Figure 1: Fluorescence visible image of a small section removed from a painting.

New capabilities such as advanced detectors and optics are improving the use of infrared and Raman spectroscopy in the chemical analysis and validation of art objects. 

The analysis of art objects is conducted for a variety of purposes, with authentication being the most visible. Before its recent restoration, the Salvator Mundi was thought to be one of a number of copies of the long-lost Leonardo da Vinci painting. It was not until layers of discolored varnish and overpaint were removed that the extraordinary paint handling and surface effects were revealed. A complete authentication process involves the analysis of provenance, technique and historical background, coupled with the application of modern scientific methods. Forgeries are common in today’s market and are increasingly challenging to detect by the use of traditional methods only. An equally important purpose is to gain an understanding of the chemical content for conservation purposes. For example, if a lacquer has been applied to a painting that has a different coefficient of thermal expansion than the paint initially applied by the artist, cracking and other degradation of the painting can occur. The analysis can also be important for the continued preservation of works of art. Understanding what wavelengths of light can be damaging to an object can provide significant guidance to the best environment for storing and displaying the object. For example, through the utilization of atomic force microscopy and Raman spectroscopy, daguerreotype photographic plates have been found to degrade significantly under ultraviolet illumination. As it is highly desired to remove as little material as possible from the work in question, microanalysis systems are best employed.

Figure 2: Representative infrared spectra collected from the paint sample shown in Figure 1. The top spectrum is from the tan layer, the middle spectrum from the yellow particle and the bottom layer from the mounting media (appears black in the visible image). The important task of analyzing works of art is usually one that involves the employment of an array of techniques including optical microscopy, X-ray diffraction (XRD), X-ray fluorescence (XRF) and scanning electron microscopy (SEM), as well as infrared (IR) and Raman spectroscopy. Essentially, the task of characterization can be divided into two areas of interest—the physical properties and the chemical properties of the sample. Optical microscopy, XRD, XRF and SEM yield information pertaining to the physical properties of the sample, while IR and Raman provide the complementary chemical information. Because IR and Raman spectroscopy access fundamental modes of vibration, they are highly specific for identification purposes. The analysis is furthered by the existence of extensive chemical reference databases. Ideally, the spectrum is collected and searched over the relevant databases for identification purposes. If a match is not found, then a careful examination between absorption band positions and intensities must be conducted to identify the compound in question.

Art analysis

Figure 1 shows the fluorescence image of a sample removed from a painting.  The size of the paint sample should be as small as possible, while being substantial enough for analysis, typically 100 x 100 microns. Many pigments undergo fluorescence upon UV illumination, thereby resolving different layers. The corresponding infrared spectra were collected using a Hyperion infrared microscope (Bruker Optics, Inc.) and are shown in Figure 2. The spectra were collected using an ATR objective that has a contact area of 100 microns. A field aperture located in a conjugate image plane defines the area of analysis. Because a germanium crystal was used as the internal reflection element, the spatial resolution was improved to 4x smaller than the wavelength of light. For the spectra shown in Figure 2, the remote aperture was set to 24 x 24 microns, yielding 6 x 6 microns at the sample. This was sufficient to resolve the suspicious-looking yellow particle. Visually, the yellow particle looked polymeric, which would be out of place for a work of art thought to be more than 200 years old. The infrared spectrum clearly indicates that the particle was not polymeric by the lack of bands associated with polymers.

Figure 3: Quartz sample with gaseous inclusions.

The downside to ATR analysis is that there is contact between the ATR crystal and the sample. It is also possible to analyze art by direct reflectance. In this case, the infrared light is bounced off the surface of the object. If the object reflectance is greater than ~4 to 6%, then the dispersion off the surface is measured. The absorbance spectrum can then be readily obtained through data processing. If the sample is less reflective, then a diffuse reflectance result is obtained. For reflectance measurements, it is best to use a high numerical aperture objective for optimal collection of the resultant light. Reflectance data acquisitions have the advantage of being non-contact, but the signal is weaker than ATR and requires post-run data manipulation. Another important consideration is whether to collect information from a single point or an area. Modern infrared microscopes allow rapid area imaging in transmission, reflectance and ATR modes.

Figure 3 shows the visual image of a quartz sample on the top left and the infrared image is shown below. The quartz sample has gaseous inclusions that appear as clear thin lines, and upon identification can provide insight into the conditions present during formation. The quartz object was polished to a thickness of 25 microns. The infrared image was collected in transmission in 2 min. Each point on the image is composed of an infrared spectrum that spans the middle infrared. The blue spectrum is a spectrum extracted from the bulk material and the red spectrum is from air pockets trapped in the quartz. It is readily determined that the inclusions contain carbon dioxide and hydrocarbon molecules. Area imaging allows not only the detection and identification of components, but an easily understood assessment of the distribution as well. The use of a full-field infrared imaging detector allows for diffraction-limited spatial resolution.

