the setup is given in Figure 2. It is made from Thorlabs cage elements with an imaging lens, a cube beam-splitter, a camera and a fiber port. The image is sampled at the exact centre of the camera over a 50 µm area using a 50 µm optical fiber. This fiber is then sent to the spectroscope that I discussed in my [»] previous post. Note that I also made a custom part to hold the setup on a telescope equatorial mount for easiness of use.

The concept is relatively straightforward once you already know how to build an [»] autocollimator and is detailed in Figure 3.

The system uses a lens to image the rays coming from infinity (see note below) and splits the signal into two parts using a cube beamsplitter, just like in an autocollimator. Both the camera and the fiber port are placed at the focal position of the lens such that the scene is imaged on both of these planes. So the fiber will collect a small area of the image that can be sent to the spectrometer. The only tricky part here is to align the fiber port with the camera such that we know which part of the scene we send to the spectrometer. I will now detail the method that I used here for the alignment.

To align the fiber port to the camera I have placed the fiber port on a XY cage-mount-compatible translation stage. I have then imaged (see note below) a 50 µm slit in horizontal position such that it is well centred vertically on the camera. Then, all you have to do is to tune the vertical actuator of the translation stage to maximize the signal in the fiber. You can use a photodetector or your eye, both technique works fine. That way, we know that the fiber port is at the same height as the camera centre. To align the fiber port horizontally now we just turn the slit 90° to align it vertically on the camera. Then, just use the second actuator until you hit the image of the slit. The fiber port should now be perfectly centred relative to the camera. The situation is illustrated in Figure 4.

Please note that when you place a fiber there is no guarantee that the fiber core will be centred perfectly in the fiber port. So, when you place a fiber, it is important to lock it as you do the alignment. If you were to remove the fiber and place it again, there will be no guarantee that the fiber will still sample the camera at the correct position. An improvement might be to use FC/PC fibers because they have an alignment key (to be verified experimentally).

I would now like to discuss a little bit about the usage of confocal spectrometers.

As described here, confocal spectrometers image scenes at infinity. They are therefore very good at sampling light from distant objects such as the street light of Figure 1. I have seen confocal spectrometers being used to image far away objects such as stars but also, and more interestingly, by the environmental police to monitor the exhaust gases of chemical plants. Indeed, some toxic chemicals like SO₂ cannot be released freely in the environment and their presence can be monitored by absorption spectroscopy. This allows people to check for the presence of toxic/forbidden gases in exhausts without the company knowing that you are performing a measurement because you can be outside of the plant, several hundred meters away.

A different way of using the confocal spectrometer setup described here is to use it as the tube lens of a microscope setup. That way, instead of looking at very distant objects you can zoom-in to analyse very small features of organic or inorganic compounds (e.g. coloured inclusions in minerals). All that is required is to place a microscopy objective in front of the confocal spectrometer and to align it just as in any regular microscope setup.

When performing the alignment of our confocal spectrometer, I said that we had to image a 50 µm slit onto the camera and the fiber port. This can be one using this microscopy version of the confocal spectrometer. To centre the slit to the camera, just put the microscopy objective on a kinematic holder and tilt it slightly to produce a lateral shift in the sensor plane.

To conclude, there is one final variation of this very same confocal spectrometer setup that I have to mention and that is often used in the earth observation business in space telescope: imaging spectroscopy. The idea of imaging spectroscopy is to replace the fiber port by a slit and to directly use that slit as the entrance of a spectrometer. Instead of analysing the spectrum at a single point, a full line is analysed at once. We could do that in our setup as well but the great interest is that, in the satellite, the slit is aligned such that it is perpendicular to the earth motion. Then, as the earth rotates, a 2D map of spectra of the ground is taken. This is a very popular technique in the space industry but it can be relatively tedious to implement due to the aberrations in the optical system that will curve the image of the slit.

I will try to make more experiments with the microscopy version of the setup in the following weeks, so stay tuned for updates!