Palladium analysis in gold items from Punic jewellery (Cádiz, Spain)

Palladium analysis in gold items from Punic jewellery (Cádiz, Spain) I. Ortega-Feliu a,b,∗ , B. Gómez-Tubío a,b , S. Scrivano a , F.J. Ager a,c , M.L. de la Bandera d , M.A. Respaldiza a,e a Centro Nacional de Aceleradores (US-CSIC-JA), C/Thomas A. Edison 7, Sevilla, 41092, Spain b Dpto. Física Aplicada III, Univ. de Sevilla, Camino de los Descubrimientos s/n, Sevilla, 41092, Spain c Dpto. Física Aplicada I, Univ. de Sevilla, C/Virgen de África 7, Sevilla, 41011, Spain d Dpto. Prehistoria y Arqueología, Univ. de Sevilla, C/María de Padilla s/n, Sevilla, 41004, Spain e Dpto. Física Atómica, Molecular y Nuclear, Univ. de Sevilla, Avda/Reina Mercedes s/n, Sevilla, 41012, Spain ARTICLE INFO Keywords: Palladium Gold X-ray fluorescence Punic jewellery Cádiz ABSTRACT Some years ago a group of jewels from the Museum of Cádiz (South Spain) was analysed by micro-PIXE (Particle Induced X-ray Emission) at Centro Nacional de Aceleradores (Sevilla). During this work two similar spiral earrings happened to have significant amounts of palladium in their composition, up to 0.7 wt%, and some hypotheses about the provenance of this palladium-bearing gold were exposed. Many similar gold earrings from the same location and time are kept in the museum, some of them permanently exhibited in the showcases. Now, by means of a portable X-ray spectrometer device, especially designed for the analysis of gold jewels, a second campaign has been accomplished to study the rest of the jewels kept at the museum. This time the equipment was moved to the museum instead of taking the gold items to the laboratory, facilitating the analysis of the samples in the showcases and reducing the costs. A total of 94 jewels have been studied and 24 of them contained palladium up to 1.1 wt%. Copper content was below 2.5 wt% in most of the cases. X Ray Fluorescence technique was employed for this study. A filter of 20 μm of Zn was used in the detector to enhance the detection of palladium, and other potential platinum group elements, by absorbing the Au-L α emission line. The effects of this method in the quantification was also compared in this work, finding higher uncertainties with the use of the filter, but demonstrating the convenience of its use to discriminate the Pd-K α line from the sum peaks of the very intense gold lines. 1. Introduction The improvement of nuclear non-destructive techniques has allowed the increase in the number of compositional analysis performed to complement the study of valuable gold jewellery all over Europe and other countries (Constantinescu et al., 2008;Tissot et al., 2013; Zucchiatti et al., 2014;Lemasson et al., 2015;Buccolieri et al., 2017). Some of the most used quantitative and non-destructive techniques employed nowadays for the analysis of this type of material are PIXE (Particle Induced X-ray Emission) and XRF (X-Ray Fluorescence). Both of them are based in the excitation of the atoms present in the sample and the subsequent emission of characteristic X-rays for each chemical excited element. The way the atoms are excited differentiates both techniques, as PIXE uses accelerated charged particles, as for example protons or alpha particles from an accelerator facility, and XRF employs X-rays produced in an X-ray tube or radioactive sources. Thus, the portability of the second one is clearly facilitating the study of valuable jewels that are usually exhibited in museums and cultural institutions. Nevertheless, some years ago, the small spot size of the particle beams reached in the accelerator facilities were chosen over the XRF devices if the study had to include a detailed compositional information of every part of the jewels, as for example, granules, filigrees, foils, unions, inclusions, etc. Other advantages of the PIXE technique are the shorter acquisition time necessary for having statistically good spectra, and the very low limit of detection (LOD) for elements that could be present in very low amounts, as traces, that could help in provenance studies. In this context, in 2005, a group of gold jewels from the Museum of Cádiz (South Spain) was analysed by micro-PIXE in the Centro Nacional de Aceleradores (Sevilla). In a previous work by some of the authors (Ontalba Salamanca et al., 2006), 