Studies on ancient silver metallurgy using SR XRF and micro-PIXE

Author’s Accepted Manuscript Studies on ANCIENT Silver MetallURGY USING Sr xrf AND Micro-pixe Angela Vasilescu, Bogdan Constantinescu, Daniela Stan, Martin Radtke, Uwe Reinholz, Guenter Buzanich, Daniele Ceccato www.elsevier.com/locate/radphyschem

PII: DOI: Reference:

S0969-806X(15)30015-3 http://dx.doi.org/10.1016/j.radphyschem.2015.07.008 RPC6872

To appear in: Radiation Physics and Chemistry Received date: 19 March 2015 Revised date: 18 June 2015 Accepted date: 14 July 2015 Cite this article as: Angela Vasilescu, Bogdan Constantinescu, Daniela Stan, Martin Radtke, Uwe Reinholz, Guenter Buzanich and Daniele Ceccato, Studies on ANCIENT Silver MetallURGY USING Sr xrf AND Micro-pixe, Radiation Physics and Chemistry, http://dx.doi.org/10.1016/j.radphyschem.2015.07.008 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

STUDIES ON ANCIENT SILVER METALLURGY USING SR XRF AND MICRO-PIXE Angela Vasilescu a,*, Bogdan Constantinescu a, Daniela Stan a, Martin Radtke b, Uwe Reinholz b, Guenter Buzanich b and Daniele Ceccato c

a Horia Hulubei National Institute for Nuclear Physics and Engineering Bucharest, str. Reactorului 30, Magurele, Ilfov, 077125 Romania; b Federal Institute for Materials Research and Testing (BAM), Richard-Wilstätter Strasse 11, 12489, Berlin, Germany; c Dipartamento di Fisica Galileo Galilei,Universita di Padova; INFN, Laboratori Nazionali di Legnaro, 35020 Legnaro (Padova), Italy *corresponding author: Fax: +40-21404-2391; Tel: +40-21404-6103; E-mail: [email protected]

Abstract This work presents the a complex evaluation of a series of Geto-Thracian silver adornments found on Romanian th

st

territory, part of the 4 century BC Agighiol (Northern Dobruja) hoard and of an ingot from the 1 century BC Geto-Dacian Surcea (Transylvania) hoard, using Synchrotron Radiation X-Ray Fluorescence and micro- Proton Induced X-ray Emission analysis and mapping in order to investigate aspects related to the elemental composition of the metal and the metallurgy implied in their manufacture. One of the samples can be linked to Laurion as the source of metal, and several items contain silver probably originated in Macedonia. The set of silver items was found to be heteregenous as composition and microstructure, and corrosion-related elements could be also identified in the X-Ray maps. Keywords: Synchrotron radiation XRF; micro-PIXE; archaeological silver 1. INTRODUCTION Synchrotron Radiation X-Ray Fluorescence (SR XRF) and micro- Particle (proton) Induced X-ray Emission (PIXE) have been used with great success in the microscopic investigation of various archaeological and art objects, as proven by a host of publications, including review articles (Bertrand et al 2012, Bertrand et al 2013, Guerra and Calligaro 2003).

The information extracted from the quantitative X-ray microanalysis of archaelogical silver can possibly answer questions about their provenance, the metallurgical techniques used in their manufacture and their corrosion -status. The study of the metallurgical aspects of old silver adornments can lead to conclusions on mining, metalw orking, crafts and trades, commercial routes.

Based on the experience of previous studies for ancient gold objects (e.g. the case of the famous Dacian gold bracelets (Constantinescu et al 2002, Constantinescu et al 2012), where the presence of certain trace elements like Sn, Sb, Te etc proved to be useful in order to identify the most probable sources of metal for their manufacture

within the historical background, and even help to authenticate the respective items, a similar approach is applied here to archaelogical silver. Both the minerals and the metallurgy of gold and silver have their specificities, but, keeping this in mind, one can try to correlate their chemical composition and the presence of certain trace -elements in the material with geological and historical knowledge.

The goal of the study was to evaluate the homogeneity of the material in relation to the metallurgical process to obtain information on the silver mineral probably used (presence of Au, Bi, Zn, Sb – provenance of the mineral?), whether there were elements added intentionally in the metallurgical process (Cu, Pb?) and to check the presence of corrosion-related elements (Br, Cl, Cu) (Hedges 1976, Wanhill 2005) in the microscopic map of their structure. This is why both the quantitative evaluation of the concentration of various elements and the mapping of the microstructure of the samples, demonstrating simultaneous or complementary presence of various elements, are important in our case.

