The degradation of cadmium yellow paint – New evidence from

Feb 1, 2019 - XXXX, XXX, XXX-XXX ... The two paints exhibit trap state emissions that differ in terms of ... In the now-brownish yellow paint, trap st...
0 downloads 0 Views 1MB Size
Subscriber access provided by University of Winnipeg Library

Article

The degradation of cadmium yellow paint – New evidence from photoluminescence studies of trap states in Picasso’s Femme (Époque des “Demoiselles d’Avignon”) Daniela Comelli, Douglas MacLennan, Marta Ghirardello, Alan Phenix, Catherine M Schmidt Patterson, Herant Khanjian, Markus Gross, Gianluca Valentini, Karen Trentelman, and Austin Nevin Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.8b04914 • Publication Date (Web): 01 Feb 2019 Downloaded from http://pubs.acs.org on February 3, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 12 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

The degradation of cadmium yellow paint – New evidence from photoluminescence studies of trap states in Picasso’s Femme (Époque des “Demoiselles d’Avignon”) Daniela Comelli1*, Douglas MacLennan2, Marta Ghirardello1, Alan Phenix2, Catherine Schmidt Patterson2, Herant Khanjian2, Markus Gross3, Gianluca Valentini1, Karen Trentelman2, and Austin Nevin4 1 Politecnico di Milano, Physics Department, Piazza Leonardo da Vinci 32, 20133, Milano, Italy 2 Getty Conservation Institute, Science Department, 1200 Getty Center Drive, Los Angeles, CA 90049, USA 3 Fondation Beyeler, Baselstrasse 77, CH-4125 Riehen/Basel, Switzerland 4 Istituto di Fotonica e Nanotecnologie - Consiglio Nazionale delle Ricerche (IFN-CNR), Piazza Leonardo da Vinci 32, 20133, Milano, Italy * Corresponding author: [email protected]

ABSTRACT: Paints based on cadmium sulfide (CdS) were popular among artists beginning in the mid-19th century. Some paint formulations are prone to degrade, discoloring and disfiguring paintings where they have been used. Pablo Picasso’s Femme (Époque des “Demoiselles d’Avignon”) (1907) includes two commercial formulations of CdS: one is visibly degraded and now appears brownish yellow, while the other appears relatively intact, and vibrant yellow. This observation inspired the study reported here of the photoluminescence emission from trap states of the two CdS paints, complemented by data from multispectral imaging, X-ray fluorescence spectroscopy, micro-FTIR, and SEM-EDS. The two paints exhibit trap state emissions that differ in terms of spectrum, intensity, and decay kinetic. In the now-brownish yellow paint, trap state emission is favored with respect to near bandedge optical recombination. This observation suggests a higher density of surface defects in the now-brownish yellow paint that promotes the surface reactivity of CdS particles and the subsequent paint degradation. CdS is a semiconductor, and surface defects in semiconductors can trap free charge carriers; this interaction becomes stronger at reduced particle size or, equivalently, with increased surface-to-volume ratio. Here, we speculate that the strong trap state emission in the now-brownish cadmium yellow paint is linked to the presence of CdS particles with a nanocrystalline phase, possibly resulting from a low degree of calcination during pigment synthesis. Taken together, the results presented here demonstrate how photoluminescence studies can probe surface defects in CdS paints and lead to an improved understanding of their complex degradation mechanisms.

Cadmium sulfide (CdS) is an important II-VI semiconductor characterized by excellent thermal and chemical stability and strong optical absorption. It is widely used for the production of solar cells, photodetectors, light emitting diodes, and lasers,1 and for the synthesis of a variety of nanostructures and quantum dots.2,3 However, its primary use is as a pigment,5 widely employed in the paint, plastic, textile, ceramics, and glass industries. The development of CdS as a pigment started in the mid1840s, initially as a single hue of yellow. By the beginning of the 1920s it was available in a broad range of colors, including both cadmium yellows and cadmium reds, based on solid solutions of CdS with ZnS and CdSe, respectively.5 With its brilliant hue and high tinting strength, paints containing cadmium yellow (here used to refer specifically to chemically pure CdS) was rapidly adopted by important painters including van Gogh, Picasso, Seurat, Matisse, Munch, and Ensor.6-16

