A comparison of near-UV continuous light sources
Photography in the near-UV range requires the use of a light source of these wavelengths. UV-A (320-400 nm) is the band closest to visible light (400-700 nm). However, the range of wavelengths of interest for near-UV photography with a consumer-level digital camera (either unmodified or with its anti-aliasing filter substituted with a quartz window) is approximately 370-400 nm. Shorter wavelengths are absorbed more easily by optical materials, and the sensor is also less sensitive to them.
UV light is dangerous, especially the shorter wavelengths (UV-B and UV-C). The following discussion contains advice on how to reduce or eliminate unnecessary risk. If in doubt, or if you already suffer from medical conditions that can be worsened by UV exposure (e.g., cataracts, light sensitivity, skin cancer and pre-cancerous alterations) you should not use UV light sources altogether. If you decide to proceed, by all means do seek additional advice on UV safety from other sources.
I cannot warn you enough: UV-C is especially dangerous. At a short range, it easily kills living cells in seconds and causes extensive DNA damage. We have no natural defences against UV-C, because this component of sunlight is quickly absorbed by oxygen in the atmosphere, and you would be exposed to it only in outer space or at extreme altitudes. You should avoid any exposure of the skin and eyes to UV-C.
Note: There are no warning signs during and immediately after overexposure to UV, so there is no way to become aware of UV exposure until it is too late to protect oneself. UV-C is known to cause burns of the cornea. Protective goggles that form a tight seal against the face all around, like the ones discussed below, should always be used in the presence of UV-C. Several literature sources specifically mention that goggles with open spaces between frame and skin are not safe, and should not be used. Some sources mention that UV-C does not penetrate the layer of dead cells on the outside of human skin. However, (1) this is a matter of range and intensity, and intense UV-C is capable of killing or damaging cells under the protective layer of dead skin, and (2) most UV-C sources also produce abundant UV-B and UV-A, which are both capable of causing damage to deeper skin layers. Therefore, be on the safe side and do not expose either your eyes or your skin to any source of UV-C radiation.
UV-A is the type of UV produced by sun-tanning lamps, but also this can be damaging. If you can avoid intense or long-term exposure to UV-A, I recommend that you do so.
When working with a UV lamp, especially if strong and/or emitting wavelengths at or below 365 nm, you should always be wearing long sleeves, gloves and protective goggles designed to block UV (the ones above are a type I selected for this use). Many types of plastic block UV, but some don't, so look for goggles designed for this purpose. Some time ago, tests of sunglasses did show that some cheap plastic sunglasses don't block UV at all. Don't skimp on protection. The goggles should have a soft edge that sits tightly on your face all around the lenses, and seals off ambient UV. Ordinary sunglass frames allow plenty of UV to pass between the frame and your face. The orange frame of the goggles shown above is made of a soft plastic that fluoresces when exposed to UV. This is a useful indicator of the presence of UV in a work environment, but fluorescence is not immediately obvious if a strong visible illumination is also used.
Polycarbonate plastics (also known as Lexan and other trade names) are optically clear in the visible range, but strongly absorb radiation above the visible range, with a very steep cutoff curve. Thus, polycarbonate is an excellent material for UV protective goggles and face masks. Other types of plastic (e.g., Teflon, PFA, PTFE, cellulose derivatives), instead, may be transparent down to 220 nm, and therefore offer no protection against UV. Dyes dispersed within a plastic substrate can turn it UV-opaque, even without affecting its visible colour.
I am not sure whether it is just my impression, but operating some UV-C sources, even briefly, seems to produce a detectable smell of ozone. Thus, a good ventilation might also be necessary for safe operation. You are hereby warned - I really should not need to say this, but you are wholly responsible for any damage to yourself or others deriving from using UV sources.
Alternative UV sources
The sun is an excellent source of near-UV, especially when high above the horizon. However, since I reside in sunny Sweden, sunlight is a rare commodity for most of the year (even more so than in sunny England). In addition, near-UV photography in the close-up and macro range is most comfortable when performed in a studio or laboratory setting. Therefore, I decided to test a few easily available artificial sources of near-UV light.
On this page, I try three types of fluorescent bulbs and a HID (High Intensity Discharge) lamp. Automotive HIDs are quite cheap, and their electonics can be powered for a 12 VDC source. There are several different types of HIDs, but those most commonly used in car headlights are of a metal-halide type. All HIDs are known to emit UV light and are typically encased in a fused silica or fused alumina envelope that transmits at least UV-A and UV-B. Automotive HIDs are usually encased in a second glass sleeve that blocks most UV (although often it transmits at least a part of UV-A). Some automotive HIDs are marketed in ways that suggest they use xenon-arc tubes, but I have been informed that it is unlikely they actually use this type of lamp.
