2008-01-28

2.2 Megalux-hours and Counting

The twenty samples that inaugurated the Aardenburg DIY light fade program just a few weeks ago (see the January 1, 2008 news item) have passed the 2 Megalux-hour exposure level in light fade testing. The first fading measurements will take place at 10 Megalux-hours in test. Megalux-hours is a technical term to describe the dose of visible light energy to which the samples have been exposed over time. Megalux-hours can be easily translated into real world extrapolations of time and light intensity more familiar to photographers. It is especially easy to relate Megalux-hours of exposure to two well-known industry assumptions about indoor light exposure. Wilhelm Imaging Research (WIR) predicts display life (i.e., the time to reach easily noticeable fading) based on an estimated 450 lux illumination level for 12 hours per day, whereas Eastman Kodak uses 120 lux for 12 hours per day. Perhaps it is by chance, but a simple relationship holds between Megalux-hours and these two industry assumptions. Ten Megalux-hours translates into approximately 5 WIR years on display. It follows then that 20 Mlux-hr equals 10 WIR display years, 30 Mlux-hr equals 15 years, etc. Therefore, to estimate accrued WIR years on display divide the Megalux-hour value by 2. For the Kodak condition, 10 Mlux-hours translates to 19 Kodak years on display, but just a slight adjustment of 120 lux to 114 lux or the 12 hour time duration to 11.4 hours of exposure time per day allows us to round the kodak prediction to 20 years for 10 Mlux-hrs exposure, 40 years for 20 Mlux-hrs, etc. Thus, to estimate accrued Kodak years on display multiply the Megalux-hour value by 2!

The first 20 samples to go into the DIY light fade test program are all made on pigment-based inkjet printers. Judging by current industry claims about the light fastness of pigment-based ink sets, all 20 samples should be very resistant to light fading. We should expect to see very little if any change as they round the 10 Mlux-hr or "5 WIR years on display" mark. Yet the measurements may very well detect some small changes, and this possibility is why it is useful to start tracking changes early in the "life of a photograph".

Megalux-hours is a useful way to define accumulated light exposure for photographers because lux meters are easily obtainable, and lux levels can also be estimated with reasonable success simply by using a camera's light meter. However, measuring lux levels over an integrated time of exposure only tracks the total incident energy in an indirect way. Incident energy outside the visible region also causes fading and other degradation to artwork. Infrared energy, for example, is the source of solar heating for objects on display, and increased heat combined with worsening air quality and/or increased relative humidity generally increases deterioration rates. It is also well known that probable light induced damage increases with shorter wavelengths of radiation. Thus, the majority of light induced fading in paintings and photographs is due to absorption of invisible UV wavelengths as well as blue wavelength energy. Because human vision is characterized by a photopic response curve that ranges from wavelengths of 380 nanometers to about 760 nanometers with peak efficiency at 550 nm, this weighting function for the human visual system ignores the UV portion of the spectrum and differs from the way that artwork "sees" (i.e., absorbs) the blue wavelength radiation. Absorption of UV and blue wavelength radiations also varies colorant by colorant, and of course, absorption is the first step to light induced degradation. That said, if one knows the spectral energy distribution of the light source then tracking the visible energy portion of the spectrum is perfectly valid and can be used to accurately estimate other energy bands for that spectrum.

An accurate spectral power distribution (SPD) of an illumination source is measured with a spectroradiometer. However, a less expensive and practical way to characterize light sources is by using a radiometer in conjunction with detectors having selective energy response curves. Aardenburg Imaging and Archives currently uses this latter approach and has just acquired a UVA energy detector (320-400nm) to use in conjunction with the photopic detector (human visual response). Hopefully, later this year a blue light detector (400-500nm) will also be acquired to help characterize the visible part of a light source most responsible for light induced damage. Also to be purchased is a UVB detector (280-320nm) which is useful to measure UVB energy emitted by artificial light sources such as fluorescent lamps. However, ordinary window glass or acrylic picture frame glazing effectively blocks UVB energy in indoor environments, so for indoor display of prints and photographs UVB energy is frequently not an issue.

The Solar light Company PMA 2100 Photometer/Radiometer with Photopic and UVA detector configuration.  Numerous detector options are available, and surprisingly, a temperature and relative humidity detector is also available for use with the PMA 2100.

Estimating the probable relative damage caused by any given light source involves more than just characterizing its spectral power distribution. It also involves the interaction of the source output with the absorption properties of all surface areas on the artwork as well as the spectral attenuating properties of surrounding walls, carpets, and other materials that reflect the source radiation towards the artwork. Error in a light fading prediction occurs when the spectral energy distribution incident at the sample plane in the accelerated light fading test is not reasonably matched by real world conditions. However, assumptions about similarity of the spectral distribution at the sample plane in the test fixture to the sample plane in the real world environment are no more perilous and perhaps less so to the predicted outcome than the basic uncertainty caused by real world variations in average daily light intensity and duration of exposure. The challenge in relating accelerated light fastness tests to real world behavior is thus twofold. We must assume the response of the artwork is to a well characterized incident spectral energy curve, and we must assume we have modeled the average daily dose (light intensity multiplied by exposure time) as it relates to the actual real world display conditions. Large discrepancies in the former factor typically cause a 2 to 5-fold error in the outcome whereas errors in the latter factor may cause more than two orders of magnitude difference in outcome from one indoor location to the next. Given this reality, light fade tests conducted under more than one spectral distribution and calculations made with user-informed choices about the actual light levels in the intended display environment seem eminently appropriate as no one choice leads to a truly universal result. Providing test results in Megalux-hour format also reinforces the understanding that light levels and hours on display are extremely large variables rather than fixed quantities in real life.

The first AaI light fading test unit is equipped with a full spectrum fluorescent lamp having correlated color temperature of 5000 degrees K and a color rendering index (CRI) of 92. The choice of full spectrum fluorescent boosts the critical blue portion of the energy output compared to typical cool white fluorescent lamps or tungsten halogen lamps, but it is still low in UVA content compared to natural daylight entering through a nearby window. Nevertheless, the unit's incident energy at the sample plane reasonably simulates the real world condition where a print is located near said window but housed in a picture frame with standard acrylic glazing. Acrylic glazing is becoming more commonplace as print sizes increase with today's large format printers due to considerations of framing weight and impact resistance. Standard acrylic glazing not advertised as a UV blocking acrylic typically filters out much more UVA content than glass (a nominal 10-15% UVA transmission compared to 85% for ordinary glass). The high 92 CRI and 5000K correlated color temperature of the selected Philips Colortone 48 inch F40T12/C50 fluorescent lamps meet ISO specification for D50 lighting and as noted above bring the blue wavelength energy more in line with natural daylight compared to cool white fluorescent lamps. The mirror box assembly described in an earlier post (see the news item for December 20, 2007) also lends itself to other lamp configurations as well that can better simulate other common viewing conditions such as natural daylight with glass picture frame glazing, unglazed artwork under natural daylight, quartz halogen gallery lighting, etc.