Ion beam analysis techniques: a powerful set ge tools for identification and sourcing of ancient gems

ION BEAM ANALYSIS TECHNIQUES: A POWERFUL SET GE TOOLS FOR THE IDENTIFICATION AND SOURCING OF ANCIENT GEMS T. C(i¡li{*aro'". J.C. Dran'". J.P. Poirot" anci G. Querré INTRODUCTION Gemstones are defined as rare minerals having a high clarity, a high hardness and showing a bright colour and a nice shining. The most important gems are diamond, sapphire and emerald. These gems are callad precious stones. Gems have fascinated man since the earliest times. They have always been sought out by man for their aesthetic and symbolic valué. Due to their scarceness and high '^alue, gems were appreciated by the mighty ones as a sign of their power (emperors ^nd kings, dignitaries of religions ...). Their high hardness and henee strong resismnce to deterioration symbolised the everlastingness of their reign. This last property is interesting from an archaeological point of view: gemstones travel through time without noticeable alteration, even buried in aggressive environment. On the other hand, gems oiay be used several times on different jewels, possibly cut and polished again. Like spices and other precious and sought-after items, gems were brought back from long distance. Accordingly, they are good tracers for ancient trading routes and influence of civilisations. Each civilisation had its own favourite gemstones: lapis lazuli in Egypt; amber and emerald in the Román Empire; gamet in early Middle Ages westem civil isations; jade, ruby and sapphire in Far-East cultures. Moreover, the provenance of some ancient gems such as emeralds or gamets is still debated. For all these reasons, the study of gems kept in museum collections is a promising research fí eld. 2. ANALYSIS OF GEMSTONES Usually, the identification of gems relies on the measurement of optical and mechanical properties: refractive index n, birefringence, visible optical spectrum, specific gravity p [1]- When gemstones are set on a jewel, these measurements are not always Centre de Recherche et de Restauration des Musées de France. Palais du Louvre. París CNRS UMR-171 . 198 T. Cmjmíako. J.-C. Dhw. J.-P. Po/koi > CJ. readily carried out. Dismounting them from the setting or sampling is of course prohibited. The provenance of gems, that means the geographical location of the deposit, is often inferred by identifying specific inclusions. Anothcr possibility is the chemical characterisation of gemstones. In this case, gems are dirccily idcntified by means of their major chemical constituents. For instance, diamond is a purc carbón crystal; niby is an AI2O3 crystal coloured by less than 1% Cr; sapphire is also an ALO* crystal but its blue colour is due to trace amount of Ti and Fe; emeraid is a bcryl of formula Be3Al2(Si03)6 coloured by a small amount of Cr. The trace clcment contcnt of gems also carries valuable information. Indeed, some of these trace elements come from the surrounding rocks, being incorporated during the génesis of the crystal. Being linked to the geological context (type of terrain and rock-forming history) trace element content may act as a fingerprint of the occurrcnce of a gemstone. A wide range of techniques have been applied in view to perform chemical analysis: EMPA. XRF, NAA, ICP/MS, SIMS. Table I summarises some features of the.se analytical techniques. Of course, careful observation of the gems and their internal features using conventional gemmological means has to be performed prior to proceeding to these chemical analy sis. Moreover, several analytical techniques have often to be combined for inferring unambiguously the provenance of a gem. Table 1. Features of several analytical techniques used for the chemical charac terisation of gems EMPA EDXRF NAA LA-ICP/MS SIMS IBA Element range > Na > Na Selected all all all LOD >1.000 ppm 1-1.000 ppm < 1 ppm < 1 ppm < 1 ppm 1-1.000 ppm Probé size pm mm whole sample 10-//m //m 10-// m Destructive no no activation P- destructive P- destructive no Preparation yes / coating no no no yes / coating no Isotopic info no no no possible yes no Sample size cm any cm any cm any EMPA: electrón microprobe analysis. EDXRF: energy-dispersive X-ray fluoresccnce. NAA: neutrón activation analysis. LA-ICP/MS: laser-ahlation lon-coupled-pla.sma mass spectrometry. SIMS: secondary ion ma.is spectrometry. IBA: ion beam analysis. 