Is there any way, shape, or form that ultraviolet light is recorded by a digital sensor? Ideally, this question is related strictly to the sensor, without even a glass lens in between which could cut or block UV light. But of course I don't take pictures with a bare sensor, I also have a lens, so I'd also like to know how much UV gets to the sensor to begin with.
Yes, digital sensors are indeed sensitive to UV light, as well as a considerable amount of the infrared spectrum. Most digital sensors are equipped with multi-coated, multi-layered filters that are designed to filter out the extended ranges of UV and IR. Generally speaking, filtered digital sensors are sensitive to a much broader range of light than the human eye, from about 250nm (the near-UV range) through visible light (400nm to 750nm), and down about 780nm (the IR range). Unfiltered, a digital sensor is sensitive to a far greater range, from deep UV (200nm, true UV) down to true IR (as far as 900nm) [#1]. It should be noted that sensitivity is not constant throughout this range, and falloff is fairly rapid and becomes significant the farther away from 380nm you go. Same goes for the IR range. Human eyesight ranges on average from about 390nm through 700nm, while some people are more sensitive and able to see from about 380nm through 750nm.
Despite the filtration applied to digital sensors, UV light is still a problem, and can affect color balance. In general, the ability to sense UV light is not a huge problem, as digital sensors have relatively weak sensitivity to blue, and the UV sensitivity is generally captured as blue. However, without proper filtration, UV dispersion can generate disruptive haze that can be captured by a digital sensor, which may result in a rather undesirable result.
It should be noted that optical glass filters out a considerable amount of UV light. Most UV wavelengths up to around 310nm are blocked by the glass of a camera lens, and the remainder from 310nm up through 380nm can be blocked with a UV/Haze filter. If one wishes to create images in the UV light range, special lenses are available. Non-standard materials such as quartz or calcium fluoride have a greater transparency to the UV spectrum. From a camera imaging perspective, most research shows the most interesting UV wavelengths probably lie between 250nm and 310nm [#2]. To get a clear UV shot, you may need to remove the UV filter that covers the sensor itself. This is similar to removing the IR filter when modifying a camera for IR work, or may involve removing the entire filter apparatus, which would remove both UV and IR filters at the same time (depends on the camera.)
- Infrared and ultraviolet imaging with a CMOS sensor having layered photodioedes
- Introduction discusses unfiltered layered CMOS sensitivity range: 200nm - 1100nm
- Layered CMOS (i.e. Foveon) tend to have a greater sensitivity range than bayer CMOS
- Interesting discussion about the individual wavelength sensitivity of each color photosite (graphs included)
- Seems a little out of date (2003/2004 period?), but still useful
- Digital Reflected-Ultraviolet Imaging
- Older article from several years ago, covers reflected-UV imaging
- Discussed the nature of UV imaging and how it differs from visual/IR imaging
- The Wratten 18A: A problematic filter for reflected-UV photography
- Interesting article that uses an original Canon Rebel and a Wratten 18A filter to image UV
- The Wratten 18A allows UV from ~290nm through 400nm
- The older Canon Rebel CMOS sensor seems to image this wavelength range well
- Visible Light CMOS Sensors
- Page 7 has a graph of CMOS vs. Human Eye sensitivity
- Stops at 400nm, but shows that the CMOS sensitivity curve is still quite high at that point, and falls off at a moderate curvature (likely ends around 250nm-290nm)