The interest of the archaeologists to study silver museum objects brought about previous work on Bronze Age artefacts, mainly of auriferous silver or a high-copper content silver alloy (Vulchitrun disk, Poduri dagger, coins) (Bondoc and Constantinescu 2003-2005, Constantinescu et al 2003, Constantinescu et al 2010). 2. EXPERIMENTAL

2.1. The samples This paper presents the compositional analysis of a series of Geto -Thracian adornments found on Romanian territory, th

st

part of the 4 c. BC Agighiol (Berciu 1969) (in Northern Dobruja) hoard and one item from the 1 century BC GetoDacian Surcea (Popescu 1972) (in Transylvania) hoard, by SR XRF and micro PIXE.

The samples were tiny (less than 1mm diameter) chips taken from the silver objects.

The study was aimed to complement and deepen previous wide-range in-situ XRF measurements. th

The Agighiol hoard (Berciu 1969) (fig.1) is of Geto-Thracian origin and was found in Northern Dobruja. It is a 4 c. BC deposit (Prunkgräber/ceremonial tomb), containing silver adornments, mainly appliques, beads, buttons (for horse harnesses). It includes, among other objects (fig. 1), a silver helmet almost identical with the Thracian helmet from the Detroit Institute of Arts and a cup/beaker similar to one from the Metropolitan Museum of Art – having the exact same tool mark, as if made by the same smith with the same tools in the same workshop. Both mentioned artefacts have been described by Farkas and Meyers (Farkas 1982, Meyers 1982). st

The Surcea Hoard (Popescu 1972), discovered in Transylvania, in Covasna County, is of later date, 1 c. BC, and represents the Geto-Dacian culture. From this hoard, we analysed a silver ingot which was found together with tools (iron anvils) and several gilded silver object (household silver, jewellery). There are no silver mines known to hav e existed in the area, but it is possible that some metallurgical skills were developed here before the arrival of the

Romans in the north of the Danube.

2.2. The experimental set-up

The Synchrotron Radiation X-Ray Fluorescence (SR XRF) investigation was performed at the BAM-line at the Helmholtz Centre Berlin for Materials and Energy HZB – BESSY II synchrotron facility (Radtke et al 2013, Reiche 2

et al 2006). The excitation-energy of 20 keV and a beam size of 100x100µm were used. The spot-size for mapping was 2.5µmx2.5µm, and the 40x40 maps were centered on previously chosen points. Quantitative evaluation was based on a combination of measurements of metallic standards and Monte Carlo simulation (fundamental parameters), described in detail elsewhere (He and van Espen 1991, Vincze at al 1995, Radtke et al 2010).

The Micro-PIXE experiment was performed at the LNL microprobe at INFN Legnaro, in the 2MeV p microbeam (Boccaccio et al 1996). The beam size was 6µmx6µm. The size of the map was chosen individually for each sample. Actually the experiment at LNL was earlier and more of the exploratory type than the more refined one at BESSY, but covered a larger amount of samples. As the beam-time required at BESSY was limited, and the fluorescence spectrum covered a wider range of energies, we insisted only on the most “interesting” samples to be investigated in more detail.

3. RESULTS AND DISCUSSION

th

Silver is known to have been used by man as early as the 4 millennium BC. However, silver objects seem to have been relatively rare in Europe before the Greek and Roman times, with only few items representing Bronze Age cultures. Apparently, in spite of the fact that technically advanced products of other metals (copper, gold, iron) were current in Europe, and that, in the Near East, eastern Mediterranean/ Aegean region, silver was abundant, the metallurgical techniques for silver and lead might not have been known or have advanced towards Europe very slowly (Bartelheim et al 2012).

The separation of silver from lead ores (practically only from argentiferous galena) - known as cupellation - was widespread during the Classical Greek and the Roman times. The procedure is very effective in producing above 95% wt% purity, but usually the metal contains minor-to-trace amounts of gold, copper, lead and bismuth (<1%), and traces of Sb, As, Te, Zn, Ni (Wanhill 2005) – their naturally geologically associated elements (if present in the ore). Thus, copper contents above 0.5-1wt % indicate deliberate addition to increase the strength of the metal.

