Nuclear Instruments and Methods in Physics Research B 273 (2012) 173–177 Contents lists available at SciVerse ScienceDirect Nuclear Instruments and Methods in Physics Research B journal homepage: www.elsevier.com/locate/nimb Non-destructive provenance differentiation of prehistoric pigments by external PIXE L. Beck a,⇑, H. Salomon b, S. Lahlil a, M. Lebon a,c, G.P. Odin a, Y. Coquinot a, L. Pichon a a C2RMF-UMR171 Centre de Recherche et de Restauration des Musées de France, 14 quai François Mitterrand, Palais du Louvre Porte des Lions, 75001 Paris, France Centre Européen d’Archéométrie, Université de Liège, Sart Tilman Bât B15, 4000 Liège, Belgium c Muséum National d’Histoire Naturelle, Département de Préhistoire, UMR 7194, 1 rue René Panhard, 75013 Paris, France b a r t i c l e i n f o Article history: Available online 26 July 2011 Keywords: PIXE Prehistoric pigment Ochre Provenance Petrography Arcy-sur-Cure a b s t r a c t The elemental analysis of minerals/rocks has been often used for the determination of their geological origin. When these natural rocks were exploited by prehistoric civilizations as objects, weapons, or pigments, the composition of the minerals can provide information on the mobility, the exchanges and the interaction between groups of population. In this paper, we will present results obtained from archaeological samples of prehistoric pigments, mainly iron and manganese oxides. PIXE analysis has been applied to samples of the prehistoric cave ‘‘La grotte du Renne’’ in Arcy-sur-Cure, France (Chatelperronian, 38,000–34,000 BP). Because most of the archaeological objects are decorated or display some use marks, it is not possible to take samples. Consequently, we have used a non-destructive technique thanks to the external beam of AGLAE (C2RMF, Paris). In order to improve the limits of detection (LOD less than 10 ppm from Cu to Sb), a metal absorber has been placed on the X-ray detector to preferentially filter the Fe–K or Mn–K lines. Based on the quantitative analysis of major and trace elements, we have obtained groups of compositions corresponding to different geological sources. We demonstrate in this study that it is possible to extend PIXE analysis to the characterization of prehistoric pigments such as iron and manganese oxides for differentiating potential sources of pigments in archaeological contexts. Ó 2011 Elsevier B.V. All rights reserved. 1. Introduction In the past years, Particle Induced X-ray Emission (PIXE) has largely demonstrated to be a powerful tool capable of analyze trace elements for determining the provenance of archaeological materials. It has been mostly applied to geological based material such as obsidian [1–3] and gems [4–6]. In these cases, raw materials have been employed by ancient civilizations without (or with minor) transformation and it is admitted that the initial chemical composition is not (or slightly) affected [7]. This is particularly recognized for trace elements which can be used as fingerprint of geological sources and thus contribute to provenance investigations. This point is an important question for historical and prehistoric studies, as it allows tracing commercial routes and cultural exchanges between ancient communities. Recent papers have focused on the study of trace elements in pigments or in raw materials from geological sources by LA-ICPMS (Laser Ablation Inductively Coupled Plasma Mass Spectrometry) [8–10], PIXE [11–15] and neutron activation analysis [16,17]. Except Refs. [13–15] all the mentioned studies have required sample preparation which is not always possible for archaeological ⇑ Corresponding author. Present address: CEA, DEN, Service de Recherches de Métallurgie Physique, Laboratoire JANNUS, 91191 Gif-sur-Yvette, France. E-mail address: lucile.beck@cea.fr (L. Beck). 0168-583X/$ - see front matter Ó 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2011.07.068 artefacts. In this paper, we present experimental PIXE configurations which allow investigating prehistoric pigments, mostly iron and manganese oxides, at the ppm level without any preparation or sampling. In order to assess the feasibility of this procedure to the study of pigment procurement, we have applied it to the prehistoric site of Arcy-sur-Cure, France (38,000–34,000 BP) [18,19]. 