16 rings of different styles, classified in four groups (G1, G2, G3 and G4) but from a similar period and provenance, were characterized, and palladium was found in every point of analysis of two of the jewels (Ref. 26080-1 and 26080-2) and in one structural element of a third one (Ref. 4181). The presence of ∗ Corresponding author. Centro Nacional de Aceleradores (US-CSIC-JA), C/Thomas A. Edison 7, Sevilla, 41092, Spain. E-mail address: [email protected] (I. Ortega-Feliu). https://doi.org/10.1016/j.radphyschem.2019.03.043 T palladium was quantified up to 0.7 wt% in those rings, and some hypotheses about the provenance of this palladium-bearing gold were established. The two rings in which palladium was present in every point of analysis are the only two items of G4, but they do not form a pair of earrings themselves, as the decoration is different. Looking for new items containing palladium, three more rings from the same collection and with the same style as G4 of the already published paper were analysed with the same experimental conditions. The results are presented in Ortega-Feliu (2008), but palladium was not found in these new items with references 23240, 26942a and 26966a. The later development of a micro-XRF portable device, already described in Scrivano et al., (2017b), opened the possibility of starting a new campaign of in-situ analysis, moving the equipment to the museum. This was possible due to the 30 μm spot diameter of the beam thanks to the use of policapillary optics. With this portable system some other gold hoards have already been studied with success, characterizing the decorative elements, like granules, filigrees, sheets, holding elements, etc. And the union procedures, brazing, solid state diffusion bonding, fusion welding, etc. (Scrivano et al., 2017a,2017b;Ortega-Feliu et al., 2018). However, this campaign is geared to the detection and quantification of palladium as a trace element, thus some technical improvement has to be tested with this equipment in order to fulfil the archaeological aim of the study: to discriminate whether the presence of Pd in gold found in those rings is an exception or not in the Cádiz workshop. As already discussed in the previous work Ontalba Salamanca et al. (2006), the very localized area and period of the Pd gold samples, and the very few references about other Pd-bearing gold objects elsewhere, suggests a short life supply of the special raw material. Based on archaeological, geological and metallogenical considerations described in Ontalba Salamanca et al. (2006), the geographical origin of the palladium-bearing gold fields could be located on auriferous alluvial deposits in the NW of the Iberian Peninsula, the former West African Gold Coast, the Eastern Africa or Anatolia. These are locations in which two scenarios can happen: scenario A, in which gold and palladium come from geologically different and spatially independent Au and PGE (Platinum Group Elements) mineral deposits but located in one same hydrographical basin; and scenario B, in which gold and palladium derive from Pd-bearing gold deposits. There exists another scenario C, in which the gold employed for the manufacture of the jewels would be the result of an intentionally mixture done by the goldsmith. For this improbable option the palladium would come from the PGE deposit located in the ultramafic massifs located in Ronda, Ojén and Carratraca, all of them geographically close to Cádiz (Ovejero Zappino, 2005; Gervilla and Leblanc, 1990;Castroviejo, 1998). Thus the overall aims of this work are: to evaluate the use of a Zn filter for the X-ray detector to improve the detection of Pd element in XRF; to characterize the composition of the gold jewels of the Punic collection found in the necropolis of Cádiz; to shed light about the presence of the palladium-bearing gold in Cádiz. 2. Materials and methods In this section the studied samples are described followed by the method used for the analysis, including the reasons for the addition of the Zn filter in the detector. 