The accepted standard method to determine the provenance of metal (Stos-Gale and Gale 2009) using lead isotopes, which would give a more precise localization for the ore source, was unfortunately not accessible to us. However, we

can discuss these objects within the limits of our approach. We have to note that, in the case of silver, due to the characteristic of the galena-type sources and the technology, possible mixing with Pb from other sources during manufacture might hide the original lead isotope fingerprint (He and van Espen 1991, Vincze at al 1995, Radtke et al 2010).

In the case of silver, gold, bismuth, zinc and antimony can be used as fingerprint elements for East European geological deposits, e.g. Bi for South Thracian and Greek, Macedonian or Aegean argentiferous galena, Sb being typical to the Northern Carpathians (Maramures, Slovakia) (Chovan et al 2011, Gale et al 1980, Gale and Stos-Gale 1981, Gitler et al 2009, Stos-Gale and Gale 2009). Actually only Au and Bi can be reliably regarded as associated with the silver source (Stos-Gale and Gale 2009) (their levels not being affected at all or only very slightly by smelting and refining). Other elements can be contaminants or intentionally added elements - the presence of copper and lead is directly related to the technology.

3.1. Composition/minor and trace elements

Our research started with in-situ measurements of the cultural heritage items at the National Museum of History of Romania, Bucharest, using a portable X-Ray X-MET 3000TX instrument, with 10% maximum uncertainty (Oxford 2006). A selection of the more significant results of the measurements is given in Table 1.

The main conclusions of the study were the following (Internal report): - Two different types of silver were identified: -

Silver with traces of Bi for a group of several objects: of which some contain lead, but also added Cu, in

order to increase the mechanical resistance of the (type A2) appliques. The rest of items do not contain Pb. -

Silver without Bi: again, with two sub-types, with and without Pb.

- The silver with Bi presents also 3 sub-classes, considering the Au content: -

Bi, Pb, Au 0.5%

-

Bi, Au 0.5%, no Pb

-

Bi, Au 1%, no Pb.

The silver composition for the Agighiol helmet is Ag 98.6%, Au 1%, Cu 0.3%, B i 0.05%, Pb 0.05%, and for the Agighiol large cup, it is Ag 99.3%, Au 0.5%, Cu 0.05%, Bi 0.1%, Pb 0.05%. They are similar with their counterparts described by Meyers (Farkas 1982, Meyers 1982), but contain more Au (about twice as much) in our case.

The identified elements are also different, but this is due to the different experimental techniques applied. Bi cannot be measured in neutron activation analysis, the method used by Meyers (Meyers 1982). Unfortunately, we could not obtain any microscopic samples for SR XRF from these items.

The SR XRF measurements yielded more precise values on the microscopic samples, and the results are given in Table 2. The maximum uncertainty levels for these results are 1% for major and minor elements and 10% for traces (<0.01 wt% concentration). Only Ti, Cr, Zn and Zr occur at trace levels in some samples; standards were measured for all of them. Sn was not measured due to the fact that we chose 20keV as excitation energy, to get better resolution in the low-energy region of the fluorescence spectrum.

To summarize, the silver from Agighiol shows the following characteristics: -

-4

-3

Ag content: 95-98.7%; Au low, as a rule, less than 1%, Au/Ag ~ 10 – 10 , comparable to the silver from

the mines in Thassos or Laurion, in Greece. -

Cu is generally <1%, a few samples ~2-3 wt%, and a single one 12.5% (for silver obtained by cupellation

from galena, Cu should be below 0.5%, unless it was intentionally added in the metallurgical process). -

The lead content is 0.01-0.5%, with one exception - 0.98%, and Bi is between 0.02- and 0.4%.

In order to compare with other silver provenance data (Gale et al 1980, Gale and Stos-Gale 1981, Gitler et al 2009), in Fig. 2 we plotted the Au/Ag versus Bi/Ag ratios for the data from Table 2. The conclusion from this graph is that only one sample is in the low Au/low Bi Laurion group, and that, as we already commented, the Agighiol hoard is very heterogeneous (different workshops and masters).

