A few weeks ago I posted some initial work which used a converted high street camera to be ale to make images at 254nm, which is down into the UVC region (see here). Frankly I was stunned to see the camera capture anything at such a short wavelength. The camera used for this was a monochrome converted Nikon d850 from MaxMax, which had had the Bayer filter and microlenses removed. Of the cameras I have, I thought this would give the best shot at seeing that far down, and indeed it did. Some of the folks on the Ultraviolet Photography forum have been looking at image UVC for a while, but until recently I had discounted it as I thought my cameras lacked the required sensitivity to be able to capture anything. After seeing an image with the monochrome Nikon d850, it got me wondering again though about looking at 254nm with a camera with the Bayer filter still in place.
For colour imaging at 254nm, I turned to a Sony A7III which was converted to multispectral imaging again by MaxMax. When this was done I’d requested the sensor coverglass be replaced with a quartz window, instead of being left in place which is normally what happens with these types of conversions. Using the came setup as discussed here, an image of the feather in a vase was captured along with the diffuse reflection standard. After white balancing this was how it looked.
As before I captured an image with a Schott WG305 filter as well to see how much of the image was coming from longer wavelengths, and here’s how that looked.
As with the monochrome camera, a faint image was seen with the WG305 in place, indicating what I was capturing at 254nm was not a pure 254nm image, but that there was some contributions from longer wavelengths. However the majority of the image was indeed coming from reflected 254nm light.
I was really surprised with the performance of the Sony A7III and the fact that I could get an image at 254nm with it, with the Bayer filter still in place. Given how much UV was absorbed at 308nm by the Bayer filter, I’d fully expected this to be the case at at 254nm, however it does not seem to be the case, and initial assessment of the images suggests that the Bayer filter is absorbing less of the light at 254nm than it does at 308nm, although this remains to be proven. Hint, hint, if anyone can deposit Bayer filter dyes on to a quartz plate at thicknesses which are representative of their usage on camera sensors let me know…..
It is possible to get some idea of how the different colours in the Bayer filter are transmitting the light from RAW files. Looking at a 308nm image in RawDigger as a Raw composite file you get a distinctly green colour cast to an image taken with a multispectral converted camera, as shown below.
The image above was taken with converted Canon EOS 5DSR camera using an Invisible Vision 308nm filter. The Raw composite file has a distinct green colour cast at this wavelength, showing the green parts of the Bayer filter have a better transmission here than the red and blue parts. Oh and do not go shining 308nm light on your skin, without doing an extensive safety assessment. You have been warned…..
The Raw file from the Sony A7III at 254nm looked like this.
At 254nm the green channels response from the diffuse reflectance standard are still slightly higher than red and blue ones, but there is much less difference between them all at this wavelength when compared to 308nm. As a result the Raw composite image has much less of a colour cast to it. This would suggest that the red, green and blue dyes have much more similar transmission to each other at this wavelength than at 308nm, although that remains to be proven.
One final thing before I end todays post. The Sony A7III has a quite amazing high ISO performance, so after I’d taken the images above I cranked up the ISO to 102,400, and took an image at 254nm with a 1 second exposure time!!!
While obviously noisy, the image at ISO102,400 is quite frankly astonishing, especially when you consider that this was a 1 second exposure at 254nm.
Yet again I have been amazed and surprised by how far into the UV region it is possible to look with what are at their heart commercial high street cameras. Even with the Bayer filter and microlenses in place, it was possible to capture images at 254nm using a converted Sony A7III camera. Thanks for reading, and if you want to know more about this or any other aspect of my work you can reach me here.
Back in April 2020 when Lockdown 1.0 kicked in in the UK, I decided it was time to try and learn a new skill. So it was that I embarked on a journey into the complex and fascinating world of microscopy. What I knew about microscopy at that point could have been written on the back of a very small envelope. Jumping in with both feet and having done practically no research at all into the area, I bought an Olympus BHB from the 1980s, which was, well, in need of a bit of love and attention. Affectionately named ‘Project Beater’ due the condition it arrived in, I started documenting my journey into microscopy here, and have been providing regular updates ever since.
Not long after getting the microscope, I started wondering whether it would be possible to make my own UV imaging system, capable of doing transmission microscopy down to and even below 300nm to help with my sunscreen research. Commercial UV transmission microscopes do exist but they are incredibly expensive to buy. To build one would be a monumental challenge, as glass absorbs short wavelength UV, and microscopes have a lot of glass in them. And when I say ‘a lot of glass’ I really was in for a shock as to how complex making a UV microscope would actually turn out to be. Since then I have been constantly modifying, tinkering, buying (oh, there has been plenty of buying) and tweaking this little Olympus BHB with the eventual aim of removing all the glass from the optical system that leads to the camera, replacing it with UV fused silica and other UV friendly materials.
This brings us to February 2021, 10 months after the initial purchase of Project Beater. With the final pieces of the UV conversion carried out, it was time to test it out and see whether it would even be able to image in the UVB region down at 313nm. 313nm was chosen as a test wavelength, as the mercury xenon lamp had a good strong line there, and the camera should still have some sensitivity there (it drops very quickly below that). For a test subject it made sense to use a sunscreen formulation which had a UVB absorber in the oil phase, and no UV absorbing ingredients. This way it should look darker at 313nm than at 365nm or in the visible. After applying a dot of product onto a quartz slide and adding a quartz coverslip, imaging in the visible region with a 32x Zeiss Ultrafluar objective gave the following.
The circular feature in the middle of the image is an air bubble in the formulation, and surrounding it is a dense network of small oil droplets in a water based continuous phase. The next wavelength to image was 365nm (UVA).
