Wide field fluorescence light source focusing on the back focal plane

Hello all. I’m curious about how to couple the illumination source into the microscope or back focal plane of the objective to create even illumination on the sample plane.

First, is it correct that I need to focus a laser source into as small a spot as possible onto the BFP, e.g. STORM or TIRF, to create even illumination on the sample plane?

If it’s a Mercury bulb for wide-field imaging, the diameter on the BFP is usually a few millimeters, not a small spot. Is it because we want to avoid hot spot from the bulb by defocusing it a little bit? Then, what’s the ideal diameter? What about Xenon lamp or LED light source, do they have an ideal spot diameter on the BFP?

If you have a light source that is spatially varying like a fiber tip or bulb filament then you want to make it conjugate to the BFP to “maximally defocus” that image at the sample. This is the principle of Kohler illumination.

One way to think about this: An ideal lens has a special relationship between the two focal planes that is the basis of Fourier optics, specifically the ray position at one focal plane transforms into ray angle at the other focal plane and vice versa. So if you place a spatially non-uniform but angularly uniform set of rays at one focal plane – e.g. the image of a hot filament at the objective BFP – then at the other focal plane you will have a angularly non-uniform but spatially uniform image at the other focal plate – e.g. uniform illumination at the objective front focal plane where the sample goes.

Making sure the image of the filament fits within the BFP is a matter of efficiently coupling the light. Different objectives have different size BFP.

In TIRF the reason you start with the image of a point source is to make sure the rays at the sample plane are collimated (tiny spatial extent at BFP means tiny angular extent at the sample plane) because in TIRF any rays at different angles might not undergo total internal reflection and hence lead to background from deep into the sample.

Kohler illumination is not the only game in town: critical illumination places the light source is conjugate to the sample. To achieve spatial uniformity you need a light source that is spatially uniform such as a liquid light guide. Kohler illumination is historically important because filaments have been a common light source and filament sources are very anguarly uniform but not spatially uniform.

Hopefully that helps.

Thanks @JonD for sharing. Is ‘angularly uniform’ the same as ‘collimated’? I can imagine filament light source being not spatially uniform, but why is it angularly uniform?

In the case of TIRF, I understand the need of collimated beam for TIRF illumination on the sample. But spatial uniform illumination is important as well. Do you mean the collimated beam is generated at the cost of even illumination?

Correct, by collimated I mean a bundle of rays traveling with the same angle so they don’t (in the ideal world) ever diverge, like a laser pointer.

Photons comes out at random directions from a filament, leading to angularly uniform illumination. I think that is what is meant by a “Lambertian” source.

If you have a Gaussian profile light source then it will stay Gaussian through lenses (the Fourier transform of a Gaussian is another Gaussian). One approach to achieve more uniform TIRF illumination is to simply expand the beam so that the FOV is only looking at the very center of a Gaussian profile (at the expense of a lot of light being wasted). Alternatively you can use a beam shaping device or apodized filter to achieve a more uniform intensity profile before refocusing that light to the BFP. Or other approaches. This is an area of active technique development that I am not involved in directly so you can find papers on it.