More detailed knowledge of combined analysis with Au/Ag, Bi/Ag ratios and lead isotope data could offer a better insight to the provenance of the silver (original ore). The high Au/moderate Bi groups discussed by Gitler et al (Gitler et al 2009) were supposed to be of mixed metal, from possible other Greek sources, Maced onia, Siphnos, Halkidiki or even Anatolia. (We did not find other reliable data containing both Au/Ag or Bi/Ag ratios and Pb isotope data on other sources.)

In our plot, the data from Agighiol with Au/Ag ratio in the region of 0.005 -0.01 can be attributed maybe to the Macedonia group from (Gale et al 1980).

The composition of the Surcea ingot was not studied in more detail, after the preliminary evaluation (Table 1).

3.2. Mapping and correlation: Corrosion

Mapping of the microstructure of the silver can bring insight in the local distribution of elements and their grouping. The surface is scanned with the microbeam and the resulting X-Ray spectra are recorded. The maps contain compositional information (actually the primary maps are spatial distributions of X-ray intensity values), so one can obtain a 3D representation of the fluorescence radiation emitted by the sample over a selected area, for every energy value in the spectrum, or a projection (2D) for each element of interest.

We have looked at the distribution of major and minor elements and also at traces or impurities. Some of these

“pictures” can be interpreted as microstructure and some can give information which can be correlated with the embrittlement of the material.

Corrosion is an important aspect for conservationists. Factors related to the object like burial time and average temperature, moisture content, pH and chemical composition of the environment, especially salt, nitrate, nitrite, Cl, Br contents are also important. Silver is known to be embrittled by the presence of Pb, Sn, Sb (Hedges 1976, Wanhill 2005). For silver artefacts, corrosion contributes to the embrittlement of the material. Corrosion -induced embrittlement is mainly due to copper segregation, but Pb conjoint with Bi can al so cause microstructural embrittlement.

Mapping in the SR XRF-beam at BAM-line was performed on a whole set of samples previously measured for composition, as point-spectra. The aim of this run was to determine whether the localization of minor or trace elements can be in anyway correlated with the major elements, and to find a possible interpretation of such correlation. The plots represent relative intensity plots of characteristic X-rays emitted by selected elements of interest, in greyscale, the darker areas representing higher intensity. 2

The spot-size for mapping was 2.5µmx2.5µm and the 40x40m maps were centred on previously chosen points (measured as point-spectra) on the whole set of samples on the holder, which were measured in a single overnight long session. Map reconstruction was done off-line. Mapping at BESSY allowed for finer probing into the structure of the material.

At the LNL micro-PIXE microprobe, scanning and mapping can be done automatically and the intensity maps are registered during the measurement. Only more elaborate evaluation implies off -line programs. The beam spot was 6µm x6µm. The size of the map was chosen individually for each sample , covering the whole area of it.

Several case studies are given below to illustrate the different kind of information which could be derived from mapping silver, focusing especially on aspects like structure, inclusions and corrosion.

3.2.1. Case 1: Agighiol bead (BESSY)

A first case is illustrated in Fig. 3.

The sample is a metal scrap from a bead, and it represents a somewhat large piece of solid silver. Both in the microphotograph (picture taken with the microscope used for positioning the samples in the beam) and in the map we can see a series of more or less parallel lines across the area, which can be explained by the laminated structure of the silver item, obtained probably by (cold?) hammering. Bi and Pb grains are along the interface, between the different layers.

3.2.2. Case 2: Agighiol applique 8742 (BESSY)

Fig. 4 presents two sets of 2D-maps of a tiny scrap of silver taken from applique 8742 (Agighiol). The size of the maps is as given before. The shapes of the grains however are quite irregular, and the mixing of elements is clearly inhomogeneous, as the surfaces overlap only partly in the grain. Fig. 4 a) shows micro-inclusions of various elements like Pb, Fe, Cu and Zn, and one more intense Bi point, associated apparently with Pb. Fig. 4 b) associates, on the other hand, Br and Cl with Ag, which points to local corrosion on a significant area of the map.

Clearly, these maps look quite different from case 1: the structure is different than in the bead sample. This can be a proof that the applique has been obtained with another technology than the bead. Maybe it has to do with the corrosion at the grain-boundaries, too, but there was no noticeable patina, to explain an amorphous aspect. 3.2.3. Case 3: Agighiol applique 8468 (LNL) – solder/repair

In the case of the Agighiol applique (button-type) 8468 we have obtained somewhat confusing results. The buttontype appliques are made of silver and have soldered “handles” made of bronze. For item 8468, this handle was broken (see Fig. 5a).