3. IBA TECHNIQUES: PIXE AND PIGE APPLIED TO THE STUDY OF GEMS It is not the place here to give a complete description of Ion Beam Analysis (IBA) techniques; numerous books and review papers will fulfil this requirement [2 and 3]. Let us just recall that IBA techniques rely on the same process: identification of ele- Til 199 CoNcRi.so Naciiiwi OI AKQi i-(nnn"RlA dental constituents by detecting products of interaction of the ion beam and the target (••e. gemstone). Among them, PIXE (particle indiiced X-ray emission) is quite similar XRF, being based on the detection of X-ray emitted by the target atoms, subsequent to an inner-shell ionisation. The only difference stems from the excitation source which a charged particle beam instead of a X-ray beam. For the PIGE technique, acronym for particle induccd y-ray emission. there is no counterpart. PIGE relies on a nuclear reaction subsequent to the collision betvveen an incident particle and a target nucleus. Pollowing this reaction, the nucleus might emit a y-ray having a specific energy. On the other hand, IBA techniques cannot be used for isotopic ratio determination ñor <^oes it give information on chemical bonds. From the point of view of gemmology, the IBA techniques exhibit several Ínteres ting features. First, in-air IBA techniques allow in sitit and non-destructive analysis ot 8cms without sample preparation. This haimlessness is a must for the study of va luable objets such as historical jewels. Using simultaneously the PIXE and PIGE methods it is possible to determine a wide range of elements. PIXE gives the composition for elements ranging from Na to U, while PIGE usefully extends the range to i-, ter ones; F, Be, Li, B. Secondly, the high sensitivity of these techniques permits inea- ^urement of elements down to a low level of concentration (e.g. — ppm wt t)r transition metáis in corundum). Finally, the use of a microbeam with a size o ess than 20-^m allows selection of inclusions for their identification. Such a small spot is also useful tor analysing a región free of inclusions. Indeed, when using a broader spot size, there is always the risk of including an inclusión in the analysed area ^at may bias the trace composition of the crystal (e.g. an Fe-containing inclusión might lead to a wrong Fe trace content in a ruby). From the point of view of the IBA techniques, gemstones are ideal samples when compared to other objects of cultural heritage such as paintings or ceramics. They have 3 simple, homogeneous composition. The sample has a polished and often fíat su ace. Owing to their very stable crystal structure, gems are nearly insensitive to beam effects (no alteration). Charging of these usually non-conductive targets is avoided when using an external beam setup. 4. EXPERIMENTAL SETUP The experimental setup and its successive improvements have been thoroughly described [4]. Only the striking features will be recalled here. The analytical system is based on an extemal beam line of the AGLAE accelerator facility (figure 1). The usual 3-MeV proton beam is focused down to 30 pm size using magnetic lenses (3 MeV is proved to be an optimum energy for analysing geo logical samples). This beam is brought on the sample by passing through a very thin SÍ3N4 foil. The detection of X-rays is achieved using two detectors. A first one is dedicated to light elements, which are the major constituents of the gems (from Na to Fe). It has a low solid angle, and a minimal filtering using an ultra thin window combined 200 T. Cmjj(íaho, J.-C. Dk.w. J.