2.1. Sample description The 94 samples studied in this work correspond with all the earrings similar to those in which Pd was previously found (Ortega-Feliu et al., 2007), and some more jewels of the same area and period. They all were found in the Phoenician city of Cádiz, one of the richest cities in the Western Mediterranean demonstrated by the enormous amount of gold jewellery found in the necropolis of the city during the last decades. This local production shows oriental influences (700–600 BCE) with an evolution to a more industrialized production corresponding to the Punic period (500–200 BCE) in which our samples were produced (de la Bandera et al., 2010;Cervera, 1923;Muñoz Vicent and Perdigones Moreno, 2000;Perdigones Moreno et al., 1990;Quintero, 1914;Quintero, 1915). The Punic jewellery analysed consisted of earrings, rings, pendants, beads and little foils dated in the 5th-4th century BCE, all of them found in burial sites, well contextualized, in the necropolis of Cádiz and exhibited in the local Museum (Fig. 1). According to the stylistic details and archaeological arguments, local production of all the analysed jewellery is assumed and there is no doubt of their authenticity. The stylistic analysis of the objects enabled the classification in 8 groups (see Table 1). 2.2. Analytical method X-ray Fluorescence was employed for the analysis of the collection of the gold jewels (Fig. 2). As said before, a beam spot size of 30 μm was achieved thanks to the use of policapillary optics. Other technical characteristics of the device are described in Scrivano et al., (2017b) and summarized in Table 2. With respect to the previous campaigns of analysis where this device was employed, this time the setup was improved for a better detection of the trace elements that could appear in the gold material. Thus, the Rh anode was substituted by a W one, in order to avoid the characteristic Rh lines in the spectra that could Fig. 1. Examples of the analysed jewels, representing each of the stylistic and functional groups found in the necropolis of Cádiz. 2 interfere with platinum group elements (PGE) emission lines. The X-ray tube output was filtered with 1 mm thick aluminium foil to attenuate the characteristic W emissions. Three gold certified standards were used for the quantification procedure, performed with WinQxas code, with the following concentrations of gold, silver and copper respectively (in wt%): 96 (96-2- 2), 90 (90-8-2) and 75 (75-15-10). In the case of standard 75, Pd was quantified by PIXE with 6 MeV protons, with a result of 0.22 ± 0.06 wt % (Ortega-Feliu, 2008). No Pd was detected in the other two standards. As it is known (Ontalba Salamanca et al. (2006) and references therein, Lemasson et al., 2015), platinum group elements (PGE), Os, Ir, Pt, Ru, Rh and Pd, are expected to be present in gold samples, but very few of them can easily been detected, due to their very low contents, usually below 1 wt%. Although the XRF technique can detect and quantify elements in such low content, the presence of high amounts of gold, silver and copper disturb the detection of these trace elements, as most of them overlap with other intense X-ray emissions. Indeed, Os overlaps with Cu-K β , and Ir and Pt emit between Cu-k β and the very intense Au-L α line, which produces a high background in the spectra (see Fig. 3). Besides, when acquiring spectra of gold samples, especially in the case of very high content of gold, X-rays of sum peaks are easily observed. These sum peaks (electronic artefacts during the acquisition of the spectra due to difficulties in discriminating two consecutive X- rays) appear at the same energies as some typical trace elements in gold matrices, which are the case of Ru and Pd. It seems that Rh is the only one that could be directly detected, as long as the X-ray tube anode is Type Ref. Samples with Pd 1a 4257; 4258 2a 26080-1; 4185; 4259; 4260; 4246; 32430/5; 32430/6; 32430/2* 3a 4170* 4a 4218; 4219; 4224; 4225 4b 32429; 32578; 32438/21 5b 32452*; 32453 64181*; 4182*; 32470/A 832530/223* Samples without Pd 1b 4130; 4131 2a 26966b; 26942b; 23240; 23241; 4227; 32430/4; 32448; 32449; 32450; 32451 2b 4220; 4221; 4132; 4133 3a 4129; 4134; 4135; 4137; 4171; 5018; 32493/107; 32493/110 3b 4172; 4173 3c 4128 4b 32438/23; 32438/30; 32569; 32494/114; 32494/115; 32494/108; 32537/221; 32537/237; 32438/31; 32442/34; 32442/38; 32496/111; 32496/113; 32444; 32495; Argolla1; Argolla2; Argolla3; Argolla4 4c 32484/94; 32484/95 5a 32454; 32455 632446; 32529; 32552; 4987; 32458; 4181; 4144; 32457/A; 32470/B; 32457/B 7a 32499; 32500; 32504; 4255; 30286 7b 32550; 32485 832530; 32555; 32530/224 * Pd content close to the LOD. Fig. 2. The micro-XRF portable spectrometer of the CNA laboratory. Table 2 Characteristics of the micro-XRF experimental set-up and manufacturers: IfG (IfG, 2019) and Amptek (Amptek, 2019). X-Ray Tube iMOX-MRF μ-Focus (IfG) Anode W Voltage 50 kV Intensity 600 μA Optical System Polycapillary (IfG) Spot size 30 μm Working distance 10 mm Detector SDD (Amptek) Resolution 127 eV (5.9 keV) Active Area 25 mm 2 Fig. 3. XRF spectra of sample 26080-1 acquired without any filter in the detection system and with a 20 μm Zn filter. The X-ray transmissión factor of 20 μm Zn and Cu are represented (x1000 for comparison). Table 1 Classification of the analysed samples (N = 94). 3 it, are presented with and without the filter in Fig. 4. Without the filter of Zn, palladium was usually quantified by the fitting code even in the standards without this element, in an amount that depends on the content of gold in the matrix, as it is related to the Au-L α sum peak. For example, in a 90 wt% gold standard up to 0.16 wt % of palladium can be quantified. On the contrary, no palladium at all was measured in the same spot if the Zn filter is used. Fig. 5 shows the Pd content quantified by the fitting code in the three gold standards employed during the analysis, with and without the 20 μm Zn filter. The most relevant conclusion was that, for the standards without Pd, the results provided high values of this element and this prevented from clarifying whether there is Pd or not in the gold matrix. However, if the filter is employed, most of the times the code Fig. 4. Details of the XRF spectra acquired for gold standards with and without Pd (top and bottom respectively) and with and without 20 μm Zn filter (right and left respectively). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) Fig. 5. Palladium contents obtained for three gold standards, one of them with about 0.22 wt% of Pd, with and without the detector filter of 20 μm Zn. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) made of a different material. As the aim of the analytical campaign was to look for traces typically present in gold, especially palladium, a technical improvement was made in the system to enhance their detection by reducing the sum peaks contribution. A 20 μm foil of Zn (Z = 30) was placed in front of the detector to absorb the Au-L α line that cause the sum peaks. The method had already been employed in PIXE but it had never been checked in portable XRF systems, as far as we know, and the effects on the different backgrounds of each technique had to be explored. As can be seen in Fig. 3, where a Pd-bearing gold jewel (museum reference 26080-1) is analysed, the spectra acquired without the filter presented small sum peaks at 19.426 keV (Au-L α + Au-L α ) , 21.155 keV (Au- L α + Au-L β ) and 22.884 keV (Au-L β + Au-L β ) . Although the total counts under these peaks were very low, they are of the same order of magnitude as the signal of palladium if its content is below 1 wt% and the one at 21.155 keV overlaps with palladium K α line at 21.177 keV. Most of the times, the palladium K β line is usually too low to be used for quantification or even just for detection. The Zn foil acts as a selective filter due to its X-ray absorption edge (9.67 keV) , just below the very intense Au-L α line (9.713 keV) , responsible of the inconvenient sum peak that overlaps with the palladium peak. The transmission coefficient of 20 μm Zn (x1000 for comparison reasons in Fig. 3) is plotted along with the spectra to show how this filter absorbs half of the peak of Au-L α line 9.713 keV, and the sum peaks disappear. Although it is true that, in order to absorb the complete Au-L α line, it would be preferable to use a copper filter (whose transmission coefficient is also plotted in Fig. 3), this filter is not used because copper itself is an important element present in gold jewels that has to be quantified. The parasitic presence of X-ray lines of the filter in the spectra is of great inconvenience when the filter is made of Cu but it is not a problem when it is made of Zn. Consequently, in order to guarantee the detection of low concentrations of palladium and discriminate it from the Au sum peaks, the use