PHOTOLUMINESCENCE SPECTROSCOPY OF NATURAL RESINS AND ORGANIC BINDING MEDIA OF PAINTINGS
LONDA J. LARSON, KYEONG-SOOK KIM SHIN, & JEFFREY I. ZINK
4 RESULTS AND DISCUSSION
PREVIOUS WORK has focused on the luminescence of dyes and pigments, natural resins, binding materials, and mixtures of linseed oil with pigments (de la Rie 1982a–c; Miyoshi 1982, 1985; Wallert 1986; Guineau 1989; Tutt 1984). Ideally, one would like to be able to distinguish among various pigments, coatings, and binding materials in the presence of one another.
Many different samples were studied, including natural resins, waxes, drying oils, proteinaceous materials, and gums as well as a few mixtures of these materials with one another and with pigments. Tables 1 and 2 list the samples studied spectroscopically in the authors' laboratory to date. Emission spectra were taken of all of the samples listed in table 1 at room temperature. In addition, several of the bulk samples as indicated in table 1 were studied at low temperatures (10K–25K). Except for dragon's blood, orange shellac (fig. 6), lemon shellac, and dewaxed, decolored shellac, all of the bulk samples exhibit broad, relatively featureless emission spectra at room temperature. For the bulk samples that were studied at low temperature, it was found that bleached beeswax (fig. 10), in addition to dragon's blood and the shellacs, exhibits structure in its emission spectrum at low temperature. Emission spectra of the films listed in table 2 were taken at low temperature (20K–25K). Several of the films, including films of elemi (fig. 11), sandarac, dragon's blood, bleached beeswax, and the shellacs, show structure to varying extents in their low-temperature emission spectra. A room-temperature spectrum of the orange shellac film cast from an ethanol solution was also obtained (fig. 6). Each spectrum was analyzed with respect to both the energy of the peak maximum and its fwhm. As mentioned previously, peak shape is also an important characteristic that was considered. The samples have large quantum yields (high intensities) of luminescence, and spectra of extremely small samples are obtained with a high signal-to-noise ratio.
The following discussion is divided into six parts:
- a consideration of the luminescence of three of the binders studied;
- a discussion of binders in the presence of lead white;
- an example of a case in which emission spectra do not provide a reliable means for differentiating between two specific bulk materials;
- a discussion of the luminescence properties of the bulk natural resins;
- a discussion of temperature effects; and
- a comparison of the luminescence spectra of films cast from various solvents and the bulk materials.
Room-temperature spectra taken by using 363.8 nm excitation of three of the binders studied—aged linseed oil (ca. 1934), aged egg yolk (ca. 1937), and bulk casein—are shown in figure 3. The emission maxima are well separated. The maximum emission in the luminescence spectrum of casein is at 21,900 cm−1, that of egg yolk is at 17,200 cm−1, and that of linseed oil is at 14,600 cm−1. Luminescence spectroscopy can thus nondestructively differentiate among these samples by the positions of their emission maxima. A second spectral feature that can be used in conjunction with the peak maxima to differentiate among these samples is the peak width. The fwhm of the peaks are 4700 cm−1, 5600 cm−1, and 4200 cm−1 for casein, egg yolk, and linseed oil, respectively.
The spectrum of casein shows a typical luminescence peak shape for a large organic molecule. It rises sharply on the high-energy side and has a longer tail on the low-energy side. In contrast, the spectra of both the linseed oil and the egg yolk are not typical of luminescence from a single component; both spectra have long tails on the high-energy side of the peak. The tail is characteristic of emission from one or more components that emit to the high-energy side of the most intense peak.
The room temperature spectrum of a modern sample of linseed oil (aged one year) also exhibits an emission spectrum characteristic of a multicomponent sample. The emission from the modern sample peaks at 20,300 cm−1. This emission is shifted to higher energy compared to the aged sample's emission by 5,700 cm−1 and is consistent with de la Rie's findings (1982b) that the luminescence of linseed oil shifts to lower energy as the oil ages. Care should be taken in this comparison, however, since the conditions under which linseed oil ages have been shown to contribute to both the intensity and the position of the emission (de la Rie 1982b).
4.2 BINDERS IN THE PRESENCE OF LEAD WHITE
It is important to determine whether or not binders luminesce in the presence of pigments. de la Rie has shown (1982c) that lead white (2PbCO3.Pb(OH)2) appears to accelerate the production of the emitting species in linseed oil (the fresh oil does not luminesce). He has suggested two possible mechanisms that could lead to this effect. First, the cause may be the accelerating effect of Pb metal on the drying process of the oil, which accelerates the production of luminescent degradation products. Alternatively, the metal may influence the luminescence of the degradation products formed as the oil ages by complex formation. Such complex formation could lead to either quenching or, as in this case, enhancement of the luminescence. The room-temperature emission spectra taken by using 363.8 nm excitation of aged samples of linseed oil, whole egg, and a mixture of oil and whole egg, all in the presence of lead white, are shown in figure 4. Pure lead white has been reported to show no luminescence, so the emission is due to the presence of the organic binder (de la Rie 1982a; Miyoshi 1982). The mixture of linseed oil and lead white has an emission maximum at 17,000 cm−1, whereas whole egg mixed with lead white has a maximum at 21,000 cm−1. The emission maximum of the mixture of linseed oil and whole egg in the presence of lead white lies between the other two at 18,200 cm−1. In addition, the width of the emission band for linseed oil alone in the presence of lead white is noticeably less (fwhm = 4,900 cm−1) than the widths of the other two spectra (fwhm = 6,000 cm−1 for egg and lead white, fwhm = 5,700 cm−1 for egg, linseed oil, and lead white). Thus oil and whole egg as well as mixtures of the two are luminescent in the presence of lead white, and the three samples are distinguishable by their luminescence spectra.