-P. Poikoi > G. Qi'erré Figure I. View of the externa! beam Une of the AGLAE accelerator-based IBA facility. P I with a helium flow. The second ene is dedicated to heavy elcments (from Ca to U) at a low level contení. It has a high solid angle and a fllter chosen in order to remove matrix X-rays. Gamma-rays are collected using a High Purity Germanium detector. The beam is monitored using a signal emitted by the exit foil: backscattered protons or X-ray line of Si, according to experimental requirements. 5. application to gemstones We will describe some recent applications of the PIXE and PIGE techniques to ancient gems from museums perfoimed with the AGLAE setup of the Centre de Recherche et de Restauration des musées de France. It should be emphasised that all gemstones studied here are well documented and have traceable history. So we are confident that these gems are the original ones. 5.1. Ruby: Ishtar's statuette This statuette which may represent Ishtar, a goddess equivalent to Aphrodite, is a perfect illustration of a IBA study of gemstones (figure 2). The sculpture is a naked woman carved in alabaster, her eyes and navel inlaid with rubies. This remarkable piece, which joined the Louvre collections in 1866, was excavated in the vincinity of Babylon in Mesopotamia [5]. Por stylistic and typological reasons it is attributed to the Parthian period (2nd century B.C. to the late 2nd century A.D.). We were able to confirm that the cabochons were indeed natural rubies and not coloured glass, as had been believed even quite recently, as the two main constituents of rubies were revealed: 99% alumina (AhOd. associated with less than 1 % chromium, the element responsible for the characteristic red colour of ruby. Several trace elements III CoNGRluSO N a( U)N \1 DI ARGIH)MI TRlA 201 Figure 2. Ishtar in being analysed using the externa! beam systern (ver cuaderniUo a co!or. pág. v^. were also measured, including titanium, vanadium, iron, copper, gallium, all being characteristic of natural stones (figure 3). The similar aspect and chemical composition of the three inlays suggest the use of a single batch of stones for the statuette. In order to determine the geographical origin of this statuette's rubies, we undertook a comparative study with contemporary rubies of known provenance [6]. A total of over 500 analyses was carried out on rubies from the main existing deposits: Afghanistan, Myanmar (Bumia), Cambodia, India, Kenya, Madagascar. Sri Lanka,Thailand and Vietnam. The trace element contents of the stones are quite different depending on their provenance and are homogeneous within a given deposit, a feature absolutely needed for a provenance study based on geochemistry. Multivariate statis- 1e5 10000 1000 100 high energy X-ray detector spectrum low energy X-ray detector spectrum TI,, Vk CfK Fe 10 15 X-ray energy (keV) Figure 3. X-ray spectra of majar and trace eiements recorded on a red eye of the statuette. 202 T. Caujc.aro. J.-C. Drw. J. P. Poiroi > G. Qi'EKR^ tical methods were used (factor analysis, hierarchical classification and discriminant analysis), incorporating all the chemical tracer elements (Ti, V, Cr, Fe, Ga). All these analyses converge and show that the statuette's rubíes havc a composition very cióse to that of rubíes from Burma (figure 4). The results of the geochemícal study were confirmed by classícal gcmmological observatíons conducted concurrently; ín particular, the rutile inclusions obscrved in the eyes and the navel are consistent with a Burmese origin. Research carried out on the rubíes of the Parthian statuette from the Louvre combining observation of the gems and their geochemícal characterísatíon by the PIXE method has led to a better ínsíght ínto thís work of art by ídcntífyíng íts constítuent materials. In addítíon, ít has províded the opportuníty to establísh a data base on the composítíons of the main ruby deposíts. Fínally, thanks to statístícal analysis, a provenance for the statuette's rubíes could be proposed: Burma. It can thus be concluded that the gems from the Parthian statuette are evídence for trade or exchange over several thousand kilometres between Mesopotamía and South-East Asía duríng the Parthian period. As far as we know, thís ís the earlíest use of rubíes ídentífíed lo date ín thís región. Thís fi gure of Ishtar ís remarkable for ít combines ínfluences derívíng from three civílísatíons: Mesopotamía (the ímage of the naked woman as a fértilíty symbol), Greece (the Hellenístíc realísm of the body), and fínally Asía (the presence of the rubíes). 