Article pubs.acs.org/ac
Combined use of Synchrotron Radiation Based Micro-X-ray Fluorescence, Micro-X-ray Diffraction, Micro-X-ray Absorption NearEdge, and Micro-Fourier Transform Infrared Spectroscopies for Revealing an Alternative Degradation Pathway of the Pigment Cadmium Yellow in a Painting by Van Gogh Geert Van der Snickt,*,† Koen Janssens,† Joris Dik,‡ Wout De Nolf,† Frederik Vanmeert,† Jacub Jaroszewicz,† Marine Cotte,§,∥ Gerald Falkenberg,⊥ and Luuk Van der Loeff¶ †
Antwerp X-ray Instrumentation and Imaging Laboratory, Department of Chemistry, University of Antwerp, Universiteitsplein 1, B-2610 Wilrijk, Belgium ‡ Department of Materials Science and Engineering, Delft University of Technology, Mekelweg 2, NL-2628CD Delft, The Netherlands § Laboratoire du Centre de Recherche et de Restauration des Musées de France, CNRS UMR 171, Palais du Louvre, Porte des Lions, 14, Quai François Mitterand, F-75001 Paris, France ∥ European Synchrotron Radiation Facility, Polygone Scientifique Louis Néel, 6, rue Jules Horowitz, F-38000 Grenoble, France ⊥ Deutsches Elektronen-Synchrotron, P06 Beamline, PETRA-III, Notkestrasse 85, D-22607 Hamburg, Germany ¶ Conservation Department, Kröller-Müller Museum, Houtkampweg 6, NL-6731AW Otterlo, The Netherlands ABSTRACT: Over the past years a number of studies have described the instability of the pigment cadmium yellow (CdS). In a previous paper we have shown how cadmium sulfide on paintings by James Ensor oxidizes to CdSO4·H2O. The degradation process gives rise to the fading of the bright yellow color and the formation of disfiguring white crystals that are present on the paint surface in approximately 50 μm sized globular agglomerations. Here, we study cadmium yellow in the painting “Flowers in a blue vase” by Vincent van Gogh. This painting differs from the Ensor case in the fact that (a) a varnish was superimposed onto the degraded paint surface and (b) the CdS paint area is entirely covered with an opaque crust. The latter obscures the yellow color completely and thus presents a seemingly more advanced state of degradation. Analysis of a cross-sectioned and a crushed sample by combining scanning microscopic X-ray diffraction (μ-XRD), microscopic X-ray absorption near-edge spectroscopy (μ-XANES), microscopic X-ray fluorescence (μ-XRF) based chemical state mapping and scanning microscopic Fourier transform infrared (μ-FT-IR) spectrometry allowed unravelling the complex alteration pathway. Although no crystalline CdSO4 compounds were identified on the Van Gogh paint samples, we conclude that the observed degradation was initially caused by oxidation of the original CdS pigment, similar as for the previous Ensor case. However, due to the presence of an overlying varnish containing lead-based driers and oxalate ions, secondary reactions took place. In particular, it appears that upon the photoinduced oxidation of its sulfidic counterion, the Cd2+ ions reprecipitated at the paint/varnish interface after having formed a complex with oxalate ions that themselves are considered to be degradation products of the resin and/or oil in the varnish. The SO42− anions, for their part, found a suitable reaction partner in Pb2+ ions stemming from a dissolved lead-based siccative that was added to the varnish to promote its drying. The resulting opaque anglesite compound in the varnish, in combination with the underlying CdC2O4 layer at the paint/varnish interface, account for the orange-gray crust that is disfiguring the painting on a macroscopic level. In this way, the results presented in this paper demonstrate how, through a judicious combined use of several microanalytical methods with speciation capabilities, many new insights can be obtained from two minute, but highly complex and heterogeneous paint samples.