4.3 DIFFERENT MATERIALS WITH SUBTLE DIFFERENCES IN SPECTRA
The position of the peak maximum may not always provide a reliable means of differentiating among materials. For example, the room-temperature luminescence spectra of gum tragacanth and gum arabic excited at 363.8 nm maximized at 22,200 and 22,100 cm−1, respectively, and are very similar in shape, as shown in figure 5. The similarity of the spectra should not be construed to mean that the structure of the materials is similar. The biogenesis of each gum appears to be quite different, and the ratio of their neutral monosaccharides, arabinose and galactose, are widely different (Twilley 1984). Therefore chemical characterization remains superior to characterization by room-temperature luminescence with ultraviolet excitation, and caution should be taken in trying to distinguish these materials by using the emission spectra taken at the one excitation wavelength and temperature.
The emission from both of these samples shifts to higher energies when the temperature is lowered to 10K. However, the spectrum of gum tragacanth, with an emission maximum at 24,000 cm−1 at 10K, shifts to a greater extent than that of gum arabic, which peaks at 22,800 cm−1 at 10K. As previously indicated, the instrument is also capable of varying the excitation wavelength to provide a multidimensional picture. These two experimentally controllable factors along with the characterization of the luminescence lifetimes may provide a method for distinguishing these two materials.
4.4 NATURAL RESINS
The luminescence spectra of 15 natural resin samples, including dammar, mastic and sandarac (tree resins), copal (fossil resin), shellac (insect derived), and rosin (resin from balsam distillation), were obtained and studied. These spectra proved particularly revealing not only of the power of luminescence spectroscopy but also of the properties of the samples. In this section, four aspects of the spectra are illustrated and discussed. First, differentiation between some of the resins will be discussed. Second, the dependence of the spectra on the wavelength of excitation is used to show how the technique can be employed to selectively excite specific components of a multicomponent system. Third, the spatial resolution of the technique is used to show that an ostensibly pure sample is actually spatially inhomogeneous. Finally, the technique reveals spectroscopic differences that may be caused by differences in the sample source or by chemical changes that occur during aging. The spectra discussed in the following sections are contained in figures 6–9.
4.4.1 Spectroscopic Differentiation
Of the 15 resins studied, dragon's blood and the shellacs can be easily differentiated from the others. The emission observed using ultraviolet excitation from each of these samples is further to the red, peaking between 15,000 and 18,000 cm−1, than the emission observed from the other resins studied. Thus they exhibit emission maxima clearly different from those observed with the other resins (figs. 7–9, 11), which emit farther to the blue. In addition to luminescing further to the red, both the bulk and film samples of dragon's blood and the shellacs show structure in their spectra. Figure 6 illustrates the structure observed in the emission spectrum of orange shellac. The origin of the structure was not thoroughly investigated, but it is most likely due to multiple emissions as opposed to being of vibronic origin.
4.4.2 Effect of Excitation Wavelength on Spectra
The excitation wavelength dependence of luminescence can be useful in characterizing the resins. As discussed previously, a one-component sample rarely exhibits wavelength dependence. However, a multicomponent sample may show wavelength dependence.
As a simple example, consider a material made up of two different components: component A, which absorbs at 351.1 nm, and component B, which absorbs at both 351.1 nm and 457.9 nm. Irradiation of the material at 351.1 nm excites both components. Thus under these conditions luminescence is observed from both components. However, irradiation of the material at 457.9 nm only excites component B. Thus under these conditions luminescence is observed only from component B. Our excitation wavelength dependence experiments indicate that the materials called sandarac, dammar, copal, rosin, and mastic consist of more than one luminescent component.
Figure 7 illustrates the specific effect of excitation wavelength on the luminescence of sandarac. As the excitation wavelength moves toward lower energy, the peak maximum shifts to lower energy. This behavior indicates that sandarac has multiple components that absorb and emit at different wavelengths.
4.4.3 Spatial Inhomogeneities
Laser-induced luminescence is a powerful technique for detecting inhomogeneities and determining the spatial distribution of the components of a sample. For example, figure 8 shows three different emission spectra from three different 100-micron regions of the same sample of what is purported to be manila copal dated 1920–23. Widely different spectra were observed, with emission maxima ranging from 16000 cm−1 to 20400 cm−1. A conventional luminescence spectrometer with a mercury discharge lamp as the excitation source does not allow the fine spatial resolution that can be achieved when using a laser. Using a conventional spectrometer, the spectra of the sample of manila copal might appear as a superposition of the spectra shown.
