FAQ: "So far, we have always talked about 3 sources of light (incandescent, natural daylight, and fluorescent), for which several CIE illuminants have been developed.
The xenon flashes used by HL’s spectrophotometers are supposed to duplicate, more or less, the CIE D65 illuminant.
How do LED sources fit in this schema? This is the question now, with the new spectrophotometers using LED sources."
For the visible spectrophotometers used to measure color today, CIE illuminants and the white light lamps in our instruments are not the same and have different functions.
CIE illuminants are fixed tables of values representing different types of while light under which a sample might be viewed. Illuminants are used to modify color values such as L*, a*, b* to quantify what the sample color would measure if viewed under different types of white light (ex: Illuminant A representing tungsten or home light at 2865 K; Illuminant F02 representing cool white fluorescent of Office Light; Illuminant D65 representing noon daylight). This calculation is done in software by selecting different illuminants in the Color Data View, and is independent of the lamp in the instrument.
The purpose of a spectrophotometer white lamp source in a spectrophotometer used for color measurement is to illuminate the sample with even power across the visible spectrum. In combination with a silicon-based detector, it is use to measure the relative %reflectance or %transmission of the sample.
In theory, an instrument lamp with a perfect white light would illuminate the sample with 100 units of power evenly across the visible spectrum. However, in practice, the white spectral output from an instrument lamp is not perfectly even across the spectrum.
In the days of instruments with a warm white tungsten halogen lamp, the output would be trimmed with a reddish IR filter to reduce the power in the red part of the spectrum (and IR to minimize sample heating) to provide a more even spectral power distribution. Typically, the trim target to match to has been the CIE D65 daylight illuminant. While D65 is an illuminant, the spectral power distribution is even across the visible spectrum and it is a recognized spectral standard. Tungsten halogen lamps are always on (energy inefficient) and must be changed every 1000 hours, adding a production expense.
With pulsed xenon lamps, there are some spectral peaks in the output and excess UV content that must be trimmed to produce an even spectral output. It is a better match to D65 than filtered tungsten halogen. The lamps are good for more than a million flashes which is typically beyond the life of the instrument. The lamp may never need changing. Highly efficient, pulsed xenon lamps only use energy when the lamp flashes.
White LED lamps hold the greatest future promise as a visible spectrophotometer lamp source. The spectral output of a white LED is not perfectly even across the visible spectrum but can be augmented or tuned with the addition of other colored LEDs to provide a very even output across the visible spectrum.
Effectively, the LED spectrum can be custom designed to match any white light output in the visible spectrum but D65 spectral power distribution still remains the target for an even power output. LEDs will probably be the lamp source that will best match it.
White LEDs should last for the life of the instrument and are an inexpensive light source. The key element is good design in the choice and mixing of LEDs in a colorimetric spectrophotometer to achieve a uniform, even spectral power output across the visible spectrum, and the ability to keep it there over the life of the instrument.
FAQ: "What about tristimulus colorimeters - isn't the lamp/detector combination supposed to match a CIE illuminant?
In color measurement, for every rule, there is an exception. In the legacy days of tristimulus colorimeters, the spectral output of the tungsten halogen lamp, in combination with 4 filtered silicon detectors, was used to closely approximate CIE illuminant/observer C/2. This allowed for fast direct reading tristimulus (3 number) color measurement but did not provide spectral data, nor was selection of the illuminant/observer combination (fixed to C/2 for all measurements) available for calculation of color values.