Color by the Numbers, Part 1: Spectrophotometry vs. Colorimetry

In our lightfastness testing of metal-complex solvent dye solutions, we see some samples that fade a lot and are very easy to qualify as “fast-fading” dyes. (See our previous post for more information on metal-complex solvent dyes and why we’re using them; as well as previous posts that detail our sample set and sample preparation) By visually comparing the test samples to the controls we can see those differences, even after as little as 24 hours in our exposure test-chamber. However, we need to quantify those changes in order to describe them in a more objective way than can be done based on human sight alone. Doing this requires some means for reproducibly recording color and calculating change. But how do we do that?

Answer: we use an instrument called a spectrophotometer to make color measurements, and then use principles of colorimetry to interpret them.

Supplier/Color/Solvent/Substrate: Kremer/Blue GN/acetone

Sample of Kremer Blue GN (CIGN Solvent Blue 67) in acetone, after testing. The A and B test plates (left and middle) were exposed to the accelerated aging chamber, while test plate C (right) was not. There is very little color fade detectable by the naked eye. AMNH/F. Ritchie

Supplier/Color/Solvent/Substrate: Kremer/Yellow 4GN/acetone

Sample of Kremer Yellow 4GN (CIGN Solvent Yellow 146) in acetone, after testing. The A and B test plates (left and middle) were exposed to the accelerated aging chamber, while test plate C (right) was not. The light fading was extreme enough to be detectable by the naked eye. AMNH/F. Ritchie

Background: A Note on the Difference Between Color and Colorants

Before describing the workings of a spectrophotometer and the principles of colorimetry, a refresher on the two different types of color theories – the way humans see color – is necessary.

1.  Subtractive color theory

subtractive color

An illustration of subtractive color theory where colorants (in this example, the pigment from oil pastels) are mixed and absorb light from certain regions of the visible spectrum, thereby subtracting them from the reflectance spectrum that is perceived by the human eye. [Photo source]

This is the type of color-mixing we are taught in grade school: whereby colorants are mixed and then absorb light from certain regions of the visible spectrum, thereby subtracting them from the reflectance spectrum that is perceived by the human eye. Under this theory the primary colors are red, blue, and yellow, and secondary colors are orange, green, and violet. For example, a blue colorant (which absorbs light in the red, orange, and yellow parts of the spectrum) mixed with a yellow colorant (which absorbs light in the red, blue, and violet parts of the spectrum) will produce green (the remaining color in the reflectance spectrum). In modern color printing, cyan, magenta, and yellow are the usual primaries. Black is produced when all colors are present because all wavelengths are absorbed.

2.  Additive color theory


An illustration of additive color theory where different light sources are shown onto a wall. Where the lights overlap, another color is perceived because visible wavelengths in the spectrum of light are mixed (or added) together, received by the eye, and interpreted by the brain as color. The white light in the center is the mixture of all three primary additive colors, mixing together to encompass the entire visible spectrum. [Photo source]

This is the type of color perception that we encounter on computer- and television monitors when projected light is combined to make different colors. In additive color theory, visible wavelengths in the spectrum of light are mixed and/or added together, received by the eye, and interpreted by the brain as color. The additive primary colors are red, green, and blue, roughly corresponding to cones in the eye that are sensitive to long, medium, and short wavelengths, respectively. Secondary colors are yellow, cyan, and magenta. For example, a mixture of equal parts red (~650 nm) and green (~510 nm) light stimulates the cones of the eye in a manner that is similar to pure spectral yellow (~570 nm), so the eye does not detect the difference between the two conditions. White light is created when red, green, and blue are all present because they encompass the entire visible spectrum.

Spectrophotometry: Making Color Measurements

Spectrophotometry is a “noninvasive technique that measures the amount of light reflected or transmitted by a material at individual wavelengths of the spectrum” (Johnston-Feller, 2001: p. 1). A spectrophotometer is the instrument that’s used to measure light reflected, or transmitted, by a material. To do this, it relies on a light source, a monochromator that isolates and directs the wavelengths, and a photodetector that detects reflected or transmitted radiation. The measurement is unit-less, or may be given in percent absorbance or transmittance. The reflectance or transmission spectrum of the material describes how a spectrum of light incident on it will be modified upon reflection from, or transmission through, it. That modified spectrum is what we observe and interpret as the material’s color under that particular light.

For this project, we purchased an X-Rite Ci62 handheld spectrophotometer with a tungsten light source, a grating monochromator, and a photodetector consisting of blue-enhanced silicon photodiodes. With this instrument we measure successive reflectance spectra from each sample at intervals during our light exposures. When a sample is measured, the amount of light reflected at each wavelength in the visible spectrum is recorded and plotted along a curve that describes the reflectance of that dye sample. By comparing successive measurements, we can see the spectral-curve change as the dye fades or discolors over time.

MSC solvent red 122 in ethyl acetate

Spectral curves of sample Solvent Red 122 during an accelerated aging test cycle. The initial color reading/spectral curve is in red (the lowest line). During each color reading, the curve flattened as the color faded and there was less to measure. The final measurement is represented by the top line (green).


Top: Spectrophotometer in the jig without quartz plate dye sample. Bottom: Quartz plate dye sample in position for color measurement when spectrophotometer. AMNH/F. Ritchie

Then, from each reflectance spectrum we can calculate parameters that uniquely describe the color we will observe under a particular light and the changes in that perceived color from our exposure test. That translation of light spectrum into color description is done with the tools of colorimetry, the scientific description of color.

Colorimetry: Interpreting Color

Colorimetry is a branch of color science that studies precisely what we need for our research: it is the quantification of color, or, as Ruth Johnston-Feller describes, a way to “describe a color unambiguously and uniquely to distinguish it from all other colors” (Ibid.; p. 15). Most colorimetric systems characterize color in three dimensions because they describe human visual perception, which is a function of the three types of photoreceptors in our eyes. A variety of these systems exist, but most of the descriptors for perceived color are derived from the CIE color system administered by the International Commission on Illumination since 1931, which is itself based on additive color mixture and the visual response to various wavelengths of light that reach our eyes.