10000 S s Ó 1000 100 10 group III group I 10 X : Ishtar A: Afghanistan B ; Burma C ; Cambodia K; Kenya M ; Madagascar S : Srí Lanka T : Thaíiand group 100 1000 10000 Fe (ppm) Figure 4. Plot of Cr vs Fe contení for the rubíes of the statuette and reference rubíes. III CoNíikl SO N ac lONM DI ARyi loMMKlA 203 Figure 5. Royal votive crown being analysed with the e.vternal beani (ver cuadernillo a color, pdg. vi)- 5.2. Emiírald: thi-: Visigothic treasure We report here an applícatíon to the analysis of emeralds of the Guarrazar treasure (figure 5). Thís famous set of votive royal crowns, made of gold with ínlaid gemstones, dates from the Visigothic period (7*^-8"' centuries). It was excavated in 1858 near Toledo and is now splít ínto two main parts kept at the National Archaeological museum ín Madrid, Spain and the National Middle-Ages museum in París, France [7]. The sources of emeralds used in ancient Egypt and the Román Empire are quite well documented; it is not so for the early Middle-Ages. This work aimed at companng an cient emeralds to those of various origins reported in ancient literature (Egypt, Aus tria, Afghanistan, Oural), on the basis of chemical analysis [8]. PIXE and PIGE techniques permitted to identify gemstones by means of their composition in major elements (Be, Al, Si) and to determine their trace element contents (figure 6). Provenance determination relies on the comparison of these contents to those of reference emeralds constituting a data base of geochemical composition. The most discriminating elements were Li, Na, V, Rb and Zn. The measured contents in these ancient emeralds are only compatible with Egyptian (Jebel Zabara) or Austrian (Habachtal) deposits (figure 7). Since it permitted to discard altérnate assumptions (e.g. Ural, Afghanistan or Pakistán), this work provided a new insight into the trade routes of gemstones at the time of the Great Invasions. 5.3. Garnet: jewels of the first Kings of Frange Gamets belong to a family of gems having variable composition. The most common type is the pyralspite family with the chemical formula X*^3AP^*2(Si04)í, X be- 204 T. CMJJdAKO, J.-C. DftW. J.-P. Po/KOI > G. QdERRÉ Figure 6. Gamma-ray spectrum of an emerald. Figure 7. Sunray plot of trace elements contení of ancient and reference emeralds. 10 10 10 10^1 10' 10' iiAL-j kJJ SI I ' I ' ' ■ ' I ' ' ' ■—I ' ■—■—' I ' ■ ' ' I PIGE spectra of an etoerald Bo 3662 koV iil Li 478 keV F 197 keV 500 1000 1500 2000 2500 3000 3500 4000 GaoMa ray Energy (keV) Statislical processing ; Sunray plot of trace elements I /|\ Afghanístan Pakistán ^Gallo-Roman Columbia _ Zambia \ ("S Egypt J Wlsigoth Zimbabwe fvladagascar Norv/ay Brazil \ ''Ja / Mg o Rb Ca Re / Cf ^ V Vaiiaules ing either Fe (almandin gamet), Mg (pyrope gamet) or Mn (spessartite gamet), or a combination of them. The question of the nature, origin, cutting and setting techniques of the gamets used in Europe during the Early Middle Ages is highiy debated. A fí rst attempt to apply IBA techniques to Early Middle Ages gamets has been performed in the C2RMF [9]. This work dealt with the identification of gamets mounted on Merovingian jewellery from Louvres-en-Parisis. The conclusión was that these archaeological gamets had a major composition compatible with gamets from Sri Lanka (almandine-pyrope type). We are currently developing a new research program on the Merovingian gamets. The aim of this work is to include trace element and near-surface analysis of the gems. III Conc;ri_s() Nac iowi di Ar(,)1 i-.omi-trIa 205 Figure 8. Merovingian jewei with garnets set in cloisonné style (ver cuadernillo a color, pág. vi). These garnets are set on the cloisonné jewels (figure 8) from the royal treasure of Saint Denis, France, comprising the famous jewels of Queen Aregonde [10]. This work will be extended with the analysis of gamets excavated from various archaeological sites in Europe. 