As Leona notes in a recent issue of PNAS (1), art and objects of cultural heritage hold information about the technology, trade, and aesthetics of a people. This knowledge can be accessed through chemical and material analysis combined with an understanding of the history of technology, science, and art. The information learned is of great value, especially for ancient objects that are so old we have no chance of finding written documentation such as bills of lading, inventories, or manuals and treatises that we may consult for descriptions of the materials used in their creation.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

The Adoration of the Shepherds, 1505/1510 by Giorgione (1477/1478–1510) from the Samuel H. Kress Collection 1939.1.289, National Gallery of Art. (Image courtesy of the Board of Trustees, National Gallery of Art, Washington, DC.)

The retrieval of this information requires detailed and accurate analysis of the materials and methods involved in the making of art and artifacts. For more than a century, museum scientists studying artworks have been able to identify single particles of many minerals and other inorganic pigments by using optical and chemical microscopy; however, the characterization of organic pigments and dyes from such small samples has been quite difficult. The ancient artists used many earths and minerals and likely just as many organic extracts to expand the range of colors. Unfortunately, many of these organic colorants have disappeared over time. However, some, in particular the red pigments formed by making insoluble complexes by reaction with aluminum salts or adsorption onto chalk or alumina, remain in good condition, especially if they have been protected from light. A fine example is the deep, rich red in the Virgin's robe painted by Giorgione (Figs. 1 and 2).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Detail from The Adoration of the Shepherds. A cross-section from a deep shadow on the Virgin's blue mantle shows that the mantle was painted over deep translucent red paint, and folds in the drapery were depicted by glazing with a purple paint mixed from the blue and red pigments. The purple is now discolored because of photo-degradation of the red pigment. The inorganic pigments ultramarine and lead white are easily identified, but the red could be one of several anthraquinone-based pigments. With a microspatula it is possible to remove enough sample from the red layer for analysis by using the SERRS technique described by Leona (1) while preserving the cross-section for further study with other methods. The dye used here to make the red lake is kermes. Scanning electron microscopy with energy dispersive spectroscopy indicates the extracted colorant was made insoluble by complexation with aluminum.

For cultural heritage studies, results that have value for elucidating our material heritage require stringent demands on the characterization of molecules. Many historical red dyes are anthraquinones. An extract from the root of madder (Rubia tinctorum) was widely used for both dyes and pigments. Three or four related anthraquinones may be present in the natural product, including the well-studied molecules alizarin and purpurin. They are similar to the colorants obtained from animal sources, kermes (also called grain), carmine, and lac. Kermes (kermesic acid; 9,10-dihydro-3,5,6,8-tetrahydroxy-1-methyl-9,10-dioxo-2-anthracenecarboxylic acid) is distinguished from cochineal (carminic acid) only by a glucoside unit at the 7 carbon and is also similar to laccaic acid A, the principal component of the red pigment lac, which contains an amide functional group on a substituent at the 7 carbon. Culturally and geographically, however, these compounds have quite separate histories of use. Cochineal was introduced to Europe a couple of decades after the Spanish conquest of Mexico (1521), whereas kermes (and “Polish” cochineal that does in fact contain kermesic and carminic acid) and lac were cultivated and imported into Europe long before the 16th century, perhaps since Biblical times (2). Thus, accurate identification and the ability to distinguish among related molecules is a prerequisite for meaningful conclusions regarding the use and trade of colorants.

HPLC was the method of choice for museum scientists to identify natural organic red pigments. Although HPLC requires relatively large samples, the slightly different molecules from different sources can be distinguished. Improvements in these methods are still needed to reduce the sample size that is acceptable for textile fibers but remains rather large for paintings and drawings. Coupling liquid chromatography with new mass spectrometric methods for dye analysis is under investigation (3).

The use of in situ UV-visible and fluorescence spectroscopy to identify organic red pigments has been investigated (4). Although this methodology does not require sampling, closely related anthraquinone pigments are not readily distinguishable in paints, and fluorescence of binding media and self-absorption effects can make acquisition of precise data problematic and the analysis of specific pigments in artifacts difficult to perform even when samples are removed from an artwork (5). It would be a great achievement if these methods could be used for chemical mapping and imaging of entire artworks, but it will take significant and sustained effort.

