- Email: [email protected]
Characterisation of archaeological pottery: The case of “Ionian Cups”
Journal of Molecular Structure 993 (2011) 142–146
Contents lists available at ScienceDirect
Journal of Molecular Structure journal homepage: www.elsevier.com/locate/molstruc
Characterisation of archaeological pottery: The case of ‘‘Ionian Cups’’ G. Barone a, V. Crupi b, F. Longo b,⇑, D. Majolino b, P. Mazzoleni a, V. Venuti b a b
Dipartimento di Scienze Geologiche, Università di Catania, Corso Italia 57, 95129 Catania, Italy Dipartimento di Fisica, Università di Messina, V.le Ferdinando Stagno d’Alcontres 31, 98166 Messina, Italy
a r t i c l e
i n f o
Article history: Available online 23 January 2011 Keywords: FT-IR SANS Ceramics Firing temperature
a b s t r a c t The aim of this study was the microscopic and mesoscopic characterisation of archaeological pottery findings addressed to the identification of the manufacturing techniques. The samples under study were ‘‘Ionian Cups’’ sherds, coming from Poira, an archaeological site in eastern Sicily (South-Italy). These cups represent a ceramic typology widely diffused in the Mediterranean Area in archaic age (VI–V century BC). The identification of the production sites of these materials, originally manufactured in Greek-Eastern area and then largely diffused in Magna Graecia, is still an open question. Here, the microscopic structural characterisation was obtained by Fourier Transform Infrared absorption (FT-IR) measurements, which permitted us to determine the mineralogical phases present in the artefacts. Furthermore, Small Angle Neutron Scattering (SANS) measurements permitted the characterisation of the size distribution and surface characteristics of the mesoscopic aggregates formed by the minerals. Ó 2011 Elsevier B.V. All rights reserved.
1. Introduction Ceramic findings are among the best preserved and the most important materials testimonies of ancient human cultures. In fact, the study of a ceramic fragment allows to obtain information concerning the artistic choices, trade pathways, cultural and technological evolution of the ancient human communities [1,2]. As is well known, the characterisation of archaeological ceramics is very complex due to the heterogeneity of the materials on different scales and to the presence of both crystalline and amorphous phases in the ceramic bodies and in the outer decorations (glaze, pigments). So the use of different analytical techniques is required to characterise the ceramic findings [2–7]. In particular, manufacture technology involves several aspects of pottery making, such as the type of raw materials used, their processing to prepare the clay paste, the surface treatment, decoration and firing to obtain the finished item. Here, we present the employment of Fourier Transform Infrared absorption (FT-IR) and Small Angle Neutron Scattering (SANS) measurements to obtain the microscopic and mesoscopic characterisation of archaeological pottery. FTIR and SANS techniques have been performed on ‘‘Ionian Cups’’ sherds, a ceramic typology widely diffused in the Mediterranean Area in archaic age (VI–V century BC). In particular, the inves⇑ Corresponding author. Address: University of Messina, Department of Physics, V.le Ferdinando Stagno d’Alcontres 31, 98166 Messina, Italy. Tel.: +39 0906765463; fax: +39 090395004. E-mail address: [email protected] (F. Longo). 0022-2860/$ - see front matter Ó 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.molstruc.2011.01.028
tigated samples coming from Poira, an archaeological site near Catania (Sicily, Italy). The production of ‘‘Ionian Cups’’ started in about 580 BC in eastern areas of Greece and was then continued in Magna Graecia until about 540 BC [8]. The presence of various production centres in the Greek colonies have been hypothesised and represents a hotly debated question. Information on the manufacture, such as the raw materials and the estimation of firing temperature of this ceramic typology, could be useful to identify the production centres.
2. Experimental 2.1. Materials The investigated samples include seven fragments of Ionian cups of type B2, labelled as POI 1, 3, 5, 13, 17, 19, 20 from the Poira necropolis (Sicily, Southern Italy) and dated back to VI–early V centuries BC. An example of this ceramic typology, together with the pictures of two investigated samples, are reported in Fig. 1. These samples belong to a large set of Ionian cups from five archaeological sites in eastern Sicily (‘‘Mendolito, Monte Castellaccio, Poira-Poggio Cocola, Piano Casazzi, Francavilla di Sicilia’’). Petrographic thin-section analyses and chemical investigations by X-ray fluorescence spectrometry (XRF) have been already performed on all these Ionian cups [8]. This study indicated that a peculiar character of Ionian cups from these archaeological sites in eastern Sicily is due to their highly homogeneous petrographic and geochemical features. This may indicate a single centre of production, characterised
G. Barone et al. / Journal of Molecular Structure 993 (2011) 142–146
143
Fig. 1. An example of (a) Ionian cup of type B2, together with the pictures of (b) POI 17 and (c) POI 19 fragments.