2. Experiment 2.1. Pigment samples Samples are composed of blocks of pigments found in the chatelperronian layers (38,000–34,000 BP) of la Grotte du Renne in Arcy-sur-Cure, France. This site represents the richest archaeological sequence evidencing the late Neanderthal activities before their complete disappearance. The excavations, conducted by André Leroi-Gourhan in the 60s, revealed a large amount of colouring materials, mainly red or black. More than 18 kg of colouring materials, i.e. about 2000 fragments and 300 worked objects were associated with fireplaces and remains of two huts, which were built with calcareous stones and mammoth tusks. All these observations indicate intense production of powder either by crushing and grinding, or by scraping and abrading. This last category of artefacts reveals consequently polished or striated surfaces (Fig. 1). 174 L. Beck et al. / Nuclear Instruments and Methods in Physics Research B 273 (2012) 173–177 Fig. 1. Some blocks of pigments and worked tools (mainly in iron oxide (red) or in manganese oxide (black)) found in the chatelperronian layers (38,000–34,000 BP) of La Grotte du Renne in Arcy-sur Cure, France. (For interpretation of the references to colours in this figure legend, the reader is referred to the web version of this paper.) Due to the presence of marks on these worked tools which testify of their use, it is not allowed to take sample for analysis. Objects are consequently directly analysed without any kind of preliminary treatment or preparation. 2.2. PIXE experimental setup: external beam and selective absorbers PIXE experiments were conducted at the AGLAE (Accélérateur Grand Louvre d’Analyse Elémentaire, Paris, France) facility using a 3 MeV proton extracted beam. The beam size is about 50 lm diameter. Because the samples are highly inhomogeneous, they were scanned in order to average over a 1 mm2 area. Objects were just set on the XYZ table in front of the extracted beam exit. Precise positioning is achieved by using the luminescence induced by the proton beam on the sample minerals. X-ray spectra are recorded by two Si(Li) detectors oriented at 45° to the beam [20,21]. One is devoted to low energy X-rays (0.1–15 keV) from major elements of the matrix, mainly Fe or Mn, Al, Si and Ca. The other detector is equipped with selective filters to reduce pileup by attenuating intense X-rays [22]. We have selected absorber according to the matrix major element. For iron based matrix, a 20 lm thick chromium absorber was mounted on the detector whereas for manganese based matrix, a vanadium absorber was used. An additional 50 lm thick aluminium filter was superimposed in order to reduce Cr or V X-rays induced by the interaction of the primary X-rays coming from the sample with the absorber. PIXE spectra were collected between 10 and 15 min (Fig. 2). Elemental concentrations have been extracted by using GUPIX [23] and TRAUPIXE [24]. Thanks to this experimental procedure, limits of detection less than 10 ppm have been achieved for most of trace elements (Table 1). These values are of the same of order of magnitude as in other studies by NAA or PIXE where samples were powdered and prepared. groups). These results were obtained after a first mineralogical determination by X-ray diffraction on a sampling of 80 blocks. Then, nine blocks without any past modification tracks were cut in order to prepare 30 lm thin sections studied with a petrographic microscope. Four groups are described as follow: Group Aa: ferrugineous hardground. This rock is characterized by benthic organisms, mainly crinoid segments and urchin remains cemented by haematite, goethite and calcite. Its formation is the result of an abrupt interruption of sedimentation lasting many million years. Group Ab: this group presents the same morphology as Group Aa, with a slight difference due to the dissolution of the carbonate phase. Small bone fragments were also identified. This pigment is formed in the same geological layer with the ferrugineous hardground but this part of the formation was submitted to water activity. Group B: ferrugineous sandstone/siltstone. This rock is characterized by 40–90% rounded quartz grains cemented by haematite or a mix of haematite and goethite. Group C: blocks of almost pure haematite. The lack of internal structure within the blocks did not allow the identification of the geological formation from which they were extracted. For black pigment materials, the results obtained by this first approach tend to show that the oxides and hydroxides of manganese form a homogeneous corpus (Group D) of black pigments along the whole archaeological assemblage. The study of the geological map and the research in the mining archives have documented the possible catchment area for these four minerals. The groups Aa and Ab are located around 40 km east to the site in the Hettangian formations (beginning to the Jurassic), whereas the group B has been identified in the Mio-Pliocene formation covering the tablelands next to the cave, around 4–40 km west. For the groups C and D, no geological layer in the area of Arcy-sur-Cure is documented. 