CIE 1931 RGB and CIE 1931 XYZ Color Spaces

The 1931 RGB and 1931 XYZ color spaces created by the CIE were the first numerical models linking the wavelength regions of the visible spectrum to human color vision. The CIE system defines a series of standard illuminants (sources) and a standard observer (mathematical representations of average human color vision), since the perception of color requires both a light source to illuminate a material and the human visual response to the light reflected (or transmitted) by that subject. A standard illuminant is characterized in terms of the spectral power distribution that distinguishes a specific light source, including the one most widely used: 6500K daylight (abbreviated D65). A standard observer is defined by a set of three mathematical color-matching functions that describe the average spectral sensitivity of the human eye at each wavelength in the visible spectrum, as derived from experimentally determined response curves for a particular field of view (such as 10 degrees, a wide field suited for viewing rough textured surfaces like fur samples). These color-matching functions are dependent on the selection of a set of three additive primary colors (for example, monochromatic red, green, and blue wavelengths, as in CIE RGB, described below) that together with a reference white point (such as D65) define a unique color space containing the gamut of all possible combinations of those primary color components. The functions are used to convert the reflectance spectrum measured with the spectrophotometer into tristimulus values, which are the relative amounts of each primary color needed to match the color sample measured.

For example, the CIE RGB color space is defined by red (700 nm), green (546.1 nm), and blue (435.8 nm) primary colors, and contains the gamut of colors that can be produced from RGB tristimulus values. A mathematically related color space, CIE XYZ, is derived from CIE RGB by means of a simple conversion, but it is useful because it makes for easier calculations than the RGB space. [For further reading on CIE XYZ, check out other on-line resources.]

CIE 1976 (L*a*b*) Color Space

There have been many developments in color science since 1931, in attempts to describe perceived color in a more straightforward and intuitive way, leading to the development of Lab color space by Richard S. Hunter, in 1945, and later the CIE 1976 (L*a*b*) color space. CIE L*a*b* color space is a mathematical derivative of the CIE XYZ color space, in which the L* parameter captures lightness, and the a* and b* parameters capture the other color dimensions. L* is proportionally related to tristimulus Y, the luminance, with darkest black at 0, and brightest white at 100. Red/green colors are represented along the a* axis, with neutral gray at 0, green at the negative values, and red at positive values. The yellow/blue colors are represented along the b* axis, with neutral gray at 0, blue at negative values, and yellow at positive values.

L*a*b* color is widely used in many industries because it offers a number of advantages over previous RGB and XYZ systems. It more closely approximates the nonlinear response of human vision, and includes a much wider color gamut than its predecessors. And it is more perceptually uniform, meaning that uniform changes in color values produce uniform changes in perceived color. Because of these capabilities, we are using L*a*b* values to describe the appearance of dye samples in our project.

Related to the L*a*b* color system is the color space CIE L*c*h°. In this system the a* and b* parameters are replaced by parameters describing chroma (or relative saturation) and hue. Unlike the L*a*b* system, which uses three-dimensional xyz coordinates to locate a point in color space, L*c*h° uses a polar coordinate system, with h° denoting a hue angle, and c* a radial distance from a central point. This color space is conceptually related to the familiar Munsell notation of color in terms of its hue, value, and chroma. It is also the color description necessary for calculating color differences with certain formulas.


Additive color theory illustration depicting how variations of color are achieved by the different combinations of red, green, and blue light (rgb values). [Photo source]

Putting It All Together

Like many spectrophotometers available today, the X-Rite Ci62 that we are using for this project has integrated software that can quickly execute the complex mathematical calculations needed to translate the measured reflectance spectrum into various colorimetric systems. From each reflectance spectrum, it computes and outputs tristimulus XYZ, L*a*b*, and L*c*h° values based on the standard illuminant (D65) and standard observer (10°) that we have selected. We then use those values to consider the overall amount of perceived color change, as well as changes in individual parameters, such as lightness.

We will go into greater depth about how we are interpreting our color measurements and tracking color changes in the up-coming Part 2 of this blog post.


Billmeyer, Jr., Fred W., and Max Saltzman. 1981. Principles of Color Technology. New York: John Wiley & Sons, Inc.

Christie, R.M. 2001. Colour Chemistry. UK: Royal Society of Chemistry.

Johnston-Feller, Ruth. 2001. Tools for Conservation: Color Science in the Examination of Museum Objects, Nondestructive Procedures. Los Angeles: The Getty Conservation Institute. Available online.

Whitmore, Paul M., Editor. 2002. Contributions to Conservation Science: A Collection of Robert Feller’s Published Studies on Artists’ Paints, Paper, and Varnishes. Pittsburgh, PA: Carnegie Mellon University Press.

Dyed Fur Samples: Part 2

Fur-Mounting System

We next needed a method of mounting the deer and fox furs so that they could be dyed and exposed in the test chamber. [See Part 1 of this series for an explanation on why we chose deer and fox furs.] Acquiring a series of meaningful color measurements from dyed fur demands a sample design that is more sophisticated than dye on quartz plate. The successful mounting system would secure a group of hairs at both ends, creating a flat sheet onto which the dye can be airbrushed. It must completely immobilize the hairs against handling, measurement, and the high rate of airflow inside the test chamber. Any loss or displacement of a dyed hair from the sample surface could be reflected as lost color in our measurements, indistinguishable from the dye fading that we are principally interested in.