Artists and paint producers quickly recognized that poor quality cadmium yellow tended to lose its color upon exposure to light.5 Several other characteristic indicators of the degradation of CdS have been reported, including the formation of superficial opaque crusts, white globules, and a tendency towards embrittlement.6-9,11,14,15 CdS produced between the late 19th and early 20th centuries appears to be particularly prone to photo-oxidative degradation as has been documented in a number of paintings dating from about 1880 to 1920.11 In some cases, the degradation appears to have become apparent quite early in the lifetime of the painting, becoming visible after only about 20 years.5,11 While these paintings have been exposed to different environmental conditions it has been speculated that the degradation of CdS may be related to the crystalline properties of the pigment and to the presence of auxiliary compounds in the paint, which may play a stabilizing role.5,11 Recent research on a set of

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

cadmium yellow pigments synthesized between 1850 and 1950 has highlighted the heterogeneity of different formulations in terms of crystal structure, crystal size, and presence of impurities, all features that can be related to the many synthesis processes and recipes available in that time period.17 Due to the non-universal, but occasionally extreme, degradation observed in paintings, the chemistry of CdS-based paints has been the focus of intensive analysis. The degradation pathway of CdS (described in detail in Supporting Information) involves the direct photo-oxidation of cadmium sulfide to cadmium sulfate, with water acting as an important trigger in the mechanism.18 Further reactions lead to the development of other chemical species, including carbonates and oxalates.6,7,19 In degraded historical cadmium paints these degradation products (sulfates, oxalates, and carbonates) have been detected with synchrotron-based analytical techniques (e.g., μFTIR, μXANES).6,7,11,13,14,15 The presence of water in the paint film can additionally promote the formation of H2SO4, leading to acid hydrolysis of the paint binding medium (often a natural drying oil) and the production of cadmium oxalate.19 CdS exhibits characteristic photoluminesce (PL), as is typical of semiconductors. In intact pigments based on hexagonal CdS (wurtzite), it is possible to detect both the fast (ps) near bandedge (NBE) emission at 2.4 eV/516 nm, as well as two longerlived (µs) emissions in the near-infrared (centered around 1.57 eV/790 nm and 1.26 eV/980 nm).17,20,21 The relatively long-lived microsecond emission is ascribed to the presence of two deep trap states (TS) located within the bandgap, whose origin – investigated by both computational and experimental methods – appears to be related to surface sulfur and cadmium vacancies that act as electron and hole traps, respectively.22-24 In comparison to the expected PL behavior in well-preserved CdS, historical cadmium yellow oil paints can exhibit TS emission that is slightly shifted in wavelength17, whereas the NBE emission is often masked by the emission of the aged binder21. In addition, degraded cadmium yellow paints can exhibit a distinct reddish-orange PL when examined under ultraviolet illumination.9,14 Presented here are the results of a study of Pablo Picasso’s Femme (Époque des "Demoiselles d'Avignon) (1907, oil on canvas, 119×93.5 cm, Fondation Beyeler, Riehen/Basel, Switzerland, Inv. 65.2), hereafter Femme. The painting contains two different CdS yellow paints that have undergone differential degradation. One yellow appears vibrant and relatively intact, while the second appears to have discolored to a brownish yellow, presumably due to degradation. The two yellow paints will be referred to throughout as ‘vibrant yellow’ and ‘now-brownish yellow’, respectively. The chemical composition and physical structure of these paints were studied using both in-situ and sample-based analyses prior to examination by PL. Results were then correlated to the trap state PL emission behavior on both the macro- and micro scale in order to determine whether there was a correlation between TS emission properties and the degree of paint degradation.