Soldering on old silver was done typically with a material containing high amounts of Ag and Pb.

Table 1 (macro XRF) shows a material composition with Cu 23.9% and Sn 15.4% (result of a lar ger area measurement at the soldered spot!), like a mix of silver and bronze. The measurement by SR XRF (Table 2) gives a local composition (a sample collected from the solder) with high Ag, some Pb, with Cu 2.4%. In this case, at BESSY, we have measured a micro-sample which can be identified from the micro-photograph as of different shape than that which was mapped at LNL. We also measured 2 points, 8468, 8468-1, on another sample taken from the same applique (presumably far from the solder), and except the presence of Cu ONLY in the solder-sample, the composition is similar. We should note however, revisiting the composition of all applique samples in Table 2, that the solder sample is not the one having the highest content of Cu! It can be supposed that this type of applique had added copper in the composition for strengthening, and that the mixing technology was imperfect. 2

A sample from the solder material was mapped at LNL by micro-PIXE. The size of the map is 1250x1250 m , and it covers the whole sample (Fig. 5b). The map shows a uniform mixing of the material and some corrosion by Br.

3.2.4. Case 4: Agighiol applique 8470 (LNL)

From the in-situ measurement of the silver objects in the Agighiol hoard, it was already s een that it contains silver objects of quite different types and also made of different material. So, it was likely that not only the detailed composition was heterogeneous, but also there could be significant differences in the metallurgical structure, depending on the craft mastered by the workshop where they were made. The Agighiol applique 8470 sample

illustrates the case of imperfect mixing of the silver (an inclusion with high Ca! has been identified in the measurement at BESSY - see Table 2), and also corrosion by Br. The structure of this sample is again different from the one in case 1. This proves the non-uniformity of the metal (technique) in various objects of the Agighiol hoard, already seen in the first investigations.

3.2.5. Case 5: The Surcea ingot (LNL)

The Surcea ingot is the only sample representing the second hoard in our study, a 1c. BC deposit and it was studied 2

only by micro-PIXE mapping at LNL. Fig. 7 shows the 312x312m maps of a more homogeneous sample and the corresponding X-Ray sum-spectrum (an average composition over the whole mapped area). From the map we can clearly notice a more or less uniform mixing with Cu and Zn, which could have been added elements for me chanical strengthening of the objects intended to be made of this ingot. An alternative speculation is that the ingot was obtained from remelted silver objects “as is”. This silver ingot contains 1.9% Cu and 2.1% Sn (Table 1). In the sum spectrum we can identify also Zn, Pb and Cu (Sn cannot be measured with the LNL micro-PIXE set-up).

4. CONCLUSIONS

The studied silver material shows a large variability in composition and micro -structure, and, even within the same hoard, various objects can be identified as coming from different material, and possibly different workshops. Basically, all silver is of high purity, with little added elements. Nevertheless various traces and minor elements are present, which would induce the idea of silver imported possibly fr om Greece, the Aegean and/or Macedonia, as ready ingots maybe, obtained by cupellation from argentiferous galena ores. Agighiol is situated in Dobruja, and the ancient Greek had colonized the nearby Black Sea coast.

In the Geto-Thracian Agighiol set, we could identify a significant amount of Bi, but also some Pb in the silver. It is interesting to remark that the two items of this set similar to objects bearing the same tool -marks (Farkas 1982, Meyers 1982) and coming most likely from the same workshop, have almost the same chemical composition (but twice the gold in our set). (There are also differences in what elements can be and were measured by the two different techniques.)

The heterogeneity in the composition of the sample set (heterogeneous as style as well) could be explained also by the hypothesis that the silver used was coming from a variety of sources.

One sample in the set can be linked to the Greek mine in Laurion.

The inhomogeneity of the composition and structure could be explained to some extent by the limits of mastering the silver smelting and/or metal-working skills by the local workshops.

Techniques have clearly improved in the later find, and were mastered in the local workshop by 1 st c. BC (Surcea).