by abundant production and an extensive distribution of products, or an extremely specialised technique that was known to several workshops. Hence, the microscopic and mesoscopic characterisation could support one among the hypotheses above mentioned. 2.2. FT-IR FT-IR measurements were performed by means of a BOMEM DA8 FT-IR spectrometer. This apparatus was equipped with a Globar lamp as source and a suitable beamsplitter and detector, depending on the part of the spectral range under study. In particular, a Hyper beamsplitter and a DTGS/FIR detector were used to collect the spectra in 200–600 cm1 range, while we used a KBr beamsplitter and a DTGS/MIR detector to register the spectra from 450 to 4000 cm1. The two measurements were combined to obtain the spectra in the range from 200 to 4000 cm1. In such configurations it was possible to use a resolution of 4 cm1. The investigated samples were prepared in pellets, about 0.5 mm thick, using small quantities (2 mg) of bulk sample dispersed in 200 mg of powdered CsI, that is transparent in the investigated IR frequency range. We remark that from each shard, about 2 mg of material was drawn from non-significant parts of the ceramic body, in order to avoid damages that could be affect the integrity and artistic content of the finding. The measurements were performed in dry atmosphere to avoid dirty contributions, 32 repetitive scans were automatically added to obtain a good signal-tonoise ratio (SNR) and a spectra reproducibility of high quality as well. The experimental spectra were compared with those of standard minerals and clays (Sadtler database ‘‘Minerals and Clays’’ [9]) and with data reported on different sources [10–12] for a reliable assignment of the bands. 2.3. SANS SANS measurements were carried out by using the PAXE spectrometer at the ORPHEE reactor of the Laboratoire Léon Brillouin
(LLB, Saclay, France). The two configurations used at large and small Q (scattering vector) range were summarised in Table 1. For a more detailed description of SANS technique the reader can refer to literature [3,13]. The investigated samples were cut in thin sections (thickness <1 mm) and therefore multiple scattering effects were minimised. No appreciable neutron activation of the samples was found after the SANS experiment. We used standard LLB SANS routines to correct the two-dimensional intensity distributions for instrumental background and to normalise them to absolute cross section per unit volume of the sample (cm1) by measuring the incident beam intensity, transmission and thickness of each sample. The structure of some samples under study turned out to be anisotropic. This may be either connected to preferential directions of the texture of mineral aggregates or to pressure effects. In such cases, Q must be treated as a vector and the xy pattern of the scattered intensity has the shape of more or less elongated ellipsoids. In this case, a longitudinal axis (Lo) along the major axis of the ellipse and, in the perpendicular direction, a transversal axis (Tr) along the minor axis, can be defined. The first corresponds to correlations in real space over shorter distances than the second. Numerical analysis is done in each direction, yielding distinct parameters. Table 1 Instrumental characteristics for SANS measurements performed by the PAXE spectrometer. Large Q Wavelength (k) Sample–detector distance Collimation Q range
6Å 2m 2m 5 103–2 101 Å1
Small Q Wavelength (k) Sample–detector distance Collimation Q range
15 Å 4.5 m 4.5 m 3 103–3 102 Å1
G. Barone et al. / Journal of Molecular Structure 993 (2011) 142–146
3. Results and discussion 3.1. FT-IR analysis
POI 1
2
1000
1500
2000
2500
3000
3500
-1
Wavenumber (cm ) Fig. 2. FT-IR absorbance spectrum of POI 1 sample.
4000
5
5 1,2,3 1
1
2
4,5 13,4
1
4 1 5 11
1000
1500
-1
Wavenumber (cm ) Fig. 3. FT-IR absorbance spectrum of POI 1 sample in the region from 200 to 1500 cm1. 1 = Quartz; 2 = diopside; 3 = muscovite/illite; 4 = bytownite/anorthite; 5 = k-feldspar.
POI 5 3
1 2 5
5
1
1
1
2
1,2,3 4
1 4,5 3,4 15 1 1 500
1000
1500 -1
Wavenumber (cm ) Fig. 4. FT-IR absorbance spectrum of POI 5 sample in the region from 200 to 1500 cm1. 1 = Quartz; 2 = diopside; 3 = muscovite/illite; 4 = bytownite/anorthite; 5 = k-feldspar.
POI 17
21 3 5
5
1,2,3
1
1
1
2 4
3,4,5 1 5 11 500
500
1
3
500
IR Absorbance (arb. units)
IR Absorbance (arb. units)
The experimental FT-IR spectra show similar features, typical of pottery materials, with a wide band ranging from 850 cm1 to 1400 cm1. In Fig. 2 the FT-IR absorbance spectrum recorded on POI 1, as example, is reported. However, as the absorption peaks of the main minerals in a typical absorbance spectrum of a ceramic occurred at wavenumbers lower than 1500 cm1 [9–12], only this region of the spectra are analysed and discussed in the following. The spectra of samples under study show similar shape, as it can be seen by an inspection of Figs. 3–5, where the spectra of POI 1, POI 5 and POI 17, as examples, were reported. The mineralogical composition obtained by the analysis of the spectra on the basis of characteristic IR wavenumbers of minerals and clays [9–12] is summarised in Table 2. First of all, the typical absorption peaks of quartz were detected in all the investigated samples: the main SiO2 stretching peak at 1080 cm1 that contribute to the large band of the spectra, the shoulder at 1170 cm1, the peaks at 505 and 692 cm1 and the distinctive doublet at 777 and 798 cm1. Furthermore in the low-frequency range (below 500 cm1) it was possible detect the other typical peaks associated to quartz (458, 395 and 370 cm1) [12]. Calcite in trace was revealed in almost the samples by the presence of the typical band centred at 1430 cm1. Beside the above mentioned minerals, it was possible identify the presence of k-feldspar, plagioclase, muscovite/illite and diopside (see Table 2 and Figs. 3–5). The revealed k-feldspar showed peaks centred at 1130, 1000, 720, 575 cm1. Regarding the plagioclase we revealed the contributions at 950, 575 and 532 cm1, that could be associated to anorthite or bytownite. These two members of plagioclase series do not show marked differences in their IR spectra, so it is very difficult to discriminate anorthite from bytownite [10]. The main contribution due to the Si–O stretching (1013 cm1) was found in all the samples. This mode could be attributed to muscovite or illite, because in this region their spectra do not show marked differences. The observed contributions at 533 and 463 cm1 allowed to suppose the presence of muscovite. This mineralogical phase was also observed by
POI 1 IR Absorbance (arb. units)
For most of the investigated samples, SANS spectra have been collected on different perpendicular sections aimed at putting into evidence the typology (linear or planar) and the spatial distribution of the preferred orientations of the mesoscopic grains.
IR Absorbance (arb. units)
144
1000
1500
-1
Wavenumber (cm ) Fig. 5. FT-IR absorbance spectrum of POI 17 sample in the region from 200 to 1500 cm1. 1 = Quartz; 2 = diopside; 3 = muscovite/illite; 4 = bytownite/anorthite; 5 = k-feldspar.