E
arly modern painters such as Vincent Van Gogh (1853− 1890) and James Ensor (1860−1949) developed an innovative and often colorful style of painting that involved exploring the new pigments and coloring agents that were © 2012 American Chemical Society
Received: July 4, 2012 Accepted: August 30, 2012 Published: August 30, 2012 10221
dx.doi.org/10.1021/ac3015627 | Anal. Chem. 2012, 84, 10221−10228
Analytical Chemistry
Article
produced for the first time in the nascent chemical industry of that period. Contrary to their predecessors who employed a traditional palette of pigments, almost all of which had a proven light-fastness and long-term durability, Van Gogh and contemporaries showed an interest in new, colorful pigments and dyes with dubious longevity. In a previous study,1 we gained insights into the alteration process of the “modern” yellow pigment cadmium sulfide (CdS). A chemical transformation led to the bleaching of the original color in several unvarnished paintings created by Ensor. In particular, the yellow discoloration on Ensor’s “Still life with coffeepot, cabbages and mask” (ca. 1920, XT511/KM105.303, KröllerMüller Museum, Otterlo) was shown to be the result of the oxidation of the sulfidic anion, under the influence of moisture, oxygen, and UV radiation, to a sulfate ion. Being excellently water-soluble, the resulting CdSO4 was leached out of the paint by the humidity in the air, a phenomenon that resulted in a porous and flaky paint layer. Variations in the climate, i.e., recurrent cycles of high and low relative humidity, gave rise to the repeated dissolution and recrystallization of the degradation product which eventually induced the formation of the globular, whitish CdSO4·H2O crystals on the surface. Also, secondary compounds such as (NH4)2·Cd2(SO4)3 and PbSO4 (anglesite) were identified, possibly stemming from atmospheric deposition of small particles on the paint surface and a lead-based component in the paint, respectively. Although “Flowers in a blue vase” (1887, F323/KM107.055, Kröller-Müller Museum, Otterlo) was painted by Vincent Van Gogh in Paris, more than 30 years before Ensor finished “Still life with coffeepot, cabbages and mask”, both paintings entered the Kröller-Müller collection during the first decades of the 20th century. Considering the fact that both pictures have been stored and displayed in a comparable environment during the last 80−90 years, the finding of altered CdS paint on both works was not surprising. Nonetheless, the morphology of the affected paint on “Flowers in a blue vase” is clearly different from the Ensor case. Instead of the discrete whitish globules and thin film composed of CdSO4·H2O that slightly reduces the intensity of the bright yellow color, a thick and heavily cracked gray-orange crust is covering the cadmium yellow in Van Gogh’s flower still life (see Figure 1). The latter disfigures the appearance of the painting to a much greater extent than the more superficial alteration in the Ensor case. In fact, the coherence of the whole paint system seems weakened in the affected areas, as the conservator documented several lacunae. These paint losses allow assessing the color as it was originally meant by the artist. In the areas where the overlying gray degradation crust is missing, the paint layer reveals a bright, warm yellow color (see Figure 1). In addition, the conservators reported that the varnish displayed a particular matte surface in the areas where it covered the affected paint. This observation suggested that the chemical transformations of the paint had also affected the overlying varnish in these specific zones. In view of this problem, the conservators supplied two samples for analysis, stemming from the edge of a lacuna (see Figure 1). The samples were analyzed by a combination of various synchrotron radiation based techniques at the European Synchrotron Radiation Facility (ESRF) in Grenoble, France and the PETRA-III storage ring at the Deutsches ElektronenSynchrotron (DESY) in Hamburg, Germany.
Figure 1. Left: Vincent Van Gogh, “Flowers in a blue vase” from the Collection Kröller-Müller Museum, Otterlo, The Netherlands. The white rectangles indicate the location of the details shown on the right. Right: details of the paint surface. The cadmium paint is severely cracked and covered with an opaque crust, while the chrome-based pigment used for the yellow flowers in the middle retained its original color. The circles indicate the area where sample material was collected and the original yellow color is disclosed. Pictures by Rik Klein Gottink (left) and Luuk Van der Loeff (right).