365nm was chosen for imaging in the UVA, as the mercury xenon lamp has a strong line there. To filter out unwanted wavelengths from reaching the camera, I used a 365nm bandpass filter from Edmund Optics. This blocks wavelengths outside the region of interest with an efficiency of OD4 (optical density 4, or in other words transmission in the out of band regions is <0.01% on average). The 365nm image looks similar to the visible one, which it should do, given the sunscreen ingredient it contains does not absorb UVA. Something to notice, in addition to the air bubble in the middle of the image, in the visible and UVA images there looks to be regions where the emulsion looks ‘lighter’, specifically to the top right and bottom right of the air bubble. I shall come back to those regions in a minute.
Once the UVA image was captured, I changed the filter for a 313nm UVB one and imaged the area again, getting the following.
At 313nm the image looks a little different. The air bubble is still visible, but the image looks darker, and those areas which were lighter in the visible and 365nm images now look darker than their surroundings. I suspect these darker areas are oil rich, and the UVB absorbing sunscreen is in the oil phase of the product.
It was great to see a UVB image appear at 313nm. However it certainly has its challenges. With the 313nm filter I have, even though it was rated OD4 for out of band regions, the degree of blocking it offered was not sufficient, and when first tried I was getting some none UVB light coming through to the camera. I had to use another 310nm OD4 bandpass filter in series with the first one, to block unwanted wavelengths (unfortunately I didn’t have a second 313nm filter, but the 310nm still let the 313nm line through). Camera sensitivity was very low at 313nm, and I had to increase the ISO setting on the camera from ISO 400 for the visible region to ISO 6400 for 313nm, and the time to capture an image increased from 1/80s to 30s!!! When you think that each image above is a stack of about 10 images, with slightly different focal points, you start to realise the logistics involved with trying to capture UV images. Oh and I really need to get better at preparing slides – this sample was not the best. One of the scariest things with making these is the knowledge that a quartz coverslip cost about 20GBP, so there is no ‘use it and throw it away’ as with glass ones. Once finished with the coverslip has to be cleaned along with the quartz slide. At 0.35mm thick to 20GBP each, cleaning a quartz coverslip is not a fun job…..
How to try and visualise how the effect of of converting the microscope to UV has had on the throughput of light to the camera? Back in June 2020 when I did my first UVA imaging of sunscreens, I also did some measurements of throughput of light through the unconverted microscope, shown here. Taking that data and normalising the irradiance to the max peak height gives the following.
In the graph above the red line shows the output from the 200W Hamamatsu Xenon lamp. It shows a long tail down into the UVB and UVC region, and there is still light present even at 250nm. The blue line is the light having passed through the microscope. This shows that below 400nm, the UV starts to be absorbed straight away. By 350nm, about 75% of the light has been absorbed by the glass in the optics of the microscope, and by 330nm, there is nothing getting through. I was actually surprised by how much UVA did go through the standard microscope, and in that form it would be suitable for UVA imaging, if used with good objectives and ideally a quartz photoeyepiece.
For the final build, I decided against using the Hamamatsu 200W light source, as logistically it was hard to use. I settled on a Zeiss 50W mercury xenon lamp, which while have a more ‘peaky’ spectrum, has strong lines at very specific wavelengths in the UV region. So how does the transmission through the microscope look now it has been converted to UV?
In this graph we now have 3 lines. The blue line is the 50W lamp alone before it is connected to the microscope. The irradiance spectra has peaks because it is a mercury xenon lamp. The red line shows the normalised spectra after it has passed all the way through the microscope to the photoeyepiece at the top, which is where the camera connects. After the conversion to UV, the light going through the microscope has a very similar profile to the original light source, showing only a small reduction down at 300nm, and even down at 250nm. This is because all the glass has been replaced with UV fused silica, and the objectives and and condenser are quartz, UV fused silica, and calcium fluoride. In this graph there is a third line in green. This is the light that comes out of the eyepieces. I added yellow filters here to block light below about 500nm as a safety precaution (the line is much more noisy than the other two due to the complexity of measuring the output from them). I will however still be using UV safety glasses whenever using this, and also using a UV blocking filter over the input whenever I use the eyepieces for focusing.
The goal when I set out on this conversion was to build a microscope which I could use for visible light imaging, as well as down to and below 300nm. Also, it had to be fairly compact as my work surface is not that big. By converting an existing microscope, and using a Zeiss 50W HBO light source, I managed to keep the size down, and retain functionality for use in the visible light region.
It has been a hugely complicated build, but very rewarding and I’ve learnt so much about microscopy and imaging in general. There are of course areas for improvement. The camera I have has low sensitivity in the UVB and by about 300nm would be unusable. At 300nm and below I’d need a UV enhanced camera, but that would require connection to a computer to run it. The Edmund OD4 filters really need to have 2 of them stacked together at 313nm and below, as individually they cannot block the out of band regions of light. This is costly though. Also, to change filters at the moment I need to remove the camera, so doing multispectral images of the same sample is pretty much impossible. Ideally the best thing to have would be a modified tube from the microscope to the camera, which had a opening to put filters in and out. This would however need to be custom made and will have to wait for now. The main thing I can do though is learn how to prepare slides better, especially for looking at emulsions.
If you’ve made it this far, thanks for reading, and if you’re interested in knowing more about this or other aspects of my research, I can be reached here.
