Vibrational Spectroscopy 65 (2013) 1–11 Contents lists available at SciVerse ScienceDirect Vibrational Spectroscopy journal homepage: www.elsevier.com/locate/vibspec A joint application of ATR-FTIR and SEM imaging with high spatial resolution: Identification and distribution of painting materials and their degradation products in paint cross sections Zofia Kaszowska a,∗ , Kamilla Malek b , Magdalena Pańczyk c , Anna Mikołajska a a Faculty of Conservation and Restoration of Works of Art, Jan Matejko Academy of Fine Arts in Krakow, 27-29 Lea Street, 30-052 Krakow, Poland Faculty of Chemistry, Jagiellonian University, 3 Ingardena Street, 30-060 Krakow, Poland c Polish Geological Institute – National Research Institute in Warsaw, 4 Rakowicka Street, 00-975 Warszawa, Poland b a r t i c l e i n f o Article history: Received 10 August 2012 Received in revised form 23 November 2012 Accepted 24 November 2012 Available online 3 December 2012 Keywords: ATR-FTIR imaging SEM-EDX Paint cross sections a b s t r a c t A series of paint cross sections from an oil painting are studied by attenuated total reflection in conjunction with Fourier transform infrared spectroscopy (ATR-FTIR) and scanning electron microscopy coupled with energy dispersive X-ray spectroscopy (SEM-EDX). The imaging modes of both methods were employed to show their potential in the detection of various painting materials and their distribution in the paint cross sections. The goal of this work is to evaluate FTIR and SEM-EDX spectroscopy in order to understand the limitations and strengths of both approaches. It has been revealed that both techniques are complementary in the identification of pigments, extenders and binding media used by an artist. FTIR spectroscopy is also a powerful tool in the studies on degradation products that are formed due to ageing or deterioration of works of art. We attempted to identify such secondary products present in the paint cross section. Cadmium oxalate and a high concentration of zinc palmitate/stearate were detected for the first time with the use of ATR-FTIR imaging technique. These results can complement studies on the conservation issue and provide an insight into understanding the mechanisms of chemical processes that appear in art works. Crown Copyright © 2012 Published by Elsevier B.V. All rights reserved. 1. Introduction For many years, Fourier transform infrared spectroscopy (FTIR) and scanning electron microscopy coupled with energy dispersive X-ray micro-analysis (SEM-EDX) have established themselves in the branch of the scientific examination of works of art. These methods are generally known for being complementary. FTIR is predisposed for the analysis of organic substances (binding media), whereas SEM-EDX allows the identification of inorganic substances (pigments and fillers) present in paint layers. Lately, IR absorption microspectroscopy has become a more attractive technique of an analysis, due to achieving the ability for multichannel, sequential and whole data collecting from micro-areas, their immediate processing and presenting in the form of two or three dimensional images [1]. Thus, it has become possible not only to identify, but also to locate materials in the micro-area, e.g. in a cross section taken from a work of art. Furthermore, FTIR microscopy is a non-destructive technique, and after the use of it a sample can be investigated by other complementary methods. The SEM-EDX ∗ Corresponding author. Tel.: +48 12 6629901; fax: +48 12 4302595. E-mail address: zekaszow@cyfronet.pl (Z. Kaszowska). method however, is capable of doing microscopic analysis from its very beginning. The combination of the scanning electron microscope and the X-ray spectrometer allows the surface scan of the sample with the use of an electron beam and the measurement of characteristic X-ray resulting in the formation of elemental maps. An analysis of cross sections with the use of FTIR imaging was introduced by Boon and co-workers who explained the molecular aspects of the ageing of paintings along with the phenomenon of forming aggregates of metal soaps (mainly lead and zinc soaps) within paint layers containing lipid binder [2–6]. Although the specular reflectance technique was successfully used by them, it has not become widespread. For the last few years, attenuated total reflection – Fourier transform infrared spectroscopy (ATR-FTIR) has been gaining popularity in the examination of cross sections [7,8]. Ricci et al. proved that thanks to the high spatial resolution (3–4 ␮m), the modern ATR-FTIR imaging technique can be applied to analyse cross sections from historic artefacts in which the stratigraphic layers have a thickness of less than 10 ␮m, e.g. albumen photographic prints [9]. Additionally, Kazarian and Chan have shown that the spatial resolution in the IR microscope, the same as used in our studies, differs between the vertical (ca. 4 ␮m) and horizontal (ca. 6 ␮m) directions due to the non-uniform illumination of aperture [10]. Spring et al. pointed out the possibility of 0924-2031/$ – see front matter. Crown Copyright © 2012 Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.vibspec.2012.11.018 2 Z. Kaszowska et al. / Vibrational Spectroscopy 65 (2013) 1–11 using chemometric methods, such as principal component analysis (PCA) and factor analysis, for a better discrimination of components on the chemical image [7]. Using ATR-FTIR imaging, degradation products such as copper soaps were identified and localized in paint cross sections [7]. They have already been detected in cross sections, but using point techniques only from much larger spatial domains though [11]. However, the number of publications presenting the potential of the ATR-FTIR