■
EXPERIMENTAL SECTION Sample A was embedded in an epoxy resin and polished with sand paper in such a way that it became possible to study its cross section. As demonstrated in Figure 2, sample A revealed a semitransparent varnish layer of ca. 50 μm on top of the original yellow layer. The morphology of this paint fragment was first characterized by means of an Olympus optical microscope (OM) and then studied by a number of complementary synchrotron radiation based microscopic analytical techniques, i.e., by a combination of microscopic Xray fluorescence (μ-XRF) and microscopic X-ray diffraction (μXRD) spectroscopy at beamline P06 of the PETRA III storage ring (DESY) and a combination of microscopic X-ray absorption near-edge spectroscopy (μ-XANES) spectroscopy and μ-XRF at beamline ID21 (ESRF). The latter X-ray microscopy end station can operate in the primary energy range from 2.1 to 9.2 keV. A fixed-exit, double-crystal Si(111) monochromator determines the energy of the beam with a resolution ΔE/E of ca. 10−4. μ-XRF and μ-XANES experiments were performed in vacuum in order to minimize air absorption, which is significant for low-energy X-ray fluorescence lines such as S Kα, and in order to avoid contributions to the spectral background due to scattering in air. Either a collimated beam of 0.2 mm diameter or a focused beam of ca. 0.9 × 0.3 μm2 (hor. × ver.), obtained by means of a Fresnel zone plate, was used. During the μ-XANES energy scans, the position of the primary beam was maintained stable within 0.5 μm. The procedure employed for correction of the beam spot motion during energy scans is explained elsewhere.2 The μ-XRF signals were collected in the horizontal plane and perpendicular to the primary beam by means of an HPGe solid-state energydispersive detector. The sample surface was oriented vertically and at an angle of 60° relative to the incident beam. Both the setup and evaluation of the XRF spectra are discussed more in detail elsewhere.3 During the μ-XRF mapping experiments, the 10222
dx.doi.org/10.1021/ac3015627 | Anal. Chem. 2012, 84, 10221−10228
Analytical Chemistry
Article
Figure 2. Summary of the scanning μ-XRF/μ-XRD results obtained at beamline P06 (PETRA III, DESY). OM: optical micrograph of sample A under visual light, after embedding and polishing. The cross section shows that a semitransparent varnish layer of ca. 50 μm was superimposed on a yellow paint layer. A white rectangle indicates the area from which the distribution maps were collected. μ-XRF: elemental maps establishing the distribution of Cd, Pb, and Zn. Map size 100 × 90 μm2, step size 1 × 1.5 μm2, primary energy 18 kV. μ-XRD: species-specific maps of the same sample area, indicating the presence of CdS in the yellow layer and an interfacial CdC2O4 film. CdCO3 was identified in both the interfacial layer and the yellow paint while PbSO4 was found in the varnish and the interfacial film. On the right: one-dimensional diffraction patterns illustrating the identification of (A) CdS, (B) PbSO4, and (C) CdC2O4.
fluorescence signals were generated by employing a monochromatic primary beam at fixed energy (around 2.5 keV at the S K-edge). PyMca was used to fit fluorescence spectra and to separate the different elemental contributions. It was in particular crucial to distinguish the partially overlapping Pb M-lines and S K-lines. This program was employed as batch fitting procedure on each pixel of two-dimensional (2D) maps.4 μ-XANES spectra were acquired by scanning the primary energy around the S K-edge (2.46−2.53 keV) with a step size of 0.2 eV. Prior to the analyses on paint samples, reference XANES spectra were acquired from pure compound powders by using a 200 μm diameter X-ray beam in transmission mode. Finely ground powder was dusted onto adhesive tape and irradiated without other form of preparation. Combined μ-XRF/μ-XRD distribution maps were obtained in the microprobe hutch of the P06 hard X-ray micro/ nanoprobe beamline (PETRA III, DESY, Hamburg) using a primary beam of 18 keV. The energy was selected by means of a Si(111) double-crystal monochromator. The beam was focused to 1.6 × 0.6 μm2 (hor. × ver.) employing a Kirkpatrick−Baez mirror optic. A Keyence optical microscope equipped with a perforated mirror allowed for positioning of the sample. In transmission geometry, diffraction signals were recorded with a 2k × 2k MarCCD camera, with 78 × 78 μm2 pixel size (165 mm diameter, MAR Research, CA, U.S.A.). Xray fluorescence signals were recorded by means of a Si drift detector (SII Nano Technology USA Inc., Northridge, U.S.A.) with a 50 mm2 active area. XRF spectral fitting was performed
