[UVC SAFETY WARNING] Microscopy at 254nm - build and images

[JMC]

Small update, as the UVC microscope imaging is still not what I would consider 'straightforward'. This time with a 32x Zeiss Ultrafluar NA 0.4 objective, and looking at a single diatom - Pleurosigma Angulatum - on my slide, with different wavelengths. The 546nm down to 313nm was done with my mercury xenon lamp and the 254nm image with the 3W low pressure mercury lamp. All images taken with the monochrome converted Nikon d850. These are the full frame image, but I have reduced resolution from 8288x5520 (the d850 resolution) down to 1000x666 for sharing here.

 

The idea was to show the effects of going from visible down into the UV. Here are the images with the wavelength given in the bottom left.

 

 

 

 

 

The 'dots' in the diatom structure aren't visible in the 546nm light image as that objective would not be expected to resolve them. However they become more and more obvious as the wavelength gets shorter.

 

At 254nm, there is a very obvious 'ring' pattern in the image with this objective, more so than the 10x one (although it was still present there). I think this is something called window etaloning, where light is reflected off the sensor and reflects back from the coverglass, bouncing back and forth. However that is currently just a hypothesis. It doesn't happen at the longer wavelengths, so could be something to do with my 254nm lamp, or it could be wavelength dependent - I'm not sure at this stage.

 

That 254nm image was a 15 minute exposure at ISO400. And that was underexposed - I had to boost it quite a bit in Darktable. Compare that with the 313nm one - 4 seconds at ISO200 - and it starts to bring home the challenges at such a short wavelength. I have no idea about my absolute sensitivity down there, and as they are different light sources it is hard to compare two directly.

[Andy Perrin]

I think the circles on the 254nm image are Newton rings, which is also consistent with your hypothesis, Jonathan. It implies that one of the surfaces must be slightly dome-shaped or warped downward.
https://en.m.wikipedia.org/wiki/Newton's_rings

[colinbm]

Wonderful seeing the different wavelengths.
Would 222nm be your next step ?

[JMC]

17 hours ago, Andy Perrin said:

I think the circles on the 254nm image are Newton rings, which is also consistent with your hypothesis, Jonathan. It implies that one of the surfaces must be slightly dome-shaped or warped downward.
https://en.m.wikipedia.org/wiki/Newton's_rings

Interesting. If it is this I wonder why it's only at 254nm, and also, I'm not sure where in the system there could be a spherical lens in contact with a flat surface. The coverglass/sensor wouldn't be the issue here - I've seen those up close and there is a definite gap in there.

 

The reason I was thinking window etaloning, was because of the reflectance of silicon which increases as the wavelength approaches that 250-270nm. This got me thinking that much more of the light would be reflected from the sensor, to then bounce back to the coverglass, and be partially reflected from that. Then back to the sensor, etc etc.

 

The 254nm light, unlike the one used for other wavelengths, doesn't have a diffuser built into the optical train, and I am wondering whether this could also be a factor.

[JMC]

7 hours ago, colinbm said:

Wonderful seeing the different wavelengths.
Would 222nm be your next step ?

Thanks Colin. Probably not, at least not yet. The higher magnification UV objectives I have seem to be good down to about 250nm, but their transmission drops quite quickly below that. For now I am going to try to optimise 254nm. I'd really like to be able to use my 100x NA 1.2 objective at 254nm, but my current cameras just aren't sensitive enough for that.

[Andy Perrin]

Jonathan, etaloning is the same phenomenon as newton rings as far as I can tell. Both are interference. Look at the picture on wiki under the “Theory” section - isn't that exactly what you described? As for what is playing the role of the spherical lens in the newton ring situation, I think one of the “flat” surfaces may not be completely flat. There may be micron scale warping. You can confirm this if you calculate the radius of curvature R from the last equation in the theory section using your picture. It should be very large compared to the gap if I’m right about warping. 
 

I found this article that mentions Newton rings in the context of a filter and a monochromatic light source, so maybe it is relevant:

https://jufa.medium.com/beware-newtons-rings-3cbfbd7228aa

 

The main difference from your theory is that the two surfaces are the filter and lens touching, not the sensor and coverglass. 

[dabateman]

Because its wavelength dependant,  I don't think its just Newton rings.

You might have an odd diffraction event going on between your camera pixels and your openings.

I would first try to adjust all diaphragms, by opening or closing them slightly. 

Also you do have other cameras with fused silica coverglass.  Do they have different pixel pitches? Try one of them to see if that may change things. 

 

Also to ensure it is wavelengths dependant and not just do to bulb alignment,  try removing the semrock 260 filters and use a Baader or larger wavelength cut off filter for the 254nm 3W bulb. It will be very dim, but it does have visible bands.

[Andy Perrin]

Dabateman- the seeming wavelength dependence may be just because one filter is making contact and the others aren’t touching. 

[dabateman]

15 minutes ago, Andy Perrin said:

Dabateman- the seeming wavelength dependence may be just because one filter is making contact and the others aren’t touching. 

He is using 2 semrock 260 filters, so its possible the spacing is too close.

Air gaps matter, as air acts like a lens in these configurations.