In spite the fact that valuable information on the silver objects can be obtained with almost no intervention to the samples by these methods, our measurements cannot give a complete description of the silver objects, as could have been obtained with more invasive methods. In cultural heritage investigations, it is of major importance to keep to the least intervention principle, so our study proved to be able to provide certain useful information for the evaluation of the archaelogical hoards in discussion.

With specific differences, but in a somewhat similar manner as in the case of gold, microanalytical X -Ray techniques can bring useful information on the provenance of silver archaeological artefacts, especially if doubled by a database on ancient metal sources, and on metallurgical aspects, to the archaeologist, and on corrosion to the restorators.

ACKNOWLEDGEMENTS

th

The research leading to these results has received funding from the European Community’s 7 Framework Programme (FP7/2007-2013) under grant agreement n.°312284 (BESSY II), EU-ENSAR-INFN Programme (LNL) and the Romanian ANCS grant PN-II-ID-PCE-2011-3-0078.

We also acknowledge gratefully the technical support of HZB-BESSY II, for providing the synchrotron beam access (BAM-line).

REFERENCES

Bartelheim, M. et al, Trabajos de prehistoria, 2012, 69(2), 293. Berciu, D., Thraco-Getic Art, Ed. Academ. RSR, Bucharest, 1969 (in Romanian). Bertrand, L. et al, Appl Phys A , 2012, 106, 377. Bertrand, L et al, J. Cult. Heritage, 2013, 14, 277. Boccaccio, P. et al, Nucl. Instrum. Methods Phys. Res. B, 1996, 109-110, 94. Bondoc, D. and Constantinescu, B., Stud. Cerc. Ist. Veche. Arh. - SCIVA, 2003-2005, 54-56, 279 (in Romanian). Chovan, M. et al, Antimony in hydrothermal mineralizations in the western Carpathians (Slovakia), http://www1.unijena.de/Antimony2011/P06%20-%20Chovan.pdf Constantinescu, B. et al, J. Anal. At. Spectrom., 2002, 27, 2076. Constantinescu, B et al, Spectrochim. Acta B, 2003, 54(4), 755. Constantinescu, B. et al, Stud. Cerc. Ist. Veche. Arh. - SCIVA, 2010, 61(1-2), 143 (in Romanian). Constantinescu, B. et al, Appl.Phys.A , 2012, 109, 395. Farkas, A.E., Metropolitan Museum Journal , 1982, 16, 33. Gale, N.H. et al, in Metcalf, D.M. and Oddy, W.A. (eds), Metallurgy in numismatics,vol. 1, The Royal Numismatic Society, London, 1980, 3. Gale, N.H. and Stos-Gale, Z.A., J. Egypt. Archaeol., 1981, 67, 103. Gitler, H. et al, Am. J. Numismat. Second Series, 2009, 21, 29. Guerra, M.F. and Calligaro, T., Meas. Sci. Technol., 2003, 14, 1527. He, F. and Van Espen, J.P., Anal. Chem., 1991, 63, 2237. Hedges, R.E.M., Studies in Conservation, 1976, 21, 44. Meyers, P., Metropolitan Museum Journal , 1982, 16, 49. Popescu, D., Bul. Mon. Ist. , 1972, XLI (1), 5, (in Romanian). Radtke, M. et al, J. Anal. At. Spectrom., 2010, 25, 631; Radtke, M. et al, Anal. Chem., 2013, 85, 1650. Reiche, I. et al, Appl. Phys. A, 2006, 83, 160. Stos-Gale, Z.A. and Gale, N.H., Archaeol. Anthropol. Sci., 2009, 1, 195. Vincze, L. et al, Spectrochim. Acta B, 1995, 50, 127. Wanhill, R.J.H., Journal of Failure Analysis and Prevention, 2005 5(1) 41. *** Oxford X-Met 3000TX XRF analyser, US Environmental Protection Agency report: EPA/540-R-06/008, http://www.clu-in.org/conf/tio/xrf 082808/cd/EPA-ORD-Innovative-Technology-Verification-Reports-Feb2006/OxfordXmet.pdf Internal report: http://www.romarchaeomet.ro

th

Fig. 1 Silver objects from the Geto-Thracian (4 century BC) Agighiol hoard

Table 1 In-situ XRF measurements (selected values) item

Ag

Au

Cu

Fe

Pb

Bi

Other

wt%

wt%

wt%

wt%

wt%

wt%

Solder 8468

59.1

0.5

23.9

0.5

0.5

-

Sn 15.4%

90 beads set

98.2

0.6

0.1

0.4

-

-

Sn trace

Applique 8467

94.5

0.5

3

0.4

0.2

Trace

-

Applique 8468 95.8

0.6

1.9

0.5

0.6

-

-

Applique 8470

98.2

0.3

Trace?

