Study of Transparency of Pigments to Near Infrared ^†

Abstract

We describe the results of a systematic study of reflectance of pigments traditionally used in Mexican cultural treasures on different substrates. We observe that underline drawings may be most easily detected in NIR when the pigment transmissivity in NIR is high. We conclude that the under-drawings of best quality are detected with illuminators with low NIR wavelengths.

1. Introduction

Mexico has a rich and diverse cultural heritage. Most of that heritage is very rather old; it has been neglected for a long time, it has been physically subject to various civil wars; and thus, it is poorly documented. We are interested in using near infrared radiation (NIR) to determine the historical value of unknown pieces of national heritage. We are attempting to classify local materials as to their transparency in NIR when used on canvas.

We appreciate that the coloring agents depend on availability of local dyes. This field has been amply studied in Italian art, with results presented in Figure 1. IR reflectography can help visualize hidden details under painting layers, which can provide valuable information for art historians researching a piece: Has it been reconstructed before? Do under-drawings indicate different objectives of the owners of the paintings during different times?

2. Materials and Methods

Pigments tend to be based on the raw materials available in the geographical location of the author’s workshop. NIR transparency depends not only on the pigment, but also on the substrate used to fix it to the canvas. Only a few pigments are well documented, so it is not always easy to choose the optimal NIR wavelength for the reflectogram.

The objective of this work is to determine the NIR transparency of several pigments typically employed in the Nueva España (Mexico) region. Additionally, we studied the reflection spectrum of these pigments to try to correlate them with their transparency.

We employed two pigment matrices: for color pigments (C) and for brown pigments (B), shown in Figure 2. On B, each column on each matrix features a different pigment and each row features a different substrate. On C, each row is a different pigment and each column a different substrate. On some elements, the under-drawing even in the visible is noticeable. This is quite noticeable for the color matrix, elements 1.2 through 1.5, and the blue 4.4. The dark blue color pixel, denoted 4,3, appears visually quite opaque.

3. Experimental

Figure 3 presents the image of color matrix as captured by Xenics-XS camera on the left and its spectral responsivity on the right.

We measure the reflectance spectra of each matrix element using a fiber optic spectrometer fitted to an integrated sphere with a tungsten halogen lamp. The spectra are compared to the reflectance spectra of a 99% reflectivity reflectance standard. We also measured the reflectance spectrum of the canvas in order to compare it with the matrix elements.

We captured NIR reflectograms of each matrix element using an InGaAs camera. To achieve multispectral imaging, we recorded several for each matrix element using different illumination sources. The narrow spectral width sources include four different 10-LED NIR lamps (centered on 970, 1050, 1070 and 1200 nm) with a FWHM of 100 nm, and three different high power LED lamps (centered on 1300, 1450 and 1550 nm) with a FWHM of 100 nm. We also captured reflectograms with an incandescent light source illuminator (blackbody-like radiation).

4. Results

The measured spectral reflectance of canvas and transparent pixels in row 1 of color matrix are shown in Figure 4a. With the exception of the first pixel that is indeed opaque in the visible, all the transparent pixels seem to have the same spectral form, except that the curves appear to be shifted to lower transmittance values with the increasing pixel number (see Figure 2b). Figure 4b shows a photo of a representative ten-LED source.

We compared the transparency of each element for each lamp in Figure 5. We present different detected reflectances from the colored pixel 4,3, indicated dark blue when illuminated with the IR radiation. The central wavelength of the illuminating lamp is indicated below the sample.

5. Conclusions

Substrate significantly alters the pigment spectrum, obscuring some spectral regions. The pigment is not transparent in the spectral regions where the pigment reflectance is much lower than that of the canvas.

Reflectance spectra of low transparency pigments tend to be flatter in form, while the spectra of high transparency paints had a shape similar to that of the canvas spectrum. The analyzed pigments exhibit decreased transparency with increasing wavelengths.