In addition, the mounting system must:

  • Be compatible with deer and fox hairs of average length
  • Consistently distribute hairs across the sample to create a relatively planar surface
  • Avoid hair breakage during sample construction, accelerated aging, or measurement
  • Be made from materials that would not slump, melt, migrate, off-gas, or otherwise degrade during accelerated aging
  • Produce samples of consistent size and shape that can be used with a jig to ensure measurements are repeatable
  • Produce samples that can be handled without accidental disruption of hairs
  • Produce samples that can be secured inside the test chamber
  • Be simple and relatively quick to execute
Title/Description of Object:

Early attempts at mounting hair for research samples that were unsuccessful.  AMNH/J. Sybalsky

Our initial experiments cast a wide net. We looked at various types of clips, from binder and bulldog clips to staples and barrettes. These attachments were quick and simple to use, lacked components that would be likely to degrade, and could be adjusted to accommodate any fiber length. But we found that they produced samples that were bulky, insufficiently secure, incompatible with the measurement aperture of our spectrophotometer, or could not easily be standardized for use with a simple jig.

We explored other non-adhesive methods, employing flat materials that can be crimped or heat-sealed to trap hairs at both ends. We laid hairs on top of strips cut from nonwoven polyester and Mylar, folded the ends of each strip over the hairs, and sealed them in place. While these mounts created visually attractive, planar samples that could be made to fit into preexisting sample holders in our test chamber, the heat seal did not pinch the hairs well enough to fully immobilize them. A similar design using strips of aluminum sheet with crimped ends was more secure, with improved tightness and rigidity. They were, however, laborious to execute and would be challenging to standardize. A further adaptation using pieces of aluminum vapor-barrier tape in place of crimping made sample construction much easier, but introduced an organic adhesive. (See images above.)

Next, we looked into designs based on the concept of an embroidery hoop: we laid hairs across the open end of a cylindrical piece of steel hardware, then secured them in place using fasteners that could be wrapped and tightened around the cylinder. Stainless steel zip ties and hose clamps both held well in general, but fibers were loose around the locking mechanism in both cases. Execution was fussy, leaving fibers inconsistently distributed. (See images above.)

Fruitful discussions with several colleagues pointed us toward polyethylene sample holders designed for X-ray fluorescence (XRF) analysis of powdered samples. These holders usually consist of several separate components: a cylindrical tube or “cell,” a collar that snaps over the top of the cell, and a cap that covers the bottom end.

To mount the fur onto an XRF sample cup, a small piece is cut from the hide. The hair on the hide is brushed using an eyebrow comb to align the hairs parallel to one another. One end of the cylindrical cell is lined with Teflon sheet to act as a stable, white backing material during color measurements in case any gaps are left between hairs. The aligned hairs are carefully placed on top of the cell and immobilized by attaching the snap-on collar in a fashion similar to an embroidery hoop. Longer hairs extending from the side of the sample cup are trimmed using a scalpel. Any hairs standing proud of the sample surface and those not trapped by the sample ring are also trimmed away using scissors and tweezers.


XRF sample cup with Teflon sheet applied to the opening of the cell; the snap-on collar before applying.  AMNH/F. Ritchie


Applying combed fox hair to the Teflon-covered end of the XRF sample cup. The snap-on collar is added after positioning the hair.  AMNH/F. Ritchie


Completed fur cup samples  AMNH/F. Ritchie

We found this method of sample construction to be labor-intensive, but it allowed us to make standardized samples more successfully than any other. Nevertheless, before committing ourselves to one method over the others, a trial run in the test chamber was in order. We subjected our two favorite mounting systems, the XRF sample cups and the aluminum sheet and barrier tape, to 600 hours of accelerated aging at 0.35 W/m2, 55% RH, 63ºC black panel temperature, and 48ºC chamber air temperature. Both systems held up to the environment and blowers well. There was no evidence of creep or adhesive migration in the aluminum mounts. In both systems, a few underfur fibers were stirred up above the sample surface by airflow inside the chamber, demonstrating the importance of early removal of fibers inclined to dislodge. This can be accomplished through application of gentle friction or canned air to the sample followed by cutting or tweezing away any loose hairs.

Because of the importance of producing samples of consistent size and shape, we ultimately adopted the mounting method based on the XRF sample cup. A selection of XRF sample cups with different dimensions and features was subsequently tested with our furs. Some were more compatible with the length of our fibers than others, and smaller cups gave us greater control over the distribution of fibers. Among the examples we tested, we felt that the Chemplex SpectroCertified Quality XRF Sample Cup No 3115 worked best.

In total, 300 sample cups were constructed (150 with fox, 150 with deer) and will be used for testing. Future blog posts will describe the testing methods and results.

Caitlin making fur cup

Project intern Caitlin Richeson preparing fur cup samples using fox hair.  AMNH/F. Ritchie

Dyed Fur Samples: Part 1

The first phase of our lightfastness testing aimed to establish the lightfastness of the Orasol® and similar Sorasolve metal-complex solvent dyes in isolation‒ that is, in the absence of a binder, and without a chemically active substrate that could potentially influence the behavior of the dyes or interfere with the measurement of color change. (See earlier posts to explain the project plan and selected dyes. Future posts will explain the results of this testing, which is currently underway.)

The second phase of testing uses accelerated aging and periodic color measurements to look at how lightfastness is affected when Orasol dyes are applied to aged, faded furs as they are used in a recoloring treatment. We expect to see substrate impacts on lightfastness for several reasons:

  • As furs age, they produce reaction products that may affect the chemical behavior of an applied dye.
  • Compared to quartz, dyes deposit very differently onto hair. There is also significant variation among fresh and aged fiber surfaces, and among fibers from different animals.
  • The optical properties of dyed fibers differ from those of dye films on quartz. Differences in how the sample reflects and/or absorbs incident light affects its total light exposure dose and its apparent color.