145
G. Barone et al. / Journal of Molecular Structure 993 (2011) 142–146
Qz
By/An
k-feld
Mu/Il
Cc
Di
POI POI POI POI POI POI POI
+++ +++ +++ +++ +++ +++ +++
+ ++ ++ ++ + + ++
+ tr + + tr + +
+ tr + tr + + +
tr tr tr tr tr tr tr
+ + + + + + +
1 3 5 13 17 19 20
petrographic thin-section analyses in a previous study on the same samples [8]. Finally, we revealed variable content of diopside, a new formed mineral, which was identified by the characteristic peaks at 1064, 960 and 467 cm1. Hence, the analysis of ceramic bodies (see Table 2 and Figs. 3–5) indicated a general homogeneity in composition among the investigated samples. The identification of the main mineralogical phases and, in particular, the presence of calcite revealed in trace, together with different contents of a new formed mineral, such as diopside, can be indicator of the firing temperature. In fact for temperatures higher than 750 °C the reaction of dissociation of calcium carbonate in calcium oxide and carbon dioxide occurs as evidenced by the thermogravimetric analysis carried out on the clays that probably were used in the production of studied Ionian cups [14]. If the temperature reaches 900 °C this decomposition is irreversible because calcium oxide reacts with clay components forming new Ca silicates (diopside and anorthite) [15–17]. Therefore, if the cooking of the manufacture occurs at approximately 900 °C, calcite disappears partially, also as function of microdomains composition [17] and is revealed in low content. So, the paragenetic composition allowed, according to literature [15–17], to hypothesise a firing temperature close to 900 °C for all the analysed samples.
(a)
5
10
4
10 -1
Sample
surface fractal dimension, Ds, by the following relation: a = 6 Ds. Rg indicates the mean-size of new formation minerals clusters/crystallites, whereas the exponent a refers to the roughness of the voids/minerals grains. In particular, since the value of Ds must be between 2 and 3, then 3 < a < 4. The two-dimensional Euclidean exponent Ds = 2 (and hence a = 4) is recovered for particles with sharp interfaces. In the following, the numbers 1–3 close to the name of each sample indicate, respectively, the sections exposed to the neutron beam perpendicular, 1 (transversal cut) and 2 (longitudinal cut), and parallel (3) to the surface of the Ionian cup. For some of the investigated samples, the structure appears to be isotropic in all the analysed sections, probably testifying not relevant consequences on the artefact connected to some external mechanical stress because of manufacturing methods. In other cases, we observed the presence of a ‘‘linear’’ anisotropy in a single section or, at the most, a ‘‘planar’’ anisotropy in two sections. This trend can be explained taking into account that the ceramic body of these very fine-grained potteries does not contain inclusions, so anisotropy can be just linked to the texture of the groundmass, as due to the forming and finishing of the manufacture. Furthermore, this is a characteristic that is not affected by all those mineralogical and structural transformations that occur during the firing process. The SANS spectra, in a log–log plot, of POI 3, together with the best-fit according to Eq. (1), are reported, as an example, in Fig. 6. The revealed low-level scattering background, related to the incoherent scattering due to hydrogen atoms, can be considered negligible.
I(Q)(cm )
Table 2 Mineralogical composition of the shards according to the infrared analysis. Qz = quartz; By/An = bytowinite/anorthite; k-feld = k-feldspar; Mu/Il = muscovite/ illite; Cc = calcite; Di = diopside. The ‘‘+’’ symbol stands for the relative abundance of the compounds in each sample; ‘‘tr’’ and ‘‘’’ symbols stand for trace and absent respectively.
3
10
2
10
1
10
0
10
-1
10
-2
10
3.2. SANS analysis
IðQ Þ ffi C 1 exp
Q 2 R2g 3
! þ C 2 Q a
ð1Þ
In this fitting law, four free parameters were introduced. The prefactors C1 and C2 contain the relative intensity of the Guinier’s and power law; the other two free parameters are the radius of gyration Rg and the exponent a, this last one is connected to the
(b)
-2
10
-1
10
-1
Q (Å )
5
10
4
-1
I(Q)(cm )
10
3
10
2
10
1
10
0
10
-1
10
-2
10
-3
10
-2
-1
10
-1
10
Q (Å )
(c)
5
10
4
10
3
10 -1
I(Q)(cm )
The interpretation of the SANS data was done using an empirical approach, that we already successfully employed in our previous work to study a set of archaeological ceramics [13]. According to this approach, the presence, in our samples, of two independent populations of mesoscopic structures with substantially different radii can be hypothesised. Their scattering contributions are supposed to be independent, and hence additive in the fitting law. The first population consists of ‘‘small’’ units, i.e. new formation minerals clusters/crystallites, whose typical dimensions can be in principle probed in the investigated spatial scale. On the contrary, the second population is composed by much ‘‘bigger’’ units, i.e. voids/mineral aggregates. Their typical sizes are out of the accessible experimental scale and can be derived by other techniques. For this second population, only surface can be probed. As a consequence, the Q-dependence of the scattered intensity I(Q) can be written as a combination of a Guinier law and a power law:
-3
10
2
10
1
10
0
10
-1
10
-2
10
-3
10
-2
10
-1
-1
Q (Å )
10
Fig. 6. Experimental (open squares) and fitted (solid line) SANS intensity for POI 3 sample analysed in section (a) 1, (b) 2 and (c) 3.