Page 2 of 12

■ MATERIALS AND METHODS Femme. After his lyrical Rose Period, Picasso’s work began to reincorporate a broad range of colors, evident in a number of works he completed in 1907 (including, for example, Buste de femme nu, Etude pour “Les demoiselles d’Avignon” [Sammlung Berggruen] and Buste de femme [private collection] as well as Femme). His work additionally took on influences from African art forms, as well as from Cézanne and Rousseau, subsequently developing what would later be called Cubism. The pinnacle of Picasso’s work in this mode is the masterpiece Les Demoiselles d’Avignon (1907, Museum of Modern Art, New York). Femme (Figure 1a) shows a similar motif, consisting of an abstract female figure with arms raised; it is thought that this painting may have been a study for Les Demoiselles d’Avignon. The painting has undergone conservation at least twice prior to the current scientific examination: once before 1953 and again between 1970 and 1989.25 A 1965 color transparency of the painting (Figure 1b) shows the painting with a brighter yellow background than is now visible (c.f. Figure 1a). At the time of the second conservation intervention, several degradation phenomena were already apparent, including a noticeable fading and brownish yellow discoloration of the background yellow paint. Today, the now-brownish yellow areas above and below the figure also exhibit a strong characteristic reddish-orange PL under UV radiation (Figure 1c), a phenomenon that has been associated with CdS degradation in works by Ensor and Matisse.9,14 In-situ analysis methods. Non-invasive X-ray fluorescence (XRF) spectroscopy was used to examine the elemental composition of the two yellow paints, using a non-contact Bruker ARTAX μXRF spectrometer (W-tube, 50 kV, 500 μA, 60 second timed assays, air path, no filtration, ~650 µm Ø spot size). Multispectral imaging was used to characterize the major colorants in the painting. Thirteen narrowband images (50 nm intervals, 400-1000 nm) were acquired using a Retiga 4000R camera (2048×2048 pixels) with a UV-Vis-NIR corrected 60 mm apochromatic macro lens (Coastal Optics). Dark current and flat field correction, and conversion to apparent reflection was carried out using the Nip2 graphical interface (libvips).26,27 The corrected image slices were registered using the TurboReg plugin for ImageJ.28,29 A multispectral image cube with dimensions 2048×2018×13 was generated using ENVI (v. 4.3). Image classification was also performed in ENVI by employing the spectral angle mapper (SAM) algorithm. The algorithm determines the spectral similarity to a reference spectrum by treating spectra as vectors and calculating the angle between them.30 Pixels with similar spectral responses are grouped and displayed in a series of false-color images. Time-gated PL imaging was used to measure the optical emission occurring at the microsecond timescale (where trap states are likely to be observed) in yellow-painted areas of the painting.31 The system is based on a ns laser excitation source (FTSS 355-50 CryLas GmbH, 355 nm, 1.0 ns pulse, 100 Hz repetition rate) and a fast time-gated intensified CCD camera