Fur Selection

Several considerations played an important role in our selection of animal fur substrates. First, we sought furs that are light in color. The principle reason for this was that a light-colored substrate was needed to control our dye application and keep reflectance spectra minima in the range of 30–40 percent (see earlier post for reflectance discussion). For consistency of samples, the furs also needed to be as uniform in color as possible. To facilitate sample mounting, longer fibers were preferred over short. The fibers themselves should present minimal color change upon exposure to accelerated aging, ensuring that the dye (and not the fiber) is the primary contributor to any color change observed in a dyed fiber. Finally, we aimed to represent some of the naturally occurring variation in mammal furs, from hollow guard hairs and bristles to awns and underfur.

arctic fox skin

Arctic fox skin used for research samples.  AMNH/F. Ritchie

We ultimately chose furs from two animals with naturally white coats: an arctic fox and a white-tailed deer. The latter is fully white but not a true albino, a variation selectively bred to be whiter than the closely related “spotted” piebald. The arctic fox represents fine, smooth-haired fur-bearing mammals, while the deer offers hollow guard hairs.

However, there is an important downside to using white furs such as these. As we pointed out in our discussion of backing materials for our samples on quartz, light-colored substrates reflect and scatter proportionately more light than dark ones. Transparent dye films applied to highly light-scattering substrates will be exposed twice: once directly by the lamps, and a second time by reflected light from below, increasing the light-exposure dose. While faded historic taxidermy may be light in color, most examples are generally still darker than these bright white furs.

deer hide

White-tailed deer hide used for research samples.  AMNH/F. Ritchie

Consequently, light-scattering plays a larger role in the fading of our samples than is expected in the treatment that we are modeling. While it could be argued that our samples represent a worst-case scenario with respect to the impact of specimen color on the longevity of recoloring treatment, we acknowledge that the use of white furs is a compromise needed to consistently produce the most light-sensitive dye application possible.


The impact of different tanning methods on the dyes under investigation is unknown and offers an interesting avenue for further research, but this is not addressed as part of our current project. Nevertheless, in order that they more closely model taxidermy in the American Museum of Natural History collection, the fox and deer pelts for this project were tanned according to methods representative of those historically used at the Museum.


Scraps of leathers prepared using different techniques and finishes. The type of tan can affect the condition of a taxidermy mount overtime.  AMNH/F. Ritchie

When considering the production and acquisition of historical taxidermy at the Museum, particularly for use in dioramas, the period of interest spans from approximately 1925 to 1965. Though we do not have a complete understanding of all the tanning methods in use at, and for, the Museum during this 40-year time frame, we were able to partner with a local tanner trained by Sinclair Clark, a renowned tanner who was on staff at the Museum around 1924–1927. Clark later set up tanneries in other locations, but maintained his relationship with the Museum tannery over a long period of time.

In general terms, Clark’s method involves the following:

  • The skin is salted to remove moisture and stabilize it prior to tanning
  • Tanning begins with rehydration in a saltwater bath until the skin is soft and pliable
  • It is next soaked in an acid pickle until swollen, and then shaved down on a fleshing machine or by hand
  • The skin is returned to the pickle, and, if needed, shaved again
  • The skin is then removed from the pickle and the acid is neutralized
  • Warmed oil is applied either by hand or with a “kicking” machine
  • The skin is left to sit overnight or for one day before being tumbled in hardwood sawdust until dry and soft
deer hide cross section

Cross section of the white-tailed deer hide used for the research samples.  AMNH/F. Ritchie

The upcoming Part 2 post will describe how we are mounting the furs to run as research samples.

It’s All In the Preparation: Part 2

With our sample set chosen, our substrate selected, and our sample mounts determined, we next needed to devise a process for applying the dyes. Since the ultimate treatment would likely involve airbrushing, we chose to spray deposit the dyes onto the quartz plates. For this we had to consider and anticipate the different requirements and challenges that might arise in the application, including:

  1. What is the best technique for mixing the dye solutions?
  2. What delivery system gives us even coverage and the most control?
  3. How much dye should be applied and how can we measure that?
  4. What shape should the dye film deposit be to allow repeatable color measurements?
  5. How should we mask the substrate to achieve the desired sample shape?
  6. How do we economize and reuse the substrates?

1.  Mixing: Our mixing procedure reflects the method reported by Ciba-Geigy (the manufacturer of Orasol® dyes before its acquisition by BASF in 2008) in the literature describing their product testing. [Ciba_orasol_brochure] Where Ciba’s protocol is unclear, we have standardized our method based on our experience of what works well.

The powdered dye is measured and then incorporated into a volume of solvent to create a 1% (weight to volume) solution. The solution is stirred using a magnetic stirrer for approximately one hour. After mixing, the solution rests for approximately 24 hours. During this period, undissolved dye particles settle, and afterwards the solution can be decanted into a new glass bottle.

Sample Preparation

Former project intern Ersang Ma decanting newly mixed dye into the spray bottle. AMNH/B. Hunan

We have observed that when left in storage, the dye solutions appear to have a shelf life. In some cases we have seen changes in the color of the solution and the formation of crystalline deposits. We haven’t looked into what these changes might mean for the behavior of the solution. We have just taken them as indications that something has changed, and in response, we prepare fresh dye solutions for each round of sample production.

2.  Delivery System: We elected to apply the dye to the quartz plates using an airbrush. Airbrushing is, in general, the delivery method that permits the greatest range of expressive use of the dyes in taxidermy restoration, where controlling color value, gradation, and blending is critical to achieving a successful result. For the same reasons, it is easier to control the amount of dye deposited onto the plate using an airbrush than by brush, roller, dropper, or dipping techniques. The spray application facilitates putting down a thin, even layer, which can be gradually built up to the desired level.

during spraying

Former project intern Ersang Ma spraying dye onto a masked quartz plate. AMNH/F. Ritchie

Our dye solutions are applied to quartz plates using an Iwata Eclipse HP-BCS airbrush at 20 psi of air pressure from a compressor. Of course, having elected to use the airbrush, our samples are always prepared in a fume hood while the researcher is wearing appropriate personal protective equipment.

