Please download to get full document.

View again

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
    Mediterranean Archaeology and Archaeometry, Vol. 18, No 5, (2018), pp. 171-178 Copyright © 20 18 MAA Open Access. Printed in Greece. All rights reserved. DOI: 10.5281/zenodo.1285906   MICRO-CHEMICAL EVALUATION OF ANCIENT POTSHERDS BY µ -LIBS SCANNING ON THIN SECTION NEGATIVES Stefano Pagnotta 1,2  , Stefano Legnaioli 1  , Beatrice Campanella 1  , Emanuela Grifoni 1  , Marco Lezzerini 2  , Giulia Lorenzetti 1  , Vincenzo Palleschi 1,3  , Francesco Poggialini 1  , Simona Raneri 2   1  Applied and Laser Spectroscopy Laboratory, Institute of Chemistry of Organometallic Compounds,  Research Area of National Research Council, Via G. Moruzzi, 1 –  56124 Pisa (ITALY)  2  Department of Earth Sciences, University of Pisa, Via S. Maria 53, Pisa (ITALY) 3  National Interuniversity Consortium of Material Science and Technology (INSTM) Received: 18/10/2017 Accepted: 28/03/2018 Corresponding author: Stefano Pagnotta (stefanopagnotta@yahoo.it) ABSTRACT In the study of ancient pottery, thin section analysis represents the basic approach to study mineralogical and petrografic features in order to obtain preliminary information about the production technology and srcin of archaeological ceramics. However, even if thin section analysis allows investigating the textural and structural characteristics of potteries, peculiar features related to clay paste and temper composition, as well as provenance issues, can be detailed addressed only by quantitative mineralogical and chemical studies. In the realization of thin sections, a negative face is always produced, similar to the thin section itself; these remains can be used for additional analyses, such as high spatial resolution micro-chemical studies using, for example, a micro-laser induced breakdown spectroscopy (LIBS) scanner. LIBS is a spectroscopic technique that, exploiting the laser radiation, is able to bring into the plasma state micrometric portions of the sample, and to analyse its content through the study of the optical emission of the plasma itself. Unlike other techniques, LIBS can detect and quantify also light elements such as aluminium and magnesium. Images produced by the micro-LIBS instrument show the spatial distribution of the chemical elements within a portion of the sample, which may have dimensions from a few hundred microns up to several centimeters. The combination of these images with algorithms derived from image processing techniques may return interesting information and supporting data to in-depth investigate pottery components detected by optical microscopy observations. In this work, we present the results of an experimental study performed on thin-section negatives with different grain size, surface treatments and aggregates, coming from some Neolithic Italian sites, exploring the potential of the LIBS method in micro-chemical studies of ancient potsherds. KEYWORDS: Elemental mapping, archaeometry, pottery, prehistory, µ -LIBS, PCA, Kohonen SOM    172 S. PAGNOTTA et al. Mediterranean Archaeology and Archaeometry, Vol. 18, No 5, (2018), pp. 171-178 1.   INTRODUCTION Petrographic analysis (Rice, 1987; Quinn, 2013), mi-crostructural analysis by scanning electron micro-scope SEM coupled with chemical micro analysis, thermal analysis, mineralogical analysis and other spectroscopic techniques (Memmi, 2004; Barone et al., 2014; Teoh et al., 2014; Maritan, 2015; Hunt, 2016; Halperin and Bishop, 2016) are well-established pro-cedures for analysing ancient pottery. However, the possibility to obtain valuable information on ceramics by using new methods requiring no sample prepara-tion and short measurement time are in high demand. Micro- laser induced breakdown spectroscopy (µ -LIBS) has been recently successfully applied on differ-ent materials (Schiavo et al., 2016; Pagnotta et al., 2017) for mapping surfaces up to one square centime-ter obtaining morphological, compositional and quan-titative data, thanks to the advantages offered by the method, namely the detection of light elements (Miziolek et al., 2006), the short time (about half an hour for sample to scan an area of 5 mm 2  at lateral resolution of 100 μ m) and the low cost of the analysis compared to other techniques (see also, Kouhpar et al., 2017; Javanshah, 2018; Hemeda, 2013). Among ancient materials, the study of prehistoric pottery is often not straightforward because of the lack of reliable sources able to support hypothesis on production technology and processes. Sometimes, this