I also tested UV LED sources, but the results were a bit too different to be directly compared with other light sources. UV LEDs are essentially monochromatic, which, depending on the subject, may be a good or a bad thing. The LEDs I tested have a power up to 3 W, which is enough to allow exposures of 1 second or more when multiple LEDs are used. I tested here a self-built lamp of this type. It is acceptable, but cheaper, brighter LEDs with shorter wavelengths are necessary before this type of light source will be competitive in cost and performance with other types.
I happened to have available also an old mercury-arc light source for a Zeiss fluorescence microscope. However, a quick web search was enough for me to exclude its use. Among the problems frequently mentioned, two are very serious. The light bulb is prone to exploding, which apparently it always does at the end of its useful life, or whenever overheated. The casing of the light source is a thick aluminium casting with cooling fins, which looks strong enough to contain a small explosion. However, it also contains a glass mirror, a condenser and a probably expensive and scarce battery of UV-pass filters. These items may also be shattered by an explosion. The second problem is that the power supply of this light source is known to generate a strong EMP (electromagnetic pulse) and a spike of several thousand Volts when switching on the bulb. There are reports of TV and computer monitors located metres away from the power supply being permanently damaged by the EMP. Naturally, I would not let this equipment anywhere near delicate electronics like digital cameras and modern computers. In case you wonder, you cannot place the power supply at a safe distance and use longer cables to power the lamp: it is exactly the cables that work like antennas and emit the EMP. The use of very short cables reduces the problem, but does not eliminate it. Modern UV sources for fluorescent microscopes are much safer to use. However, their cost is out of proportion to alternative UV sources.
None of the UV light sources discussed on this page are designed with photography in mind. This is what makes them all rather cheap. I also tried an unmodified Nikon SB-800 flash in the same setup. Even when placed 3 cm from the subject, it provided far less near-UV, at the same lens aperture, than the other sources described on this page. Nonetheless, the short exposure time may make this source useful with non-static subjects that can be approached at a close range. Damage of the subject by overheating, however, is a distinct risk.
Using fluorescent tubes - general advice
Long, straight fluorescent tubes are not convenient UV sources for close-up and macro photography. The whole tube surface emits UV in all directions, and it is difficult or impossible to concentrate this emission onto a small surface.
Compact fluorescent tubes are folded into two or more parallel segments (or a spiral), and are much more adequate (but still not a point-source of light). Maximum intensity of illumination is in a plane perpendicular to the long axes of the tube segment. In most cases, this means, in practice, that a lamp with straight segments should illuminate the subject from the side of the lamp. "Pointing" the tube segments toward the subject would achieve a much lower illumination (unless a parabolic reflector is used around the lamp).
A simple, curved aluminium reflector (even the mirror side of an aluminium kitchen foil) placed a few cm behind the tube can almost double the intensity of the illumination toward the subject. A parabolic reflector can concentrate light onto a relatively small area and further increase its intensity. A reflector can also act as a lamp shade and avoid unwanted exposure of radiation to your body (albeit reflected radiation will always be present).
Fingerprints strongly absorb UV. By touching the surface of a UV fluorescent tube (or any other UV light source or UV-pass filter) you are likely to create UV-dark spots and to reduce its efficiency. You can remove skin oils with a very diluted solution of dishwater detergent in water. Ethyl- or propylalcohol can also be used, but not the mixtures of alcohol and oils used for skin rubbing (they are likely to make things much worse). Alcohols and solvents may damage the external coating of fluorescent tubes and some types of plastics.
Short-wave water-sterilizing tube
This lamp is a short-wave fluorescent tube of the type used for sterilizing pond water. This type of lamp is, essentially, a fluorescent tube made of quartz (to transmit UV) and devoid of internal phosphor coating (which in normal fluorescent tubes converts UV into visible light, and filters out any remaining UV). This lamp provides relatively high amounts of UV-A, UV-B and UV-C.
The tube I chose can be mounted and operated in an ordinary table-top lamp. A lamp shade made of thick cardboard or plastic can be added to restrict light to the subject area, thus reducing the risk of indirect exposure to your skin. This tube comes with a warning saying that it is meant to be used inside a completely sealed water filter, with no radiation allowed to leak. In addition to UV-cut laboratory goggles, I would recommend to wear a cotton face mask, long sleeves and cotton gloves when using this lamp for an extended time.