of the Zn filter is mandatory. The XRF spectra of two of the gold standards, one with Pd (0.22 ± 0.06 wt%) and the other one without 4 provided zero % of Pd. The highest value of the Pd content in the standards with no Pd was chosen as a limit of detection (LOD) of the procedure, without the filter, 0.17 wt%, and with it, 0.07 wt% (the very high value at 0.131 wt% was been discarded as a singularity). Finally, in the case of the standard with some Pd the result was slightly overestimated if no filter was used in the detection system. As a drawback, the acquisition time must be increased when using the filter in order to keep an acceptable uncertainty. Indeed, to keep a relative uncertainty of about 8% the acquisition time was increased up to 600 s, otherwise the relative uncertainty can be as high as 14.0%, as can be checked in Table 3, where the results and uncertainties obtained with different acquisition times are detailed for samples analysed in the same area with and without the filter. It can also be checked that the quantification of Pd concentration was overestimated in about 0.1 wt% if the filter was not employed in the detection system. In this work, some spots were analysed without the filter of Zn. In a few of them the Pd content was so high that even the K β line could be detected. Although the technical conclusions of the methodology described above show that the Pd content will be slightly overestimated, these spectra were fitted and the results were used, with caution, for the interpretation of the historical conclusions. 3. Results A total of 94 gold jewels of the necropolis of Cádiz were characterized. The results of obtained with the new XRF system and those previously obtained by PIXE in the same analysed spots are in good agreement. For example, PIXE results were 0.40 wt% in sample 4181 (0.40, and 0.369 wt% in similar spots in this work) and 0.26 wt% in sample 26080-1 (0.29 and 0.318 wt% in similar spots in this work). Fig. 6 shows the results for Au, Ag and Cu for the analysed jewels. The copper content is below 2.5 wt% in the great majority of the samples, thus probably copper was not intentionally added to the alloy during the manufacture of most of them (Karydas et al., 2004;Ortega- Feliu et al., 2007). Silver content varied in these samples without following a relation with the type or specific use of the jewels, but systematic differences were found in different parts of a piece (for example, the flowers in groups 1 and 2 in Fig. 1). This is related with the manufacturing procedure followed by the goldsmith and the soldering methods but it was not possible to extract reliable conclusions with the number of points analysed in this campaign, so they were not included in this paper. In most of the samples, the Au concentration was over 88 wt%. With only one exception (with Pd content very close to the LOD), Pd is found in those samples with Au concentration over 88%, but this could be explained by the fact that the lower the content in Au, the lower the content in Pd, as will be discussed later, and hence the corresponding Pd content of samples with Au < 88 wt% could fall under the LOD. The new analyses confirm the presence of palladium and add several new jewels to the list of the palladium-bearing gold objects of the Punic necropolis of Cádiz up to 24 pieces, which is 25% of the analysed jewels. In the following discussion, those points of analysis in which the Pd content reported by the fitting code is below the LOD established in section 2.2 are classified as samples without Pd. Besides, some of the samples analysed without the filter have Pd concentration very close to the LOD, and taking into account that the estimation of the Pd content when not using the filter is overestimated in about 0.1 wt%, these samples might be, actually, below the limit of detection, and hence classified as samples without Pd. This is the case, for example, of jewel 4133, with a Pd content without filter of 0.231 and 0.186 wt%, in two different spots, but the result with filter is zero. For the rest of the samples, Pd and Au contents are represented in Fig. 7. Some samples are classified as having Pd, but this conclusion should be taken with caution because the content of Pd is very close to the LOD of the procedure and more analyses should be