3.  Controlling Dye Application: It is important to control the dye application because the color produced by a medium-value application is inherently more light-sensitive than a heavy or light application of the same color. The sensitivity of the medium-value color derives from its reflectance spectrum having values that are not pinned at the low values (for very dark colors) or high values (for very light). It has also been found that for a given amount of dye loss, reflectance values around 30-40% show the largest increase. So when we spray our dye solutions on the quartz plates, we are attempting to produce samples whose reflectance spectra have minima that fall in that range. Doing so also avoids pinning large parts of the spectrum at the very high or low reflectance levels. (Specifics of reflectance spectra will be described in an upcoming post.)

However, controlling the amount of dye deposited onto the plate, even in approximate terms, presents significant challenges. Theoretically it would be possible to accurately measure the weight of the dye deposited, but adjusting one’s delivery to repeatedly match this measurement is fanciful. A more realistic approach to dealing with this uncertainty is to aim for the most sensitive application in which color change can be accurately measured and described; that is, the lightest application that is not so light as to make impossible the observation and measurement of fading.

Experiments conducted for the paint industry have demonstrated that such samples have a minimum reflectance of 30-40%; that is, in the spectrum of light reflected from the sample surface, the amount of light reflected at the most readily absorbed wavelength is 30-40%. When this condition is met, the entire spectrum will fall between the extremes of light and dark, so that there is room for it to gradually rise or fall as the colorant fades or darkens.

relfectance percentages

Mockup demonstrating how much dye corresponds to different reflectance values. AMNH/F. Ritchie

Since it can be difficult to visually gauge whether a particular dye sample on a clear plate meets these criteria, we created a set of small mockups with white backgrounds that reflect a range of minimum reflectance values (14% – 76%). These give us a sense of what each of those dye loadings look like. To arrive more precisely at the 30-40% minimum reflectance value in our samples, in the course of spraying, each sample is placed on a white membrane filter backing and measured using the spectrophotometer. More dye is added, if needed, until the desired loading is achieved.

color reading during spraying

Current project intern Caitlin Richeson measuring the percent reflectance on a freshly-sprayed quartz plate dye sample using a handheld spectrophotometer. AMNH/F. Ritchie

4.  Sample Shape: As our last post mentioned, our experimental procedure calls for taking periodic color measurements from each replicate throughout the duration of the exposure. At each time index, the values obtained from three locations on the plate are measured and averaged, to produce an overall measurement for the sample. In order to accurately calculate color change (which we will discuss in depth in an upcoming post), it is important that these three locations are reproducible, i.e., they do not change from measurement to measurement.

jig positions

Jig used to hold the quartz plate dye samples during color measurements. The samples are placed in each of the three positions for the three readings obtained using a spectrophotometer. AMNH/F. Ritchie

This requirement informed the design of a small jig with a wide T-shaped window into which the quartz plate can be inserted in any of three positions: left, right, and bottom. The spectrophotometer is placed in a fixed position atop the window. As each set of measurements is taken, the plate is moved through the three arms of the T. As long as the plate is inserted into the jig in the same orientation each time, the three measurement locations are constant. To ensure proper orientation of the quartz plates for each round of measurement, the dyes are applied to form an inverted T-shaped deposit on the plate. This irregular shape ensures the plate is properly orientated to provide the three viable measurement locations. If the plate were positioned in the jig in an incorrect orientation, at least one measurement would include an undyed area on the plate, giving the researcher an immediate cue that she has made a positioning error.

5.  Masking: A metal mask easily clips on to the plate to expose the same T-shape sample area in each application. After application, the metal mask is removed and cleaned before applying the next dye solution.

assembling samples

Assembling the bare quartz plate, metal mask, and backing material (to help see how much dye has deposited onto the clear plate) before spraying with dye. AMNH/F. Ritchie

6.  Reusing substrates: Since our quartz plates are in limited supply, their use does not end after a test cycle. After final color measurements and final photography, the plates are cleaned so that they can be reused for the next test round. Our cleaning procedure is intended to ensure that the plates are free of contaminants prior to the application of dye. Although our substrate is chemically inert, contaminants or dye degradation products can interfere with the results. Through testing and troubleshooting, a streamlined process is now in place:

Quartz Plate Cleaning Procedure:

  1. The quartz plates are first wiped with a solvent to remove grease and/or remaining dyes left on the surface from the last completed test cycle. The solvent we have chosen to use for this process is acetone, as it is effective on grease and all of the dyes.
  2. Solvent cleaning alone is not sufficient to fully remove the dye and all residues from the plates. The initial solvent cleaning step can redeposit dye in the roughened edges of the plate, where it becomes stubbornly embedded. Furthermore, we found that many dyes as they age create a thin, transparent, insoluble film on the plate – a “ghost” pattern on the surface, which accepts new dye solution differently than the virgin quartz surface. To remove these residual deposits and “ghosts,” we next polish the plates with Bueller MetaDi Monocrystalline Diamond Polish, 1 μm. The monocrystalline diamond particles are single grain particles with sharp edges. These particles are suspended in a fluid, which we can spray onto the plates and gently buff the surface of the quartz using cotton pads. The polish removes the residual dye and insoluble deposits on the plates.
  3. Next, the quartz plates are transferred to a bath of detergent and water. The detergent, Sparkleen, (Sodium Carbonate 10 to 25%, Sodium Dodecylbenzenesulfonate 1 to 10%, non-ionic detergent 1 to 10%) is a conventional laboratory glassware detergent that aids in the removal of both organic and inorganic deposits.
  4. Finally, after the quartz plates have been rinsed of the detergent, they are allowed to dry and then receive a final wipe with acetone. This final round of solvent cleaning is intended to ensure that no contaminants are left on the plates from handling. Once the plates are clean, they are ready to be sprayed with dye once again.

The preparation of the quartz plates is both time-consuming and stringent, but is an essential part of the experimental procedure.

