I’ve received some questions about my design, and rather than keeping it private I’ve decided to make a post about my design decisions and more importantly; my mistakes.
My Raman spectrometer is a two box setup. One contains the laser and the other contains the spectrograph. I believe it will be very difficult to have the laser in the same box as the spectrograph because the Raman signal is so weak, so any stray laser light is likely to completely drown the signal. The two boxes are connected with fiber optics.
I use a similar setup for irradiating the sample and collecting the Raman scattered radiation as Mohr in his article.¹
I use a rebranded Nikon microscope objective, a 5x5mm rectangular prism instead of mirror (I also have a 3x3mm prism, and I should lose less of my signal with it, but it spreads the laser a lot). I use a spherical lens (d = 25mm, f = 25mm) to focus the light collected from the microscope objective onto the fiber optic.
This last part is actually not ideal, because the spherical lens projects a cone of light with a maxium angle of tan⁻¹ 12.5mm/25mm, which is slightly steeper than the numerical aperture of the fiber dictates, so some light is probably lost. However the diameter of the microscope aperture is not 25mm but more something like 12mm, so the collimated Raman scattered will hit the fiber a lower angle than the maximum angle the spherical lens can achieve (this was entirely a coincidence, sometimes you are just lucky).
The lesson here is you need lenses that match the numerical aperture of your fiber optics .
The fiber itself is a 200µm multimode fiber. With a larger fiber I would be able to get a bigger numerical aperture and not lose as much light in the collection setup, and it would definitely be easier to focus the spherical lens properly onto the fiber.
At the other end of the fiber sits an achromatic doublet. My initial idea was to have a slit between the fiber and the achromat, but to actually benefit from the entire length of an optical slit, you need a cylindrical lens and the achromat is spherical, so most of the light exiting a slit would be lost.
Without a slit between the fiber and the achromat, the resolution limiting factor in the spectrograph becomes the diameter of the fiber (at least in theory), so this speaks for a small diameter. I don’t know the golden middle way between resolution and loss of signal, but a 200µm slit width does not seem excessive at this point (I may get wiser about this at some point, time will tell).
The job of the achromat is to collimate the light from the fiber optic onto the diffraction grating. The diameter of the achromat is matched to the size of my grating. You get the idea here:
The focal length of the achromat is also not matched to the fiber. It’s 40mm, and with a diameter of 25mm that gives a numerical aperture of 0.30, while the fiber has an NA = 0.39 (so some light is definitely lost here). If I were to shop for another achromat I would watch out for this.
After the light is diffracted it’s focused by a 90mm f/4 m-rokkor lens I had lying around. To collect as much diffracted light as possible it has to sit almost right next to the grating. On ebay you can find old 85mm f/1.8 lenses for next to nothing and they probably do a better job than the m-rokkor I had in my drawer – if nothing else they have larger apertures.
Here’s a diagram to illustrate some of the above:
You can read more about it in my post Explanations and equations
These are the immediate thoughts that come to mind of the mistakes I’ve made so far. I hope you can learn from them.
¹ Mohr C., Spencer C.L. and Hippler M., J. Chem. Edu., 2010, 87, (3), pp 326–330