Ultraviolet C (UVC) is something that until recently the vast majority of us haven’t had to be concerned about. UVC is short wavelength UV that is even more energetic than UVB which is what is responsible for burning us when we go out in the sun for too long. Although present in the light emitted by the sun, our atmosphere absorbs UVC before it reaches the ground, so acts as a natural filter for it. As a result of this efficient filter process we are not exposed to it from sunlight. However UVC can be generated on Earth from a some high energy processes such as when arc welding. Recently there has been a huge increase in the number of UVC lamps in the market place, because UVC offer a route to sterilising surfaces, an area which has received much more attention since Covid arrived on the scene last year. Now if you go into a drug store or surf the web there are a massive range of ‘UVC lights’ which claim to be able to clean and sterilise surfaces that are exposed to them. With these becoming more widely available, there is however a bigger risk of exposure to the wider population and that is because of the effects that they can have on skin and eyes. UVC is highly energetic and can result in painful burns on the skin along with damage to the eye if exposed.
This got me wondering whether topical cosmetics products which are designed to protect us from the UVA and UVB in sunlight could also offer protection from UVC light?
Recently I developed an imaging setup which can be used to capture high resolution photos at 254nm which is in the middle of the UVC region (see here). To test this out this question about sun protection products, I took a couple of moisturisers, one with no SPF ingredients, and one with was SPF 30 rated, and applied them to test plates used for in-vitro SPF testing. These plates are made of a polymer called Polymethylmethacrylate (PMMA) which is transparent to UVA and UVB, and slightly transparent to UVC. I dosed the products at about 2mg/cm2 which is the dose used for SPF testing.
Once applied, I then photographed the plates under 254nm light, to see how they compared and got the following.
What is the image above telling us? This a 254nm (UVC) reflected light photo, so the brighter areas are reflecting more 254nm light, and the darker areas are reflecting less. The plate with the moisturiser without SPF, in the middle of the image, looks ‘medium grey’ indicating some absorption of the UVC. The test plate itself absorbs some of the 254nm light, however the none SPF moisturiser is absorbing some of light as well – this not surprising as UVC is absorbed by organic material, and the moisturiser is an oil in water emulsion. The SPF30 moisturiser is much darker than the none SPF moisturiser though. This indicates that the SPF30 moisturiser is absorbing much more of the UVC radiation than the SPF free version. Although designed to protect the skin against UVA and UVB, the SPF30 test product is also absorbing UVC.
While I had the PMMA plates with the moisturisers on them, I thought I would run the transmission spectra through them on my Ocean Optics FX spectrometer, just to see what they looked like. First set of scans is between 250nm and 800nm.
So what is going on with these transmission spectra? The blank PMMA plate has a fairly flat transmission spectrum between about 300nm and 800nm. Below 300nm though, transmission drops rapidly and by 250nm it is almost opaque. It should be noted that the surface of the PMMA plates are rough, which is why the max transmission isn’t very high. Actually when the moisturisers are put onto a plate, max transmission in the visible part of the spectrum becomes higher than the blank plate. This is because the moisturiser smooths the surface that the light comes into contact with, and thereby reduces the scattering, and increasing the amount of light that can pass through the plate.
Below 400nm, the moisturisers behave differently, and this can be seen more easily if I just plot the 250nm to 400nm region.
Below 400nm, we can see how the SPF product absorbs much more UV between about 275nm and 400nm when applied to the PMMA plate. Both products absorb some of the light at 250nm, although as can be seen the PMMA plate itself is also absorbing most of the UVC that far down. The transmission spectra shown here don’t fully explain the photograph of the moisturiser coated plates, in which the SPF product looked darker than the none SPF one at 254nm. The photos are looking at reflection of light from the subject, while the transmission spectra are just that – transmission – and this may be accounting for the difference between what the photo and spectra are saying. What’s even more obvious is that the PMMA plate is not really suited for transmission measurements when you get down to 250nm. This it not a huge surprise as it was not designed for that. It does mean though that these are not ideal for use as a transparent substrate at 254nm, but I have some ideas about what to use down there.
The last 12 months have seen rapid changes to our lives, some of which brings additional risks which we need to think about. The availability of UVC lights to the wider population has exploded, and along with that there comes to risk of skin damage from exposure. UVC photography enables this invisible light to be imaged and also to visualise how it interacts with surfaces, and as can be seen here whether topical skin care products can be used to absorb it.
Thanks for reading, and if you’d like to know more about this and any other aspect of my work, you can reach me here.
With some experiments, the end result is fairly easy to predict. Others, not a chance. With this piece of work I had no real idea what to expect, or even whether I’d be able to see anything at all. Essentially, it was this – can I image at 254nm using the equipment that I already have?
To start with some background. I have an interest in UV photography because of my work on sunscreens, and as part of my research in the area, I’m always looking to push the limits of what can be photographed with the kit that I have. My monochrome converted Nikon d850 camera has a really nice sensor for UV, visible and IR imaging, and I was able to measure the spectral response of it down to 280nm which is the limit of my measurement capability. At 280nm there was still a little bit of sensitivity there, but it had dropped to almost nothing and I’d assumed (always dangerous) that below 280nm it would be essentially zero. When it comes to UV imaging, very often specific wavelengths are discussed, as they correspond to lines in the mercury emission spectrum. 365nm and 313nm are quite strong emission lines, and they lay within the UVA and UVB regions respectively. However drop below 280nm and you get in to the murky world of UVC. While there is UVC in light coming from the Sun, it does not reach ground level as the atmosphere blocks it. However there are artificial sources of UVC. Welders are often exposed to it due to ionisation of the air and metal, where it can result in bad burns to skin and damage to the eyes, and in our new Covid related world UVC sterilisation of surfaces is becoming more and more widespread. This got me wondering whether I could break through into the UVC region and capture photos at 254nm which is another one of the strong mercury emission lines.