146
G. Barone et al. / Journal of Molecular Structure 993 (2011) 142–146
Table 3 Values of radius of gyration Rg and fractal exponent a as obtained by the fitting procedure for all the investigated samples. Error bars on Rg values are, on average, of the order of 7% (4% and 10% for the lower and the higher Rg values, respectively). Error bars on a values are of the order of 1%. Longitudinal and transversal configurations are indicated as (Lo) and (Tr), respectively.
using clays from sediments cropping out near the archaeological site. 4. Conclusions Application of FTIR spectroscopy together with SANS to the analysis of pottery shows great potential to understand the manufacturing and firing techniques of ceramic artifacts. Here, these two techniques allowed us to characterise the ceramic body of a set of Ionian cups, coming from Poira (Sicily, Italy) and dated back to VI– early V centuries BC. FT-IR spectra recorded on ceramic bodies revealed a general homogeneity in composition among the investigated samples. Quartz, k-feldspar, plagioclase, muscovite/illite, diopside and traces of calcite were detected in all the samples. The presence of diopside and bytownite/anorthite allowed to hypothesise a approximately firing temperature of 900 °C. On the other hand, SANS measurements allowed to characterise at mesoscopic scale the ceramic body of the investigated samples.
Sample
Section
Rg (Å)
a
POI 1
1 3
293.1 218.8 (Lo)–264.8 (Tr)
3.49 3.47 (Lo)–3.42 (Tr)
POI 3
1 2 3
279.2 309.2 298.9
3.79 3.85 3.79
POI 5
1 2 3
195.2 243.9 (Lo)–273.2 (Tr) 311.0
3.69 3.57 (Lo)–3.61 (Tr) 3.50
POI 13
1 2 3
288.3 300.3 298.0
3.91 3.92 3.68
POI 17
1 2 3
302.5 293.1 370.3
3.40 3.48 3.23
POI 19
1 2
193.4 (Lo)–272.4 (Tr) 138.0 (Lo)–194.1 (Tr)
3.54 (Lo)–3.54 (Tr) 3.67 (Lo)–3.62 (Tr)
Acknowledgments
POI 20
1 2 3
215.2 268.4 332.7 (Lo)–238.0 (Tr)
3.67 3.69 3.73 (Lo)–3.75 (Tr)
The research was supported by Ministero dell’Università e della Ricerca Scientifica e Tecnologica Grant MURST-PRIN2007. ‘‘Identification of the application fields of innovative non-destructive and microdestructive methods for the analysis of historical archaeological ceramic findings through the systematic comparison with the traditional methodologies’’. The authors are grateful to LLB (Laboratoire Léon Brillouin Saclay, France) for providing beam time and technical assistance.
In Table 3 we report the values of the radius of gyration Rg and the exponent a, obtained by the above described fitting procedure, for all the investigated samples. It is worth remarking that according to the used approach, Rg and a are uncorrelated since linked to different types of structures. By an inspection of the table, it can been observed that the obtained values of fractal exponent a are between 3 and 4, in agreement with the fractal surface model [13], for all the investigated samples. Regarding the fractal exponent a, higher values usually indicate mesoscopic structures (aggregates/voids) with a rougher surface [3]. The obtained values of Rg, ranging from 138 to 370.3 Å, can be associated to the mean-size of new formation minerals clusters/ crystallites. Taking into account the mineralogical composition obtained by FT-IR analysis, we could suppose that the extracted values of Rg, accounts for the mean sizes of new formation minerals, such as diopside and bytownite/anorthite. It is worth underlining that the mesoscopic characterisation could be useful to attribute the provenance of the investigated samples. In fact, under the assumption that samples manufactured from the same clay (i.e., in the same area) with the same firing history exhibit the same features at mesoscale level, these mesoscopic properties can be then considered parameters sensitive to the provenance of the artifacts. Hence, in order to identify the production sites of Ionian cups, the obtained data represent a preliminary analysis that we propose to study in depth by means of a comparison with SANS measurements on samples produced by
References [1] C. Casieri, C. Terenzi, F. De Luca, J. Appl. Phys. 105 (2009) 034901. [2] C. Terenzi, C. Casieri, A.C. Felici, M. Piacentini, M. Vendittelli, F. De Luca, J. Archaeol. Sci. 37 (2010) 1403. [3] A. Botti, M.A. Ricci, G. De Rossi, W. Kockelmann, A. Sodo, J. Archaeol. Sci. 33 (2006) 307. [4] S. Akyuz, T. Akyuz, S. Basaran, C. Bolcal, A. Gulec, Vib. Spectrosc. 48 (2008) 276. [5] B. Łydzba-Kopczyn´ska, E. Zych, C. August, G. Rusek, A. Pankiewicz, J. Mol. Struct. 887 (2008) 41. [6] A. Raškovska, B. Mincˇeva-Šukarova, O. Grupcˇea, Ph. Colomban, J. Raman Spectrosc. 41 (2010) 431. [7] C. Papachristodoulou, K. Gravani, A. Oikonomou, K. Ioannides, J. Archaeol. Sci. 37 (2010) 2146. [8] C.M. Belfiore, E. Aquilia, G. Barone, G. Lamagna, B.E. Mcconnwll, P. Mazzoleni, Period. Mineral. 79 (2010) 57. [9] Sadtler Database for FT-IR, BioRad Laboratories. [10] G.E. De Benedetto, R. Laviano, L. Sabbatini, P.G. Zambonin, J. Cult. Heritage 3 (2002) 177. [11] W. Griffith, Advances in the Raman and infrared spectroscopy of minerals spectroscopy of inorganic-based materials, in: R.J.H. Clark, R.E. Hester (Eds.), Wiley, Chichester, 1987 (Chapter 2). [12] E.L. Kendix, S. Prati, R. Mazzeo, E. Joseph, G. Sciuto, C. Fagnano, e-PS 7 (2010) 8. [13] G. Barone, V. Crupi, D. Majolino, P. Mazzoleni, J. Teixeira, V. Venuti, J. Appl. Phys. 106 (2009) 054904. [14] G. Barone, P. Mazzoleni, personal communication, 2010. [15] J. Capel, F. Huertas, J. Linares, Miner. Petrog. Acta 29 (1985) 563. [16] P. Duminuco, M.P. Riccardi, B. Messiga, M. Setti, Ceramurgia 5 (1996) 281. [17] M.P. Riccardi, B. Messiga, P. Duminuco, Appl. Clay Sci. 15 (1999) 393.