2 ACS Paragon Plus Environment

Page 3 of 12 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Figure 1. (a) Color image of Picasso’s Femme; (b) image of the painting taken in 1965; (c) UV-induced visible fluorescence (note: dark grey spots on the white background indicate modern retouching); (d) false color SAM composite image showing the distribution of the now-brownish yellow (turquoise) and vibrant yellow (red) paints (C9546-03, Hamamatsu Photonics) with a gate width adjustable from 3 ns to the continuous mode, and a spectral sensitivity from 350-850 nm. In this work the laser beam was enlarged to 20 cm in diameter, achieving a typical excitation fluence per pulse of 0.03 μJ/cm2 (equivalent to an average power density of 3 μW/cm2). Time-gated images of TS emission were recorded by employing a gate width of 1 μs, set at a delay of 1 μs after laser excitation. Steady-state laser-induced PL spectroscopy was performed at selected points on the painting.32 The system employs a compact spectrometer (TM-CCD C10083CA-2100, Hamamatsu Photonics), with spectral sensitivity from 3501000 nm, and UV-laser excitation with a 1 mm Ø spot size. To study the dependence of the emission on the excitation intensity the system was equipped with a variable neutral density filter in the excitation path and with two laser sources, which work in different modes but have the same average power density: a pulsed Q-switched laser (FTSS 355-50 CryLas, 10 μJ/mm2 fluence/pulse, 1 mW/mm2 average power density, 100 Hz repetition rate) and a CW laser (DL–375-015, CrystaLaser, 1 mW/mm2 power density). It is worth noting that the pulsed laser provides an instantaneous excitation intensity about 7 orders of magnitude higher than the CW laser. The maximum power density employed in this study is below the damage threshold expected in paints and no visual damage was observed in the analyzed paints. Microsamples and microscopy methods. Two multi-layered paint samples were prepared for analysis in cross-section by embedment in Technovit LC2000 resin (a light-curing acrylic copolymer) followed by dry hand polishing. Samples were selected from representative areas within the vibrant and the now-brownish yellow paints (samples 1 and 2, respectively; sampling sites are shown in the Supporting Information). Optical microscopy was performed with a Leica DM4000 equipped with a Flex camera (Diagnostic Instruments) under visible light (with crossed polarizing filters) and under UV excitation with detection of visible fluorescence. Scanning electron microscopy with energy dispersive spectroscopy (SEM-EDS) was performed using FEI-Philips XL30 ESEM-FEG working in H2O mode (10 mm working distance, 20 kV, 1 torr H2O, Oxford INCA software suite).

MicroFTIR spectroscopy was performed on representative sample particles placed on a diamond window, flattened with a metal roller and analyzed with a Bruker Hyperion 3000 FTIR microscope (transmission mode, ~100×100 µm2 aperture, 15× objective, 64 scans at 4 cm-1 resolution). The PL properties of the microsamples were probed with a TRPL (time-resolved PL) microscopy system, fully described elsewhere.33 The system is based on the same pulsed laser excitation and time-gated detector employed for in situ timegated PL imaging. The image detector is coupled to an epifluorescence microscope equipped with 10 bandpass filters centered between 400-850 nm in 50 nm intervals. The microscope employs a 50× (NA=0.5) refractive objective, providing emission imaging from a circular field of view 300 μm Ø with a spatial resolution of 0.6 μm. By using a timegated approach it is possible to reconstruct multi-spectral images of the PL emission at the microsecond timescale. For lifetime-resolved measurements, PL images in selected spectral regions are detected at different delays with respect to the pulsed excitation. The emission decay kinetics of selected areas of the field of view are fitted with a three-exponential decay model. The average lifetime, , is calculated as the mean of the lifetimes weighted over the relative amplitude originating from each decay pathway.

■ RESULTS The results of the characterization of the vibrant and nowbrownish yellow paints in Femme, performed using a range of non-invasive and sample-based analytical techniques, are summarized in Table S1 in Supporting Information. Both paints were determined to be cadmium sulfide-based. Paint chemistry and degradation. Figure 1d is a false color composite of the painting generated by multispectral imaging, which shows the distribution of the vibrant (shown in red) and now-brownish (in turquoise) cadmium yellow paints across the entire painting. This composite image indicates that the now-brownish yellow paint is not only present in the background, but also in localized areas within the figure itself. In contrast, the vibrant yellow is only present within the figure, in the chest, arms, and in the face. XRF spectroscopy (not shown) carried out at numerous points across both yellow areas revealed the presence of cadmium, barium, and lead,

3 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

suggesting the use of cadmium yellow (CdS), barium white (barium sulfate, BaSO4) and lead white (basic lead carbonate, 2PbCO3·Pb(OH)2) in the paints and/or the ground layer, all materials commonly found in paints used by Picasso.34-36

Page 4 of 12

enrichment of oxalates and carbonates, as well as a depletion of fatty acid (stemming from the oil binding medium)19 in comparison to the more intact vibrant yellow (sample 1) (see also Figure S1, Supporting Information). Photoluminescence properties. The PL properties of the vibrant and now-brownish yellow paints were studied at both the macro and sub-micrometer scale. In addition, we examined a small area of the now-brownish yellow that was protected from light by the application of opaque paper tape along the upper edge of the painting, and thus avoiding degradation and retaining the brighter color shown in Figure 2b (the location of the covered-background yellow is shown in Figure S3, Supporting Information). Hereafter this area will be referred to as the ‘covered-background paint’. This covered-background paint fortuitously provided a valuable opportunity to examine the now-brownish background yellow paint at two different stages of degradation.