done. Besides, these very low Pd content samples do not follow a general correlation that can be observed between Pd and Au contents. Indeed, with a few exceptions, Sample Zn filter t (s) Pd content (wt%) Uncertainty Absolute (wt%) Relative (%) 26080-1 No 300 0.38 0.04 10.5 400 0.43 0.04 9.3 800 0.42 0.03 7.1 Yes 400 0.29 0.04 14.0 600 0.318 0.025 7.9 4185 No 600 0.37 0.03 8.1 Yes 600 0.252 0.025 9.9 800 0.259 0.019 7.3 4133 No 200 0.231 0.019 8.2 200 0.186 0.017 9.1 Yes 600 < LOD - - Fig. 6. Ternary diagram with the contents of Au, Ag and Cu of the analysed samples. Fig. 7. Pd vs. Au contents in the analysed samples. Table 3 Example of the results and uncertainties obtained in some of the samples. 5 4. Conclusions The correlation between Pd and Au indicates a specific origin for the raw material, characterized by natural content in Pd. The presence of Pd in several local manufactured jewels with different functionality, although most of them are earrings, suggests the use of imported raw material, or the recycling of foreign gold artefacts for their manufacture, with a common provenance. The high number of pieces with Pd (at least the 25% of the analysed samples) suggests that the access to the specific source of the Pd-bearing gold was frequent for a certain period. Besides, the samples with Pd do not share any other similarity between them, as neither the style, the decorative pattern, the functionality, the gold content, the tomb or exact point of discovery inside the big necropolis of Cádiz are common in this group of samples. Thus, it seems that the special characteristic of this gold was probably unknown by the goldsmith manufacturing the jewels. Finally, it was demonstrated that, to guarantee the detection of low concentrations of palladium and discriminate it from the Au sum peaks, the use of the Zn filter is mandatory, although the acquisition time must be increased when using the filter to keep a reasonable uncertainty, with the only disadvantage of the number of points of analysis that can be done in a day, as the gold jewels do not suffer any damage under the X-ray beam. Acknowledgments This paper is dedicated to the memory of Mr. Gobain Ovejero, who gave us an essential support in the discussion about the provenance of the palladium. Work partially supported by the project HAR2015-67113-P from the Spanish Ministry of Science and Innovation. The authors wish to thank the staff of Museo de Cádiz. S. Scrivano and I. Ortega-Feliu acknowledge the financial support from the “V Plan Propio de Investigación” of the University of Seville. References Amptek, 2019. 14 DeAngelo Drive. Amptek, Inc, Bedford, MA 01730 USA URL: http:// amptek.com/, Accessed date: 26 March 2019. de la Bandera, M.L., Ortega, I., Gómez, B., Ontalba, M.A., Respaldiza, M.A., 2010. Caracterización del taller de orfebrería de Cádiz mediante técnicas de análisis nucleares. Mainake (0212-078-X) 32 (1), 37–59 (in Spanish). Buccolieri, A., Castellano, A., Degl'Innocenti, E., Cesareo, R., Casciaro, R., Buccolieri, G., 2017. EDXRF analysis of gold jewelry from the Archaeological Museum of Taranto, Italy. X-Ray Spectrom. 46, 421–426. https://doi.org/10.1002/xrs.2761. Castroviejo, R., 1998. Nuevas aportaciones a la tipología de metales preciosos para exploración en España. Boletín Geológico y Minero 109 (5–6), 497–520. (in Spanish). http://oa.upm.es/10244/. Cervera, F., 1923. Excavaciones en extramuros de Cádiz. Memorias Junta Superior Excavaciones Arqueológicas, Madrid (in Spanish). Constantinescu, B., Bugoi, R., Cojocaru, V., Radtke, M., Calligaro, T., Salomon, J., Pichon, L., Röhrs, S., Ceccato, D., Oberländer-Târnoveanu, E., 2008. Micro-SR-XRF and micro-PIXE studies for archaeological gold identification – the case of Carpathian (Transylvanian) gold and of Dacian bracelets. Nucl. Instrum. Methods Phys. Res. B 266, 2325–2328. https://doi.org/10.1016/j.nimb.2008.03.054. Gervilla, F., Leblanc, M., 1990. Magmatic ores in high-temperature alpine-type Iherzolite massifs (Ronda, Spain and Beni-Boussera, Morocco). Econ. Geol. 85, 112–132. https://doi.org/10.2113/gsecongeo.85.1.112. IfG, 2019. IfG - Institute for Scientific Instruments GmbH. Rudower Chaussee 29/31, D- 12489 Berlin. URL. www.ifg-adlershof.de, Accessed date: 26 March 2019. Karydas, A.G., Kotzamani, D., Bernard, R., Barrandon, J.N., Zarkadas, Ch, 2004. A compositional study of a museum jewellery collection (7th–1st BC) by means of a portable XRF spectrometer. Nucl. Instrum. Methods Phys. Res. B 226 (1–2), 15–28. https://doi.org/10.1016/j.nimb.2004.02.034. Lemasson, Q., Moignard, B., Pacheco, C., Pichon, L., Guerra, M.F., 2015. Fast mapping of gold jewellery from ancient Egypt with PIXE: Searching for hard-solders and PGE inclusions. Talanta 143, 279–286. https://doi.org/10.1016/j.talanta.2015.04.064. Muñoz Vicent, A., Perdigones Moreno, L., 2000. Actas del IV Congreso Internacional de Estudios Fenicio y Púnicos, vol. II. Cádiz, pp. 881 (in Spanish). Ontalba Salamanca, M.