Contents lists available at ScienceDirect
Journal of Molecular Structure journal homepage: www.elsevier.com/locate/molstruc
Characterisation of archaeological pottery: The case of ‘‘Ionian Cups’’ G. Barone a, V. Crupi b, F. Longo b,⇑, D. Majolino b, P. Mazzoleni a, V. Venuti b a b
Dipartimento di Scienze Geologiche, Università di Catania, Corso Italia 57, 95129 Catania, Italy Dipartimento di Fisica, Università di Messina, V.le Ferdinando Stagno d’Alcontres 31, 98166 Messina, Italy
a r t i c l e
i n f o
Article history: Available online 23 January 2011 Keywords: FT-IR SANS Ceramics Firing temperature
a b s t r a c t The aim of this study was the microscopic and mesoscopic characterisation of archaeological pottery findings addressed to the identification of the manufacturing techniques. The samples under study were ‘‘Ionian Cups’’ sherds, coming from Poira, an archaeological site in eastern Sicily (South-Italy). These cups represent a ceramic typology widely diffused in the Mediterranean Area in archaic age (VI–V century BC). The identification of the production sites of these materials, originally manufactured in Greek-Eastern area and then largely diffused in Magna Graecia, is still an open question. Here, the microscopic structural characterisation was obtained by Fourier Transform Infrared absorption (FT-IR) measurements, which permitted us to determine the mineralogical phases present in the artefacts. Furthermore, Small Angle Neutron Scattering (SANS) measurements permitted the characterisation of the size distribution and surface characteristics of the mesoscopic aggregates formed by the minerals. Ó 2011 Elsevier B.V. All rights reserved.
1. Introduction Ceramic findings are among the best preserved and the most important materials testimonies of ancient human cultures. In fact, the study of a ceramic fragment allows to obtain information concerning the artistic choices, trade pathways, cultural and technological evolution of the ancient human communities [1,2]. As is well known, the characterisation of archaeological ceramics is very complex due to the heterogeneity of the materials on different scales and to the presence of both crystalline and amorphous phases in the ceramic bodies and in the outer decorations (glaze, pigments). So the use of different analytical techniques is required to characterise the ceramic findings [2–7]. In particular, manufacture technology involves several aspects of pottery making, such as the type of raw materials used, their processing to prepare the clay paste, the surface treatment, decoration and firing to obtain the finished item. Here, we present the employment of Fourier Transform Infrared absorption (FT-IR) and Small Angle Neutron Scattering (SANS) measurements to obtain the microscopic and mesoscopic characterisation of archaeological pottery. FTIR and SANS techniques have been performed on ‘‘Ionian Cups’’ sherds, a ceramic typology widely diffused in the Mediterranean Area in archaic age (VI–V century BC). In particular, the inves⇑ Corresponding author. Address: University of Messina, Department of Physics, V.le Ferdinando Stagno d’Alcontres 31, 98166 Messina, Italy. Tel.: +39 0906765463; fax: +39 090395004. E-mail address: [email protected] (F. Longo). 0022-2860/$ - see front matter Ó 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.molstruc.2011.01.028
tigated samples coming from Poira, an archaeological site near Catania (Sicily, Italy). The production of ‘‘Ionian Cups’’ started in about 580 BC in eastern areas of Greece and was then continued in Magna Graecia until about 540 BC [8]. The presence of various production centres in the Greek colonies have been hypothesised and represents a hotly debated question. Information on the manufacture, such as the raw materials and the estimation of firing temperature of this ceramic typology, could be useful to identify the production centres.
2. Experimental 2.1. Materials The investigated samples include seven fragments of Ionian cups of type B2, labelled as POI 1, 3, 5, 13, 17, 19, 20 from the Poira necropolis (Sicily, Southern Italy) and dated back to VI–early V centuries BC. An example of this ceramic typology, together with the pictures of two investigated samples, are reported in Fig. 1. These samples belong to a large set of Ionian cups from five archaeological sites in eastern Sicily (‘‘Mendolito, Monte Castellaccio, Poira-Poggio Cocola, Piano Casazzi, Francavilla di Sicilia’’). Petrographic thin-section analyses and chemical investigations by X-ray fluorescence spectrometry (XRF) have been already performed on all these Ionian cups [8]. This study indicated that a peculiar character of Ionian cups from these archaeological sites in eastern Sicily is due to their highly homogeneous petrographic and geochemical features. This may indicate a single centre of production, characterised
G. Barone et al. / Journal of Molecular Structure 993 (2011) 142–146
143
Fig. 1. An example of (a) Ionian cup of type B2, together with the pictures of (b) POI 17 and (c) POI 19 fragments.