0.5

-

0.1

-

Surcea ingot

92.6

0.7

1.9

0.5

0.8

-

Sn 2.3%

Agighiol

Zn 1.1%

Table 2 SR XRF results on Agighiol samples

Sample

Ca

Ti

Cr

Fe

Cu

Zn

Br

Zr

Ag

Au

Pb

Bi

Agighiol wt%

wt%

wt%

wt%

wt%

wt%

wt%

wt%

wt%

wt%

wt%

wt%

8470

48.8

0.010

0.009

0.026

0.083

0.002

0.003

0.012

50.6

0.223

0.017

0.174

Bead 1

-

0.034

0.026

0.047

0.387

0.009

0.032

0.019

98.5

0.837

0.017

0.063

Bead 2

-

0.107

0.029

0.077

0.311

0.049

0.201

0.011

98.4

0.725

0.014

0.035

8497

-

0.033

0.060

0.102

1.36

0.009

0.006

0.071

97.5

0.492

0.124

0.251

8472

-

0.050

0.032

0.156

0.100

0.021

0.008

0.042

98.7

0.692

0.058

0.192

8436

-

0.056

0.587

0.245

2.45

0.021

0.004

0.009

95.1

0.983

0.120

0.410

8436-1

-

0.013

0.010

0.029

2.02

0.008

0.002

0.012

96.8

0.637

0.115

0.327

8465

-

0.042

0.020

0.103

2.89

0.026

0.081

0.013

95.2

1.04

0.395

0.188

8471

-

-

0.339

0.472

0.081

0.056

0.036

0.291

98.7

0.031

0.011

0.019

8467

-

0.171

0.031

0.087

12.5

0.040

0.003

0.019

85.4

0.504

0.976

0.281

8468

-

-

0.077

0.156

2.42

0.045

0.267

0.070

95.5

0.775

0.534

0.534

8468

-

0.022

0.019

0.066

1.80

0.005

0.263

0.009

96.1

1.33

0.358

0.049

8468-1

-

0.023

0.024

0.069

0.938

0.008

0.318

0.020

97.0

1.37

0.199

0.023

solder

Fig. 2 Au/Ag versus Bi/Ag wt% ratios

a)

b)

c)

Fig. 3 Bi, Au, Pb associated in silver mineral: Agighiol bead 5 (BESSY): a) microphotograph of the sample and the 2

measurement point (central spot size in photo is 50x50m ); b) 3D map of relative X-Ray signal intensity [a.u.] for Ag 2

2

40x40m , with 2.5x2.5m beam-spot; c) 2D (projection) maps for Ag, Au, Bi and Pb, as in (b)

a)

b)

Fig. 4 Agighiol applique 8742: a) metallurgy/micro-structure; b) corrosion (Br and Cl); map size as in Fig. 2

a)

b)

Fig. 5 a) Button-type appliques: 8468 and 8469; b) Solder material and corrosion of Ag due to Br: Agighiol applique 8468; 2

map size: 1250x1250 m

2

Fig. 6 Applique 8470 (LNL): imperfect metallurgy and corrosion (map-size is 625x625 m ): Ag and Br superposition, Ca and Fe micro-inclusions in the matrix

a)

b) 2

Fig. 7 Surcea ingot (LNL): A 312x312 m map sum-spectrum (a) and mapping of X-Ray signal intensities for Ag, Cu and Zn homogeneous spatial localization over the sample area (b) HIGHLIGHTS (Studies on Ancient Silver Metallurgy using SR XRF and Micro-PIXE): 

SR XRF and PIXE were used to investigate the composition and structure of ancient silver.



Au and Bi are fingerprint elements, useful to identify metal sources.



A Geto-Thracian silver sample is linked to the Laurion mine in Greece, by the Au/Ag and Bi/Ag ratio.



Microstructure was correlated with metal-working techniques.



Corrosion related traces of Br and Cl was identified by characteristic X-ray mapping.