This would be no simple task. Even if the camera was sensitive enough to capture an image there, the sensitivity it has would be extremely low, meaning that the presence of any other wavelengths would completely swamp the signal from 254nm. Therefore really good filtration would be needed. For the filter I used a 254nm bandpass filter I got from a Sirchie forensics camera a while back. While it has relatively low transmission at 254nm, it has good out of band blocking, and it meant I didn’t need to spend a fortune on a new filter. Any glass in the lens would absorb the UVC, so I chose the 105mm Rayfact UV lens for this experiment – it has quartz and calcium fluoride lens elements instead of glass and is good for using down to around 220nm. Camera was my monochrome converted Nikon d850 from MaxMax. When I had this made I requested a fused silica sensor window, in case I ever wanted to go below 280nm. Light source was an interesting one. In the end I got a pair of 8W 254nm UVC fluorescent tubes from ‘a well known internet auction site’. These are normally used for sterilisation of water, but they fit my UVP UV lamp so I thought I’d try them out. Subject wise, a small vase with a couple of feathers and a Spectralon 20% diffuse reflectance standard was chosen.
For imaging, everything was done in a dark room, and the subjects were placed in a box painted with Semple Black 3.0 paint which has low reflectance and low fluorescence, to cut down on stray light. I wore UV protective glasses (in fact that point deserves capitals – I WORE UV PROTECTIVE GLASSES – please do not play around with UVC lights without wearing these). In addition to imaging the subject using the 254nm filter, I also took another image with a Schott WG305 long pass filter in front of the 254nm filter, to check for stray light getting through.
Right, that is far too much waffle. Was it possible to capture a UVC image with this setup? Well, yes it was, and this is what it looks like.
Just for comparison, here it is when imaged in visible light.
The first thing to note was the vase – the glass looks black in UVC. This is because the glass strongly absorbs the UVC. How did I know this was a UVC image? It’s very easy to fall into the trap of assuming that this is a UVC image, but by using a Schott WG305 long pass filter as well I could block the UVC while letting longer wavelengths through, and this is what I got with that.
While it is hard to see online, there is a really faint image here with the Schott WG305 filter in place, so it is fair to say that my 254nm image, while predominantly UVC is not all UVC and there is a tiny amount of contamination of the image by longer wavelengths. However it is fair to say that the image with the 254nm filter alone is mainly a UVC image.
At this point it is worthwhile looking at the properties of the filters and the light source to help explain what is going on. Here they are between 230nm and 800nm.
And now between 230nm and 430nm.
What are these graphs telling us? The light source (the purple line) has a strong emission at 254nm, with a few very small peaks at longer wavelengths. There may well be more lines 800nm but I cannot see up there. The peak transmission of the 254nm band pass filter (the green line) coincides with the 254nm lamp which is good. The red line (WG305 filter) is the filter I used to check for leaks and as you can see there is a little bit of an overlap with the tail of 254nm filter at around 300nm, and this could be source of the faint image in the WG305 photo above. I may try a slightly longer wavelength long pass filter to check that. The blue line is a ‘red herring’, as I haven’t included images with that filter here, but it is another long pass filter and was part of my other tests on the setup. Here’s a snap of the setup I used to capture the images.
The observant amongst you will notice that the camera in the image above is not a Nikon d850. It is actually my Sony A7III which has been converted to multispectral imaging. This camera, unlike the d850, still has it’s Bayer filter though, so can capture colour UV images. More to come on that, as I’ve been able to get colour images down in the UVC region with it…..
When this experiment started, I honestly did not know what to expect. I’d assumed that the camera wouldn’t be sensitive enough to see anything this far down, but I was wrong and the camera is indeed sensitive enough to see 254nm light. Always nice to be surprised and this opens up its possible uses to new areas such as coronal imaging. Filtering will need some more work, as it’d be good to get a more narrow band pass 254nm filter, but for now this one will do the job.
If you’ve made it this far, well done and thanks for reading. If you’d like to know more about this or other aspects of my work, you can reach me here.
When looking out for interesting imaging equipment, I normally have a bit of an idea what I’m after – usually my focus is on UV imaging, and that then has a bearing on the types of things that I am looking for. Since getting into microscopy though my horizons have been broadened as there is some really quirky and interesting stuff out there. This leads me to today post – the Leitz Ks series microscope objectives. These were part of a set of lenses that Leitz produced for a very specific job – looking at the grain in photographic emulsions. Having spent a fair bit of time inhaling fumes in a dark room when I was younger, I thought this was a cool idea and set about trying to find some of these objectives….
Finding these was not too hard, but finding them for a reasonable price was a bit more difficult. In the end, one came from Germany, one from a dealer in Sweden, and two from a lab equipment reclamation company here in the UK. Here they are.
In the range there were 4 objectives. From smallest magnification to largest, there was the 22x Oel (oil immersion) A 0.65.
Then we have the Ks 45x A 0.65 dry objective.
Next is the Ks 53x A 0.95 fluorite oil immersion objective.
And finally the Ks 100x A 1.32 apochromatic oil immersion objective.
All of these objectives were designed for Leitz 170mm finite tube length microscopes. Normally, on an objective after the tube length is the thickness of the coverslip that they are designed for, however these are different. With these there are different numbers and then a ‘mu’ symbol.
So what’s going on with these objectives? At this point it is a good point to dig back through the archives for some information. There was a Leitz brochure produced for these objectives, and I downloaded a copy from online. Here’s the front page.
The first page of the brochure give a bit more of a background to the objectives in the range.
As can be seen, the full range had 4 different Ks objectives, and two others. The numbers after the tube length, refer to the maximum layer thickness of the photographic film that they were designed for. Oddly on the 22x objective it says 0-2300, instead of the 2200 it has in the table above, while on the other objectives, the numbers correspond to the ‘max layer thickness’ in the table.
The next page in the brochure gives more details on the objectives.