by abundant production and an extensive distribution of products, or an extremely specialised technique that was known to several workshops. Hence, the microscopic and mesoscopic characterisation could support one among the hypotheses above mentioned. 2.2. FT-IR FT-IR measurements were performed by means of a BOMEM DA8 FT-IR spectrometer. This apparatus was equipped with a Globar lamp as source and a suitable beamsplitter and detector, depending on the part of the spectral range under study. In particular, a Hyper beamsplitter and a DTGS/FIR detector were used to collect the spectra in 200–600 cm1 range, while we used a KBr beamsplitter and a DTGS/MIR detector to register the spectra from 450 to 4000 cm1. The two measurements were combined to obtain the spectra in the range from 200 to 4000 cm1. In such configurations it was possible to use a resolution of 4 cm1. The investigated samples were prepared in pellets, about 0.5 mm thick, using small quantities (2 mg) of bulk sample dispersed in 200 mg of powdered CsI, that is transparent in the investigated IR frequency range. We remark that from each shard, about 2 mg of material was drawn from non-significant parts of the ceramic body, in order to avoid damages that could be affect the integrity and artistic content of the finding. The measurements were performed in dry atmosphere to avoid dirty contributions, 32 repetitive scans were automatically added to obtain a good signal-tonoise ratio (SNR) and a spectra reproducibility of high quality as well. The experimental spectra were compared with those of standard minerals and clays (Sadtler database ‘‘Minerals and Clays’’ [9]) and with data reported on different sources [10–12] for a reliable assignment of the bands. 2.3. SANS SANS measurements were carried out by using the PAXE spectrometer at the ORPHEE reactor of the Laboratoire Léon Brillouin
(LLB, Saclay, France). The two configurations used at large and small Q (scattering vector) range were summarised in Table 1. For a more detailed description of SANS technique the reader can refer to literature [3,13]. The investigated samples were cut in thin sections (thickness <1 mm) and therefore multiple scattering effects were minimised. No appreciable neutron activation of the samples was found after the SANS experiment. We used standard LLB SANS routines to correct the two-dimensional intensity distributions for instrumental background and to normalise them to absolute cross section per unit volume of the sample (cm1) by measuring the incident beam intensity, transmission and thickness of each sample. The structure of some samples under study turned out to be anisotropic. This may be either connected to preferential directions of the texture of mineral aggregates or to pressure effects. In such cases, Q must be treated as a vector and the xy pattern of the scattered intensity has the shape of more or less elongated ellipsoids. In this case, a longitudinal axis (Lo) along the major axis of the ellipse and, in the perpendicular direction, a transversal axis (Tr) along the minor axis, can be defined. The first corresponds to correlations in real space over shorter distances than the second. Numerical analysis is done in each direction, yielding distinct parameters. Table 1 Instrumental characteristics for SANS measurements performed by the PAXE spectrometer. Large Q Wavelength (k) Sample–detector distance Collimation Q range
6Å 2m 2m 5 103–2 101 Å1
Small Q Wavelength (k) Sample–detector distance Collimation Q range
15 Å 4.5 m 4.5 m 3 103–3 102 Å1
G. Barone et al. / Journal of Molecular Structure 993 (2011) 142–146
3. Results and discussion 3.1. FT-IR analysis
POI 1
2
1000
1500
2000
2500
3000
3500
-1
Wavenumber (cm ) Fig. 2. FT-IR absorbance spectrum of POI 1 sample.
4000
5
5 1,2,3 1
1
2
4,5 13,4
1
4 1 5 11
1000
1500
-1
Wavenumber (cm ) Fig. 3. FT-IR absorbance spectrum of POI 1 sample in the region from 200 to 1500 cm1. 1 = Quartz; 2 = diopside; 3 = muscovite/illite; 4 = bytownite/anorthite; 5 = k-feldspar.
POI 5 3
1 2 5
5
1
1
1
2
1,2,3 4
1 4,5 3,4 15 1 1 500
1000
1500 -1
Wavenumber (cm ) Fig. 4. FT-IR absorbance spectrum of POI 5 sample in the region from 200 to 1500 cm1. 1 = Quartz; 2 = diopside; 3 = muscovite/illite; 4 = bytownite/anorthite; 5 = k-feldspar.
POI 17
21 3 5
5
1,2,3
1
1
1
2 4
3,4,5 1 5 11 500
500
1
3
500
IR Absorbance (arb. units)
IR Absorbance (arb. units)
The experimental FT-IR spectra show similar features, typical of pottery materials, with a wide band ranging from 850 cm1 to 1400 cm1. In Fig. 2 the FT-IR absorbance spectrum recorded on POI 1, as example, is reported. However, as the absorption peaks of the main minerals in a typical absorbance spectrum of a ceramic occurred at wavenumbers lower than 1500 cm1 [9–12], only this region of the spectra are analysed and discussed in the following. The spectra of samples under study show similar shape, as it can be seen by an inspection of Figs. 3–5, where the spectra of POI 1, POI 5 and POI 17, as examples, were reported. The mineralogical composition obtained by the analysis of the spectra on the basis of characteristic IR wavenumbers of minerals and clays [9–12] is summarised in Table 2. First of all, the typical absorption peaks of quartz were detected in all the investigated samples: the main SiO2 stretching peak at 1080 cm1 that contribute to the large band of the spectra, the shoulder at 1170 cm1, the peaks at 505 and 692 cm1 and the distinctive doublet at 777 and 798 cm1. Furthermore in the low-frequency range (below 500 cm1) it was possible detect the other typical peaks associated to quartz (458, 395 and 370 cm1) [12]. Calcite in trace was revealed in almost the samples by the presence of the typical band centred at 1430 cm1. Beside the above mentioned minerals, it was possible identify the presence of k-feldspar, plagioclase, muscovite/illite and diopside (see Table 2 and Figs. 3–5). The revealed k-feldspar showed peaks centred at 1130, 1000, 720, 575 cm1. Regarding the plagioclase we revealed the contributions at 950, 575 and 532 cm1, that could be associated to anorthite or bytownite. These two members of plagioclase series do not show marked differences in their IR spectra, so it is very difficult to discriminate anorthite from bytownite [10]. The main contribution due to the Si–O stretching (1013 cm1) was found in all the samples. This mode could be attributed to muscovite or illite, because in this region their spectra do not show marked differences. The observed contributions at 533 and 463 cm1 allowed to suppose the presence of muscovite. This mineralogical phase was also observed by
POI 1 IR Absorbance (arb. units)
For most of the investigated samples, SANS spectra have been collected on different perpendicular sections aimed at putting into evidence the typology (linear or planar) and the spatial distribution of the preferred orientations of the mesoscopic grains.
IR Absorbance (arb. units)
144
1000
1500
-1
Wavenumber (cm ) Fig. 5. FT-IR absorbance spectrum of POI 17 sample in the region from 200 to 1500 cm1. 1 = Quartz; 2 = diopside; 3 = muscovite/illite; 4 = bytownite/anorthite; 5 = k-feldspar.
145
G. Barone et al. / Journal of Molecular Structure 993 (2011) 142–146
Qz
By/An
k-feld
Mu/Il
Cc
Di
POI POI POI POI POI POI POI
+++ +++ +++ +++ +++ +++ +++
+ ++ ++ ++ + + ++
+ tr + + tr + +
+ tr + tr + + +
tr tr tr tr tr tr tr
+ + + + + + +
1 3 5 13 17 19 20
petrographic thin-section analyses in a previous study on the same samples [8]. Finally, we revealed variable content of diopside, a new formed mineral, which was identified by the characteristic peaks at 1064, 960 and 467 cm1. Hence, the analysis of ceramic bodies (see Table 2 and Figs. 3–5) indicated a general homogeneity in composition among the investigated samples. The identification of the main mineralogical phases and, in particular, the presence of calcite revealed in trace, together with different contents of a new formed mineral, such as diopside, can be indicator of the firing temperature. In fact for temperatures higher than 750 °C the reaction of dissociation of calcium carbonate in calcium oxide and carbon dioxide occurs as evidenced by the thermogravimetric analysis carried out on the clays that probably were used in the production of studied Ionian cups [14]. If the temperature reaches 900 °C this decomposition is irreversible because calcium oxide reacts with clay components forming new Ca silicates (diopside and anorthite) [15–17]. Therefore, if the cooking of the manufacture occurs at approximately 900 °C, calcite disappears partially, also as function of microdomains composition [17] and is revealed in low content. So, the paragenetic composition allowed, according to literature [15–17], to hypothesise a firing temperature close to 900 °C for all the analysed samples.
(a)
5
10
4
10 -1
Sample
surface fractal dimension, Ds, by the following relation: a = 6 Ds. Rg indicates the mean-size of new formation minerals clusters/crystallites, whereas the exponent a refers to the roughness of the voids/minerals grains. In particular, since the value of Ds must be between 2 and 3, then 3 < a < 4. The two-dimensional Euclidean exponent Ds = 2 (and hence a = 4) is recovered for particles with sharp interfaces. In the following, the numbers 1–3 close to the name of each sample indicate, respectively, the sections exposed to the neutron beam perpendicular, 1 (transversal cut) and 2 (longitudinal cut), and parallel (3) to the surface of the Ionian cup. For some of the investigated samples, the structure appears to be isotropic in all the analysed sections, probably testifying not relevant consequences on the artefact connected to some external mechanical stress because of manufacturing methods. In other cases, we observed the presence of a ‘‘linear’’ anisotropy in a single section or, at the most, a ‘‘planar’’ anisotropy in two sections. This trend can be explained taking into account that the ceramic body of these very fine-grained potteries does not contain inclusions, so anisotropy can be just linked to the texture of the groundmass, as due to the forming and finishing of the manufacture. Furthermore, this is a characteristic that is not affected by all those mineralogical and structural transformations that occur during the firing process. The SANS spectra, in a log–log plot, of POI 3, together with the best-fit according to Eq. (1), are reported, as an example, in Fig. 6. The revealed low-level scattering background, related to the incoherent scattering due to hydrogen atoms, can be considered negligible.
I(Q)(cm )
Table 2 Mineralogical composition of the shards according to the infrared analysis. Qz = quartz; By/An = bytowinite/anorthite; k-feld = k-feldspar; Mu/Il = muscovite/ illite; Cc = calcite; Di = diopside. The ‘‘+’’ symbol stands for the relative abundance of the compounds in each sample; ‘‘tr’’ and ‘‘’’ symbols stand for trace and absent respectively.
3
10
2
10
1
10
0
10
-1
10
-2
10
3.2. SANS analysis
IðQ Þ ffi C 1 exp
Q 2 R2g 3
! þ C 2 Q a
ð1Þ
In this fitting law, four free parameters were introduced. The prefactors C1 and C2 contain the relative intensity of the Guinier’s and power law; the other two free parameters are the radius of gyration Rg and the exponent a, this last one is connected to the
(b)
-2
10
-1
10
-1
Q (Å )
5
10
4
-1
I(Q)(cm )
10
3
10
2
10
1
10
0
10
-1
10
-2
10
-3
10
-2
-1
10
-1
10
Q (Å )
(c)
5
10
4
10
3
10 -1
I(Q)(cm )
The interpretation of the SANS data was done using an empirical approach, that we already successfully employed in our previous work to study a set of archaeological ceramics [13]. According to this approach, the presence, in our samples, of two independent populations of mesoscopic structures with substantially different radii can be hypothesised. Their scattering contributions are supposed to be independent, and hence additive in the fitting law. The first population consists of ‘‘small’’ units, i.e. new formation minerals clusters/crystallites, whose typical dimensions can be in principle probed in the investigated spatial scale. On the contrary, the second population is composed by much ‘‘bigger’’ units, i.e. voids/mineral aggregates. Their typical sizes are out of the accessible experimental scale and can be derived by other techniques. For this second population, only surface can be probed. As a consequence, the Q-dependence of the scattered intensity I(Q) can be written as a combination of a Guinier law and a power law:
-3
10
2
10
1
10
0
10
-1
10
-2
10
-3
10
-2
10
-1
-1
Q (Å )
10
Fig. 6. Experimental (open squares) and fitted (solid line) SANS intensity for POI 3 sample analysed in section (a) 1, (b) 2 and (c) 3.
146
G. Barone et al. / Journal of Molecular Structure 993 (2011) 142–146
Table 3 Values of radius of gyration Rg and fractal exponent a as obtained by the fitting procedure for all the investigated samples. Error bars on Rg values are, on average, of the order of 7% (4% and 10% for the lower and the higher Rg values, respectively). Error bars on a values are of the order of 1%. Longitudinal and transversal configurations are indicated as (Lo) and (Tr), respectively.
using clays from sediments cropping out near the archaeological site. 4. Conclusions Application of FTIR spectroscopy together with SANS to the analysis of pottery shows great potential to understand the manufacturing and firing techniques of ceramic artifacts. Here, these two techniques allowed us to characterise the ceramic body of a set of Ionian cups, coming from Poira (Sicily, Italy) and dated back to VI– early V centuries BC. FT-IR spectra recorded on ceramic bodies revealed a general homogeneity in composition among the investigated samples. Quartz, k-feldspar, plagioclase, muscovite/illite, diopside and traces of calcite were detected in all the samples. The presence of diopside and bytownite/anorthite allowed to hypothesise a approximately firing temperature of 900 °C. On the other hand, SANS measurements allowed to characterise at mesoscopic scale the ceramic body of the investigated samples.
Sample
Section
Rg (Å)
a
POI 1
1 3
293.1 218.8 (Lo)–264.8 (Tr)
3.49 3.47 (Lo)–3.42 (Tr)
POI 3
1 2 3
279.2 309.2 298.9
3.79 3.85 3.79
POI 5
1 2 3
195.2 243.9 (Lo)–273.2 (Tr) 311.0
3.69 3.57 (Lo)–3.61 (Tr) 3.50
POI 13
1 2 3
288.3 300.3 298.0
3.91 3.92 3.68
POI 17
1 2 3
302.5 293.1 370.3
3.40 3.48 3.23
POI 19
1 2
193.4 (Lo)–272.4 (Tr) 138.0 (Lo)–194.1 (Tr)
3.54 (Lo)–3.54 (Tr) 3.67 (Lo)–3.62 (Tr)
Acknowledgments
POI 20
1 2 3
215.2 268.4 332.7 (Lo)–238.0 (Tr)
3.67 3.69 3.73 (Lo)–3.75 (Tr)
The research was supported by Ministero dell’Università e della Ricerca Scientifica e Tecnologica Grant MURST-PRIN2007. ‘‘Identification of the application fields of innovative non-destructive and microdestructive methods for the analysis of historical archaeological ceramic findings through the systematic comparison with the traditional methodologies’’. The authors are grateful to LLB (Laboratoire Léon Brillouin Saclay, France) for providing beam time and technical assistance.
In Table 3 we report the values of the radius of gyration Rg and the exponent a, obtained by the above described fitting procedure, for all the investigated samples. It is worth remarking that according to the used approach, Rg and a are uncorrelated since linked to different types of structures. By an inspection of the table, it can been observed that the obtained values of fractal exponent a are between 3 and 4, in agreement with the fractal surface model [13], for all the investigated samples. Regarding the fractal exponent a, higher values usually indicate mesoscopic structures (aggregates/voids) with a rougher surface [3]. The obtained values of Rg, ranging from 138 to 370.3 Å, can be associated to the mean-size of new formation minerals clusters/ crystallites. Taking into account the mineralogical composition obtained by FT-IR analysis, we could suppose that the extracted values of Rg, accounts for the mean sizes of new formation minerals, such as diopside and bytownite/anorthite. It is worth underlining that the mesoscopic characterisation could be useful to attribute the provenance of the investigated samples. In fact, under the assumption that samples manufactured from the same clay (i.e., in the same area) with the same firing history exhibit the same features at mesoscale level, these mesoscopic properties can be then considered parameters sensitive to the provenance of the artifacts. Hence, in order to identify the production sites of Ionian cups, the obtained data represent a preliminary analysis that we propose to study in depth by means of a comparison with SANS measurements on samples produced by
References [1] C. Casieri, C. Terenzi, F. De Luca, J. Appl. Phys. 105 (2009) 034901. [2] C. Terenzi, C. Casieri, A.C. Felici, M. Piacentini, M. Vendittelli, F. De Luca, J. Archaeol. Sci. 37 (2010) 1403. [3] A. Botti, M.A. Ricci, G. De Rossi, W. Kockelmann, A. Sodo, J. Archaeol. Sci. 33 (2006) 307. [4] S. Akyuz, T. Akyuz, S. Basaran, C. Bolcal, A. Gulec, Vib. Spectrosc. 48 (2008) 276. [5] B. Łydzba-Kopczyn´ska, E. Zych, C. August, G. Rusek, A. Pankiewicz, J. Mol. Struct. 887 (2008) 41. [6] A. Raškovska, B. Mincˇeva-Šukarova, O. Grupcˇea, Ph. Colomban, J. Raman Spectrosc. 41 (2010) 431. [7] C. Papachristodoulou, K. Gravani, A. Oikonomou, K. Ioannides, J. Archaeol. Sci. 37 (2010) 2146. [8] C.M. Belfiore, E. Aquilia, G. Barone, G. Lamagna, B.E. Mcconnwll, P. Mazzoleni, Period. Mineral. 79 (2010) 57. [9] Sadtler Database for FT-IR, BioRad Laboratories. [10] G.E. De Benedetto, R. Laviano, L. Sabbatini, P.G. Zambonin, J. Cult. Heritage 3 (2002) 177. [11] W. Griffith, Advances in the Raman and infrared spectroscopy of minerals spectroscopy of inorganic-based materials, in: R.J.H. Clark, R.E. Hester (Eds.), Wiley, Chichester, 1987 (Chapter 2). [12] E.L. Kendix, S. Prati, R. Mazzeo, E. Joseph, G. Sciuto, C. Fagnano, e-PS 7 (2010) 8. [13] G. Barone, V. Crupi, D. Majolino, P. Mazzoleni, J. Teixeira, V. Venuti, J. Appl. Phys. 106 (2009) 054904. [14] G. Barone, P. Mazzoleni, personal communication, 2010. [15] J. Capel, F. Huertas, J. Linares, Miner. Petrog. Acta 29 (1985) 563. [16] P. Duminuco, M.P. Riccardi, B. Messiga, M. Setti, Ceramurgia 5 (1996) 281. [17] M.P. Riccardi, B. Messiga, P. Duminuco, Appl. Clay Sci. 15 (1999) 393.