[Andy Perrin]

Yes, I think systematically checking for contact points all through the setup is a good idea here. 

[JMC]

Cheers all, I get the feeling this will take a bit of tracking down.

 

I can look at filters when I am a bit more awake, but I'd be surprised if that is the issue. I have 2x 365nm filters taped together, and 2x 313nm taped together. They are each in their original metal rings, so the glass surfaces would be closer together than the Semrock 260 and Sirchie 254nm filter I use for the 254nm images. Putting other filters (such as the Baader U) in the way is also problematic as they need to be small enough diameter to fit on the photoeyepiece, so I am limited to 1 inch filters there.

 

The only other camera I have with a fused silica coverglass is not monochrome (which obviously impacts exposure time), but I will try that when I can.

 

I do tend to think that adding a diffuser to the light might help, but again that increases exposure time so is not something I can quickly test.

 

If I have any luck tracking down the issue I will of course update on here.

[Andy Perrin]

Jonathan the issue would more likely be lens-filter or lens-SOMETHING contacts rather than filter-filter. There is a reason those are circle -shaped interference and not some other shape, a curved surface must be involved somewhere in this. 

[JMC]

Ok, cheers Andy.

[Andy Perrin]

Could you please post a photo of just the fringes (with scale bar, though)? I would like to try plotting them against the Newton ring formula and see if they are spaced according to the theory.

[Alaun]

The rings seem centered around the probably thickest part of the diatom. Does the embedding material shrink? If so, the shrinkage might be restricted (or less) there. So the cover-glass might get a very slight bulb structure? 

[JMC]

Done some quick follow up tests this morning. The rings are there even when the diatom slide is not, so it's nothing to do with the slide.

 

Andy, I'll drop you a message, when I have had chance to get you an image to play with. This was where I was reading about window etaloning;

https://www.azom.com/article.aspx?ArticleID=14009

http://www.astrosurf.com/aras/fringing/schlatter/ripple.htm

 

[colinbm]

You can remove the cover glass & just have the bare sensor, then you have no reflections.
This technician suggests that a bare sensor is an option.
https://www.monochromeimaging.com/

[JMC]

2 hours ago, colinbm said:

You can remove the cover glass & just have the bare sensor, then you have no reflections.
This technician suggests that a bare sensor is an option.
https://www.monochromeimaging.com/

That is an option, but it makes cleaning the sensor a nightmare - high risk of damage to surface and the wiring. Another option is the use of an antireflective coating on the sensor coverglass which is optimised to let UV through. But finding one with the right thickness, and overall size, without spending huge amounts on a custom one will be the challenge there.

[dabateman]

Or just order a HQ pi monochrome sensor from MaxMax,  they are bare sensor and very sensitive. 

You would just need a Raspberry pi to run it.

[JMC]

What I have learned this morning.

 

I get the ring pattern with both my monochrome d850 and my multispectral A7III (with Bayer filter and microlenses attached). These both have fused silica coverglasses.

I get the ring pattern if I use my Sirchie 254nm filter alone, and if I use it in combination with the Semrock Brightline 260 filter. It does not matter if the Semrock is above or below the Sirchie. I cannot use the Semrock filter alone as it does not block IR.

I get the ring pattern with the Sirchie filter irrespective of its orientation.

 

Essentially with everything I have tried so far, I get the ring pattern at 254nm.

[Andy Perrin]

Jonathan, the cheapest solution may end up being image processing to remove the pattern. That is how I fixed the gain irregularities in my triwave. 

[dabateman]

The coverglass thickness might be different between the A73 and the D850. Did the ring pattern change in thickness, spacing and number of rings?

If not, than its not due to back reflection from the coverglass on the camera sensor,  as you thought.

It might be due to the bulb or the optics within the light pathway. Can you see if you get the same ring patern using your two 365nm filters taped together,  using the same 3W germacidal bulb? The peak is very weak there, but the camera sensitivity increase should help. If no ring pattern, than it is wavelength dependant.  If a ring pattern,  than it might be due to the bulb orientation or still some of the optics.

[Andy Perrin]

I keep coming back to the fact that the rings are circular. I truly think something curved touching something flat must be involved here, and whatever it is is aligned with the optical axis. It may be internal rather than external since so many external things have been changed without affecting the results. Newton rings are the same as etaloning with one or both surfaces having curvature. The difference is just nomenclature. 

[JMC]

Well, David, it looks like you're on to something.

 

I used the double stack of 365nm filters with the lamp for 254nm (rather than the usual lamp) and got the following.

 

As you can see, the ring artefacts are present. So it does indeed look like the light source is an important factor in these rings. The 254nm lamp does not have a diffuser on it, while the other lamp does, so this will be the next thing to try.

[Andy Perrin]

Jonathan, I found this:

https://www.olympus-lifescience.com/es/microscope-resource/primer/techniques/darkfieldtrouble/

Quote

Problem: 
At high magnifications, colloidal particles display incomplete Newton rings and appear to be unevenly illuminated.
Solution: The condenser is probably off-center. Change to a lower power objective and re-center the condenser. Carefully check to determine whether the higher power objective has a different center, and if so, center the objective to the microscope optical axis.