A subset of 9 of the Orasol dye colors, dissolved in propylene glycol monomethyl ether (PGME) and airbrushed to the quartz plate substrate. AMNH/F. Ritchie


It’s All In the Preparation: Part 1

As earlier posts have mentioned, the first phase of our research aims to establish the lightfastness of the Orasol® and similar Sorasolve metal-complex solvent dyes in isolation‒ that is, in the absence of a binder, and without a chemically active substrate that could potentially influence the behavior of the dyes or interfere with the measurement of color change. To accomplish this we needed a substrate for the dye film that would not be altered physically or chemically by the accelerated aging exposure and would not interfere with our reflectance measurements (reflectance measurements will be discussed further in up-coming posts). The required substrate would need to be:

1.  Optically pure.

Any color or visual irregularities in a substrate would affect our color measurements, and impede an accurate calculation of the dyes’ lightfastness.

2. Durable.

The substrate needs to withstand a degree of handling throughout its use. For example, in one test cycle, the samples are measured for color change 8 times, in addition to before-, during-, and after-treatment photography.

3. Chemically inert.

The substrate must not be altered physically or chemically by the accelerated aging process, to ensure that it does not discolor (impeding accurate measurement of color change in the dye itself) or produce reaction products that could potentially alter the behavior of the dye.

Before making our final choice, we considered and rejected several candidates:

pink dye on wool

BASF dye Pink 478 (CIGN Solvent Red 127) sprayed on wool textile substrate. Although wool readily accepts dye, it was not chosen as the substrate because of its chemical instability during accelerated aging. Note also that the color of the dye is different depending on solvent. Solvent, top to bottom: acetone, ethyl acetate, ethanol, isopropanol, PGME. AMNH/F. Ritchie

Wool textile: inexpensive, readily available, and resilient to handling. It has the great benefit of accepting the dye in a manner very similar to the fur/hair on our taxidermy specimens because it is made from the same material, keratin fiber. However, it is not chemically inert, and is known to bleach or yellow during accelerated aging.

Acid-free paper: durable and inexpensive, but, like wool, is not chemically inert and may undergo color change or influence the stability of the applied dye. In addition, it accepts the dye solutions in a manner quite different from hair/fur, as they are readily absorbed deep into the sheet.

Glass slides or plates: much more chemically stable than wool or paper, though they may still produce reaction products in high temperatures and humidity. Dye applied to glass sits on the surface, similar to keratin fiber. Glass is inexpensive, but could be cleaned and reused if necessary. It is, of course, more fragile than wool or paper.

Ultimately we decided to use thin quartz plates, because they best fulfill our optical, physical, and chemical requirements for a substrate. Quartz plates consist of pure silicon dioxide (SiO2). They are not only optically pure (being transparent and colorless) and have a high damage threshold; they are also non-absorbent and reusable. However, as we came to appreciate, they are typically used in very sensitive research applications, which require high standards of dimensional exactness and chemical/optical purity. As such, they aren’t cheap. Finding affordable quartz plates was very challenging, and eventually took us to a Chinese company that was able to provide us with custom-cut plates that easily met the needs of our project, without the cost that accompanies the exacting standards required by other research applications.

Drawing on the procedure outlined in the ASTM D4303 standard, our testing of each dye solution and set of lighting conditions uses three replicate samples (A, B, and C). Two plates (A and B) are exposed in the aging chamber during the test cycle, and the third (C) is held in dark storage as a control. Throughout each test cycle, color measurements are taken periodically from each replicate. Each plate is measured in three locations, and these results are averaged to produce an overall measurement for that sample at that time index.

Having selected a transparent substrate, it was necessary for us to then identify a material to place behind our samples during our reflectance measurements and accelerated aging cycles. For the former, a standard white backing material was needed — something opaque (to eliminate any influence on the measurement from the tabletop or work surface behind this backing), and either durable or easily replaced. It would also need to be extremely white (i.e. with a uniform high reflectance throughout the visible spectrum) so that it would reflect the light transmitted through the sample back up into our reflectance spectrophotometer without distorting the spectrum of the dye. We chose to use plain white nitrocellulose membrane filters stacked three layers deep to ensure opacity. While a white ceramic tile could also have been used, we felt that the membrane filters would be more easily incorporated into the wooden jig that we are using to standardize our measurement locations.

jig with sample__1

Jig built for the project to hold quartz plate dye samples during color measurements. Note the white nitrocellulose film used to back samples. AMNH/F. Ritchie

For the same reason that a white backing material is helpful in color measurement, it would be problematic in our accelerated aging exposures: a white surface would act as a mirror, reflecting light from the xenon lamp that had already passed through the sample back up toward the plate. This would be exposing the dye films twice, once by direct exposure of the lamps, and a second time by reflected light from below, increasing the light exposure dose by some unknown percentage. This would make our determination of a rough equivalence between accelerated and real time aging periods in our dioramas more difficult. Eliminating the backing altogether would have the same effect: the plates would rest directly on our reflective aluminum sample holders. We needed to identify a material that would absorb the light transmitted through the sample plates but also would withstand the extreme environment within the accelerated aging chamber. We would need to use a black backing inside the chamber.

To realize this light-absorbing backing, we initially lined our aluminum sample holders with black archival matte board. We also added bumpers along the edges of the backing to create a small space for air circulation between the board and the plates, which helps with heat dissipation. Unfortunately, we discovered that the matte board generated a sticky condensate on the underside of the plates during aging. We altered our design to instead use black textile masks and corrugated blue board. These have proven to be free of the condensate problem of the matte board. The masks slowly fade during our exposures that include UV radiation, so we designed our sample mounts in such a way that new masks can be inserted when replacement is needed.

black panel_plate_holder

Clockwise: Quartz plate without dye, sample backing of black textile mask with cardboard bumpers covered in black textile mask, metal sample holder. AMNH/F. Ritchie

sample holders

Top: Dye samples resting on black backing. Bottom: Dye samples assembled in sample holder, ready for testing. AMNH/F. Ritchie







Part Two will describe how we prepare the quartz plate dye samples.