Figure 2. (a) and (c): (left) visible light (with crossed polarizing filters) and (right) UV-induced visible fluorescence images of samples 1 and 2, respectively. Note that the ground is visible below the yellow dashed line. (b) and (d): SEM-EDS maps of samples 1 and 2, respectively, with distributions of cadmium (red), barium (green), and lead (blue) overlaid onto the BSE.

The multi-layered samples were examined by light microscopy and SEM-EDS to better characterize the stratigraphy and chemical compositions of the individual layers. As shown in Figure 2, both samples consist of two layers: a lower white ground layer (inferred to be lead carbonate with a trace amount of alumina [Al2O3]) and an upper, pigmented layer. In the vibrant yellow the upper paint layer (sample 1, top row in Figure 2) was inferred to be composed primarily of lead carbonate, cadmium sulfide, and barium sulfate based on EDS data (Figure 2b). Orange-red particles present in very trace quantities throughout this upper layer contain Hg and S, suggesting the presence of the red pigment vermilion (HgS) as well. While the source of this mixture cannot be determined from these data alone, it should be noted that several shades of commercial CdS-based pigments were available at the time, and some of them may have contained phases such as vermilion and lead white. In the upper paint layer of the now-brownish yellow (sample 2, bottom row in Figure 2) EDS analysis indicates the presence of cadmium sulfide and barium sulfate only (Figure 2d). In contrast to sample 1, lead was not detected in the pigmented layer of sample 2. Additionally, a thin (~5 µm) discontinuous, cadmium-rich surface crust is also visible at the top of the upper pigmented layer of sample 2, possibly the result of redeposition of cadmium depleted from deeper inside the layer. In agreement with results from UV-induced fluorescence imaging of the entire painting, a strong reddishorange emission is visible at the paint surface of sample 2, that is less intense as it extends into the bulk of the sample (Figure 2c). Additional differences between the vibrant and now-brownish yellow paint samples were revealed by µFTIR analysis. The upper pigmented layer in sample 2 was found to have an

Figure 3. Time-gated PL images of the μs emission (1 μs delay after pulsed laser excitation; 1 μs temporal gate width) detected in three areas of the painting containing both the vibrant and nowbrownish yellows. The now-brownish yellow areas show an intense emission ascribed to radiative recombination from trap states.

The results of time-gated PL imaging of regions containing both the vibrant and now-brownish yellows are shown in Figure 3. Microsecond timescale emission, indicative of optical recombination from TS in inorganic materials,37 is detected strongly throughout the now-brownish yellow paint. This emission is significantly less intense (by a factor of 30) in the vibrant yellow paint, and, therefore, is not visible in figure 3. The covered-background paint displays a detectable TS emission, but with an intensity reduced by a factor around 2.5 with respect to the nearby unprotected and now-brownish paint (detail shown in Figure S3 in Supporting Information). PL emission spectra were collected from representative points of the painting using Q-switched and steady state (CW) laser excitation at different excitation fluence. In semiconductor materials the method is employed to probe different recombination paths.37 In CdS-based materials, it is well known that the mechanism for carrier recombination is

4 ACS Paragon Plus Environment

Page 5 of 12 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

strongly influenced by electron trapping in TS21,38. When employing low-fluence excitation, emission predominantly occurs through radiative recombination from TS. Instead, when the excitation intensity (i.e., photon flux) is increased, TS are filled, and the density of available empty TS decreases. When the available TS are saturated, electrons newly promoted to the conduction band most likely relax through other available recombination paths with faster lifetimes (picosecond or nanosecond timescale).

Figure 4 – In-situ steady-state PL spectra acquired at different excitation intensities of (top) the vibrant yellow paint, (middle) the now-brownish yellow paint and (bottom) the covered background yellow paint. For better visibility and easier comparison, spectra are shown following normalization to maximum intensity.