imaging method has still been small [7–9,12,13]. In all of the above-mentioned examples, the ATR-FTIR imaging was accompanied by the SEM-EDX analysis, carried out in the background as an instrument to specify and obtain credible results by using IR spectroscopy. This technique has been extensively applied in examining the objects made of a great variety of materials such as stone, ceramic, glass, metal, paper, parchment, wood or textile [14]. For many years, SEM-EDX has also been a powerful tool in the characterization of paint cross sections, giving information related to the elements present in the inorganic pigments or extenders as well as their distribution in the different structural layers [15]. Based on an inspection of optical microphotography of the cross section, scientists select an area for a point and/or line analysis, and then choose a set of elements, to some extent subjectively, for the creation of distribution maps. Therefore, the principal goal of this work is to show strengths and limitations of both techniques, ATR-FTIR imaging as well as SEM-EDX, for the identification and location of in/organic components of paint cross sections. Our field of interest includes traditional painting materials such as pigments, fillers and binding media used in the process of executing a work of art, as well as secondary products as a result of their interaction under the influence of environmental factors. Based on the studies of three cross sections, we present paint compounds that can be identified directly and confidently and compounds which can be identified indirectly, through the detection of secondary products. Both methods applied here provide a valuable data set for the analytical application in the museum laboratory but as our study shows, their careful analysis is required to understand the artists’ use of materials. Although we do not propose here a new analytical methodology for the investigations of paint cross sections using ATR-FTIR/SEM-EDX imaging, we intend to compensate the lack of publications presenting the results of this approach in the studies on Cultural Heritage. We also suggest a somewhat different way of using the SEMEDX method, different to the earlier mentioned examples [7,8]. Even though, SEM-EDX analysis is based on the coupling of two devices: the scanning electron microscope and the X-ray spectrometer, researchers often perceive the microscope only as a source of inducing X-ray fluorescence. Hence, they underestimate its basic ability to create detailed images of the examined surfaces. Here, the high quality backscattered electron (BSE) images of the chosen micro-areas of the paint cross section are generated, with a higher spatial resolution than in a standard EDX analysis. The examination of the shape and size of grains allows distinguishing the phases in paint cross sections. Next, the atomic number contrast emphasized on the images is a useful indicator for the identification of the materials. Thanks to the BSE images, an electron beam during the elemental analysis can be focused directly on a single pigment or filler grain (the point analysis). The interpretation of the SEM-EDX spectra consists in correlating the detected set of elements with the material (qualitative analysis). However, because of a heterogeneous structure of paint cross sections, it is more of a semi-quantitative analysis. A small oil painting (14 cm × 29 cm) presenting a landscape with a ploughman (Fig. 1) was chosen as a subject of our study. The painting is a private property. It was purchased in the 1980s at an antiquities auction in Arnhem, the Netherlands. Although there is a signature Vincent in the bottom left hand corner, the authorship by Fig. 1. The landscape with a ploughman (a private property). Vincent van Gogh has not been confirmed up to now. The landscape is based on two opposing colours: blue and light brown (yellow) expanded by the painter into a wide range of shadows. The composition contains two culminating points, a dynamic figural scene with a man leading a pair of harnessed horses and a static arrangement of traditional Dutch farm buildings. 2. Experimental 2.1. Sample preparation Due to the small size and good condition of the painting, only four paint chips were taken from the edge of the painting. The samples were mounted in DURACRYL® PLUS acrylic resin. The surface of the cross section was polished by hand with Micromesh® polishing cloths of various grades, up to the smallest no. 12000. The stage of preparing samples has an important influence on the results of the analysis and the quality of the chemical images acquired by the ATR-FTIR spectroscopy [16–18]. The surface of the cross sections has to be completely flat and smooth so that an ATR crystal can stick directly to the surface. The porous structure of the sample makes the analysis impossible. In turn, the embedding resin smeared on the surface provides fake results if polished too intensively. After taking microscopic photographs, three cross sections were selected for the examinations: No. 1 from the grey-blue sky area (x = 29, y = 11.5 cm), c.f. Fig. 2a; No. 2 from the green-yellow field area (x = 0, y = 4.5), c.f. Fig. 5a; No. 3 from the violet-brown field area (x = 28, y = 0), c.f. Fig. 7a. 2.2. ATR-FTIR spectroscopy A liquid nitrogen cooled MCT FPA detector comprising 4096 pixels arranged in a 64 × 64 grid format was used to measure ATR-FTIR images with a Agilent 670-IR spectrometer and 620-IR microscope operating in rapid scan mode. Images were collected with an 8 cm−1 spectral resolution in the range 3800–850 cm−1 using 512 and 256 scans for background and samples, respectively. In the ATR imaging configuration, the spectrometer and the FPA detector were coupled with an infrared microscope with a 15× Cassegrain objective and a Ge ATR crystal. The use of a germanium crystal (refractive index = 4) allows to achieve a spatial resolution of about 3–4 ␮m in the fingerprint region of the mid infrared (MIR) spectrum [7]. The area of a sample analyzed each time was ca. 70 ␮m × 70 ␮m with each pixel sampling from an area of 1.1 ␮m × 1.1 ␮m. No ATR correction to the measured FTIR spectra was applied before an analysis. Chemical images were created from ATR-FTIR spectra by plotting integrated band intensities as a function of x–y pixel position. A colour scale from high value (red) to low value (blue) was employed. Band absorbance was calculated by the integration method, with a linear baseline drawn through the peak edges. The spectrum above this line was integrated over the wavenumber range of the band. Z. Kaszowska et al. / Vibrational Spectroscopy 65 (2013) 1–11 3 Fig. 2. Cross section No. 1: (a) visible microphotograph. The green, pink and orange squares indicate the selected areas for the ATR-FTIR imaging analysis. The size of the FTIR images is ca. 70 ␮m × 70 ␮m. The blue and the red in FTIR images denote low and high absorbance, respectively (b)–(e) FTIR images from the green area representing the integrated absorbance of the following IR bands at: (b) 1173, 1106, 1065 cm−1 –barium sulphate; (c) ca. 2920, 2850, 1730 cm−1 – a lipidic binder; (d) 1539 cm−1 – zinc stearate/palmitate; (e) 1589 cm−1 – carboxylates; (f) ca. 1400 cm−1 – carbonates; (g)–(j) FTIR images from the pink area representing the integrated absorbance of the following IR bands at: (g) ca. 1400 cm−1 – carbonates; (h) 1612, 1311 cm−1 – cadmium oxalate; (i) 995 cm−1 – an unidentified compound; (j) 1171, 1110, 1070 cm−1 – barium sulphate; (k)–(l) FTIR images from the yellow area representing the integrated absorbance of the following IR bands at: (k) ca. 1400 cm−1 – calcite; (l) 3290, 1635, 1531 cm−1 – a proteinaceous material; (m) FTIR spectra extracted from the pixels labelled in (b) and (e), the inset shows intensity changes of 1539 and 1589 cm−1 bands; (n) FTIR spectra extracted from the pixels labelled in (h), (i), and (l). (o) FTIR spectra of reference substances. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.) Single FTIR spectra of reference substances were recorded by using the same instrumentation as above in the spectral range of 3800–700 cm−1 . A spectral resolution of 4 cm−1 and 128, and 64 scans for background and samples, respectively, were set for measurements. of 20 kV were used. The beam diameter was focused to about 1 ␮m. To improve surface conductivity, the samples were carbon coated. 2.3. SEM-EDX analysis Wavelength dispersive X-ray (WDX) maps of distribution of selected elements were performed by Cameca SX 100 instruments. The Zn K␣ and Ba L␤ lines were measured on LLIF diffraction crystals (high intensity, large size diffraction crystals) whereas the S K␣ line was measured on LPET diffraction crystals. The maps of distribution of selected elements were processed using the image analysis software PeakSight SX 100. The samples were analyzed on a LEO 1430 scanning electron microscope, equipped with an energy dispersive X-ray spectrometer of Oxford instruments (ISIS 300). For most of the elemental analyses of selected homogeneous mineral crystals and synthetic substance grains, a beam current of 50 nA and an acceleration voltage 2.4. WDX analysis 4 Z. Kaszowska et al. / Vibrational Spectroscopy 65 (2013) 1–11 Fig. 3. Cross section No. 1: (a) BSE image of the paint and ground layers, (b) BSE image of the blue paint layer, (c) EDX spectrum of selected cobalt blue grain, (d) EDX spectrum of selected chromium pigment grain, (e) BSE image of the layer 3, (f) EDX spectrum of not-identified pigment grain in the layer 3, (g) EDX spectrum of another not-identified pigment grain in the layer 3. Mineral symbols after Kretz: Brt – barite; Cal – calcite; Qtz – quartz [30]. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.) Z. Kaszowska et al. / Vibrational Spectroscopy 65 (2013) 1–11 Scanning electron microscopy and microprobe analyses were carried out in the Laboratory of Microanalyses in the Polish Geological Institute – National Research Institute in Warsaw. 3. Results and discussion 3.1. Paint cross section No. 1 The cross section No. 1 taken from the edge of the painting from the grey-blue sky, illustrates all stratigraphic layers present in the selected area. The microphotograph (Fig. 2a) shows the following layers from top to bottom: a slightly blue paint layer (Layer 1), a white paint layer (Layer 2), a specific “dirty red” layer (Layer 3), an outer ground layer – thin and white (Layer 4), and a bottom ground layer – thick, significantly transparent and homogenous (Layer 5). Using the ATR-FTIR method to examine the first two stratigraphic layers (Fig. 2a, the green area), strong signals from the spectral bands at 1173, 1106, 1065 cm−1 are observed (Fig. 2b and m). The 1065 cm−1 band is characteristic for barium sulphate (Fig. 2o). Its distribution in the FTIR image (Fig. 2b) clearly shows that this material was used in the execution of the light blue layer (Layer 1). This result is confirmed by the SEM-EDX analysis that revealed barium, sulphur, and zinc elements in many spots of the layer. Barium sulphate that can occur naturally or be artificially precipitated is used until now as a pigment as well an extender by the commercial production of organic pigments and tube oilcolours. From the 1870s, a white pigment called lithopone gained popularity. It is a mixture of barium sulphate and zinc sulphide (BaSO4 + ZnS). Its FTIR spectrum is given in Fig. 2o. The bands shape and positions suggest the presence of lithopone