The Dye Is Cast

The initial sample set undergoing accelerated aging in our Q-SUN chamber is expected to be the largest. It includes all of the Orasol dyes that we have been able to acquire to date from the manufacturer BASF and retail supplier Kremer Pigments, as well as Sorasolve dyes available from Museum Services Corporation (MSC). This post includes a listing of all of our dye samples. For a comprehensive concordance of materials by manufacturer, current and former product names, and Colour Index Generic Names (CIGN), please see the table at the end of the post.

Orasol dyes, manufactured by and available from BASF

BASF does not sell small quantities of its product directly to consumers, but samples may be requested from the company for testing purposes. Materials obtained this way are understood to be of relatively recent production.

  • Orasol® Red 330 (CIGN Solvent Red 130)
  • Orasol® Red 335 (CIGN Solvent Red 122)
  • Orasol® Red 355 (CIGN Solvent Red 119)
  • Orasol® Red 363 (CIGN Solvent Red 125)
  • Orasol® Red 365 (CIGN Solvent Red 160)
  • Orasol® Red 395 (CIGN Solvent Red 122)
  • Orasol® Red 471 (CIGN Solvent Red 118)
  • Orasol® Pink 478 (CIGN Solvent Red 127)
  • Orasol® Orange 245 (CIGN Solvent Orange 56)
  • Orasol® Orange 247 (CIGN Solvent Orange 11)
  • Orasol® Orange 251 (CIGN Solvent Orange 54)
  • Orasol® Orange 272 (CIGN Solvent Orange 99)
  • Orasol® Yellow 081 (CIGN Solvent Yellow 79)
  • Orasol® Yellow 141 (CIGN Solvent Yellow 81)
  • Orasol® Yellow 152 (CIGN Solvent Yellow 88)
  • Orasol® Yellow 157 (CIGN Solvent Yellow 82)
  • Orasol® Yellow 190 (CIGN Solvent Yellow 89)
  • Orasol® Blue 825 (CIGN Solvent Blue 67)
  • Orasol® Blue 855 (CIGN Solvent Blue 70)
  • Orasol® Brown 324 (CIGN Solvent Brown 43)
  • Orasol® Brown 326 (CIGN Solvent Brown 44)
  • Orasol® Black X45 (CIGN Solvent Black 28)
  • Orasol® Black X51 (CIGN Solvent Black 27)
  • Orasol® Black X55 (CIGN Solvent Black 29)

Orasol dyes, manufactured by BASF, and available from Kremer Pigments

Kremer sells small quantities of BASF Orasol dyes retail. To do so, they purchase dyes in bulk from BASF and warehouse the stock until it is sold. As a result, dyes purchased from Kremer have an unknown production and storage history. Orasol products listed in Kremer’s product catalog are identified using both old and newer Orasol naming systems, reflecting the name in use by the manufacturer at the time Kremer made its purchase from BASF.

  • Orasol® Red 395 (CIGN Solvent Red 122)
  • Orasol® Orange 247 (CIGN Solvent Orange 11)
  • Orasol® Yellow 152 (CIGN Solvent Yellow 88)
  • Orasol® Yellow 4GN (CIGN Solvent Yellow 146)
  • Orasol® Yellow 2RLN (CIGN Solvent Yellow 89)
  • Orasol® Blue 825 (CIGN Solvent Blue 67)
  • Orasol® Brown 324 (CIGN Solvent Brown 43)

Sorasolve dyes, manufactured/supplied by First Source Worldwide LLC, and available from Museum Services Corporation (MSC)

Museum Service Corporation’s retail product-catalog lists their metal-complex solvent dyes using the old BASF naming system for the Orasol® brand; however, the dyes are supplied to MSC under the brand name Sorasolve by First Source Worldwide LLC, an American chemical company based in Neenah, Wisconsin. Orasol® and Sorasolve dyes with the same Colour Index Generic Name share an essential colorant with the same chemical constitution.

  • Yellow 2RLN (Sorasolve/CIGN Solvent Yellow 89)
  • Yellow 4GN (Sorasolve/CIGN Solvent Yellow 146)
  • Yellow 2GLN (Sorasolve/CIGN Solvent Yellow 88)
  • Orange G (Sorasolve/CIGN Solvent Orange 11)
  • Red BL (Sorasolve/CIGN Solvent Red 122)
  • Red G (Sorasolve/CIGN Solvent Red 125)
  • Pink 5BLG (Sorasolve/CIGN Solvent Red 127)
  • Blue GN (Sorasolve/CIGN Solvent Blue 67)
  • Brown 2GL (Sorasolve/CIGN Solvent Brown 42)
  • Brown 2RL (Sorasolve/CIGN Solvent Brown 43)
  • Brown 6RL (Sorasolve/CIGN Solvent Brown 44)
  • Black CN (Sorasolve/CIGN Solvent Black 28)
  • Black RLI (Sorasolve/CIGN Solvent Black 29)

We have characterized all of these dyes using fourier transform infrared spectroscopy (FTIR). Our results confirm that samples of old and new Orasol® dyes, as well as the related Sorasolve dyes, have very similar infrared spectra (see image below).

Yellow 89

FTIR spectra for Solvent Yellow 89 samples acquired from BASF (purple line), Kremer (blue line), and MSC (red line). Note the likeness between spectra, indicating that the three dyes sold by three companies under the same CIGN are chemically very similar, if not the same.

All of these dyes are soluble in a wide selection of solvents. As mentioned previously, during dye testing conducted in conjunction with the renovation of the dioramas in the Jill and Lewis Bernard Family Hall of North American Mammals, we observed that solvent choice can have subtle effects on both the color and lightfastness of the dyes. In order to explore these effects more extensively, five different solvents were selected for our present tests. They reflect a range of properties with respect to dye solubility and volatility/evaporation rate. Among them are solvents commonly used in the conservation of art and artifacts, as well as some others used in previous testing by Ciba-Geigy. Unusually toxic solvents or solvents that present other problems precluding their general use in restoration work were excluded.