In this case, the studied paints’ response to variable excitation fluence varies between the vibrant yellow and the nowbrownish yellow (Figure 4). Under high-fluence nanosecond pulses (10 μJ/mm2 equivalent to 1 mW/mm2 mean power density – blue traces in Figure 4) the emission peaks around 550-570 nm for all the three paints, with the now-brownish paint and the covered-background paint showing broader spectral emissions. In contrast, under low-fluence the emission shifts to another band in all three yellows, peaked at longer wavelengths and optimally probed by CW excitation at the lowest power

density (20 μW/mm2 – green traces in Figure 4). The new band peaked at 1.42 eV/870 nm in the vibrant yellow paint (Figure 4-top), at 1.88 eV/660 nm for the now-brownish yellow paint (Figure 4-middle), and at 1.72 eV/720 nm for the covered-background paint (Figure 4-bottom). Interestingly, in the now-brownish yellow paint and the covered-background paint these higher wavelength bands display similar spectral shape, though they peak at different wavelengths. We finally note that PL spectra excited with intermediate excitation fluence (red and yellow traces in Figure 4) have an intermediary behaviour with emissions falling between the two cases discussed in detail. The emission bands observed from high-fluence pulsed excitation (i.e., those between 500-600 nm) are ascribed to fast recombination paths, which do not exhibit saturation phenomena. The emission peaks are red-shifted with respect to the band-edge of hexagonal CdS (expected at 2.4 eV/516 nm), have a broad spectral emission profile, and agree with published PL studies on aged cadmium yellow oil paints21. Based on this, we speculate that these emissions mainly originate from fluorophores in the aged oil binder39, whose blue-green emission is spectrally filtered by the absorption spectrum of the yellow paints,40 with a possible further contribution from the NBE radiative recombination of CdS. In contrast, the emission occurring at longer wavelengths (i.e., those peaked between 660-870 nm) and resulting from excitation using the lower fluence laser excitation is attributed to optical recombination from TS for all three areas examined, likely resulting from crystal defects. The observed differences in the peak position of the TS emission band between the three areas indicates that the three CdS paints differ in terms of nature or density of crystal defects. Figure 5 shows the results of microscale PL measurements performed on the samples removed from the vibrant- and nowbrownish yellow paints (no sample of the covered-background yellow was taken). Time-gated emissions occurring on the microsecond timescale (Figure 5a) show that the nowbrownish yellow is characterized by a TS emission centered around 650-700 nm, whereas the vibrant yellow paint is characterized by a peak emission centered beyond 800 nm, in good agreement with the TS emission profiles using low fluence CW excitation. In addition, TRPL microscopy shows that the TS recombination emission lifetime of the vibrant and now-brownish paints are different: ~ 1.3 µs, and ~ 0.6 µs, respectively (Figure 5b). On the basis of the PL studies, we propose that the reddishorange optical emission observed in the now-brownish paints of Picasso’s Femme, and similarly in other degraded cadmium yellow paints,9,14 is ascribed to TS radiative recombination peaked around 660 nm and displaying an emission lifetime close to 0.6 µs. The PL emission profiles and decay kinetics of other chemical species present in the yellow paints from Femme were also measured in order to evaluate the possibility that they contribute to optical emission from TS, as was speculated in previous research.9,14 The results are summarized in Table S2 and Figure S4 in Supporting Information. Both cadmium carbonate and cadmium sulfate are poorly luminescent materials. Barium sulfate has an intense TS emission, peaked at 650 nm, ascribed to substitutional ions,41 but the mean

5 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

lifetime is around 320 µs, far exceeding that observed in any of the Picasso samples. Similarly, both of the mineral phases of lead white, (cerussite, PbCO3, and hydrocerussite, 2PbCO3·Pb(OH)2), have a TS emission peaking around 595 nm with a mean lifetime of >15 µs, again, much longer than the lifetime decay measured for either cadmium yellow in Femme.