in the sample No. 1, however, a mixture of barium sulphate with the other white pigment like zinc white (ZnO) cannot be excluded. Next, the SEM-EDX analysis exhibits that cobalt blue (CoO·Al2 O3 ) makes the layer blue (c.f. Fig. 3a–c). Apart this, single cadmium yellow (CdS) grains were detected between the barium sulphate grains (data not shown). In one of the examined spot, a chrome-containing particle was identified. Probably, there is viridian (Cr2 O(OH)4 ) or chromium oxide (Cr2 O3 ) (Fig. 3a, b, and d). All pigments were identified through correlation of discovered elements in the examined points with known painting materials. Moreover, FTIR spectra of the analyzed area show the presence of bands at ca. 2920, 2850, and 1730 cm−1 that are assigned to triglycerides (Fig. 2m). They are distributed in the entire surface, suggesting that a lipid-containing binder was applied for painting as well as for preparing the ground (Fig. 2c). In the second stratigraphic layer – the white paint layer (Fig. 2a, Layer 2) – ATR-FTIR spectra reveals products of a chemical process between a pigment and a lipid-rich binding medium resulting in the formation of salts of carboxylic acids (Fig. 2d and e). SEM-EDX points out the presence of zinc and lead elements in this layer. Most probably, the 1539 cm−1 band pictured in Fig. 2d and m originates from zinc palmitate or zinc stearate, or even both of the salts. Due to their similar chemical structure, it is difficult to differentiate these salts based on the IR spectra only. Robinet and Corbeil have reported the presence of a characteristic band for the COO− asymmetric stretches of zinc stearate at 1540 cm−1 [19], while Mazzeo et al. have presented a zinc palmitate spectrum with a band at 1538 cm−1 [20]. Stearic and palmitic acids are the most common saturated mono-carboxylic acids present in plant oils used in painting as binding media. Zinc stearate was also used as a surfactant in mechanically grind oil paints. Unfortunately, we are unable to definitely determine a substance responsible for another band of medium intensity, located at 1589 cm−1 (Fig. 2e and m). Its spatial distribution coincides with the presence of zinc carboxylates. IR spectrum of the paint sample based on the zinc white and 5 linseed oil, which was naturally aged for 10 years, also shows a band of this wavenumber [20]. However, a width of this band in the 1650–1520 cm−1 region suggests the presence of a mixture of carboxylates with different molecular weight. This suggestion has been confirmed by van der Weerd, who examined a zinc whiteand linseed oil-based paint sample that was prepared in 1941 [21]. The second derivative of its IR spectrum showed that the broad 1589 cm−1 band is accompanied by the two shoulders at 1620 and 1540 cm−1 . Here, the inset in Fig. 2m shows the co-existence of both bands at 1589 and 1539 cm−1 in various points of the cross section No. 1. The change of their relative intensities clearly shows that they must be attributed to different carboxylates. These observations suggest a similar mechanism of the triglycerides decomposition or ageing of art objects due to mixing a lipid binding medium with zinc pigments. The formation of metal carboxylates has been widely reported in the literature [4,6,19,20]. This process is caused by the hydrolytic degradation of triglycerides leading to the formation of free fatty acids, aldehydes, ketones or lactones. However, we do not observe a significant red shifting of the C O stretching mode in FTIR spectra recorded here, which should be expected for the mentioned degradation products. This vibration appears at ca. 1710 cm−1 for free fatty acids, whereas FTIR spectra of the cross section No. 1 show its presence in the range of 1728–1732 cm−1 . This may suggest a low concentration of hydrolysis products (undetectable here) or a direct chemical reaction between free fatty acids and metal ions. Hence, we observe bands of drying oil and metal carboxylates only. The spectrum extracted from the chemical image of the white layer 2 (Fig. 2e and m) shows the presence of an intense band at 1392 cm−1 that is characteristic for carbonates (the asymmetric stretching mode of CO3 2− ). Since SEM-EDX results suggest the presence of a lead element (Fig. 3f and g), the carbonate band likely originates from a white pigment – lead white (2PbCO3 ·Pb(OH)2 ). FTIR spectrum of standard lead white is shown in Fig. 2o. Traces of strontium were also found by SEM-EDX in the examined area. The analysis of the third stratigraphic layer (Fig. 2a, Layer 3) reveals very interesting results. SEM-EDX analysis shows that the matrix of this layer is made up of lead and calcium compounds (Fig. 3e). On the other hand, ATR-FTIR spectra from the pink area in Fig. 2a confirm the presence of carbonates exhibiting a band at ca. 1400 cm−1 (see Fig. 2g, n, and o). We conclude that lead white and calcite (CaCO3 ) are the building components of this layer. FTIR spectra of lead white and calcite are shown in Fig. 2o. Both of them show the presence of an intensive absorbance at ca. 1395 cm−1 , whereas the second band at 872 cm−1 only appears in the spectrum of calcite. The latter is observed in FTIR spectra shown in Fig. 2n, however, we found also spectra typical for lead white. The BSE image reveals (c.f. Fig. 3e) that the calcite grains are relatively large, of irregular shape and with sharp edges, whereas grains of lead white are smaller and shaped like scales. In the BSE image they exhibit white colour because of high atomic number of lead. Because of the colour of the layer 3 we cannot exclude the presence of red lead (Pb3 O4 ). We should also consider the possible presence of an organic red pigment. Quicksilver and sulphur were found in a few spots by SEM-EDX. Therefore, there is no doubt that particles of vermillion (HgS) are scattered in the layer 3 of the cross section No. 1 (Fig. 3e). Blue grains are also visible in this layer and they are responsible for the violet shade of the red layer (c.f. Fig. 2a). Unfortunately, we cannot definitely match the composition of the analysed grains to the type of the blue. The result of the one point analysis however is worth noting. The SEM-EDX spectrum shown in Fig. 3f contains an interesting set of elements. Can tin and cobalt be correlated with the presence of a blue pigment called cerulean blue (CoO·nSnO2 )? Why is there such a significant amount of magnesium? Could magnesium indicate the presence of Indian yellow? Similar questions have no definite/conclusive answers. We also do not know what kind of material consists of the following 6 Z. Kaszowska et al. / Vibrational Spectroscopy 65 (2013) 1–11 set of elements (Al, Fe, K, Mg, Si, Ti, Pb, Zn) that was identified at another grain (Fig. 3g). On the other hand, ATR-FTIR imaging shows that there is a grain characterized by bands at 1612 cm−1 and 1311 cm−1 (Fig. 2h and n) in the referred layer. We could assign these bands to calcium oxalate in the form of whewellite minerals (CaC2 O4 ·H2 O) since this type of oxalate has been often reported in studies on works of art. However, its IR bands are observed at 1622 and 1321 cm−1 and are attributed to the symmetric stretch of the C O group and asymmetric stretch of the C O group, respectively [22]. So far, calcium oxalate has been detected on calcium-rich surfaces like wall paintings or polychromed limestone sculptures and in the paint layers containing organic pigments (e.g. red lake pigments). They have also been classified as degradation products in oil paint layer with smalt [23]. On the other hand, copper oxalates have been identified in a 16th century Cypriot wall painting by using FTIR and Raman spectroscopy [22]. The characteristic strong IR bands of copper oxalates have been found at 1614, 1363 and 1319 cm−1 . This shows that wavenumbers of the oxalate stretching modes are sensitive to the metal type. Our SEM-EDX analysis of the cross section No. 1 shows that the cadmium element is also present in the examined point. From that reason we propose assign the observed ATR-FTIR bands at 1612 and 1311 cm−1 to cadmium oxalate. IR bands for the neat Cd(C2 O4 )·3H2 O appear at 1613 and 1314 cm−1 [24]. Although, the result seems improbable in terms of painting material technology, we attempt to explain this phenomenon. Kittel has reported that in the early period of cadmium yellow (CdS) production, one method of precipitating cadmium sulphide required the presence of oxalic acid [25]. In that way a yellow of light shade was obtained that contained a significant amount of cadmium oxalate. In turn, Keim has detected cadmium oxalate in cadmium pigments available on the market by the end of the 19th century [26]. These results have been mentioned in the specialist literature by Eibner in 1909 [26] and by Fiedler and Bayard in 1986 [27]. The Keim’s analysis indicates that a pigment can contain more than 30% oxalate in some cases, but only 1.5% sulphur. To our best knowledge, the presence of cadmium oxalate in a cross section from a real painting has not been reported up to now. Additionally, our SEM-EDX reveals the presence of cadmium sulphide grains (cadmium yellow) in a few other points of the red layer (Fig. 3e). The identification of components in the two-layered ground is straightforward. We established from ATR-FTIR analysis that barium sulphate (Fig. 2j and n) is present in the outer ground layer (Fig. 2a, Layer 4), while calcium carbonate (Fig. 2k and n) is found in the bottom ground layer (Fig. 2a, Layer 5). All spectra show the presence of bands at ca. 1400 and 872 cm−1 . SEM-EDX analysis allowed us to complete that data. Zinc, along with barium and sulphur elements, was discovered in the outer ground layer. In order to determine the chemical composition of this layer, a mapping of these elements was performed using the SEM–WDX system. The distribution of these elements shows clearly (Fig. 4a and d) that there is lithopone (BaSO4 + ZnS), and not a mixture of barium white and zinc white. A small amount of zinc white (ZnO) was found in the layer 5. The electron images of this layer also reveal the presence of fossilized fauna and flora belonging probably to planktonic species like foraminifera, coccoliths, genus Biscutum, Quadrium or Vatznaueria. Moreover, we identified a protein layer on the bottom of the cross section No. 1 (c.f. Fig. 2a, Layer 6). Amide A, I and II modes are attributed to IR bands at 3290, 1635 and 1531 cm−1 , respectively, as presented in Fig. 2l and n. Likely, an animal glue was used to impregnate a canvas before application of the ground. 3.2. Paint cross section No. 2 The paint cross section No. 2 was taken from the edge on the left of the painting (the landscape area). A visible microphotograph, shown in Fig. 5a, exhibits two main paint layers (Layers 1 and 2) Fig. 4. WDX maps of distribution of the selected elements identified in the outer ground layer of the cross section No. 1 (area 4 in Fig. 3a) (a) Distribution of the Zn K␣-line, (b) Distribution of the Ba L␤-line, (c) Distribution of the S K␣-line. A color scale from high value (red) to low value (blue) is employed (d) Compilation of map of distribution of the Ba L␤- and the Zn K␣-line. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.) and a two-layered ground (Layers 3 and 4). Paint layers are not homogenous. We present below the FTIR analysis of the layers 1 and 2 represented by the areas marked with the violet and red squares in the microphotograph in Fig. 5a. Yellow is the dominant colour of the first stratigraphic layer. However, white and blue streaks blend into the yellow matrix. A similar set of materials was identified in the yellow as well as white areas. ATR-FTIR analysis indicates the presence of basic lead carbonate – lead white (Fig. 5b, f, k, and l). SEM-EDX identified lead and zinc whites (Fig. 6b, d, and e), while single cadmium yellow grains are observed in the yellow paint layer (Fig. 6b). The layer 1 of this paint cross section exhibits a very intense colour but the relatively small amount of cadmium sulphate grains is found. Therefore, we assumed that an organic yellow pigment is responsible for this colour. Even though, mid-IR spectroscopy is a powerful tool for an analysis of organic compounds, in the case of examining paint cross sections, its application is limited in some cases. The concentration of organic pigments on the cross section surface is usually disproportionately small compared to the concentration of the remaining components like inorganic pigments, fillers, mordants of dyes, and even binding media. Thus, the magnitude of absorbance for low concentrated analytes is too small to be observed in IR spectrum. FTIR spectra collected for the area depicted in Fig. 5a with the violet square show the presence of bands with a maximum at 1570–1581 cm−1 , which could originate from unspecified degradation products (Fig. 5c and k). Their largest concentration is mainly located within the yellow and blue fragments of the layer 1 and overlaps the presence of the oil binding medium (Fig. 5d). The blue in that layer was classified as Prussian blue (Fe4 [Fe(CN)6 ]3 ) (Fig. 5e) since its IR band is observed at 2090 cm−1 (C N stretching vibration), see Fig. 5k. According to our discussion in the previous paragraph, degradation products can originate from the formation of salts with metal ions. Here, the SEM-EDX results also showed the presence of lead, zinc and cadmium elements (Fig. 6a and b). Lead and cadmium soaps are identified in FTIR spectra by the presence of a band at 1530–1540 cm−1 , whereas zinc carboxylates give absorption at ca. 1590 cm−1 [20]. Thus, the ca. 1580 cm−1 band observed in the FTIR spectra of the cross section No 2 rather originates from zinc soaps (Fig. 5k). Z. Kaszowska et al. / Vibrational Spectroscopy 65 (2013) 1–11 7 Fig. 5. Cross section No. 2: (a) visible microphotograph. The violet and red squares indicate the selected areas for the ATR-FTIR imaging analysis. The size of the FTIR images is 70 ␮m × 70 ␮m. The blue and the red in FTIR images denote low and high absorbance, respectively (b)–(e) FTIR images from the purple area representing the integrated absorbance of the following IR bands at: (b) ca. 1400 cm−1 – lead white, (c) 1570–1580 cm−1 – an unidentified carboxylate, (d) 2924, 2854, ca. 1735 cm−1 – a lipidic binder, (e) 2090 cm−1 – Prussian blue (f)–(j) FTIR images from the red area representing the integrated absorbance of the following IR bands at: (f) ca. 1400 cm−1 – lead white, (g) 1535, 1581 cm−1 – unidentified carboxylates, (h) ca. 2920, ca. 2850, ca. 1730 cm−1 – a lipidic binder, (j) 3687, 3619, 1026, 1003, 910 cm−1 – kaolinite, (j) 1169, 1068 cm−1 – barium white and silicate, (k) FTIR spectra extracted from the pixels labelled in (b) and (e), (l) FTIR spectra extracted from the pixels labelled in (g), (i), and (j), the inset shows the second derivative of a spectrum from (g) in the region of 1800–1300 cm−1 . (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.) 8 Z. Kaszowska et al. / Vibrational Spectroscopy 65 (2013) 1–11 Fig. 6. Cross section No. 2: (a) BSE image of paint and ground layers, (b) BSE image of yellow paint layer, (c) BSE image of green-brown paint layer, (d) BSE image of white-blue streaks in the yellow paint layer, (e) EDX spectrum of selected zinc white grain. Mineral symbols as in Fig. 3. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.) The appearance of the second stratigraphic layer is different from the first one (Fig. 5a, Layer 2). Wide yellow-brown and green streaks blended together are observed in that layer. They contain numerous pigment particles of variable colour and size. ATR-FTIR spectra revealed the presence of: basic lead carbonate (Fig. 5f and l), an oil binder (Fig. 5h and l), kaolinite (Al2 Si2 O5 (OH)4 ) (Fig. 5i and l) [28,29], barium sulphate and silicates (SiO2 ) (Fig. 5j and l), and further unspecified degradation products (Fig. 5g and l). For the latter, some bands with maxima centred around 1580 cm−1 are observed. The second derivative of IR spectrum extracted from the FTIR image in Fig. 5g exhibits the presence of the three IR bands at 1628, 1581, and 1535 cm−1 (see the inset in Fig. 5l). A similar set of bands has been observed for the aged paint sample of zinc white in oil, as mentioned above [21]. Our SEM-EDX analysis showed that apart from lead and barium whites, zinc white is also present. Additionally, single grains of cobalt blue (CoO·Al2 O3 ) and cadmium yellow are found. The point analysis also shows the presence of iron. Iron oxides like ␣-Fe2 O3 – red-violet haematite, ␣-FeO(OH) – yellow-brown goethite, ␥-FeO(OH) – yellow-orange lepidocrocite are the main components of natural and synthetic ochres. In natural ochre, one of the above-mentioned components is present in a mixture with other minerals. The latter are most often clay minerals such as kaolinite, illite or smectite as well as calcite, gypsum, and quartz. Kaolinite is sometimes treated as an indicator to determine the ochre’s origin. It has been reported that the main component of French ochre is kaolinite, whereas Spanish ochre usually contains sulphates like gypsum [28]. On the other hand, synthetic ochres