UV (365nm) and visible (546nm) microscopy of a diatom

[JMC]

I've been playing around with building a UV microscope over the last year or so in between paid work (thanks Covid), which has been a fun build but very challenging. Most of my work with it is on sunscreens for a client and I cannot share those images yet, but I thought I would share a couple of of quick images of a diatom, from a Diatom Lab 2.0 test slide (see here - http://www.diatomlab.com/diatom-test-slide-version-2.0.html)

 

The microscope is based on an Olympus BHB, and I've swapped all the glass in it for fused silica. In theory it can image down to about 250nm, but I'd need different camera/filters/light etc to get that low. With my setup I can image down to 313nm. I used a 100x NA0.85 Zeiss Ultrafluar objective, with glycerine immersion, and imaged using a monochrome converted Nikon d800. The images are stacks for 5 individual photos. The images were 'cleaned' (although looking at them now, not very well I hasten to add) and sharpened, but treated equally.

 

The test slide mounting media lets some UV through, but is only good for 365nm (it is pretty much opaque below that). So I did visible (filtered 546nm light) and 365nm from a mercury xenon lamp.

 

I thought I'd image Diatom 3 from the slide, which is Gyrosigma reimeri, and has about 18-22 longitudinal striae per 10um, or in other words about 500nm per striae. This is a severe test even for a high magnification, high NA objective, so should give my home built microscope with a 100x NA0.85 a real challenge. UV transmission microscopy was first developed in the pursuit of resolution, as resolution is directly related to wavelength. As such I expected the UV image to be higher resolution.

 

Firstly, the visible image (546nm). This is the whole frame of the image.

In visible light the striae are barely visible, but can just about be seen in the middle of the diatom.

 

Secondly at 365nm, again as a full frame image.

At 365nm the difference is pretty striking and the striae become visible.

 

Cropping the 365nm image gives the following.

Counting the striae gives 9 over the 5 micron distance or about 18 per 10 micron which is in keeping with the quoted number from the maker of the slide.

 

It's a shame the mounting media in the test slide blocks the shorter wavelength UV, otherwise I'd try it at 313nm as well.

[Andy Perrin]

Nyquist criterion says that,

 

frequency < 2*sampling frequency

or
1 stria/500nm < 2/wavelength 

 

so wavelength < 1000nm is the absolute cutoff. In visible light (500nm wavelength say) you are at half the ultimate limit and at 365nm it’s about a third of the limit. 

[JMC]

Andy, I use the Abbe equation for microscope resolution to give me an idea of theoretical best case scenario - d= λ/2 NA

 

For 546nm illumination (with this objective) the theoretical limit would be 321nm, and for 365nm it would be 214nm. I'm sure though that it will not get close the theoretical limit. I've swapped out factory made lenses for 'closest approximation' fused silica ones, I've replaced glass elements in the trinocular head with fused silica blocks, I've got filters in the way both on the light source and between the photoeyepiece and camera, and the condenser, objective and photoeyepiece are all from manufacturers different to the microscope itself. Add all these together I'm just happy it works.....

[diant]

Some ten years ago I was solving alike task, but as I worked with Violet light (400-450nm) I made emphasis on NA of microscope lenses.

At that time I synthesized a gelatin photo emulsions with necessary (for me) properties and needed to see their crystals (form, size etc.)

All my troubles was solved by Carl Zeiss 2mm Apochromatic HI 90 1,40 oil immersion lens.

 

[JMC]

Diant, very nice. Yes using 400-450nm light is a great way to improve resolution, and to be able to use standard cameras. Your post just reminded me, I have some 100x objectives with higher NA than my Ultrafluar. I must check to see whether they let UV through at 365nm and try them with the diatom slide.

 

At some point I'd like to find someone who can make me a diatom slide using glycerin as the mounting fluid and with a fused silica slide and coverslip. It'd be cool to take a look at it in UVB and see how the resolution improves.

[Andy Perrin]

I think Abbe and Nyquist are the same thing. I probably just did it wrong in that case. 
 

ETA: no, it seems Abbe is a tighter limit than Nyquist based on some fast googling. Anyway you are right, I’m sure you aren’t diffraction limited. 

[diant]

22 minutes ago, Andy Perrin said:

Anyway you are right, I’m sure you aren’t diffraction limited. 

Andy, are you mean a spherochromatism? (as main limiting factor in such cases)...

Or something else?

[Stefano]

Very nice. It would be interesting if these diatoms absorb at 313 nm and are colored with the TriColour technique.

[Andy Perrin]

1 hour ago, diant said:

Andy, are you mean a spherochromatism? (as main limiting factor in such cases)...