Figure 5. PL emission profiles of the now-brownish (red trace) and vibrant (blue trace) yellow paint films: (a) normalized PL time-gated spectra (0.2 μs delay, 10 μs gate width). To facilitate comparison, the spectrum of the vibrant yellow paint has been scaled 25×. (b) PL decay-kinetics detected in the spectral band of maximum emission (830-870 nm for the vibrant yellow paint and 630-670 nm for the now-brownish yellow paint). The mean lifetime of the decay-kinetic profile is calculated from a three-exponential decay model.

Page 6 of 12

yellow paint. The covered-background paint was protected from these effects; it provides a snap-shot of how the nowbrownish paint may have once appeared. The combination of in situ and sample-based PL studies suggest that the TS emission from both the vibrant and nowbrownish yellow paints originates from crystal defects (i.e. cadmium and/or sulfur vacancies) in the CdS itself, as opposed to other species present. Further, the more intense TS emission behavior observed in the now-brownish yellow paint suggests a higher density of crystal defects in the degraded nowbrownish paint relative to the vibrant, more intact, yellow paint. This intense TS emission is a strong indication of high surface reactivity, characteristic of semiconductor nanoparticles,42 which have PL features strongly dependent on their surface states and defects. In nano-scale samples, a large fraction of atoms are located at or near the surface, forming dangling bonds that act as trap states for excited electrons.43 In particular, when considering nanometer-sized CdS quantum dots, it has been demonstrated that the TS emission increases in intensity and shifts to shorter wavelengths as the particle size decreases.24 Here, in a similar way, we have shown that the TS emission from the now-brownish yellow paint is more intense and occurs at shorter wavelengths than what was measured in the vibrant yellow paint.

■ DISCUSSION

Figure 6. A schematic view of the vibrant, intact, yellow (left), the covered-background yellow (middle), and the now-brownish degraded yellow (right) paints found on Picasso’s Femme in terms of: (top) possible CdS crystal grain size and surface reactivity, with straight black lines depicting dangling bonds at the crystal surface; (bottom) radiative recombination paths as NBE (1) and TS (2) recombination. TSs characterized by a higher density of levels are depicted with darker grey shades.

On the basis of the identification of distinctly different chemical compositions in the two yellow paints - one which includes lead white and very trace vermilion and one which does not - we infer that Picasso used two different commercially-produced cadmium-based yellow paints in Femme. The physical characteristics of the now-brownish yellow paint, enriched in sulfate, carbonate, and oxalate species, indicate it has degraded compared to the vibrant yellow paint. Since they occur together on the same painting, it can be reasonably assumed that both yellow paints have been exposed to the same environmental effects over the life of the painting, and thus the observed differential deterioration is likely related to the chemical reactivity of each cadmium

Therefore, we speculate that the now-brownish background yellow paint contains nanocrystalline pigment particles, while the vibrant yellow paint is composed of larger CdS grains. The smaller grains would be expected to result in a higher density of reactive dangling bonds at the particle surfaces, influencing the CdS chemical reactivity overall (i.e., sensitivity to photooxidative effects), and shifting the PL properties. The CdS in the covered-background paint (similarly susceptible to photo-activation as the exposed, now-brownish yellow) has been protected from light. As a result it has fewer dangling bonds, and exhibits PL properties intermediate between the vibrant yellow and exposed, now-brownish yellow paints. Future exposure to light would create further surface defects

6 ACS Paragon Plus Environment

Page 7 of 12 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

and result in the stronger TS emissions, and additional shifting to higher energies, as observed in the nearby degraded nowbrownish paint (Figure 6). That pigments containing two different sizes of CdS particles would have been commercially available is supported by the findings of previous studies, which suggest that a subset of CdS pigments produced in the late 19th to early 20th centuries were left un-calcined after synthesis, a procedure that would lead to reactive nanocrystalline CdS grains.8,12 It is worth noting that in Edvard Munch’s The Scream (c.1910, Munch Museum, Tøyen, Oslo, Norway) it has been shown that an altered cadmium yellow paint consists of small (