The major components of bulk dammar, mastic, and rosin luminesce at room temperature under 363.8 nm excitation with band maxima at 22,000 cm−1, 20,600 cm−1, and 19,300 cm−1, respectively. Spectra from these samples are shown in figure 9. However, small regions of the samples could be found in which the band maxima were shifted from those in figure 9. Thus in the samples with closely spaced emission maxima the presence or absence of spatial inhomogeneity must be verified before spectra can be used to identify samples.
Chemical changes that occur with aging might be expected to cause spatial inhomogeneity in a sample. We have noted that the naturally aged resins obtained from the reference collection at the Fogg Art Museum, Harvard University, show spatial inhomogeneity to varying extents. The effects of the atmosphere would be predominant on the surface of and in cracks on a sample. Luminescence spectroscopy can probe the spatial location of these processes.
4.4.4 Effects of Sample Source or Aging
The effects of sample origin and/or aging are exemplified in figure 6. All three spectra shown are of orange shellac. Figure 6a is a spectrum of a modern bulk sample. The emission spectrum shown in figure 6b was taken from a modern film cast from an ethanol solution of the same bulk material source as used to take the spectrum shown in figure 6a. Figure 6c is the spectrum of an aged film cast from ethanol. Both the modern bulk and modern film samples exhibit emission peaks or shoulders at 16,500, 15,800, and 14,700 cm−1, but the relative intensities of these features are different between the bulk and the film spectra. The aged film shows only two features in its emission spectrum; a peak at 17,000 cm−1 and a shoulder at 16,000 cm−1. The emission spectrum of the aged orange shellac film is shifted to slightly higher energies with respect to the newer sample. This finding is contrary to de la Rie's finding (1982b) that the emission from resins tends to shift to lower energies with age. Again, care must be taken in these comparisons since the conditions under which the sample was aged will contribute to its luminescence characteristics. The differences between the spectra of the modern and aged films are probably related to chemical differences between 6b (ca. 1988) and 6c (ca. 1932), either caused by changes that occur during aging or by differences in the sample source.
4.5 TEMPERATURE EFFECTS
Studies of the temperature dependence of the luminescence showed that the intensity increased with decreasing temperature. This increase is due to a decrease in the rate of nonradiative decay. One advantage in using a low temperature is to increase the signal-to-noise ratio. For the resins orange shellac, copal, rosin, sandarac, mastic, and dammar the intensities increased approximately tenfold on going from room temperature to 10K. A second advantage in using low temperature is that, in general, the quantum yield of phosphorescence is increased to a greater extent relative to fluorescence. For many compounds phosphorescence is absent at room temperature but easily detected at low temperatures. New phosphorescence bands that may grow in at low temperatures may be useful in identifying materials. Bleached beeswax (fig. 10) is an example of a sample that not only shows an intensity increase at low temperature but also exhibits this effect of new bands growing in at low temperature.
4.6 COMPARISON OF BULK AND FILM SAMPLES
Comparison of bulk and film samples allows the evaluation of differences that result from solvation of the different components that make up the bulk materials. Comparisons were made between bulk and film sample spectra taken at the same temperature. In general, the film samples had luminescence spectra with larger fwhm than the corresponding bulk materials. In some cases only one solvent was used and compared to the bulk, while in other cases three different solvents were used (table 2).
When one solvent was used, little difference was observed between the bulk and film samples of the shellacs, bleached beeswax, and the copals. The peak positions for the bulk shellacs and bulk beeswax were the same as in the corresponding films. For orange shellac, different relative intensities of the peaks were observed in the bulk as compared to the film material. This comparison is shown in figure 6a–b. For bleached beeswax the relative peak heights could not be easily compared due to inhomogeneities detected in both the bulk and film samples. The bulk copals were inhomogeneous, but parts of the bulk samples looked similar to the film. In the case of gum tragacanth and gum arabic, the overall shape of the bulk and film spectra were similar, but the films show maxima that were red shifted by 700 and 1,200 cm−1 for gum tragacanth and gum arabic, respectively, compared to the bulk.
The multisolvent cases yield several interesting observations. Dammar and rosin films showed slightly solvent-dependent emission maxima, although the overall shape of the spectrum changed little between the bulk and the different film samples. Mastic and sandarac, in addition to showing solvent-dependent maxima, also exhibited slightly different band shapes depending on which solvent was used. Elemi and dragon's blood are more extreme cases where a large solvent effect was observed. Figure 11 shows this effect for elemi where the emission spectrum of the bulk sample along with films cast from turpentine, ethanol, and mineral spirits are compared. A total of five different bands can be observed in the film spectra. The bulk spectrum is a superposition of these various bands.
The differences observed between emission from the bulk materials and various films could be attributed to selective solvation of components of the resin in a given solvent. Another possible explanation is related to the inhomogeneity of the bulk samples. Components within the bulk sample may not contribute to the luminescence because of self-absorption of the exciting or emitted radiation. These components could become more accessible after dissolution in a solvent. Solvent retention or the reaction of the solvent with the material may also contribute to the observed differences to some extent.