Figure 4: Visible image of pen ink crossing (top left), Raman chemical images of paper (1088 cm-1 integration) (upper right), z-grip pen (1647 cm-1 integration) (lower left), and image of the ball point pen (1146 cm-1 integration) (lower right).Raman spectroscopy has been used for many years in the art and forensic communities for the identification of materials that have a weak infrared signal or characteristic bands below the cutoff of typical infrared detectors.  Photoconductive mercury-cadmium-telluride (MCT) detectors that are typically employed have a long wavelength cutoff of about 600 cm-1. Modern Raman microscopes routinely allow the detection and identification of inorganic pigments that frequently have spectral signatures down to ~50 cm-1. Because Raman spectroscopy, like infrared spectroscopy, accesses fundamental modes of vibration, it is highly specific for chemical identification. This can be vital in the analysis of pigments and dyes, which frequently have important vibrations below this wavelength. It is worth noting that Raman analysis is non-destructive and allows in-situ analysis even when applied to investigations in aqueous solutions.

Documentation analysis

Document analysis can be broken down into a few categories of investigation—authentication, detection of alteration and determination of origin. The detection and characterization of ink applied after the creation of a document is important in determining whether a document has been altered. Confocal Raman microscopy can be applied such that a depth profile can be performed on the ink in question to determine which was applied first. The depth resolution for confocal Raman microscopy is defined by the Rayleigh criterion and is about one micron. If ink is applied first and allowed to dry, subsequent application of ink would sit on top of the earlier deposited ink. If an ink applied later comprises different pigment, Raman would readily differentiate between the inks. Figure 4 shows a micrograph of an ink crossing from a document containing two different inks and its corresponding Raman images. Figure 5 shows the raw Raman spectra collected from the ball point pen (top) and the z-grip pen (bottom). The Raman spectra were collected using a Senterra Raman microscope (Bruker Optics, Inc.) at 532 nm with a 1 sec integration in the confocal mode of operation at a resolution of ~3 cm-1. The Raman microscope was mounted on a z-stage to allow access to large samples. The Raman spectra shown have undergone removal of the fluorescence by utilization of a concave baseline correction method. Prior to fluorescence removal, the overlaying Raman spectra are barely observable. The 1648 cm-1 band was integrated for the z-grip image and the 1146 cm-1 band for the ball point pen image. The ball point pen was readily identified as methyl violet and the z-grip ink as methyl violet with an additive. It is clearly evident that the z-grip pen was applied over the ball point pen.Figure 5: Raw Raman spectra of z-grip pen (top, offset) and ball point pen (bottom) collected at 532 nm.

We can extend the investigation further by bringing the document through the field of focus to depth resolve each of the inks at the junction point, as shown for the z-grip pen in Figure 6. The depth profile was conducted by stepping in 1-micron steps from the top surface focus position into the sample for a total of 200 microns. As expected, the spectrum at the interface was composed of a combination of the spectra representative of both inks. This example demonstrates the facility of confocal Raman microscopy for document analysis, so long as the fluorescence can be effectively removed.

Another document was examined with surface enhance Raman spectroscopy (SERS) to determine if the fluorescence could be effectively quenched while achieving a significant signal enhancement. This document contained phthtalo blue and methyl violet inks. The 488 nm spectrum showed some peaks of both over a strongly fluorescent background. Both the paper and methyl violet ink contributed to the fluorescence signal. Using 785 nm excitation did not improve the signal much because, as the fluorescence of methyl violet decreases, phthalo blue becomes slightly fluorescent. In this case, a droplet of silver colloid suspension—prepared using a method where microwave-assisted reduction is employed—was deposited onto the document using a piezo device prior to analysis. The silver colloid droplet size was about 20 microns in diameter. Figure 7 shows the Raman spectrum without correction (top) and the SERS result (bottom). It is clear that the fluorescence was effectively quenched and a very high quality SERS Raman obtained. Figure 6: Depth profile of z-grip pen as a function of depth. This striking example demonstrates the promise of using SERS microscopy for the detection of very small amounts of dyes and pigments, even on fluorescing substrates such as paper or skin.

In conclusion, while infrared and Raman microscopy have been employed for many years in the field, new capabilities, such as the employment of FPA detectors and ATR immersion optics in the infrared and confocal depth profiling, fluorescence reduction and SERS for Raman, have dramatically improved the ability of infrared and Raman microscopes to be used in the increasingly challenging area of art analysis. Lastly, infrared and Raman microscopes are readily available with user-friendly software interfaces and self-validation capabilities that ensure good instrument performance.