6. CONCLUSION IBA techniques and specifically the PIXE/PIGE combination in air appear as valuable tools for analysing gems mounted on historical jewels. They usefully complement classical gemmological investigations. The main advantages are 1) completely non-destructive analysis, 2) broad range of elements detected, 3) \0-f4m size spatial resolution 4) ppm level sensitivity and 5) good accuracy. They were applied to deter mine the nature and provenance of ancient mbies. For Middle Ages emeralds and gar nets, the results already obtained are promising. More reference data are needed and the combination with other analytical techniques has to be tried. 7. REFERENCES [1] Anderson, B.W. and Jobbins, E.A., 1990, Gem testing, Butterworths, London. [2] Bird, J.R. and Williams, J.S., 1989, Ion beams for materials, Academic Press, London. 206 T. Caumíamo. J.-C. Dka.s. J -I*. I^oihoi > G. Querré [3] Campbell, J.L. and Malmqvist, K.G., 1995, Particle-induced X-ray spccírometry (PIXE), John Wiley & Sons, Lxjndon. [4] Calligaro, T., Dran, J.-C., Hamon, H., Moignard, B. and Salomón, J.. 1998, An exlemal milli-beam for archaeometric applications on ihe AGLAE IBA facility of the Louvre museum. Nucí. Instr. and Meth. B 136/138, 339-343. [5] Tallen, F., 1999, Actes du colloque: Comaline ct pierrcs prccieu.scs - 1995, Musée du Louvre, Paris. [6] Calligaro, T., Mossmann, A., Poirot, J.-P. and Querré, G., 1998, Provenance study of ru bíes from a Parthian statuette by PIXE analysis. Nucí, hustr. and Meth. B 136/138, 846- 850. [7] Perea, A. and Balmaseda, LJ., 1999, Las tres historias del tesoro de Guarrazar y el proyecto Au, Revista de arqueología 221, 37-45. [8] Calligaro, T., Dran, J.-C., Poirot, J.-P., Querré, G., Salomen, J. and Zwaan, J.C., 2000, PDCE/PIGE characterisation of emeralds using an extemal microbeam. Nucí. In.str. and Meth. B161/163, 769-774. [9] Farges, F., 1998, Mineralogy of the Louvres Merovingian gamet cloisonné jewelry: Origins of the gems of the first kings of France, Am. Mineral 83, 323-330. [10] Fleury, M. and France-Lanord, A., 1998, Les tré.sors mérovingiens de la basilique de Saint-Denis, Gérard Klopp eds, Woippy, France. APROXIMACIÓN AL ESTUDIO DE PROCEDENCIA Y TECNOLOGÍA DE FABRICACIÓN DE LAS CERÁMICAS CAMPANIFORMES DEL YACIMIENTO DE LA EDAD DEL COBRE DE LOS MILLARES Josefa Capel''', Feniatido Molina'", Trinidad Nájera*", José Linares'-' V Francisco Huertas'" Resumen En este trabajo se lleva a cabo el estudio analítico mediante Espectrometría de Masas, Difracción de Rayos-X y Microscopio Estereoscópico de un conjunto de vasijas cerámicas de estilo campaniforme procedentes del yacimiento de la Edad del Cobre de Los Millares. El análisis químico se ha centrado en el estudio de los elementos denominados tierras raras que constituyen un indicador exce lente del origen de los materiales. El estudio mineralógico así como la observa ción óptica han aportado información sobre procesos tecnológicos de manufacturación. Palabras clave: cerámica. Edad del Cobre, procedencia, manufacturación. 1. INTRODUCCIÓN Los trabajos sobre la movilización de elementos químicos durante los procesos de hidrólisis y alteración de rocas ocupa un importante campo en la investigación desa rrollada en el ámbito de las ciencias del suelo y ciencias de la Tierra en general. La mayor o menor movilidad de estos elementos, en el caso de los minerales, está in fl uenciada por la estructura del cristal. Es igualmente obvio que existen numerosos factores fisicoquímicos tales como pH, temperatura y presión de la solución, textura y permeabilidad de los sólidos, área superficial, etc., que influyen de igual forma. No obstante un mismo elemento puede comportarse, dependiendo de su posición cristalo- Dpto. de Prehistoria y Arqueología. Facultad de Letras. Campus Universitario de Cartuja, 18071 Granada Dpto. de Ciencias de la Tierra y Geoquímica Ambiental. Estación Experimental del Zaidín, CSIC. Profesor AIbareda, s/n, 18008 Granada.