can also contain kaolin, gypsum or silica. These materials are popular as extenders for oil-colours at their mass-production. As mentioned above, kaolinite is identified in the cross section No. 2 (Fig. 6a and Z. Kaszowska et al. / Vibrational Spectroscopy 65 (2013) 1–11 9 Fig. 7. Cross section No. 3: (a) visible microphotograph. The blue square indicates the selected area for the ATR-FTIR imaging analysis. The size of the FTIR images is 70 ␮m × 70 ␮m. The blue and the red in FTIR images denote low and high absorbance, respectively (b)–(f) FTIR images from the blue area representing the integrated absorbance of the following IR bands at: (b) ca. 1390 cm−1 – lead white, (c) 1038 cm−1 – silicate, (d) 3687, 3618, 1026, 1003, 906 cm−1 – kaolinite, (e) 1535, 1450 cm−1 – zinc palmitate/stearate, (f) 2924, 2850, ca. 1735 cm−1 – a lipidic binder, (g) FTIR spectra extracted from the pixels labelled in (c)–(e). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.) c). However, it is difficult to determine whether it represents natural ochre or acts as a paint extender. We do not discuss here the role of barium sulphate and the type of the ground, since their function and composition are identical like for the cross section No. 1. 3.3. Paint cross section No. 3 ATR-FTIR analysis of the cross section No. 3 is limited to the area marked in a microphotograph visible in Fig. 7a, whereas SEM-EDX analysis was performed for the area marked in the BSE image in Fig. 8a. The chemical image for a band at ca. 1392 cm−1 represents a spatial distribution of carbonates (Fig. 7b). The lack of a band at 872 cm−1 that is characteristic for calcite, suggests the presence of lead white. It is located in the outer paint layer and in the layer situated directly on the ground (Fig. 7a, Layers 1 and 3, respectively). Both of these layers are reddish in the colour, although the layer 3 does not possess orange-red grains as large as the layer 1. Unfortunately, the FTIR method does not identify chemical compounds responsible for this colour. But two grains of a larger size presented in the layer 1 are analysed. The IR image for a band at 1038 cm−1 correlated with the location of the first grain is depicted in Fig. 7c. This absorption is characteristic for silicates (the in-plane Si O stretching mode) (Fig. 7g). The next grain is identified as a kaolinite mineral (Fig. 7d and g). Similarly to silicate found in the cross section No. 2, this material exhibits IR bands at 3687 and 3618 cm−1 (the OH stretches of outer and inner hydroxyl groups), 1026 cm−1 (the stretching mode of Si O Si), 1003 cm−1 (the stretching mode of Si O Al), and 908 cm−1 (the stretching mode of Al O H). SEMEDX confirms this result. Furthermore, the SEM-EDX point analysis also shows to what extend the elemental composition of the paint layer changes from point to point. For instance, the presence of cobalt and aluminium elements in one of the examined points clearly suggests the use of 10 Z. Kaszowska et al. / Vibrational Spectroscopy 65 (2013) 1–11 numerous dark gray areas visible in the BSE image (Fig. 8c) can be just correlated with the presence of zinc stearate/palmitate. 4. Conclusion Fig. 8. Cross section No. 3: (a) and (b) BSE images of paint and ground layers (c) BSE image of red and blue paint layers. Mineral symbols as in Fig. 3. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.) cobalt blue by the artist (Fig. 8c). However, it is difficult to verify if arsenic and cobalt identified in the other point originate from cobalt red (alkaline cobalt arsenate: Coy (AsO4 )x ) or how to interpret the following groups of elements: Fe, Pb, Si, As; Al, Si, Zn; and Fe, Si – up to now, iron has been an indicator of ochre. An interesting result is found from the spectroscopic analysis of the white-blue “isle” situated between the reddish paint layers (Fig. 7a, Layer 2). The main material that is identified is zinc stearate/palmitate with absorption bands at 1535, 1450, 1396 cm−1 as the chemical image in Fig. 7e shows. The IR spectrum depicted in Fig. 7g shows a spectrum of the pure substance without any bands originating from pigments or extenders [19]. It has not been reported so far that a paint layer can contain so high concentration of a metal soap. Additionally, the image for the C H and C O vibrations at 2917, 2850, and 1730 cm−1 exhibits the same distribution of lipids originating from binding medium (Fig. 7f). Probably, A series of the below presented paint materials or their components were identified in the stratigraphic layers of the painting by using the vibrational and X-ray spectroscopic methods. They belong to various groups of painting materials like pigments, extenders or binding media. We detected lead white, zinc white, calcite, barium white, lithopone, quartz, kaolin, cadmium yellow containing cadmium oxalate, pigments containing Fe (probably natural and/or synthetic ochres), vermilion, cobalt blue, Prussian blue, and traces of chrome pigments (emerald green or chrome oxide). We also do not rule out the presence of red lead, cobalt red as well as organic pigments, especially yellow and red. Secondary products were also found in the paint layers. They are soaps formed from the interaction between some pigments with a lipid binding medium. We definitely confirmed the presence of zinc palmitate/stearate as well as the use of oil as a binding medium in the paint and ground layers. Our studies do not set a definite date of the execution of the painting presented here, but they give a few important arguments for the further discussion on this issue. All of the identified painting materials and their components were being used in painting at the turn of the 19th and 20th century. Cadmium yellow was produced