3. Results 3.2. PIXE for iron oxide pigments 3.1. Mineralogical and petrographical characterization of the pigments Based on mineralogical studies, Salomon et al. [19] have demonstrated that the red pigment materials found in the cave of Arcy-sur-Cure have three different geological provenances (A–C Twenty-seven samples corresponding to13 worked tools and 14 blocks were analyzed by PIXE. Around 20 of these samples were already characterized by petrography and/or XRD (see Section 3.1) to be used as group references. Concentrations have been obtained 175 L. Beck et al. / Nuclear Instruments and Methods in Physics Research B 273 (2012) 173–177 Si Ca 10000 1000 22 Tl L pile up Ce 100 77 Counts 1- Low X-ray energy detector Fe Pb L (a) 100000 18–45 3–9 26–80 1000000 La 60 10 1 15 20 25 30 35 40 2- High X-ray energy detector with 20 µm Cr and 50 µm Al Sb Zn Rb 10000 Zr Mo 100 Sn Sr Sb Ba 25 30 Mo 10 7 2–8 11 As 1000 3 1–2 3 Counts Fe Ba Mn 100000 14 10 24 4–10 56 5 8 0 1000000 3 1–2 4 17 Zr Y Sr 2 1–2 2 2 1–3 3 Rb As Ga Zn Co Ni Cu for major and trace elements. Due to the irregular distribution of fossil remains, totally or partially preserved, Fe and Ca cannot be used to characterise the pigments. However, the Si:Al ratio gives information on the alumino-silicate phase (Fig. 3). The alumino-silicate phases for groups Aa and Ab are composed of the same clay mineral. Group B shows a high Si content which is due to the large amount of quartz. Group C is not well defined. In order to establish geochemical groups relative to iron oxide, bivariate plots of iron and trace elements have been performed. Among the 15 detected trace elements, three of them (Mo, As and Sb) are correlated to iron with different ratios according to Table 1 PIXE limits of detection (LOD) for elements heavier than iron. Fig. 2. Typical PIXE spectra for prehistoric pigments composed of (a) iron oxide or (b) manganese oxide. 1—Low X-ray energy detector (in blue). 2—High X-ray energy detector (in red) equipped with a 20 lm thick selective filter and 50 lm Al. The selective filters are Cr for iron matrix and V for manganese matrix. [Spectra ref: 09jun016-R437 and 18oct015- R598]. (For interpretation of the references to colours in this figure legend, the reader is referred to the web version of this paper.) 12 2–4 12 (b) 3 3–4 6 40 3 5–10 7 35 7 8–9 11 20 Energy (keV) 19 15 57–100 30–54 145 10 Iron oxide matrix (this study) Iron oxide matrix [12] Manganese oxide matrix (this study) 5 0 4 1–2 10 1 176 L. Beck et al. / Nuclear Instruments and Methods in Physics Research B 273 (2012) 173–177 900000 800000 y = -2.2194x + 635203 2 R = 0.221 Aa 700000 SiO2 600000 Ab 500000 C 400000 B 300000 y = 2.65x R² = 0.92 200000 100000 0 0 20000 40000 60000 80000 10000 120000 140000 160000 0 Al2O3 Fig. 5. SiO2 concentration as a function of Al2O3 concentration (in weight%) for prehistoric pigments composed of manganese oxides-hydroxides (Arcy-sur Cure, France). Fig. 3. SiO2 concentration as a function of Al2O3 concentration (in weight%) for prehistoric pigments composed of iron oxide (Arcy-sur Cure, France). the geological group (Fig. 4). Group C is characterized by a high contents of Mo, As and Sb. Groups Aa and Ac have the same contents of As, Sb and Mo with intermediate values, confirming the same geological origin. Group B has very low content in trace elements. These bivariate plots corroborate the source distinction based on petrographic analysis, only performed on non-worked blocks. Thanks to the PIXE results, it has been possible to precise doubtful attribution of several archaeological samples. In particular, some attributions previously obtained by optical observation and XRD were changed and two unidentified worked tools were attributed to group B and group C (Fig. 4). These results provide strong evidence that iron oxide sources in the Arcy-sur-Cure area can be discriminated using external PIXE. present in amounts not exceeding 10 wt%. Fig. 5 shows these oxides are closely correlated showing the homogeneity of the corpus of samples. The minor and trace elements (Ce, Co, La, Y, Sr, Ni, Zn, Cu, As, Zn, Mo) confirm this result. Four of them are presented in Fig. 6. These data further demonstrate the uniformity of the manganese oxide corpus and evidence a common origin for all the samples studied. 