There was a question from Jerry in a previous post:
"A simple question for which I can't find an answer: Why do you have to collimate light spectroscopy? What would happen if you didn't collimate it?"
The basic idea of spectroscopy is to look at the different colors of light coming from some source. Typically light from a source (like an excited gas) is passed through a diffraction grating that makes different colors bend different amounts (like a prism, only better).
Collimated light is basically light that is all going the same direction. How about I go ahead and draw a picture. This is light from a lightbulb that is collimated then going through a diffraction grating.
Now, what happens if you don't have the slit to make the light collimated?
Technically, that is still collimated light. However, in this case the problem is that you would get multiple overlapping diffraction patterns. Suppose you have a lightbulb that you want to look at with a diffraction grating. The diffraction grating would create a whole bunch of overlapping lightbulb images of different color. The typical way to fix this is to either use a line-type light source, or pass the light through a slit.
I tried to take a picture of this (a line bulb and a normal bulb), but I don't think it turned out too well. Well, here is the diffraction image for those two bulbs:
Maybe you can notice a little more detail in the bottom image. Think of this as a whole bunch of different colored vertical filaments that don't really overlap. Also, maybe you can notice the dark area around green-red for the tube bulb. This might be because they were in tinted glass.
I've enjoyed most of what I've read of your blog, but this answer oversimplifies to the point of wrongness. In fact, your second picture looks a lot closer to what actually is going on inside a spectrometer than your first.
No. He didn't oversimplfy anthing. He did confuse collimated light and coherent light.
Do you have to collimate light to do spectroscopy? I'm not sure exactly what you mean by spectroscopy, but in general, no.
For example, if I have a set of good absorptive color filters (a set of standard Schott glass filters, for example), by swapping out the filters in front of my detector I can do some crude spectroscopy (with resolution limited by how good the absorptive color filters are).
A very crude example of this kind of spectroscopy would be a color camera, with red green and blue filters.
If, on the other hand, you want a more traditional spectroscopic instrument, where the incoming light is "split up" according to its wavelength, then you have more restrictions on your incoming light.
Basically, you need the different colors to "go different places". For this to work, it means you can't have your original beam of light going everywhere - if all the light is going everywhere, you can't separate one part from the other. Hence, your light must somehow be restricted.
This restriction has to be either in its spatial profile (like passing a thin slit) or in its direction (collimated), or some combination of the two (which is usually the case). And that's why, typically, light needs to be collimated for spectroscopic instruments. What kind of collimation you need depends on what kind of spectroscopic tool you're using.
I though he did a fantastic job