Á., Gómez-Tubío, B., Ortega-Feliu, I., Respaldiza, M.Á., de la Bandera, M.L., Ovejero Zappino, G., Bouzas, A., Gómez-Morón, A., 2006. Externalbeam PIXE spectrometry for the study of Punic jewellery (SW Spain): The geographical provenance of the palladium-bearing gold. Nucl. Instrum. Methods Phys. Res. B 249, 622–627. https://doi.org/10.1016/j.nimb.2006.03.172. Ovejero Zappino, G., May 2005. Personal Communication. (in spanish). Ortega-Feliu, I., Gómez-Tubío, B., Ontalba Salamanca, M.Á., Respaldiza, M.Á., de la Bandera, M.L., Ovejero Zappino, G., 2007. Gold and electrum jewellery in the strategic area of Gadir in Phoenician period. Nucl. Instrum. Methods Phys. Res. B 260 (1), 329–335. https://doi.org/10.1016/j.nimb.2007.02.077. Ortega-Feliu, 2008. Thesis of Inés Ortega-Feliu, 2008. PhD. Universidad de Sevilla. Ortega-Feliu, I., Scrivano, S., Gómez-Tubío, B., Ager, F.J., de la Bandera, M.L., Respaldiza, M.A., Navarro, A.D., San Martín, C., 2018. Technical characterization of the necklace of El carambolo hoard (Camas, Seville, Spain). Microchem. J. 139, 401–409. https:// doi.org/10.1016/j.microc.2018.03.024. Perdigones Moreno, L., Muñoz Vicente, A., Pisano, G., 1990. Studia Punica, vol. 7. Roma, pp. 7. Quintero, P., 1914. Boletín de la Sociedad Española de Excursiones, vol. 22. pp. 161 (in Spanish). Quintero, P., 1915–1919. Excavaciones en extramuros de Cádiz. Memorias Junta Superior Excavaciones Arqueológicas, Madrid 1925–1934 (in Spanish). Scrivano, S., Ortega-Feliu, I., Gómez-Tubío, B., Ager, F.J., de la Bandera, M.L., Respaldiza, M.A., Ontalba-Salamanca, M.A., 2017a. Non-destructive micro-analytical system for the study of the manufacturing processes of a group of gold jewels from “El Carambolo” treasure. Radiat. Phys. Chem. 130, 133–141. https://doi.org/10.1016/j. radphyschem.2016.08.008. Scrivano, S., Ruberto, C., Gómez-Tubío, B., Mazzinghi, A., Ortega-Feliu, I., Ager, F.J., Laclavetine, K., Giuntini, L., Respaldiza, M.A., 2017b. In-situ non-destructive analysis of Etruscan gold jewels with the micro-XRF transportable spectrometer from CNA. J. Archaeol. Sci.: Rep. 16, 185–193. https://doi.org/10.1016/j.jasrep.2017.09.032. Tissot, I., Tissot, M., Manso, M., Alves, L.C., Barreiros, M.A., Marcelo, T., Carvalho, M.L., Corregidor, V., Guerra, M.F., 2013. The earrings of Pancas treasure: Analytical study by X-ray based techniques –A first approach. Nucl. Instrum. Methods Phys. Res. B 306, 236–240. 2013. https://doi.org/10.1016/j.nimb.2012.11.054. Zucchiatti, A., Climent Font, A., Gutiérrez Neira, P.C., Perea, A., Fernández Esquivel, P., Rovira Llorens, S., Ruvalcaba Sil, J.L., Verde, A., 2014. Prehispanic goldwork technology study by PIXE. Nucl. Instrum. Methods Phys. Res. B 332, 160–164. 2014. https://doi.org/10.1016/j.nimb.2014.02.052. the higher the Au content is, the higher the Pd in the material. This fact leads to the conclusion that most probably Pd is, in fact, a trace element naturally associated to the gold, and hence linked to its provenance. Some samples were analysed only in one spot, as the access to the jewels in the museum was limited in time, and this study is focused on the detection of Pd. In order to perform a detailed study of the manufacturing procedures and soldering techniques, a second campaign should be done in the future. Nevertheless, some samples were analysed in more than one point and differences can be seen in the composition of the different decorative elements, for example in 4185, 26080-1 and 4246. In a few other cases, characterized by very high Pd and Au concentrations, the results are much more similar in the different elements. This is the case of samples 4257, 4258, 4259 and 4260. 6