using the PyMCA software package4 while the ensuing XRD data was analyzed with XRDUA software.5 Sample B was taken from a different spot, as demonstrated by Figure 1, but showed a composition and alteration similar to sample A. Sample B was not embedded or prepared in order to avoid any contamination of organic media, but crushed in between two diamond cells. Pressing the samples caused some disturbance in the stratification (see Figure 4), but it reduced the sample thickness to a few micrometers. In this way, it was rendered thin enough for IR, thus allowing analysis in transmission. The FT-IR spectromicroscope at ID21 (ESRF) is composed of a Thermo Nicolet Nexus infrared bench associated with an infrared Thermo Continuum microscope and a liquid nitrogen-cooled mercury cadmium telluride detector from Thermo Scientific, with a measuring range of 650−4000 cm−1. More details about the setup can be found elsewhere.6,7 The acquired FT-IR spectra were made of an average of 20 scans, and the spectral resolution was set to 8 cm−1. Distribution maps were collected using an aperture of 8 × 8 μm2 and a step size of 10 μm. Spectra were converted to absorbance using Thermo Omnic 7.1 software (Thermo Scientific Inc.). Raw spectra were processed by 13 point smoothening, baseline-corrected, and normalized using Bruker OPUS software. A new background was taken every 30 min to correct for atmospheric alterations or changes in the beam. 10223
dx.doi.org/10.1021/ac3015627 | Anal. Chem. 2012, 84, 10221−10228
Analytical Chemistry
■
Article
= 1.5 × 10−8) instead of the SO42− ions (KS = 5.3 × 10−1) and therefore give rise to crystalline cadmium oxalate, the most stable salt. In the past, oxalate complexes of various natures have been encountered commonly on outdoor surfaces exposed to weathering, such as stone materials (marble, sandstone, granite, etc.),8,9 exterior wall paintings,10 rock art,11 and even stained glass windows.12 In these cases, the oxalate ions are either formed by an oxidative degradation of organic material or produced as a metabolic byproduct of microbiological organisms such as lichens, algae, or bacteria cultivating on organic material. For both the chemical and biological pathway, the particularly stable, inorganic oxalate crystals result from the reaction of oxalic acid with a substrate rich in Ca, Cu, Fe, and/ or Pb. Although the occurrence of oxalate films is less expected on easel paintings due to the protected indoor environment in which they are conserved, several authors recently reported the presence of opaque oxalate crusts on easel paintings.13−15 In all cases, the authors indicated that the oxalic acid was predominantly stemming from varnish layers, a finding that is constituent with the hypothesis formulated for this Van Gogh case. Surprisingly, the μ-XRF elemental maps demonstrated that Cd was not only found in the paint and the oxalate layer, it appeared to be distributed in a lower concentration throughout the varnish as well. Since a deliberate addition of a cadmium compound to the varnish can be considered as highly improbable, it was assumed that the cadmium in the varnish is stemming from the underlying paint layer. Considering the high solubility of CdSO4 (see Table 1), it is reasonable to accept that the upper part of the degraded paint layer was at least partly dissolved during the application of a varnish (containing solvents) and dispersed throughout the coating. During the drying of a varnish film, the solvent slowly migrates to the surface where it evaporates. During its upward drift, the solvent is expected to take along any dissolved components, which explains the observed distribution of Cd ions throughout the varnish.16,17 This theory is also consistent with the fact that the distribution of Cd in the varnish is most dense near the paint/varnish interface, but that its concentration gradually decreases toward the surface as the Cd2+ ions encounter oxalate reaction partners on their way up. The nature of the Cd in this “diluted” area could not be established by means of μ-XRD, but it is reasonable to assume that it is present as cadmium oxalate as well, but in a concentration below the detection limit (typically >1000 ppm). Alternatively, Cd can be present in an amorphous form. Apart from cadmium oxalate, also CdCO3 was found in the intermediate layer. Several explanations can be put forward to account for the occurrence of this cadmium compound. The carbonate may have been formed as a secondary degradation product following the primary photodegradation of CdS, perhaps by the capture of atmospheric CO2, as suggested by Mass.18 Alternatively, a tertiary process involving the further breakdown of cadmium oxalate into cadmium carbonate cannot be excluded, while finally its presence might also be explained as simply a residue of the original raw material used for the industrial production of CdS, as explained in the previous article.1 This last explanation is given credibility by the fact that, as Figure 2 demonstrates, the presence of CdCO3 is not limited to the intermediate, oxalate layer; it is present in the (original) yellow paint as well. In addition, a combination of several of the above-mentioned pathways is possible as well. We hope that
RESULTS AND DISCUSSION Optical Microscopy. An important difference with the Ensor case is the fact that Van Gogh’s flower still life, at some point, was covered with a varnish, while the previous case, Ensor’s “Still life with coffeepot, cabbages and mask”, had always remained free of a protective coating. On the basis of the observations of the conservators, it was concluded that the varnish coating on Van Gogh’s work was applied after the onset of the CdS degradation process. Moreover, the varnish was most probably applied as part of a conservation treatment, in an attempt to consolidate the flaking CdS paint and/or to protect it from further deterioration. Apart from that, the study of sample A through an optical microscope indicated the presence of a film of