After getting these I thought I’d see what type of images I could get from them, so dug out an old slide and the immersion oil (for the 3 objectives that needed it). My Olympus microscope has a 160mm tube length (as opposed to the 170mm tube length that these were designed for), but I have used other 170mm tube length objectives with success so thought it was still worth a try. At this point I should be sharing some lovely images showing the grain structure in the slide, however I have nothing for you, sorry. The images were actually not that good, from any of them, and I got sharper ones with my normal objectives. Not sure why this is the case to be honest. I struggle to believe that they are all damaged. Online, I was only able to find one thread talking about using them, and they were struggling too, so perhaps there is something odd about their design which makes settling them up and using them difficult. So for now they remain an interesting piece of history.
Thanks for reading, and if you’d like to know anything about this or other aspects of my work, you can reach me here.
Some of the experiments I do fall into the ‘Please feel free to try this at home’ category. Others however are of the ‘Do not attempt to do this yourself without full understanding the risks involved. No really, I mean it, stop what you are doing now and step away from the work bench’. Today’s post is firmly from the second category. Please do not attempt to do anything like this unless you understand the risks of working with high intensity UV light. With that out of the way, let’s begin….
For my UV microscope build, almost everything has needed modification in one way or another, in order to be usable down in the short wavelength range of UV below 300nm. My main UV light source – a 200W Hamamatsu LC8 mercury xenon lamp – gives out plenty of UV down to just below 300nm, but is really awkward to use with the microscope, due to the position of the output from the lamp. I have a couple of mercury xenon lamps for microscopes – a 100W Olympus one, and a 50W Zeiss HBO – but when I initially checked the output from both of these there was a bit of a problem. While they both had a good output at 365nm, down below 320nm there was essentially nothing. If I’m wanting to look down at 300nm and below, this is a bit of an issue as I need light down there. Initially it struck me as a bit odd, and had me wondering what was going on. Both microscope lights were using mercury xenon lamps like the Hamamatsu LC8, so why was output so restricted? Perhaps the bulbs were made of different materials which didn’t have as good a UV transmission? This could be possible, but it would be very odd. So was it something else?
For now I have settled on trying to use the Zeiss 50W HBO setup rather than the Olympus one. This is mainly driven by space constraints – the Olympus light source is huge, and I just don’t have space to have it out on the bench all the time. The Zeiss HBO is dinky, so is much less of an issue to leave out. Here it is on the bench.
If you look down the output port of the light (with it switched off obviously), there is a metal tube which can be moved in and out by turning a knob on the side of the light housing.
At the far of the tube near the bulb itself is a lens, and the movement of this tube and lens actually focuses the light. But the lens is a problem. It is highly unlikely that this lens is a UV fused silica or quartz one, so it’s going to be glass, and glass absorbs short wavelength UV. This lens will be blocking the light that I want to see. To determine if this was indeed what was happening, the first thing to do was remove the tube and lens. This was actually pretty simple, by loosening the small screw near the focus knob. Once removed, the tube looked like this.
The lens is held in with a simple spring clip, but this was down at the bottom of the tube and was really, really tricky to get out. Once out, the lens itself could be removed, and this is what it looked like.
It’s a odd looking, asymmetric design, with the highly curved side facing the bulb. Focal length was about 20mm, which make sense given where it sits in relation to the bulb. Once removed I could measure the irradiance spectra of the light, with and without the lens to see what effect it had on the output. This aim here was not to determine absolute irradiance, but to understand the relative output with and without the lens to understand what wavelengths it was absorbing. For this I used my Ocean Optics FX spectrometer, and got the following.
The really obvious difference between the two plots is in the UV region – with no lens in place there are a lot more lines there than when the lens is in the way, especially below 350nm. To try and show this better I normalised the lines to the 578nm peak (which should be pretty similar whether or not the lens is there), and replotted the graphs with an emphasis on the UV and near-visible region.
Plotting the data like this really emphasizes the effect of the lens. With the lens in place the 365nm line is starting to be attenuated. By the time you get to 330nm, most of the light is absorbed, and then below 310nm there is nothing, all the UV has been blocked. With the lens removed, there are still well defined UV peaks at 313nm, and even below 300nm, down to where the spectrometer reaches the edge of its calibration at 250nm. This is not a huge surprise – that lens is thick, and there’s no wonder it is absorbing loads of the UV. Success then, removing the lens brings back the peaks I’d expected from a mercury xenon lamp. Yay.
Simply removing the lens means that this light source can be used for UVB imaging at around 300nm and even below. What is the downside here though? Removing the lens means losing some light intensity, as the light is no longer focused. But in theory I could replace the lens with a planoconvex lens of around 20mm focal length, so give me some rudimentary focusing capability. I could even use a custom UV fused silica condenser lens, but this will be an expensive option. While probably not as good as the complex shaped lens originally used, a simple planoconvex lens should offer improvement over no len, and is something I may consider in the future if light intensity isn’t sufficient.
What have I learned here? Well I certainly know a lot more about the inside of mercury xenon lamps now. Not everything works the first time (and no, this is not the first time I have realised that). However think logically and approach a problem systematically and a solution can usually be found. While I have gone quickly through the description of what I did here, I must emphasize that working with these types of UV lights is not without risk. Exposure to the skin can quickly result in burns, and to the eyes can result in blindness. I made sure I was well protected when doing this work.
Another part of my UV microscope build can now be ticked off, and I am one step closer to being able to image in the 300nm region and below for my sunscreen work. 2021 is starting to look better already……. Thanks for reading, and if you’d like to know more about this or any other aspect of my work, you can reach me here.