If extensive use as a UV-A source is planned, I would investigate whether it is possible to use a glass filter on the lamp to eliminate UV-C and (unless necessary for photography) most UV-B. This would substantially reduce any health risk. In this context, Pyrex is reported to absorb UV-C, and could be a cheap filter for this purpose. It is typically transparent to UV-B and longer wavelengths, unless pigments have been added.
In the visible range, this lamp produces a moderate amount of bluish/purplish light.
This is a type of fluorescent lamp meant to be used in reptile terraria. Some reptiles need moderate amounts of UV-B to remain healthy, and these lamps are designed for this purpose (see this site for an extensive discussion, as well as general information on UV and artificial UV sources). They produce also reasonable amounts of UV-A. The above specimen is new, but only a small portion of the tube lights up, near the base of the tube. This does not improve even when the tube is left on for an extended time. According to descriptions on the web, this is normal and does not indicate a defective lamp. This type of lamp has a built-in electronic starter, and needs only an ordinary bulb socket.
In the visible range, this lamp produces a rather dim white light. It does not come with safety warnings. I assume that the amount of UV-B it produces is very limited. It produces insufficient amounts of UV at 325 nm to be used for framing and focusing, unless the subject is very reflective. It is slightly better at 340 nm, and usable at 365 nm.
This type of lamp is designed for small dance halls and parties, and is a moderately strong source of UV-A only. It comes with no notice of health hazards, and I assume that exposure to its light is less dangerous than sunlight. It also has a dark blue/purple coating that cuts visible light. As a result, it emits a very small amount of violet light, but stimulates a bright blue UV-induced fluorescence of white paper and other sensitive materials up to a few metres away. Most of its emission seems to be in the 380-390 nm band.
Also this type of lamp fits in a normal bulb socket. The dark coating is on the external side of the tube, and appears to be delicate and easily scratched.
HID lamps are often sold as a kit for fitting in car headlights. This kit consists of a small bulb, a metal box containing the electronics and a set of cables and connectors. It is powered by 12 V DC. I use a switching power supply to run it from mains. The electronics generate a pulsating high voltage when powering up the lamp, which emits a buzzing sound for up to about one minute. The lamp lights up gradually, and is ready to use once the noise has mostly disappeared. It does not generate a strong EMP, and seems safe to use near electronic devices (after all, it is designed for use in close proximity with automotive electronics). Use only the cables and connectors included with the kit, and you should be safe. You need a type of HID designed to replace a headlight halogen lamp with a single filament, which is much cheaper than the types intended to replace a two-filaments lamp in normal headlights (the latter HIDs usually have a servo-mechanism in the base of the lamp to shift the position of the bulb).
There are several types of lamps for this use, and most of them are rated by their nominal light temperature. 4,700 K to 5,000 K are frequently used, and produce a white or slightly yellowish light. 6,000 K gives a so-called daylight. 8,000 K lamps are used when a light with a definite blue cast is desired. Higher temperatures are not normally used, but I was able to locate kits containing lamps rated at 11,000 K and 16,000 K. The 11,000 K lamps produce a bluish-indigo light (rather unpleasant, in fact) that contains quite a bit of UV-A. The 16,000 K lamps in my possession, instead, produce a whiter light with a hint of yellow (this does not make much sense in terms of color temperature, but nonetheless I can confirm my observations) and slightly more UV-A.
I have seen automotive HIDs advertised as 30,000 K, but I think this is only a fantasy specification. Anything in this range probably would produce soft X-rays and UV-C, rather than visible light. These lamps emit a strongly blue or purple light, but I don't know how much UV they emit. Blue light corresponds to a color temperature of 8,000-10,000 K, while the extreme purple of twilight is around 12,000 K.
The subject is a stray flower growing in my garden in the wrong season. This species normally displays a strong UV pattern, with a dark central area of the flower. However, perhaps because of the late season (mid-November with snow and below-freezing temperatures), this particular flower has a less evident pattern. It also insisted in slowly turning away from the camera, which resulted in the test pictures being taken at different angles.
For this test, I used a UV Rodagon 60 mm lens and a Schuler UV-pass filter. This filter transmits a little also in the indigo, so pictures taken with it, at least in principle, are not monochromatic but may contain a little colour information. Filters with a narrow banpass in the near-UV only, like the Baader UV, produce instead strictly monochromatic images (except for a small transmission peak in the near-IR, which virtually all UV-pass filters seem to suffer from in various degrees). The camera is a modified Nikon D70s with a quartz window in place of the antialiasing and IR filter.
Exposure times were around 15-30 seconds and the aperture used ranged between f/8 and f/11. The results were not processed, except for an increase in gamma and reduction in the red channel in the third picture (black light tube).