kind of information appears quite relevant in the reconstruction of social dynamics, as in the transition between phases (Maritan et al., 2017) and manufac-ture routines (Dal Sasso et al., 2014; Antonelli et al. 2018). A significant example is represented by Neo-lithic potteries, whose production can be framed in the streamline of the overcoming technological evo-lution represented by metallurgy; this event shifted the artefacts manufacture from hand-made to spe-cialized production requiring control of the firing processes and a special conditioning of raw materi-als (Pessina and Radi, 2003). Due to the relevance of this ceramic class, some representative potsherds of Southern Italy Neolithic productions, namely from Abruzzo and Puglia, have been sampled for µ -LIBS analysis. These materials exhibit quite peculiar char-acteristics: very thin walls, high purified clay paste (with inclusions barely visible to the naked eye) and a glossy black external and internal surface, which sometimes tends to show a metallic aspect (Cre-monesi, 1965, 1973; Cremonesi and Tozzi, 1987). The technology for realizing these surfaces is currently of great interest to prehistoric archaeologists (Agostini et al., 2003), interpreting the finishing treatment as an imitative routine of metal prototypes that in the same period began to circulate in the areas of the Near East. In this contribute we therefore explore the potential of µ -LIBS in supporting archaeological in-vestigations on ancient pottery by providing in mi-cro-destructive, fast and short time qualitative chem-ical maps which interpretation can supply informa-tion on bulk mineralogical composition and manu-facture of surface finishing. 2.   SAMPLES AND TECHNIQUES Five samples of black gloss ceramics from Fucino - Paterno (Abruzzo region, central-eastern Italy) and one from S. Anna di Oria (Puglia region, south-eastern Italy) have been analysed in this study (Fig. 1).  Figure 1. The thin section negatives used for the analysis.   MICRO CHEMICAL EVALUATION OF ANCIENT POTSHERDS BY MEAN OF MICRO-LIBS SCANNING  173 Mediterranean Archaeology and Archaeometry, Vol. 18, No 5, (2018), pp. 171-178   The samples were selected on the basis of the sta-tistical representativeness of this particular ceramic class. Details about samples are provided in Table 1 . Table 1. Summary table of the analyzed samples. Id. Provenance Shape paste Finishing SAOFN1 S.Anna di Oria (Puglia) Tronco-conical vessel Semi-depurated Black-gloss POFN1 Paterno (Abruzzo) Tronco-conical vessel Semi-depurated Black-gloss POFN2 Paterno (Abruzzo) Tronco-conical vessel Semi-depurated Black-gloss POFN3 Paterno (Abruzzo) Tronco-conical vessel Semi-depurated Black-gloss POFN4 Paterno (Abruzzo) Tronco-conical vessel Semi-depurated Black-gloss POFN5 Paterno (Abruzzo) Tronco-conical vessel Semi-depurated Black-gloss Data were acquired using a Modì smart LIBS sy s-tem (Bertolini et al., 2006) equipped with a Zeiss Ax-ioplan A1 microscope and a Thorlabs’s XY stage. The Modì Smart was equipped with an Nd:YAG Laser (  λ= 1064nm), focused on the sample through a dedi-cated 10X microscope objective (Figure).  Figure 2. Scheme of the Micro-  Modì system.   The plasma light was focused with a ball lens and sent through an optic fibre to an AvaSpec Dual Spec-trometer, covering a spectral range from 200 nm to 900 nm (resolution of 0.1 nm in the UV and 0.3 in the VIS-IR region). A double pulse laser (first pulse= 5.4 mJ; second pulse= 8.7 mJ with 1 ms retard between them) was used to realize raster of 50x50 shots at a lateral resolution (pixel dimension) of 100 μ m for the acquisition of compositional data on a 5 mm 2 . In order to synchronize the laser pulses with the motorized sample holder, a dedicated software de-veloped with the NI LabVIEW 8.5 was used. Collected data were processed by an in-house rou-tine on MATLAB ® , able to select the intensity lines of interest for each detected element (Al, Si, Ca, Mg, Fe, Na, K) in the square matrix associated to the scanned surface. At the same time, the software per-formed a normalization procedure point-by-point based on the total intensity of the LIBS spectrum (Figure), to minimize effects caused by changes in the focusing of the laser beam.  Figure 3. A typical LIBS spectrum of a potsherd. 200 400 600 80020004000600080001000012000140001600018000    I  n   t  e  n  s   i   t  y   (  a .  u .   ) Wavelenght (nm) LIBS Spectrum  174 S. PAGNOTTA et al. Mediterranean Archaeology and Archaeometry, Vol. 18, No 5, (2018), pp. 171-178   Table 2. Elements of interest for this study and their wave-length. Element Wavelength (nm) Na 819.48 Mg 279.55 Al 309.24 Si 288.16 K 766.5 Ca 422.68 Fe 538.72 Starting from the intensity matrix based on the µ -LIBS spectral lines, gray-scale elementary maps were therefore obtained each of them accounting the dis-tribution of a specific element over the wavelength interval 200 - 900 nm The treatment of elementary maps as multi-spectral images allows to obtain a multispectral “cube” in which the contribute of d e-tected elements can be simultaneously visualised (Figure).  Figure 4. Iper-spectral “  cube ”   , composed by the elemental maps. To process to the compositional maps generated by the µ -LIBS system, the typical multispectral imag-ing approach has been used (Legnaioli et al., 2013). The recombination of some elementary maps, re-levant in the analysis of archaeological ceramics, al-lows thus to obtain false color images describing specific compositional variation into a sample; in this case, RGB images accounting the detection of Si (Red), Al (Blue) and Ca (Green) were processed. Moreover, a self-organized map (SOM) of Koho-nen with four neurons were provided to evaluate the potential of un-supervised segmentation process in assisting the interpretation of chemical data. The use of SOM network allows to reduce the dimensionality of the classification problem, preserving the topol-ogy of the training set (inputs). The kohonen algorithm, starting from each input image, generates a random centroid for each output neuron. The randomly generated centroid updates with the weights of the pixels most similar to that centroid, reaching a mean value after several itera-tions (generally 1000). At the end of the learning process the images will be segmented into areas with similar properties. In this case, the process started from a seven-dimensional space (inputs), consisting of the maps of all the analysed elements, and was reduced to a maximum 4-dimensional (output neu-rons) space (Figure). The four output segments can be therefore recombined in a final image able to de-scribe the contribute of the obtained segments, by assigning to each of them one channel in a CMYK color space (Cyan-Magenta-Yellow-Black).  Figure 5. Structure of the SOM network utilized. 3.   RESULTS The interpretation of RGB colour maps, based on the normalized intensities (a.u.) of Si (Red), Al (Blue) and Ca (Green) (Figure), can provide information on the possible differences in chemical composition be-tween clay paste and surface layers. First of all, it can be observed that almost all the samples tend to a purple colour; only the sample POFN4 tends to a reddish colour, suggesting higher level in Si-rich phases. Green areas account Ca-rich phases, mainly attributed to minerals phases con-taining calcium; blue areas, on the contrary, deline-ate the distribution of minerals phases enriched in aluminium. In the case of studied samples, it is interesting to note that in some specimens (SAOFN1, POFN1 and POFN3) the upper part of the scanned area (that cor-responds to the surface of the potsherds) turns from purple/magenta to lighter or darker colour, due to different proportion of aluminium and silica content (Figure). This result could support the hypothesis on composition as well as technology employed to make the surface finishing. In fact, a homogenous composition between ceramic body and surface should indicate a simple lustre of the surface. Alter-natively, a different composition should possibly suggest the application of high-depurated layers of clays to obtain black gloss.   MICRO CHEMICAL EVALUATION OF ANCIENT POTSHERDS BY MEAN OF MICRO-LIBS SCANNING  175 Mediterranean Archaeology and Archaeometry, Vol. 18, No 5, (2018), pp. 171-178    Figure 6. Micro-photographs of the scanned area of about 5 mm  2 (top) for the analysed samples; composed RGB colour maps of distribution of Si (Red), Ca (Green) and Al (Blue)(bottom). In order to overcome interpretation based only on three elements and obtain information based on the overall chemical composition, SOM approach has been applied. The use of the SOM network gener-ated (as results of training stage) different seg-mented images for each sample (Figure), with black and white areas accounting the different weights related to each input element (Figure). SAOFN1 POFN1 POFN2 POFN3 POFN4 POFN5 Segment 1 Segment 2 Segment 3 Segment 4  Figure 7. Binary images showing the distribution of the various segments obtained for each of the analysed samples. 
Related Search
Similar documents
View more
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks