First attempt to obtain the bulk composition of ancient silver–copper coins by using XRF and GRT

Nuclear Instruments and Methods in Physics Research B 358 (2015) 93–97

Contents lists available at ScienceDirect

Nuclear Instruments and Methods in Physics Research B journal homepage: www.elsevier.com/locate/nimb

First attempt to obtain the bulk composition of ancient silver–copper coins by using XRF and GRT A.I. Moreno-Suárez a,b,⇑, F.J. Ager a,b, S. Scrivano a, I. Ortega-Feliu a, B. Gómez-Tubío a,c, M.A. Respaldiza a,d a

Centro Nacional de Aceleradores (Universidad de Sevilla-CSIC-Junta de Andalucía), Sevilla, Spain Departamento de Física Aplicada I, Universidad de Sevilla, Seville, Spain c Departamento de Física Aplicada III, Universidad de Sevilla, Seville, Spain d Departamento de Física Atómica, Molecular y Nuclear, Universidad de Sevilla, Seville, Spain b

a r t i c l e

i n f o

Article history: Received 19 January 2015 Received in revised form 26 April 2015 Accepted 29 May 2015

Keywords: XRF GRT Enrichment Silver Coin

a b s t r a c t Archeological silver–copper pieces often show surface enrichments in silver, either intentional or fortuitous. When this happens, non-destructive techniques like PIXE (Proton Induced X-ray Emission) and XRF (X-Ray Fluorescence) are not sufficient to access the whole bulk pieces because their penetration depths are typically of a few tens microns. If the archeological pieces cannot be cut or polished, it is necessary to apply other non-destructive techniques to access the bulk pieces. That way, archeological bronze pieces have been successfully studied combining XRF (or PIXE) with GRT (Gamma-Ray Transmission). In this work, the bulk composition of five silver Roman coins have been indirectly measured by combining XRF and GRT. These results were compared with previous works made by our group using the same coins by direct means of PIXE and XRF, so the accuracy of this indirect method could be tested. Ó 2015 Elsevier B.V. All rights reserved.

1. Introduction

2. Materials and methods

The archeological silver–copper pieces sometimes can have a surface enrichment of silver [1–15]. This enrichment can reach some hundreds of microns in depth [16] so it is not possible to determinate the fineness of the pieces by superficial techniques like XRF or PIXE because their penetration depths are shorter [3,4,17–20]. In these cases it is necessary to combine those superficial techniques with other techniques that allow in-depth analysis. Thus, the combination of XRF (and PIXE) with GRT has been successfully used in archeological bronzes [22,23]. Therefore, the combination of XRF with GRT has been used in the present work to study the bulk composition of silver–copper Roman coins. It is important to note that the bulk composition of the coins used in the present work was already studied by micro-XRF and micro-PIXE in a previous investigation [16]. With this knowledge, the results obtained in the present work can be compared with the previous ones, checking in that way the feasibility of the combination of XRF and GRT.

2.1. Coins

⇑ Corresponding author at: Centro Nacional de Aceleradores (Universidad de Sevilla-CSIC-Junta de Andalucía), Sevilla, Spain. E-mail addresses: [email protected], [email protected] (A.I. Moreno-Suárez). http://dx.doi.org/10.1016/j.nimb.2015.05.038 0168-583X/Ó 2015 Elsevier B.V. All rights reserved.

A set of six Roman Republican coins from 211 BC to 86 BC (four denarii and two victoriati) [16] were analyzed in the present work (Table 1). The classification of the coins was done according to Crawford [24]. Their surface elemental compositions were determined by means of XRF; meanwhile the GRT technique was used to correct in depth (volume) the concentration of the surface compositions obtained by XRF. All coins except the one named N1 were cut in halves. In a previous work [16], a cross-section of each coin was polished with a diamond solution up to 1 lm and then analyzed by micro-PIXE and micro-XRF (direct bulk measurements).

2.2. Experimental The experimental conditions of XRF, micro-XRF and micro-PIXE can be found elsewhere [16], so only a short description of XRF will be given here. The surface of the coins were measured by XRF using a portable X-ray tube with a W anode and a 12.5 lm thickness Be window. Its output was filtered with a 1 mm thick aluminum foil and the analyses were performed at 30 kV and 590 lA. A silicon drift detector (Ketek) with a 8 lm thickness Be window and a Zr

94

A.I. Moreno-Suárez et al. / Nuclear Instruments and Methods in Physics Research B 358 (2015) 93–97

element (lmixture) was theoretically calculated using the concentrations Ci given by XRF in the surface (Eq. (2)).

Table 1 Set of six Roman Republican coins analyzed [16]. Coin

Reference

N1 N2 N8 N9 N10 N11

Cr. Cr. Cr. Cr. Cr. Cr.

Monetary valor

218 350 113 114 53 83

Denarius Denarius Denarius Denarius Victoriatus Victoriatus

147 86 206–195 206–195 211 211–210

internal collimator was used. The acquired XRF spectra were analyzed using the WinQxas software [25]. Certified standards with similar compositions to these coins have been used to calibrate the system. The experimental GRT setup consisted of a 241Am (principal photon energy 60 keV) point gamma source (6 mm diameter) properly shielded with Pb and collimated (4 mm diameter), and a shielded NaI(Tl) detector surrounded by copper plates to absorb Pb X-rays from the shielding. A Canberra 2020 amplifier and a Canberra S100 multichannel analyzer were connected to the detector and the spectra were acquired with Genie-2000 software [26]. The same points analyzed by XRF were measured by GRT. The mass attenuation coefficient (l) needed for the GRT correction was calculated for each point (Eq. (1)) and its average values were then used to make the corrections in depth,

l¼

1 ln qx

P

Date (BC)

  I I0

ð1Þ

where I0 is the intensity of the incident photon beam, I is the intensity of the transmitted photon beam, x is thickness of the point analyzed and q is the mass density of the sample. The mass density q was obtained by an experimental method based in the principle of Archimedes and using a He pycnometer [27]. The samples were considered to be a binary compound of silver (first element) and a mixture of all other elements present in the alloy (second element called ‘‘mixture element’’). Under this hypothesis [21,22], the mass attenuation coefficient of the mixture

lmixture ¼ Pi–Ag

li C i

ð2Þ

i–Ag C i

Then the average mass attenuation coefficient (l) measured by GRT allows us to correct the concentration of the elements in the bulk. If a binary alloy of silver and the other elements (mixture) is considered, the final bulk concentration of silver is given by Eq. (3).

C Ag ¼

lmixture  l lmixture  lAg

ð3Þ

It is important to note that the density of the coins has to be determined accurately for a proper combination of XRF and GRT, which is no an easy work when the coins are very porous. 3. Results and discussion The XRF results of the surface of the coins given by Ager et al [16] were corrected in depth by using GRT and both are shown in Table 2 (‘‘surface’’ from [16] and ‘‘bulk’’ from this work). The average mass attenuation coefficient l used for each coin is given in Table 3. When the surface results from [16] are compared with the bulk results obtained in this work it is clear that the combination of the techniques XRF + GRT always predicts surface enrichment in silver since the elemental composition of the bulk (XRF + GRT) is lower than the elemental composition of the surface (XRF). In addition, Ager et al [16] demonstrated that not all coins had surface enrichment in silver: the three denarii (N2, N8 and N9) were homogeneous pieces as shown in Figs. 1 (coin N2) and 2 (coin N8), meanwhile the two victoriati (N10 and N11) presented a surface enrichment in silver as shown in Figs. 3 (coin N10) and 4 (coin N11). The micro-PIXE and micro-XRF Ag and Cu results of the bulk composition of the coins from [16] and the indirect results of the bulk composition made by combining XRF and GRT are shown in the Table 4. By comparing both results two main conclusions can be extracted:

Table 2 Mean concentrations and standard deviations obtained by XRF [16] of the surface of the coins (surface) and of the bulk composition of the coins indirectly obtained by combination of XRF and GRT (bulk). Coin

Concentrations (wt. %) Ag

Au

Cu

Fe

Zn

Pb

Bi

Hg

Mn

Br

N1

Surface Bulk

97.8 ± 0.9 83 ± 3

0.25 ± 0.25 2.2 ± 0.6

0.13 ± 0.05 1.2 ± 0.2

0.7 ± 0.4 4.9 ± 0.9

0.033 ± 0.010 0.27 ± 0.05

0.05 ± 0.03 0.45 ± 0.10

0.02 ± 0.02 0.16 ± 0.04

– –

0.02 ± 0.05 0.12 ± 0.04

1.0 ± 0.5 7.5 ± 1.4

N2

Surface Bulk

97.9 ± 0.5 90 ± 7

0.28 ± 0.02 1.4 ± 1.0

1.1 ± 0.3 5.4 ± 3.8

0.08 ± 0.03 0.44 ± 0.34

0.017 ± 0.011 0.10 ± 0.08

0.51 ± 0.14 2.5 ± 1.8

0.08 ± 0.02 0.37 ± 0.27

0.011 ± 0.007 0.05 ± 0.05

– –

– –

N8

Surface Bulk

97.5 ± 1.6 86 ± 5

0.6 ± 0.4 4.7 ± 2.0

1.0 ± 0.6 5.2 ± 1.9

0.048 ± 0.010 0.36 ± 0.14

0.037 ± 0.019 0.25 ± 0.10

0.5 ± 0.5 2.11 ± 0.86

0.04 ± 0.04 0.13 ± 0.06

0.003 ± 0.006 0.009 ± 0.007

0.23 ± 0.30 0.78 ± 0.40

0.04 ± 0.05 0.14 ± 0.07

N9

Surface Bulk

98.3 ± 0.4 90 ± 5

0.9 ± 0.3 5.0 ± 2.5

0.56 ± 0.03 3.3 ± 1.6

0.067 ± 0.014 0.39 ± 0.19

0.024 ± 0.005 0.15 ± 0.07

0.16 ± 0.02 0.91 ± 0.45

0.03 ± 0.02 0.36 ± 0.08

– –

0.005 ± 0.006 0.03 ± 0.02

0.02 ± 0.02 0.08 ± 0.06

N10

Surface Bulk

95.5 ± 1.5 60 ± 6

0.28 ± 0.03 2.6 ± 0.5

3.8 ± 1.5 32.8 ± 5.3

0.046 ± 0.008 0.44 ± 0.08

0.024 ± 0.005 0.23 ± 0.05

0.34 ± 0.06 3.10 ± 0.58

0.069 ± 0.012 0.64 ± 0.12

0.005 ± 0.006 0.04 ± 0.01

0.003 ± 0.005 0.03 ± 0.01

– –

N11

Surface Bulk

94.2 ± 0.6 64 ± 3

0.49 ± 0.04 3.1 ± 0.3

4.8 ± 0.6 30.3 ± 2.1

0.11 ± 0.07 0.70 ± 0.06

0.031 ± 0.008 0.19 ± 0.02

0.29 ± 0.06 1.81 ± 0.20

0.026 ± 0.005 0.16 ± 0.02

0.010 ± 0.007 0.07 ± 0.01

0.009 ± 0.006 0.059 ± 0.009

0.02 ± 0.02 0.10 ± 0.01

Table 3 Average mass attenuation coefficients of the coins used in the present study.

l (cm2/g)

N1

N2

N8

N9

N10

N11

5.22 ± 0.14

5.49 ± 0.35

5.42 ± 0.14

5.543 ± 0.088

4.32 ± 0.43

4.424 ± 0.071

A.I. Moreno-Suárez et al. / Nuclear Instruments and Methods in Physics Research B 358 (2015) 93–97

95

Fig. 1. Micro-PIXE elemental maps of Ag and Cu of the cut cross-section of coin N2 (analyzed area: 2.5  2.5 mm2) [16].

Fig. 3. Micro-PIXE elemental maps of the cross-section of coin N10 (analyzed area: 250  250 lm2) [16].

(a) For those coins having silver surface enrichment (N10 and N11) the results provided by XRF + GRT are in good agreement with those obtained by micro-PIXE and micro-XRF. The differences between the direct (micro-PIXE and micro-XRF) and the indirect (XRF + GRT) measurements could be attributable to the porosity of the coins. Fig. 5 shows a SEM (Scanning Electron Microscope) image of a region of the cross-section of coin N11 where porosity is clearly observed. It has to be noticed that the mass density used in the calculation is very important for a suitable correction by GRT. It has been checked that the higher is the precision achieved in the measurement of the densities, the better are the XRF + GRT results (they are more close to the direct results given by micro-PIXE and micro-XRF). The method to measure the density of the samples must be improved more. The surface enriched layers were also seen in the SEM images, as shown in Fig. 6, where a clear difference in thickness between layers for coin N11 can be observed. This difference in thickness between enriched layers in coin N11 can also be observed in Fig. 4, where two different enriched layers were found: one 250 lm thick in

one side and another one about 200 lm thick in the other side. For coin N10 those enriched layers were calculated to be about 150 lm and 200 lm thick. The differences between the thicknesses of enriched layers in the same coin could be attributable to the manufacture process. For example, each side of the coin could have been contacted with materials of different thermal conductivities during the solidification or each side could have received different treatments or the same treatment but with different durations. (b) In the case of the homogeneous coins (N2, N8 and N9), the results provided by XRF and GRT (indirect measurement) are different from those obtained by micro-PIXE and micro-XRF in, at largest, 15% Ag. The works by Cesareo and Mancini [28,29] suggest that the cause of this discrepancy between surface and bulk composition for homogeneous coins could be in the elemental mass attenuation coefficient l of Ag (5.766 cm2/g), Bi (5.233 cm2/g), Pb (5.021 cm2/g), Hg (4.683 cm2/g) and Au (4.528 cm2/g) [31], as they are similar for a 60 keV radiation, the energy used with the 241Am gamma source in the GRT set-up. This similarity among

Fig. 2. Micro-XRF elemental profiles of Ag and Cu across the cross-section of coin N8 [16].

96

A.I. Moreno-Suárez et al. / Nuclear Instruments and Methods in Physics Research B 358 (2015) 93–97

Fig. 4. Micro-XRF elemental profiles of Ag and Cu across the cross-section of coin N11 [16].

Table 4 Ag and Cu concentrations in bulk compositions obtained by micro-PIXE and microXRF [16] and XRF + GRT. Coin

N1 N2 N8 N9 N10 N11

Bulk concentrations (wt.%) Micro-PIXE [16]

Micro-XRF [16]

Ag

Cu

Ag

Cu

XRF + GRT Ag

Cu

– 97.8 ± 0.8 97.4 ± 0.4 – 63.7 ± 0.4 56.7 ± 0.2

– 1.35 ± 0.02 1.96 ± 0.01 – 35.6 ± 0.2 42.5 ± 0.1

– 97.1 ± 0.3 97.2 ± 0.4 96.2 ± 0.2 61.3 ± 2.4 58.3 ± 1.9

– 1.7 ± 0.2 1.8 ± 0.3 2.4 ± 0.2 38.6 ± 2.4 41.6 ± 1.9

83 ± 3 90 ± 7 86 ± 5 90 ± 5 60 ± 6 64 ± 3

1.2 ± 0.2 5.4 ± 3.8 5.2 ± 1.9 3.3 ± 1.6 32.8 ± 5.3 30.3 ± 2.1

the elemental mass attenuation coefficients would give an incorrect calculation of lmixture and consequently, an incorrect result of Eq. (3). However, a further study is needed to understand this issue. On the other hand, the coin N1 represents a special case. The previous work by Ager et al. [16] could not demonstrate whether it was homogeneous or not since it was not cut. In the present work this coin N1 is considered as homogeneous because three

Fig. 6. SEM image of a region of the cross-section of coin N11. Different thicknesses were observed in surface enriched layers.

reasons: its surface Ag concentration is similar to N2, N8 and N9 (homogeneous coins), it is a denarius as N2, N8 and N9 and finally, its bulk Ag concentration is similar to N2, N8 and N9.

Fig. 5. SEM image of a region of the cross-section of coin N11. Porosity is clearly observed both in surface enrichments (shown here) and in the bulk composition.

A.I. Moreno-Suárez et al. / Nuclear Instruments and Methods in Physics Research B 358 (2015) 93–97

4. Conclusions The bulk composition of silver coins shown by Ager et al [16] has been indirectly measured by a combination of XRF and GRT. The GRT correction gives results in agreement with those obtained by micro-PIXE and micro-XRF when the coins have surface enrichments, although the method to measure the density of the coins has to be optimized. For the homogeneous coins, the results show a discrepancy not larger than 15% Ag compared to those obtained by micro-PIXE and micro-XRF. This discrepancy needs a further study to be fully understood. Considering the results presented in this work and previous results [21,23,30], it can be concluded that the combination of XRF and GRT gives good results for bulk concentrations of coins if: (1) surface and bulk show different concentrations (there is an enrichment and/or corrosion of the surface of the coin) and (2) bulk is an approximately homogeneous mixture. However, further studies are needed to improve and optimize the method. Acknowledgments We thank Pierluigi Debernardi for supplying most of the samples. Work partially supported by the project HAR2009-07449 from the Spanish Ministry of Science and Innovation and Criolab Lda. Company (Portugal). References [1] L.H. Cope. Surface-silvered ancient coins, in: E.T. Hall, D.M. Metcalf (Eds.), Methods of Chemical and Metallurgical Investigation of Ancient Coinage. RNS Special Publication 8, Cambridge (1972) 261–278. [2] S. La Nièce, Silvering, in: S. La Nièce, P. Craddock (Eds.), Metal Plating and Patination, Butterworth Heinemann, Oxford, 1993, pp. 201–210. [3] L. Beck, S. Bosonnet, S. Réveillon, D. Eliot, F. Pilon, Silver surface enrichment of silver–copper alloys: a limitation for the analysis of ancient silver coins by surface techniques, Nucl. Instr. Meth. B 226 (2004) 153–162. [4] L. Beck, E. Alloin, C. Berthier, S. Réveillon, V. Costa, Silver surface enrichment controlled by simultaneous RBS for reliable PIXE analysis of ancient coins, Nucl. Instr. Meth. B 266 (2008) 2320–2324. [5] E.T. Hall, Surface enrichment of buried metals, Archaeometry 4 (1961) 62–66. [6] G.F. Carter, Reproducibility of X-ray fluorescence analyses of Septimius Severus denarii, Archaeometry 19 (1977) 67–73. [7] Mehmet Basutçu. Contribution à l’analyse par réactions nucléaires à l’aide de particules alpha de basse énergie (<3.5 MeV): étude comparative du dosage de l’argent par des réactions nucléaires non destructives en numismatique (Doctoral Thesis), University of Paris VI, 1980. [8] Z. Smit, P. Kos, Elemental analysis of Celtic coins, Nucl. Instr. Meth. B 3 (1984) 416–418. [9] D.A. Scott, Gold and silver alloy coatings over copper: an examination of some artefacts from Ecuador and Colombia, Archaeometry 28 (1986) 33–50. [10] Marie.-Anne. Meyer, Guy. Demortier, Nonvacuum analyses of silver coins (9th to 15th century A.D.), Nucl. Instr. Meth. B 49 (1990) 300–304. [11] R. Linke, M. Schreiner, Energy dispersive X-Ray fluorescence analysis and XRay microanalysis of medieval silver coins, Microchim. Acta 133 (2000) 165– 170. [12] G. Weber, J. Guillaume, D. Strivay, H.P. Garnir, A. Marchal, L. Martinot, Is the external beam PIXE method suitable for determining ancient silver artifact fineness?, Nucl Instr. Meth. B 161–163 (2000) 724–729.

97

[13] Virginia. Costa, The Deterioration of Silver Alloys and Some Aspects of their Conservation, vol. 2, Maney Publishing Ó W. S. Maney & Son Limited, 2001, pp. 18–34 (Rewiews in conservation). [14] Haim Gitler, Matthew Ponting, The silver coinage of Septimius Severus and his family (193-211 a. d.). A study of the chemical composition of the Roman and Eastern Issues. Glaux 16, Edizioni ennerre S.r.l., Milan, 2003. [15] R. Linke, M. Schreiner, G. Demortier, M. Alram, Determination of the provenance of medieval silver coins: potential and limitations of x-ray analysis using photons, electrons or protons, X-Ray Spectrom. 32 (2003) 373–380. [16] F.J. Ager, A.I. Moreno-Suárez, S. Scrivano, I. Ortega-Feliu, B. Gómez-Tubío, M.A. Respaldiza, Silver surface enrichment in ancient coins studies by micro-PIXE, Nucl. Instr. Meth. B 306 (2013) 241–244. [17] Vasiliki Kantarelou, Francisco José Ager, Despoina Eugenidou, Francisca Chaves, Alexandros Andreou, Elena Kontou, Niki Katsikosta, Miguel Ángel Respaldiza, Patrizia Serafin, Dimosthenis Sokaras, Charalambos Zarkadas, Kyriaki Polikreti, Andreas Germanos Karydas, X-ray fluorescence analytical criteria to assess the fineness of ancient silver coins: application on ptolemaic coinage, Spectrochim. Acta Part B (2011) 681–690. [18] M. Rodrigues, F. Cappa, M. Schreiner, P. Ferloni, M. Radtke, U. Reinholz, B. Woytek, M. Alram, Further metallurgical analyses on silver coins of Trajan (AD 98–117), Journal of Analytical Atomic Spectrometry, Themed issue: Synchrotron Radiation in Art and Archaeometry 26 (2011) 984–991. [19] M. Rodrigues, M. Schreiner, M. Melcher, M. Guerra, J. Salomon, M. Radtke, M. Alram, N. Schindel, Characterization of the silver coins of the Hoard of Beçin by X-ray based methods, Nucl. Instr. Meth. B 269 (2011) 3041–3045. [20] R. Linke, M. Schereiner, G. Demortier, M. Alram, H. Winter. The provenance of medieval silver coins: analysis with EDXRF, SEM/EDX and PIXE, in: Nondestructive Microanalysis of Cultural Heritage Materials 42 (2004) 605-633. [21] M.A. Respaldiza, F. Barranco, J. Gómez-Camacho, B.M. Gómez-Tubío, M.M. Ruíz-Delgado, Combining PIXE and XRF with Gamma-Ray Transmission to get accurate analysis of archaeological bronzes, Nucl. Instr. Meth. B 50 (1990) 226–230. [22] M.A. Respaldiza, B.M. Gómez-Tubío, A. Sánchez del Junco, F. Barranco, C. SáizJiménez, Non-destructive analysis of archaeological bronzes by nuclear techniques, Nucl. Instr. Meth. B 89 (1994) 109–113. [23] A.I. Moreno-Suárez, B. Gómez-Tubío, M.A. Respaldiza, F. Chaves, I. OrtegaFeliu, M.Á. Ontalba-Salamanca, F.J. Ager, Combining non-destructive nuclear techniques to study Roman leaded copper coins from Ilipa (II-I centuries B.C.), Nucl. Instr. Meth. B 269 (2011) 3098–3101. [24] M.H. Crawford, Roman Republican Coinage, Cambridge University Press, 1982 (Reprint of 1974 original). [25] WinQXAS – Quantitative X-ray Analysis System for MS Windows operating system, version 1.40, copyright Ó 2002 International Atomic Energy Agency. [26] Available at: . [27] Centro de Estudios y de Experimentación de Obras Públicas (CEDES). La picnometría del helio. Fundamentos físicos, normativa, descripción del equipo y procedimiento experimental. Ministerio de Fomento y Ministerio de Medio Ambiente y Medio Rural y Marino. Gobierno de España. (In Spanish). [28] R. Cesareo, C. Mancini, Non-destructive analysis of silver alloys by means of low energy c-ray and neutron transmission measurements, Int. J. Appl. Radiat. Isot. 30 (1979) 589–594 (Pergamon Press Ltd.). [29] Carlo. Mancini, Neutron and gamma ray transmission: methods and accuracy for the analysis of silver and gold alloys, Int. J. Appl. Radiat. Isot. 36 (6) (1985) 489–494 (Pergamon Press Ltd.). [30] I. Ortega-Feliu, A.I. Moreno-Suárez, B. Gómez-Tubío, F.J. Ager, M.A. Respaldiza, S. García-Dils, O. Rodríguez-Gutiérrez, A comparative study of PIXE and XRF corrected by Gamma-Ray Transmission for the non-destructive characterization of a gilded roman railing, Nucl. Instr. Meth. B 268 (2010) 1920–1923. [31] National Institute of Standards and Technology. .