While learning about microscopy during the last year, I’ve also been on the look out for information on historical items which were designed for use in the ultraviolet (UV) region. Although the vast majority of lenses are based on glass optical lenses, manufacturers have occasionally made UV lenses using materials such as quartz and calcium fluoride instead of glass, in order to let short wavelength UV light through. Leitz made a number of objectives and condensers for UV, which I’ve previously written about here, here and here. Lomo made a wide range of UV objectives, photoeyepieces and condensers, containing quartz lenses and their 10x objective is discussed here and photoeyepieces here. Zeiss have made probably the widest range of lenses and objectives for UV, marketed under the Ultrafluar label. These were available for the 160mm finite tube length microscopes, and continue to be made for the more modern design infinite tube microscopes. I have a few of these Zeiss objectives, and will write further on them after a bit more work with them. Before the Ultrafluars though they also made a quartz condenser and objectives, however it is harder to track down information on them given their age.
A few months ago I came across someone selling what was advertised as a Zeiss quartz condenser, and after a bit of discussion I bought it. For nearly 6 months it remained sitting on a shelf in the US, and I finally got it a few weeks ago. It consisted of 3 parts – the main body of the condenser with an adjustable iris, 2 additional lenses, all contained in a wooden box as shown below.
A search online revealed nothing about this, so my first check was a transmission measurement, which showed the following.
It was a huge relief to see it confirmed that these were definitely not glass but quartz lens elements (glass transmission would drop in the UV while these remained constant). Phew.
The history of the equipment I have is important to me, as I like to learn about what things were used for before and when they were made. To try and find out about this condenser set I approached the Zeiss Archives, with the hope that they could cast some light on it and when it was made. Unfortunately they couldn’t find record of this one, but that wasn’t a huge surprise as these would have been rare items with very few being made. To try and find something out about it, the next step was to get back in touch with the vendor. They pointed me towards a Zeiss Microscopes catalogue from 1934 which listed it. This was a period of time when catalogues were real hard back books, and I managed to track down a copy here in the UK. On examination it was indeed in the catalogue as shown below (with the part number given as 11 42 50).
Interestingly it also appears in a 1927 Zeiss book with the same part number.
Something with the same part number (this time shown as 11.4250) but slightly different numerical apertures for the lenses even appears as far back as the 1913 catalogue.
Looking back through some pre-1913 ones I couldn’t find any mention of it. It looks therefore that mine is probably from somewhere between 1913 and 1934, although I’ve not been able to verify how long it was made for after 1934. They really did make these well – the iris moves and adjusts as if it were made yesterday.
In the research world it is very easy to become fixated on the future, the goal of the work. We should not forget the great science and engineering that has already been done in order to get us to where we are. As with all historical research, it was fascinating to dig into the archives to try and find out about this condenser. It introduced me to the Zeiss Archives, a vast online database of equipment (although only in German, which makes things a little more challenging), and along the way I’ve been able to learn about other areas which will help with my work in the future.
If you’ve made it this far, thanks for reading, and if you’d like to more about this or any other area of my research, you can reach me here.
In my work on UV imaging, I’m always on the lookout for suppliers who can offer solutions for UV lighting. Normally, I’d use xenon flash systems, xenon and mercury xenon continuous light and fluorescent sources. Problem with all these is that you get a lot of unwanted wavelengths, and when using a camera for UV photography, what you don’t want is lots of visible and IR light that needs to be filtered out. As such light emitting diode (LED) technology would seem to be an ideal solution as it can offer relatively narrow bandwidths in a small package without the need for additional filtration.
UV LED technology is evolving rapidly. A few years ago 365nm was pretty much all that was available, but the range of wavelengths available is now extending further and further into the short wavelength UV regions. This is being driven in no small part by the use of UV-C light for cleaning and sterilising materials, an area which has become even more important given what is going on globally at the moment.
I was recently searching for a collimated UV source that would be able to provide light in the 250nm to 300nm range for my UV microscope build, and this lead me to AquiSense Technolgies, and their PearlLab BeamTM system. This is a compact, collimated beam device which employs UV LEDs and emits radiation in the germicidal range. Its small size makes it ideal for use as a table top system, and it is available in a range of options from 1 to 3 different wavelengths. Normally used for imaging Petri dishes, I can see this type of technology becoming more and more widespread for the UV imaging and microscopy world, for both reflectance and fluorescence work, especially as LED powers and efficiencies continue to rise.
AquiSense Technologies is the global leader in UV-C LED systems design and manufacture. They work with leading LED manufacturers to evaluate their devices and then design efficient disinfection products. Using a combination of patented technology and in-depth know-how, AquiSense integrates LED devices into products that solve real world problems in water, air, and surface applications. You can find out more about their range of LED UV lighting solutions here.
2021 is well under way now, and of course that means that here in the UK we are in another lockdown. On the plus point with the nights being long and the weather cold, I have time to share something with you on one of the recent lenses to arrive here – the Astro Berlin 120mm f2.1 Quarz Objektiv. For those of you who haven’t heard of Astro Berlin they were optical company based Berlin, and were founded in the early 1920’s (I’ve seen 1921 and 1922 mentioned for this). While originally they produced astronomical optics, they soon changed to camera and cinecamera lenses. Remaining active in East Berlin after WWII the company finally went into liquidation in 1991. They have gained a bit of a legendary reputation amongst photographers and cinematographers, due in part to their combination of long focal lengths and fast apertures with the optical characteristics of older designs of lenses. As such they tend to demand high prices from those wanting to use them as well as the usual array of historical optics collectors.