The HID lamp produces the best results in my judgement. It also allows the shortest exposure and the aperture to be closed down, which results in a sharper picture. The uncoated sterilizing tube comes second, but requires the longest exposure. Apparently, it emits mostly UV-C, which the camera cannot record. The two coated fluorescent tubes produce low contrast, probably as a result of abundant near-IR emissions that pass through the Schuler filter. The black light tube is better than the ReptiGlo in exposing the UV pattern of the flower. However, it also displays the type of central flare that is often caused by near-IR in lenses not designed for this type of ligh. It is quite possible that the UV-pass filter external coating of the tube is also transparent to near-IR. The UV Rodagon is definitely not designed for IR (in fact, not even for visible red), and the modified camera used for this test is very sensitive to near-IR. The HID does not produce an IR flare - apparently it emits only little in the red and near-IR, although its deep-IR emission is likely strong.
To verify that light from the HID contains little or no near-IR, I compared pictures taken with the HID and with a normal incandescent lamp, without using a filter. This makes the camera sensitive to near-UV, visible and near-IR (although, in practice, only a little sensitive to near-UV and very sensitive to near-IR). The difference is very remarkable (above pictures). Note that the two pictures were taken with the same preset white-balance. The HID picture (taken at 1/30 s) shows no detectable red and near-IR information, which is otherwise very evident (as red and purple) with the incandescent lamp (for this reason, exposure time with the latter had to be shortened to 1/125 s). Because of the short exposure times, there is practically no UV information in these pictures. Subsequently, I performed a more rigorous test by placing an IR-pass filter in front of the lens and re-taking the pictures. In this case, the HID produced an almost completely black picture, while the incandescent lamp produced a picture quite similar to the one shown above. Pictures taken with an incandescent lamp and a variety of filters filters are shown here.
Except for the black light tube, all tested sources produce evidently dichromatic pictures, with very distinct red and indigo areas. This may be undesirable if you intend to record only near-UV radiation but, with most subjects, I do prefer to have this additional colour information. The black light tube differs from the other light sources tested here in producing very little visible light.
As discussed here, special care should be taken to eliminate other illumination sources that may contaminate the picture with near-IR. All UV-pass filters (even the best ones) have a big or small transparency peak in the near-IR, to which digital cameras are very sensitive. I found at my own expense, for instance, that an incandescent 40 W lamp that I was using just to be able to see what I was doing (placed much farther away from the subject than my UV lamp) was contributing enough near-IR to substantially reduce contrast (above pictures).
A HID lamp is especially suitable for macro photography and photomicrography because it is essentially a point source of light. In my case, the lamp itself is a 4 mm sphere with two tungsten electrodes, enclosed in a slightly wider quartz tube for safety. Fluorescent tubes are much larger, and difficult to place close to a subject. The main drawback of a HID is that it produces also abundant visible light and heat, albeit less heat than an incandescent lamp. The visible light is so intense as to cause pain and temporary blindness if viewed at a close range. For this reason, you must place a HID lamp in a suitable shade (don't forget to allow for suitable ventilation). For the lamp, so far I have used improvised shades, depending on the need. The cables between the HID bulb and electronics are short, but not prohibitively so.
Compared to HIDs, the power UV LEDs I tested so far produce lower amounts of UV radiation and are limited to a narrow bandwidth of around 10-15 nm. Low-cost power LEDs are so far unavailable at shorter wavelengths than 365 nm. HIDs cost less, produce more UV and emit a broader UV spectrum and shorter UV wavelengths. Their main drawback is that they also emit large amounts of visible light and modest amounts of near-IR.
HID bulbs for automotive lights are permanently encased in a second glass sheath, which protects the bulbs from thermal shocks and water leaks but may absorb some UV (intentionally and as a safety measure). This protective sheath is permanently sealed around the HID bulb, but it might be possible to remove by breaking it off. I attempted to do so by applying a moderate pressure with a bench vise but did not succeed. A higher force or an impact are likely to burst also the internal HID bulb, so I did not make further attempts.
None of the tested UV sources provide high levels of UV that allow short exposure times. HIDs are useful for framing and focusing in live view with UV-pass filters mounted on the lens, but a studio strobe with uncoated tube, or even a battery-operated electronic flash with uncoated tube, are far better UV sources for the actual exposure (albeit, naturally strobes are not a continuous source).
High-K HID lamps are convenient continuous sources of near-UV, and superior to UV LEDs in several respects. However, their principal usefulness is for framing and focusing in live view, not as UV sources during the actual exposure. Fluorescent tubes of various types emit lower amounts of near-UV, but some types produce shorter wavelengths than HIDs.