in Germany from 1825, while chrome oxide was introduced into artistic painting in the twenties of the 19th century. In turn, lithopone gained popularity in the seventies of the 19th century. In the twenties of the 20th century, a new product – titanium white has begun to supersede toxic lead white, which was used for centuries. The imaging mode of techniques, SEM-EDX and ATR-FTIR spectroscopy allowed us to achieve spatial and complementary information on the identified materials. SEM-EDX is a powerful tool for the identification of inorganic components such as pigments and extenders. In that way it compensates the deficiency of ATRFTIR imaging that band below 850 cm−1 cannot be detected. One of the factors hindering a proper interpretation of SEM-EDX results is the strong fragmentation of the chemical species in a micro area, whereas the size of the material particles is smaller than the beam current diameter and interaction volumes. The fact that many pigments can be characterized by the same set of elements makes their unambiguous identification impossible. Moreover, it is sometimes difficult to verify a particular pigment or a mixture. For instance, whether the pigment called lithopone (BaSO4 + ZnS) is identified or a mixture of barium sulphate (natural barite or synthetic blanc fixe) and zinc white (ZnO) is present. Also, it is not always possible to determine the function of an identified substance. We discussed above the function of kaolinite as a component of natural ochre and an extender for oil paints. A similar question concerns barium sulphate that can be used intentionally or as an extender. Although, MIR spectroscopy is usually considered to be a suitable method for the detection of organic compounds present in works of art, especially binding media, modes of polyatomic inorganic groups are also IR active in the range of 850–1800 cm−1 . It allowed us to identify the inorganic groups like carbonates, sulphates, silicates, and aluminosilicates. In our investigations, the ATR-FTIR imaging has primarily shown its potential as a tool for an analysis of secondary products formed in the paint layers due to chemical processes between pigments and binding media or due to an effect of external factors. It is not always possible to identify directly the type of binding medium, because the weak IR bands of binding media are often overlapped by strong bands of degradation products. However, an indirect identification is possible through detection of degradation products. In order to do that, the knowledge of a possible degradation mechanism is essential. In this work, Z. Kaszowska et al. / Vibrational Spectroscopy 65 (2013) 1–11 ATR-FTIR imaging reveals for the first time the presence of cadmium oxalate and its distribution within the paint layer as well as a high concentration of “neat” zinc palmitate/stearate. [13] [14] [15] [16] References [17] [18] [1] E.N. Lewis, P.J. Treado, R.C. Reeder, G.M. Story, A.E. Dowrey, C. Marcott, I.W. Levin, Anal. Chem. 67 (1995) 3377. [2] R.M.A. Heeren, J.J. Boon, P. Noble, J. Wadum, ICOM committee for conservation, in: 12th Triennial Meeting, Lyon, 1999, p. 228. [3] J. Van der Weerd, H. Brammer, J.J. Boon, R.M.A. Heeren, Appl. Spectrosc. 56 (2002) 275. [4] J. Van der Weerd, J.J. Boon, M. Geldof, R.M.A. Heeren, P. Noble, Z. Kunsttechnol, Konservierung 16 (2002) 36. [5] J. Van der Weerd, R.M.A. Heeren, J.J. Boon, Stud. Conserv. 49 (2004) 193. [6] K. Keune, Binding medium, pigments and metal soaps characterised and localised in paint cross-sections, PhD dissertation, University of Amsterdam, 22 June 2005 (available online). [7] M. Spring, C. Ricci, D.A. Peggie, S.G. Kazarian, Anal. Bioanal. Chem. 92 (2008) 37. [8] R.A. Godali, J. Hall, R.J. Sharer, L. Traxler, L. Rintoul, P.M. Fredericks, Appl. Spectrosc. 62 (2008) 10. [9] C. Ricci, S. Bloxham, S.G. Kazarian, J. Cult. Heritage 8 (2007) 387. [10] S.G. Kazarian, K.L.A. Chan, Appl. Spectrosc. 64 (2010) 135A. [11] R. Mazzeo, E. Joseph, S. Prati, A. Millemaggi, Anal. Chim. Acta 599 (2007) 107. [12] E. Joseph, C. Ricci, S.G. Kazarian, R. Mazzeo, S. Prati, M. Ioele, Vib. Spectrosc. 53 (2010) 274. [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] 11 S. Prati, E. Joseph, G. Sciutto, R. Mazzeo, Acc. Chem. Res. 43 (2010) 792. M. Schreiner, M. Melcher, K. Uhlir, Anal. Bioanal. Chem. 387 (2007) 737. G. Elzinga-Ter Haar, Stud. Conserv. 16 (1971) 41. C.J.G. van Hoek, M. De Roo, G. van der Veer, S.R. van der Laan, Microsc. Microanal. 17 (2011) 656. J.J. Boon, S. Asahina, Microsc. Microanal. 12 (2006) 1322. Ch.M. De Fonjaudran, A. Nevin, F. Piqué, S. Cather, Anal. Bioanal. Chem. 392 (2008) 77. L. Robinet, M.C. Corbeil, Stud. Conserv. 48 (2003) 23. R. Mazzeo, S. Prati, M. Quaranta, E. Joseph, E. Kendix, M. Galeotti, Anal. Bioanal. Chem. 392 (2008) 65. J. Van der Weerd, Microspectroscopic analysis of traditional oil paint, PhD dissertation, University of Amsterdam, 2005 (available online). A. Nevin, J.L. Melia, I. Ostricioli, G. Gautie, M.P. Colombini, J. Cult. Heritage 9 (2008) 154. M. Spring, C. Higgitt, D. Saunders, Natl. Gall. Tech. Bull. 26 (2005) 56. A. Moses Ezhil Raj, D. Deva Jayanthib, V. Bena Jothy, Solid State Sci. 10 (2008) 557. H. Kittel, Pigmente, Herstellung, Eigenschaften, Anwendung, Wissenschaftliche Verlagsgesellschaft M.B.H. Stuttgart, Angwand. Chem. (1960) 197. A. Eibner, Malmaterialienkunde als Grundlage der Maltechnik, Springer, Berlin, 1909. I. Fiedler, M. Bayard, in: R.L. Feller (Ed.), Artists’ Pigments: A Handbook of their History and Characteristics, Cambridge, National Gallery of Art Washington, 1986. C. Genestar, C. Pons, Anal. Bioanal. Chem. 382 (2005) 269. J. Madejova, P. Komadel, Clays Clay Miner. 49 (2001) 410. R. Kretz, Amer. Miner. 68 (1983) 277.