Or something else?

Not just that, but like Jonathan said, the various other non-ideal things will limit him before diffraction is the main issue. 

[diant]

Andy, my opinion is that we are stay just at diffraction limit here. My reflection is the next:

 

I know (and many times verified it myself) that usual photographic lenses do not draw a dot ("a star" as astronomers say) as a dot. They have not diffraction limited quality being wide open. It is so.

 

But I think (though I did not check it myself) that with microscopic lenses (especially a high quality apo ones with high magnification and high NA) the things should be quiet different. Otherwise "NA" value would lose its sense.

 

D = 0,61 lambda / NA   - (I always use 0,61; Jonathan use 0,5) - is truly diffraction limited resolution formula.

And if we use 1,4NA condensor (I use such) and place it in a right position (enough close to the slide), and if we have 1,4NA high quality lens, you should have in its focus really 200nm resolution picture for 450 nm light (as it point us the formula). And I had it in my images as you can see above, as I can see there a most thinnest microcrystalls in form a small spheres exact ~0,2 mkm diameter.

 

[dabateman]

Diant sure, if you calibrate and optimize the cone of light perfectly to the slide and the opening of the objective lens. Than you will have an excellent system for microscopy. 

However,  Jonathan's point is he didn't do that. So he has losses. But it works good enough for him.

 

He should definitely take the 2 months needed to fine tune those lens elements to get that extra 10%. Even better if he pulls it all apart into a light bench to really optimized the spacing and distances to the true optimum. Then he can push beyond Abbe with point source calculated center deconvolution.

Or not, and just be happy with his compact setup that probably good enough for the clients. 

 

[JMC]

David, yes indeed. Could the setup be made better? Yes, probably. However is it good enough for me at the moment without doing that? Yes, it is. I got into microscopy in March 2020. I bought myself a microscope which was in need of repair to learn a new skill, because a lot of my paid work dried up due to Covid (many of my clients went out of furlough, and with them not at work, they had no need for me). Before that I had pretty much zero experience with a microscope, and certainly didn't know anything about the details of how they worked.  Of course after fixing the one I got, I got to wondering about whether I could turn it into a UV microscope that would work down to 300nm or even below. This was where things got really complex.

There were a couple of key points I needed to keep in mind when doing the build. It needed to be compact - no bigger than a normal microscope - as I have limited space for it. It needed to be safe. It needed to be standalone - no room for a computer next to it. It needed to be cheap - nobody was funding this but me. It met the first three criteria, but while it was a lot less than buying a new UV microscope, it could hardly be called cheap. Even buying second hand equipment, the build still cost 1000's of GBP. Some things were bought to try and didn't work, others were just rare and didn't come up for sale very often, others needed to be bought new or custom made.

Now with hindsight, it looks as though 365nm microscopy is relatively simple to do - you can use 'standard' objectives, condensers and even eyepieces, and without any major modifications to the microscope. I've even done it with using a Nemo torch as the light source. However as is often the case with UV, if you want to go deeper that is where the problems start.

[diant]

Jonathan, tell me, what is your ultimate goal now (excuse me, sometimes it is difficult for me to clear see all modern English...) - to reach really UV (may be even UBV) microscopy or to reach a max possible resolution?

[dabateman]

I was thinking Jonathan should buy a Nitrogen laser (337nm), build a moving scanning head using a Raspberry pi. Then he could push it to 60nm resolution using the point spread function and deconvolution. 

 

Or he could tell the sunscreen manufacturers to blend in Alexflor dyes into the formulation.  Then using excitation /deexcitation at the center he could go below 20nm.

 

However that might just change the point of looking at the sunscreen.  But if microencapsulated correctly,  would be fun.

 

Ok I will stop here.

 

I think he is interested in spread and reflectance quality of the true sunscreen formulation.  Time to buy some pig skin and look at burn protection from UVB. 

[JMC]

Diant, when I went into this, I had no real goal - I simply wanted to see if I could make a UV microscope. As it became apparent that it would be possible to build, I started wondering if I could make something which could be used to image sunscreen topical formulations to help with my work. I work in the dermatology area, including sunscreen development, and thought it would be nice to have a microscope which could look at the formulations in the UVA and UVB regions to see where the active ingredients are. With the setup I have now I can image sunscreens at 313nm and 365nm as well as in the visible region.

 

Absolute resolution is not my goal, but I have noticed that resolution does indeed improve as wavelength decreases, even with my home built system.

 

In the future I may try going deeper into the UV with it. In theory my build should be good down to 250nm and perhaps even below. But for that I'd need a different light source, filters and camera. This would be less for my skin research though, and more just because I am curious.