The solvents we are testing include:

  • Acetone
  • Ethyl acetate
  • Ethanol
  • Isopropanol
  • Propylene glycol monomethyl ether (PGME).

Dye sample Red BL (CIGN Solvent Red 122) purchased from Kremer. Note the difference the solvent choice makes in surface texture/coverage between the sample dissolved in acetone (left) and the same sample dissolved in PGME (right). AMNH/F. Ritchie

Dye concordance table

Concordance of dye materials by current and former product names, Colour Index Generic Names (CIGN), chemical composition, and source.


Accelerated Aging Chamber, Part 1

Q-Lab Corporation, manufacturer of the Q-SUN Xe-3 accelerated aging chamber, promotes this machine as “the simplest, most reliable, and easiest to use full-sized xenon arc chamber available.” Before purchasing ours in February 2014, we began making upgrades to water and electrical systems in our lab to meet its basic requirements. Perhaps naively, we had planned to install the chamber and begin our testing promptly once those upgrades were complete. Throughout following months, we encountered a series of unexpected challenges in the set up and operation of our new chamber. This is the first in a pair of posts that will introduce the Q-SUN Xe-3, its capabilities and some of the theory behind its use, explore the challenges we have had, and suggest some key issues that you might consider when planning to acquire a large piece of new equipment for your laboratory.


2014-03-10 15.47.49

Q-SUN Xe3 Accelerated Aging Chamber with Q-Lab training specialist Alan Boerke for size comparison. AMNH/B. Nunan


Much to the disappointment of some of our curious colleagues in other departments who wondered what one does with an accelerated aging chamber, the Q-SUN Xe-3 can not be used to expedite troublesome developmental phases in your toddler, nor be run in reverse to reunite you with your youth. Too bad. Instead, this machine is used to rapidly reproduce the damage to materials that is caused by light, temperature, and humidity in real environments over longer periods of time.

The tester is a bit bigger than a refrigerator, and contains three powerful xenon arc lamps that expose samples to bright, daylight-imitating light inside of a compartment roughly the size of an oven. The spectrum of light produced can be adjusted with the installation of various filters above the sample compartment. Light output is measured in irradiance (W/m2), and can be controlled at either 340nm (ultraviolet) or 420nm (visible) depending on what filters are in use. The tester also maintains set points for relative humidity, chamber air temperature, and the temperature of a black panel placed inside the sample compartment.

2014-03-10 16.06.51

Q-Lab training specialist Alan Boerke discusses the Q-SUN aging chamber with project conservators and other conservation scientists from neighboring NYC institutions. AMNH/B. Nunan

On March 10, 2014 the American Museum of Natural History (AMNH) hosted a training session on accelerated aging and use of our new Q-SUN Xe-3 with Alan Boerke, Technical Sales and Training Specialist at Q-Lab Corporation. The training was attended by selected museum staff and colleagues from Yale’s Institute for the Preservation of Cultural Heritage, the Metropolitan Museum of Art, and the Museum of Modern Art, as well as students from New York University’s Institute of Fine Arts Conservation Center.

Accelerated aging makes use of the principle that exposure to high intensity light for a short time can produce deterioration similar to that caused by low intensity light over a longer time. However, in order to correctly interpret one’s results, one must understand that for many reasons, accelerated aging does not occur in a way that is strictly reciprocal. In part this is due to the inability of any aging chamber to exactly replicate every aspect of real-world exposures: wet/dry, thermal, or light/dark cycling, the spectrum of incident light, and the presence of air pollutants, dust, or adjacent materials may be impossible to simulate. This non-equivalence is also a consequence of thermal chemistry that unfolds simultaneously alongside light damage, but can’t easily be differentiated from it or accelerated proportionally.

Alan emphasized benchmarking as a way of managing this problem. To create a benchmark, materials aged in real-time are used to define the mode and extent of change taking place over a known duration. When a comparable degree of change is observed in the accelerated test, a correlation factor can be identified to be used in calculating an approximate relationship between accelerated and real-time aging. However, benchmarking has some obvious drawbacks, not the least of which is that a material that ages well may take many many years to fail in a real-world exposure environment. If one is conducting accelerated aging on that material, it’s usually because one needs information promptly and can’t afford to wait.

The success of this approach depends on the selection of a standard that exhibits deterioration behavior similar to the samples being tested- both in the real world and accelerated aging environments.  However, since the samples being tested have unknown aging behaviors one standard is usually insufficient.  So instead of choosing a single standard, it is better to select a series of standards that will hopefully bracket the behavior of the samples.  For lightfastness testing, a common set of standards is the Blue Wool scales, wool swatches dyed with eight different dyes that exhibit a range of different lightfasnesses.  By including the Blue Wool scales in our accelerated tests we can determine which of the eight standards our samples behave most similarly to.

Blue Wool card_annotated

Blue Wool scale assembled by the team using blue wool reference standards 1-8 obtained from SDC Enterprises Lmtd and mounted onto card stock. A new Blue Wool reference scale will be used with each test round. AMNH/F. Ritchie

Our training session also included a discussion of other factors that could affect the results of our testing: the color and cleanliness of the sample, whether it is mounted at an angle, or over a backing board, and its height inside the sample compartment; variations in sample handling and measurement technique; breaks in our test cycle for sample measurement; and the age of the xenon lamps. Getting repeatable results hinges on limiting variation in these influences.

We concluded with a tutorial in which Alan showed us all the basics for running and maintaining the machine: how to load and rotate samples, install and calibrate lamps, change light filters, and program the desired parameters.

With all of this new knowledge in hand, we promptly began the process of translating our research plan into an actual method for mounting and testing our dye samples. Very quickly we observed that doing so would not simply be a matter of plugging in the Q-SUN, programming the ASTM D4303 test parameters, and pressing the ON button. Many unanticipated challenges were yet to come…

Calibrating the lamps

Conservation intern Ersang Ma prepares to calibrate the Q-SUN lamps, according to procedures learned during the one-day training session by Q-Lab. AMNH/B. Nunan