During my historical research, I’ve come across a couple of Astro Berlin lenses which were designed for UV imaging – the 44mm f2.5 Quarz-Tachar, and 135mm f4.5 Quarz-Anastigmat – and as you can imagine, if the normal glass Astro Berlin lenses are rare, these are almost mythical. However a few months ago I saw an advert for an Astro Berlin 120mm f2.1 Quarz Objektiv, which was a new one to me. This looked to have been designed for large format photography and the seller mentioned it fitted a Compur II shutter, and that they believed it was a quartz lens as they had used it to capture images in UV light (I’ll come back to these points later). After a bit of discussion, I made my offer for it which was accepted, and it arrived just before Christmas. Here’s a picture of the lens (along with a nice reflection of my kitchen in the front element).
The specifications of this lens were quite impressive – 120mm focal length and f2.1 aperture. Compare this with the UV Nikkor (105mm f4.5), UV Sonnar (105mm f4.3), Asahi Ultra Achromatic Takumar (85mm f4.5), and even the Zeiss UV Objektiv (60mm f4), and as you can see this Astro Berlin lens has an extremely large aperture. Image size was mentioned as 6x6cm or even 6x9cm for macro work. As you can imagine, I was rather excited when this arrived, and the first thing to do was test it for UV transmission. The lens itself easily dismantles to produce a single front element, a central section for Waterhouse stops, and then a rear section which is actually a cemented doublet (which started to get me a bit nervous as cemented elements for a UV lens, especially an old one, would be unusual). What did the transmission measurement show? Let’s take a look.
Hmmm, what is going on here? The front element is not glass and it’s safe to assume this is quartz based on the transmission spectrum which is still looking good down at 280nm. However the rear group does not look to be quartz as it is showing a drop in transmission from 400nm down to being essentially opaque by 320nm. Don’t get me wrong, this is pretty impressive transmission for a glass lens, but obviously when the 2 parts of the lens are used together, this limits the overall transmission to wavelengths longer than 320nm.
What about this doublet at the rear of the lens? This rear element can be seen in the image below.
The image above shows the side view of the rear cemented doublet on the left, and the lens retaining ring on the right. You can see it is a doublet, due to the depression in the black coating about a third of the way in from the left hand side. Whether one (or both, but that would be odd) of the two parts of this doublet are quartz will have to remain a mystery for now, as I don’t want to scratch off the black coating from around the edge to check. It could also be that the adhesive holding the lens elements together (most likely Canada Balsam) is blocking some of the UV, so that is something I’ll be testing in the future as I haven’t been able to find anything definitive on that.
The lens does not have an aperture at the moment, and is built for Waterhouse stops, so that is next on the list of things to make, which will enable me to stop down the lens.
How to capture images with this lens? The lens mount was apparently that of a Compur II shutter. Problem with that is that the Compur II shutter design parameters (thread diameter and pitch) were never actually standardised. Measuring thread diameter is fairly easy, and this was 50mm. Thread pitch is a little harder to measure (well it is for me anyway), and as far as I could tell this was a 1mm, or close to 1mm. To adapt this to a camera, I took the following approach – a M50x1 to M65 adapter from RafCamera, a M65 helicoid, a M65 to Hasselblad bayonet adapter (also from RafCamera), and then a Noveflex adapter to go from Hasselblad to Canon or Nikon. Most of these parts I had already from other projects, but the M50x1 to M65 needed to be ordered. Putting all these together and mounting the lens on a camera looks as shown below.
The length of the helicoid I have limits this to macro work only at the moment (need to get a shorter M65 helicoid at some point). Also the M50x1 adapter seems to have a very slightly different thread pitch to the lens. I mentioned above the thread pitch is hard to measure accurately and I now suspect the lens is threaded with something slightly different to M50x1. It is close enough to mount it well, but just doesn’t screw all the way on. Despite that this setup at least allows me to capture some photos with it. The sun came out today, so I got a few shots in the garden, just using visible light with a normal colour camera, which are shown below.
As can be seen above the lens is quite ‘glowy’, which is not surprising as I am using this wide open without any stops, and my finger covering the hole in the barrel of the lens, where the stops would go. Below is a shot showing the entire 35mm frame (reduced in resolution from the whole 50Mp for sharing).
Cropping this and showing the central part of the image at the original resolution it was captured at gives the following.
There certainly is plenty of glow there, but that thin thread is a strand of spider silk, showing the lens is certainly capable of capturing plenty of detail.
The nice thing about this lens is that the rear doublet can be removed, and the lens still fits the same M50x1 adapter. This allows imaging using just the quartz singlet at the front of the lens, and an example image is given below.
Funnily enough the image with just the quartz singlet gives quite a similar image to the one taken with the complete lens shown earlier in the post, at least on my 35mm full frame digital camera. The good thing about this is that being quartz it can be used for imaging in the UVB region below 320nm.
The lens does not have an adjustable aperture, but a slot for taking Waterhouse stops. While these are normally made from brass, I thought I would make a set from 1mm thick black plastic sheet.
I made a set which should in theory go from wide open to about f11, although this is just approximate, and the hole sizes were determined by what size drills I had as much as any maths. How do they work? To try this out I got a set of shots of a pine tree in the garden, firstly without any stops in place and then from largest to smallest.
Going from ‘no stop’ to the largest stop, produced only a very slight increase in depth of field, and I suspect this due to the metalwork on the inside of the lens. After that, as the hole of the Waterhouse stop decreases in size the depth of field increases, just as it would when you make the aperture smaller on a more modern lens. Also the images become much less ‘glowy’. Success. As they are at the moment, they are a bit loose in lens when slotted in the lens, so I may add a layer of black tape to one side to just increase the thickness slightly and make them more snug in the slot.