4. Conclusion External PIXE has been applied to analyse prehistoric blocks of red and black pigments. Due to the archaeological importance of the artefacts (38,000–34,000 BP), the analysis was performed directly on the samples, without any kind of preparation. The use of selective filters has given access to limits of detection less than 10 ppm for iron-based pigments as well for manganese-based pigments. The quantification of trace elements provides chemical 3.3. PIXE for manganese oxide pigments All the black pigment samples contain more than 70 wt% of manganese oxide. Other major elements such as Si or Al can be 900 Mo 800 Aa 700 600 Ab 500 C 400 300 B 200 100 0 0 200000 400000 600000 800000 1E+06 Fe2O3 800 10000 As 700 Sb 600 1000 500 400 100 300 200 10 100 0 0 200000 400000 600000 Fe2O3 800000 1E+06 1 0 200000 400000 600000 800000 1E+06 Fe2O3 Fig. 4. Concentration (in ppm) of Mo, As, Sb trace elements as a function of the concentration in Fe2O3 for the prehistoric red pigments of Arcy-sur Cure, France. Two unidentified samples (photographies) have been attributed by PIXE to groups B and C. L. Beck et al. / Nuclear Instruments and Methods in Physics Research B 273 (2012) 173–177 177 Fig. 6. As, Cu, Zn trace elements as a function of MnO2 for the prehistoric black pigments of Arcy-sur Cure, France. markers which allow establishing distinctions between different types of red pigments. These results provide strong evidence that prehistoric pigment sources can be discriminated using external PIXE. It is one of the main archaeological issues, in particular in the field of the chronological attribution of rock art and in the determination of colouring provenance. References [1] L. Bellot-Gurlet, G. Poupeau, J. Salomon, T. Calligaro, B. Moignard, J.-C. Dran, J.A. Barrat, L. Pichon, Nucl. Instrum. Meth. B 240 (2005) 583–588. [2] T. Calligaro, J.-C. Dran, S. Dubernet, G. Poupeau, F. Gendron, E. Gonthier, O. Meslay, D. Tenorio, Nucl. Instrum. Meth. B 240 (2005) 576–582. [3] D. Juárez-Cossío, E. Terreros, J. Quiroz-Moreno, S. Romero-Sánchez, T. Calligaro, D. Tenorio, M. Jiménez-Reyes, M. De Los Rios, Nucl. Instrum. Meth. B 267 (2009) 1149–1152. [4] T. Calligaro, A. Mossmann, J.-P. Poirot, G. Querré, Nucl. Instrum. Meth. B 136138 (1998) 846–850. [5] T. Calligaro, J.-P. Poirot, G. Querré, Nucl. Instrum. Meth. B 150 (1999) 628–634. [6] T. Calligaro, S. Colinart, J.-P. Poirot, C. Sudres, Nucl. Instrum. Meth. B 189 (2002) 320. [7] L. Wilson, A.M. Pollard, The provenance hypothesis, in: D.R. Brothwell, A.M. Pollard (Eds.), Handbook of Archaeological Sciences, John Wiley and Sons Ltd, Chichester, UK, 2001, pp. 507–517. [8] M. Gil, M.L. Carvalho, A. Seruya, A.E. Candeias, J. Mirão, I. Queralt, Nucl. Instrum. Meth. A 580 (2007) 728. [9] R.L. Green, R.J. Watling, J. Forensic. Sci. 52 (2007) 851. [10] E. Iriarte, A. Foyo, M.A. Sánchez, C. Tomillo, J. Setién, Archaeometry 51 (2009) 231. [11] B. David, E. Clayton, A.L. Watchman, Aust. Archaeol. 36 (1993) 56–57. [12] P. Nel, P.A. Lynch, J.S. Laird, H.M. Casey, L.J. Goodall, C.G. Ryan, R.J. Sloggett, Nucl. Instrum. Meth. A 619 (2010) 306–310. [13] M. Sanchez del Rıo, P. Martinetto, C. Solıs, C. Reyes-Valerio, Nucl. Instrum. Meth. B 249 (2006) 628–632. [14] L. Beck, M. Lebon, L. Pichon, M. Menu, L. Chiotti, R. Nespoulet, P. Paillet, X-Ray Spectrom. 40 (2011) 219–223. [15] M. Lebon, L. Beck, S.Grégoire, L. Chiotti, R. Nespoulet, M. Menu, P. Paillet, Open J. Archaeometry, in press. [16] R.S. Popelka-Filcoff, J.D. Robertson, M.D. Glascock, C. Descantes, J. Radioanal. Nucl. Chem. 272 (2007) 17–27. [17] R.S. Popelka-Filcoff, E.J. Miksa, J.D. Robertson, M.D. Glascock, H. Wallace, J. Archaeol. Sci. 35 (2008) 752–762. [18] C. Couraud, Galia Préhistoire 33 (1991) 17–52. [19] H. Salomon, C. Vignaud, Y. Coquinot, S. Pagès-Camagna, M.-P. Pomiès, J.-M. Geneste, M. Menu, M. Julien, F. David, Technè (2009). [20] T. Calligaro, J.D. MacArthur, J. Salomon, Nucl. Instrum. Meth. B 109/110 (1996) 125. [21] J. Salomon, J.-C. Dran, T. Guillou, B. Moignard, L. Pichon, P. Walter, F. Mathis, Nucl. Instrum. Meth. B 266 (2008) 2273. [22] C.P. Swann, S.J. Fleming, Nucl. Instrum. Meth. B 49 (1990) 65. [23] J.L. Campbell, N.I. Boyd, N. Grassi, P. Bonnick, J.A. Maxwell, Nucl. Instrum. Meth. B 268 (2010) 3356. [24] L. Pichon, L. Beck, Ph. Walter, B. Moignard, T. Guillou, Nucl. Instrum. Meth. B 268 (2010) 2028.