Smith and Clark (6) cogently described how advances in Raman microscopy have allowed that analytical method to become an important tool for studying cultural heritage. The microscope lowered detection limits and improved specificity of analysis of organic materials in artworks. A particular benefit noted is the ability to focus on individual particles. Additionally, the use of the technique for in situ analysis, that is surface analysis without resorting to removing a sample, is an advantage over older methods. A library of spectra is available for identification of unknowns by comparison (www.chem.ucl.ac.uk/resources/raman). However, sensitivity remains a problem, as does fluorescence of the analyte and its matrix that often contains many chemical species that contribute to the Raman signal.

The fluorescence of oil and resinous binding media that were used widely in the past markedly interferes with the collection of useful Raman spectra of colorants on artifacts. Contributions to the Raman signal from fibers and mordants affect our ability to identify dyestuffs. In surface-enhanced Raman spectroscopy (SERS) the Raman signal of a molecule is multiplied many-fold by interaction with colloidal metal particles, often silver. SERS also mitigates signal interference through substantial quenching of fluorescence and has been effectively used to identify pigments. SERS has already been shown to have potential for identification of red lakes, but high-quality results for the whole variety of red lake pigments have been tantalizingly difficult to obtain. Using red and infrared lasers for Raman microscopy has reduced fluorescence of the samples (7), but there is a concomitant loss of sensitivity.

Although SERS usually requires that a sample be removed from an object for study, the increased sensitivity means that in the best cases very little material, perhaps as little as a single grain of pigment, is required (8). Often, however, the complexity of artists' materials means that this is not the case. Successful characterization of a pigment in paint is possible when the weight ratio to the binder is high, but for many samples the percentage weight of the molecules of interest is low. Additionally, deterioration over centuries decreases the amount of unaltered material present in a work and available for analysis.

The advances made by Leona's study (1) refocus attention on developing Raman spectroscopy as perhaps the best method for analyzing minute quantities of organic pigments. Interfering materials, such as an oil binder in paint, are destroyed by hydrolysis using vapor phase hydrofluoric acid. The simplicity of the nonextractive, almost lossless method for sample work-up effectively preserves the analyte.

The method used for preparation of the suspension of colloidal silver needed for the SERS scattering effect not only provided a stock that was stable for months rather than hours or days, but judging from the narrowness of the band in the visible region of the spectrum, it also produced a suspension that must have a narrow dispersion of shapes and sizes of silver nanoparticles. The nanoparticles aggregate in the presence of the anthraquinone colorants and signal is increased by an effect that occurs because the energy levels of the surface plasmon resonance produced in the colloid by this novel preparative method are just about right for creating the condition suitable for resonance Raman spectroscopy. This resonance further augments the enhancement in signal developed by the proximity of nanoparticles and their aggregates, thus enabling surface-enhanced resonance Raman spectroscopy (SERRS). Now, minute samples of red paints can be characterized despite the presence of oils and resins in the original sample. The resonance Raman spectra of related colorants such as madder and kermes are distinct and the related compounds can be distinguished. An extremely important result is the ability to detect carminic acid from cochineal; for some reason this molecule can be difficult to detect.

Advances in the SERRS methodology, while driven by curiosity about the pigments and dyes on works of art and artifacts combined with the requirement of analyzing small, multicomponent samples, will surely be adopted for situations where fluorescence and the need for low detection limits currently makes SERS difficult. The advantages of this new method mean that in the museum laboratory SERRS could replace chromatography for analysis of organic red pigments when instruments with 488-nm lasers are acquired. It would be a boon to the field of cultural heritage science if this work can be extended to the analysis of flavanoid-based yellow lakes. This may take much more research into facile and reproducible synthesis of suitable silver colloids. A theoretical understanding of the modes by which resonance enhancement of Raman scattering occurs would be most helpful. Libraries of SERS spectra are also needed. With these developments Raman spectroscopy can be used to identify both inorganic and organic colorants in cultural objects because it would require minute amounts, such as only a few particles.

Clearly, the desire to glean the information on the practice of the ancients that is contained within artifacts was the motivation for developing improvements in the application of Raman scattering spectroscopic techniques. The broader desire by curators and conservators to know more about all of the various colorants and binders in many works will drive progress in the prediction and interpretation of surface-enhanced resonance Raman spectra of organic molecules. Such innovation in science is motivated by our insistence on knowing more about our cultural heritage and is a reflection of the dyers and painters' experimentation from hundreds, even thousands, of years ago.