After all this, where have we got to with the Astro Berlin lens? While seemingly not an all quartz lens, this certainly has good UV transmission, and has a quartz element at the front. Using some off the shelf adapters, it can be mounted to a range of different cameras including 35mm SLR, and medium format – it’d be lovely to try this on a UV sensitive Hasselblad at some point. It also can be dismantled and even when using just the front quartz singlet, can still produce a usable image which would enable it to be used for UVB imaging. When there is a bit more UV around I shall get some images with it and see how it performs.
Thanks for reading, and if you’d like to know more about this or any other aspect of my work, I can be reached here.
Happy New Year everyone. 2021 brings promise of a new start and I hope to bring you plenty of cool science over the next 12 months. I have some pretty special lenses and microscope kit to review, so watch this space. Today though, we start with an update on the UV microscope build. This has been a project I’ve been working on for about the last 6 months, and has required some major modification of my Olympus BHB to make happen (see here for the previous summary of the work to date). Today an update on the modification of the binocular/trinocular head, and the results of some major mechanical surgery.
When I took the Olympus head (I’ll just refer to it as a ‘head’ from now on, rather than repeating myself every time) apart, I realised that the construction was going to be quite difficult to adapt for UV. There was a lot of thick glass in there, along with a prism which acted as beam splitter – see here for the initial assessment. Instead of trying to replicate the prism exactly, I made the executive decision to cut the prism in half, and replace half of it with a simple block of UV fused silica. This would mean that when the prism was in the position for taking a photo, the light would go straight through to the camera, rather than being split between the camera and the eyepieces. This would result in more light for the camera (a good thing for UV imaging), and with most cameras having some form of live view, I don’t need to look down the eyepiece for the final focus. The window in the bottom part of the head also needed replacing, but this would just need a simple disk of UV fused silica. However both parts needed to be custom made and the prism would need cutting in half to make this all work. To supply the fused silica components and cut the prism down, I went to a company I’ve worked with quite a bit over the last year – UQG Optics – who were able to supply the parts I needed and do the work on the prism. When anything needs to be custom made the costs go up dramatically, and given the tolerances involved were tight, it was a case of ‘measure many times’ before the dimensions went to UQG for cutting.
When the two UV fused silica parts arrived, along with the cut down prism beam splitter, I assembled everything and glued the circular window into its metal housing, as you can see below.
Transmission of the new parts, compared with the old ones is given below.
Note the absolute transmission here is probably slightly higher than in real life (by a couple of percentage points). Even after collimation of the light beam for the transmission test, it will still be slightly divergent. Including such thick pieces of glass / fused silica in the path of the beam means that some more light will reach the spectrometer detector than would happen without the glass / fused silica in place. I have tried to minimise this with the setup of the device, but can’t guarantee that it has corrected it fully. The important thing to note is that the new fused silica components have a flat transmission curve as a function of wavelength, unlike the original glass ones, which block varying parts of the UV. The new UV fused silica parts will allow UVB (and even UVC) imaging down to and below 280nm.
Once the glue had set on these, the head needed re-assembling. Not a simple task, and akin to putting a quart into a pint pot. The picture below give you and idea of the parts that needed to go back into it.
Once back together (after a lot of loosening and tightening screws to get everything aligned, and cleaning of any glass after putting fingerprints on it) it looked like this.
The next step was to put the yellow filters in the base of the tubes for the eyepieces – the idea here being to eliminate any UV that could otherwise reach the eyes. This is part of my ‘belt and braces’ approach to safety when using this device. Not only would any UV have to pass through the glass part of the beam splitter, and the glass eyepieces, but now has to go through yellow filters which have a transmission of <0.01% of anything below 400nm. As mentioned before, I’d still use UV protective glasses when using this, the yellow filters just another layer of protection. Once it is up and running, I’ll be measuring the spectral throughput to the eyepieces to see how low it is. If in any doubt, I’ll just use live view on the camera for all the focusing work. Here’s a view down the tube of one of the eyepieces to the yellow filter.
One thing I haven’t got right was the diameter of the filter needed. The eyepiece tubes are 23.2mm, and I assumed 23mm filters would be fine. However at the bottom of the tube the diameter opens up and the filters rattle around a bit unless the eyepiece tubes are screwed all the way down. This can be rectified by using larger diameter filters, or by gluing the ones I have in place, and I’ll see if I can get access to where I need to to be able to do this. For now the tubes will be screwed fully down, locking the filters in place.
There was a lot of glass in the head, all aligned very precisely, and I had real concerns that this would simply not work well enough, leaving me with a very big bill for a new paperweight. Once assembled, for a quick test I did a visible light image of one of the Diatom Lab strewn diatom slides using an Olympus 20x SPLan objective, and the normal tungsten light source on the microscope. Stacked in Zerene Stacker, here is how it looks (I used a Schott S-8612 filter as this was taken with my multispectral camera and I wanted to remove IR from the image).
The image above has been reduced in resolution for sharing on the web, and is a crop from the full area imaged, but demonstrates that this modified head can at least give a usable image in visible light. Note, the fuzzy areas in the largest diatom are more down to my stacking technique than anything else.
In theory this is the final part of the UV microscope build – all the glass parts have now been converted to fused silica, and the next stage is to try imaging in the UV. To start with I’ll use 365nm as this is simpler and safer, and I have a couple of light sources (LED and mercury xenon lamp) to try out. I also have a Diatom 2.0 test slide on order from Diatom Lab, so I can look to see how the resolution of the imaging changes when I go from visible to UV. See, I told you 2021 would be exciting.
The original plan was to use this for sunscreen formulation imaging – to be able to image UVA and UVB absorbing components separately to see where they are in the formula – and I’ll certainly be using it for that first. After that we shall see, as there are plenty of other applications for UV imaging, in biology, forensics and other fields.
Thanks for reading, and if you’d like to know more about this or any other aspect of my work, I can be reached here.