University of Pennsylvania Museum of Archaeology and Anthropology

Blue lines: multispectral imaging for pigment identification


January 30, 2015

Visible-induced infrared (IR) luminescence is the invisible light that some materials produce when they are excited with visible light. We can capture that invisible light with a modified camera and use it to identify those materials and find out where they are. For those of you who follow the Artifact Lab Blog, this technique will sound familiar. Molly Gleeson used it to identify Egyptian blue on a shabti box that she was working on. Guess what? The technique can be used to see other materials too! In examining a 12th-century Islamic manuscript, NEP27, I needed to find a way to distinguish between indigo and ultramarine. Both are blue in visible light. Both are made with elements that are below the detection limits on our pXRF, our usual go-to method for elemental analysis.

Detail of marginal elements on page 360 of NEP-27. Notice the blue outlines around the gilded designs.
Detail of marginal elements on page 360 of NEP-27. Notice the blue outlines around the gilded designs.

It was starting to look as if I’d need to take a sample to examine with our polarized light microscope and possibly send out for x-ray diffraction, when my colleague, Tessa de Alarcon, remembered a single sentence in an article on Egyptian blue and Han blue, the same reference that Molly used for her work. What did it say? “The emission of indigo is reported to be at c. 750 nm while that of lazurite at c. 830 nm.” (Verri, 1012) There it was in black and white. The secret to a non-destructive method for answering the question of indigo or ultramarine. Those numbers are the wavelengths of light that they produce when excited. Both are in the infrared range and invisible to our naked eye, but we can see them with our camera. Lazurite is the color component of ultramarine, so if it has emissions that are different from indigo, maybe we could use filters to separate them.

To find out if this would work, I found a couple of dry pigment samples, mixed them up with gum Arabic, and painted them out on a swatch of paper. With this set of knowns, I set up the Mini-CrimeScope, our tunable light source, and the modified camera to try different light and filter combinations. It turned out that only the ultramarine luminesced with exposure to light centered around 455 nm. Using this set up, I tried it out on the manuscript, and sure enough, the blue decorative elements fluoresced just like the natural ultramarine sample.

Detail of page 360 illuminated with 455nm light captured in infra-red, grey scale. The blue lines seem to disappear and are hard to distinguish from the paper. They didn’t disappear though; they’re fluorescing!
Detail of page 360 illuminated with 455nm light captured in infra-red, grey scale. The blue lines seem to disappear and are hard to distinguish from the paper. They didn’t disappear though; they’re fluorescing!

The false color image helps to visualize which areas are luminescent. In the photo, the information from the IR photo replaces the red channel, so the red lines around the decorations are the luminescent parts . Or in this case, the ultramarine!

Detail photographs of page 360 illuminated with visible light and with 455nm light captured in infra-red combined into a false color image. The blue lines now appear red, helping to show where the luminescence took place.
Detail photographs of page 360 illuminated with visible light and with 455nm light captured in infra-red combined into a false color image. The blue lines now appear red, helping to show where the luminescence took place.

This wasn’t true across the board, but the blue decorations don’t look the same throughout either. I’m not sure if this means that ultramarine was used in some areas, but not others, or if different sources of the pigment have enough variation to produce different results. My samples included different sources for the ultramarine, which were also variable with their luminescence, so I’m thinking that the luminescent component might be variable depending from batch to batch. I’m still working on the details to see if we can use this as a diagnostic technique, but this is a promising development!


© Penn Museum 2018 Sitemap | Contact | Copyright | Disclaimer | Privacy |