 

David, fancy techniques are beyond me unfortunately. I'm a KISS kind of guy. Give me a UV light, some UV lenses and a sensitive camera, and that is about as advanced as my brain can do.

[dabateman]

In all seriousness,  Jonathan if you do want to try UVC or deeper into UV than the Raspberry pi HQ camera from MaxMax might just be your best option.  It should easily work with your scope.

Also this software will do everything you might want and just a click to run it:

https://github.com/Gordon999/Pi-Camera-GUI

 

I could give you the code for a desktop shortcut and tell you how to set it up more simply if you needed.

 

The pi HQ and Raspberry pi take up no space 

[colinbm]

My desk looks like that too....

[Andrea B.]

Those desks are a sign of serious adventurers in the UV realms !!

*****

 

Jonathan, I've been so tied up with the forum software upgrade that I'm just now stopping by to tell you how amazed I was by this microscopy work !! It is a striking example showing how the shorter wavelengths bring out surface details.

 

Keep in mind that if we have enough interest, I can always create a Micro tag and/or open a UV Microscope section.

[JMC]

David, yeah UVC microscopy for me will have to wait for now, but thank you.

 

Andrea, thanks. Yes it is fun and certainly brings out more details. The original pioneers in UV microscopy worked at 275nm and and 257nm. I'd be interested to know how many had skin and eye problems with all those cadmium and mercury UV sources.....

[Guest]

On 10/5/2021 at 4:24 AM, JMC said:

Now with hindsight, it looks as though 365nm microscopy is relatively simple to do - you can use 'standard' objectives, condensers and even eyepieces, and without any major modifications to the microscope. I've even done it with using a Nemo torch as the light source. However as is often the case with UV, if you want to go deeper that is where the problems start.

 

Can you expand on this? A bill of materials would be a nice addition.

 

Thanks

[dabateman]

Many of the Swift, AmScope, Omax trinoccular microscopes use glass in the path that has transmission above 350nm.

My Swift trinoccular microscope in my image is similar.  All the focusing glass above the stage transmits above 350nm. Its just the Led light source and the condenser that block UV,  light below 400nm. There is a plastic dome over the led box. So you can modify or remove them or just not use them.

I have enough work distance that I can do UV reflectance shinning a Convoy led flashlight from the side or underneath at angle. Fortunately its RMS mount and many objectives work well with it.

 

I got this scope for $200 in January 2020, just before COVID for my daughter.  So I have had to reverse some of my modifications and play with it, as she has shown an interest in using it again.  Its actually not on my desk in the photo, but a table in the family room.

So I no longer have a fiber coming out of one eye piece for spectral measurements of samples. And removed just after I fixed the parallax issue and tests. Oh well, dark slides also work well with 10x objectives for dark field. 

[dabateman]

Its the Swift SW350T currently $300 on Amazon. 

But this one might work too:

https://www.amazon.com/dp/B095372D8W/ref=cm_sw_r_cp_apa_glt_fabc_4T4WCX7HKCYH51233TYT?_encoding=UTF8&psc=1

 

[Guest]

Ok, I see. So this topic was more or less to show that JMC could build something on his own through his own research.  I can fully respect such a journey. 
 

However, are you telling me that he built the equivalent of a $300 microscope package or is it more than that?

[Andy Perrin]

Blazer, no! He did not build the equivalent of a $300 microscope for many times that amount. Jonathan’s scope is good out to UVC in principle. He just doesn’t have the LIGHT SOURCES for that realm. 
 

In UV imaging, it gets increasingly hard to image at shorter wavelengths. All the problems multiply below 350m. Materials don’t want to transmit light. The light itself is dangerous and can make ozone in some cases. Lenses made for visible light are not corrected enough to work. The camera sensor must not have the dreaded AR coatings and preferably should be debayered. Etc etc. 

[dabateman]

I wouldn't be surprised if Jonathan in the end spent $10000 or a little over.

But he switched out all the glass for fused silica, got a custom made condenser and bought some really nice quartz objectives. Then may have over and started collecting some really cool objectives. Quartz objectives can be $500 to $5000 each.

I got a 15x and 40x mirror objectives which can also see UVC. Much cheaper than quartz ones.

He also got a really cool vacuum UV objective,  which can be used at 193nm.

So not your typical build.

 

My point was an off the shelf Chinese scope are typically ok for UVA. Below that and you need to add more zeros to your budget. 

 

I did get a really cheap 1x and 2x RMS objective set, and the 2x is good to 300nm. So I lucked out with that purchase.