Fast mapping of gold jewellery from ancient Egypt with PIXE: Searching for hard-solders and PGE inclusions

Talanta 143 (2015) 279–286

Contents lists available at ScienceDirect

Talanta journal homepage: www.elsevier.com/locate/talanta

Fast mapping of gold jewellery from ancient Egypt with PIXE: Searching for hard-solders and PGE inclusions Quentin Lemasson a,b,n, Brice Moignard a,b, Claire Pacheco a,b, Laurent Pichon a,b, Maria Filomena Guerra c a

C2RMF, Palais du Louvre-Porte des Lions, 14 quai François Mitterrand, 75001 Paris, France Fédération de recherche NewAGLAE, FR3506 CNRS/Ministère de la Culture et de la Communication/Chimie ParisTech, Palais du Louvre, 75001 Paris, France c ArchAm-UMR 8096 CNRS-University Paris 1 Panthéon -Sorbonne, MAE, 21 allée de l’Université, 92023 Nanterre, France b

art ic l e i nf o

a b s t r a c t

Article history: Received 27 January 2015 Received in revised form 16 April 2015 Accepted 21 April 2015 Available online 5 May 2015

A new PIXE setup at the external beamline of the AGLAE accelerator is assessed for fast mapping the joining regions and the PGE inclusions of nine Egyptian gold items from the Louvre museum collection, dated to the end of the 2nd Intermediate Period and to the New Kingdom. The setup is composed of a cluster of SDD detectors divided in two “super detectors” dedicated to analyse the matrix and the trace elements. It provides the possibility to realise large and/or fast maps on artefacts by scanning the beam over the sample surface. Different softwares have been developed or updated to visualise, process, and quantify the data. By using this setup, we could determine the elemental distribution of major elements Au, Ag and Cu on the different joining regions, estimate the composition of the brazes, and show that they were produced by adding Cu to the base gold alloy. By fast mapping the PGE inclusions we could reveal a large variety of compositions within a single object. In addition to the expected Ir–Os–Ru system inclusions, we could also show for several inclusions the presence of another element, Pt. For a region where PGE inclusions overlap the joining area we could show that fast mapping allows to determine the compositions of the inclusion, the brazing alloy, and the base-alloy. & 2015 Elsevier B.V. All rights reserved.

Keywords: PIXE Fast mapping Egypt Gold PGEs Brazing

1. Gold work in ancient Egypt The few publications on the analytical study of goldwork attributed to the 2nd Intermediate Period (c. 1800–1550 B.C.) and to the New Kingdom (c. 1550–1070 B.C.) in Egypt show the use of polychrome effects within the same objects by using gold alloys with different colours [1–4] and sometimes native aurian silver [5]. A variety of gold colours could be obtained by adding Ag and Cu contents to gold from various sources. Before the introduction of parting, in the first millennium B.C. [6,7], gold objects were made with native gold from different sources, and so containing different concentrations of Ag, to which Cu could be added to enhance the colour and the properties of the alloys. Recent publications showed that gold jewellery is made with a variety of gold colours within a same tomb during the New Kingdom [2] and from high carat gold alloys to high silver

n Corresponding author at: C2RMF, Palais du Louvre-Porte des Lions, 14 quai François Mitterrand, 75001 Paris, France. Tel.: þ 33 1 40 20 24 82. E-mail address: [email protected] (Q. Lemasson).

http://dx.doi.org/10.1016/j.talanta.2015.04.064 0039-9140/& 2015 Elsevier B.V. All rights reserved.

electrum, even for items belonging to a single individual during the 2nd Intermediate Period [4,8]. The majority of the objects were constructed by joining different parts. Hard-soldering or brazing seems to be widely used in the Egyptian workshops since at least the Middle Kingdom [9], and may attain a very high quality for some particular jewellery items belonging to important personages [10,11]. The brazes or brazing alloys are in this case gold alloys whose melting point is lower than the parts to be joined. In order to lower the melting temperature of a gold alloy, the concentration of Ag and/or Cu must be intentionally increased (by changing the native base alloy or by adding the metals). Egyptian goldworkers should have had an empirical knowledge on the melting ranges of the binary and ternary alloys of gold, copper, and silver and well practise brazing as represented on a wall painting dating from about 1475 B.C. in the tomb of the Vizier Rekh-mi-re at Thebes [12]. Gold brazes can be produced with the same technique as polychrome gold alloys, which means by addition of Cu to native gold containing different Ag contents. It is however difficult to estimate the original composition of the brazes in small objects. When running at high temperature the filler between the parts to be joined, a region where the alloys mix up tends to be formed. By

280

Q. Lemasson et al. / Talanta 143 (2015) 279–286

scanning these joining regions with a good spatial resolution technique it is possible to construct an elemental distribution map of the filler constituents and provide data allowing an estimation of the original composition. The new fast mapping setup developed at the AGLAE accelerator of the C2RMF (Centre de Recherche et de Restauration des Musées de France) provides an opportunity to approach the composition of the brazes used in Egypt in the 2nd Intermediate Period and in the New Kingdom. This same setup could be applied to another important question on ancient Egyptian gold work, which is the gold origin. The presence of Platinum Group Elements (PGE) inclusions in Egyptian gold objects was observed for the first time on the surface of objects from the excavations of Naqada and Ballas [13] and discussed by a few authors [14,15], and recently published work on Egyptian gold jewellery emphasises the presence of PGE inclusions on the majority of the objects [3,4,8]. We must still cite the Egyptian case belonging to the Louvre collection where Berthelot identified the presence of gold, silver and platinum [16]. The presence of PGE inclusions on the surface of gold objects is related to the use of alluvial gold and could be an indicator of the provenance of the metal [17]. In the case of other civilisations, the presence of Pt in gold [18–21] and the presence and composition of PGE inclusions [22] has been used with success for gold provenancing. However, the PGE mineralogy of Egypt is not very well known, in spite of the identification of Os-rich and Os–Ir alloy minerals in the Eastern Desert [23]. The PGE can be separated in two groups [24], the first containing Ir, Os, and Ru (non soluble in gold and usually associated with chromites) and the second containing Rh, Pd, and Pt (usually associated with sulphides, but Pd and Pt are more soluble in gold). The association of the PGE to particular minerals can provide hints on the differentiation of the sources of gold even if the very few published data on Egyptian objects tend to show that their composition remains quite heterogeneous [15]. Like the thin joining regions in gold objects, the form and dimensions of the PGE inclusions may be quite variable. Their very uneven volume and their heterogeneity are analytical difficulties for PIXE that can be surmounted by the use of mapping with the PIXE experimental setup described in this work. The experimental difficulties connected to the variable unknown thickness of the regions of interest are presented and the data obtained is used to discuss the role of this new setup to solve questions of this type.

2. Fast mapping with PIXE at AGLAE The extracted beam line of the accelerator AGLAE is totally dedicated to the non-invasive analytical study of cultural heritage objects. The different setups that have been developed at AGLAE involve the determination of the elemental and structural composition of ancient materials, providing data on the manufacture technologies of the objects, on the provenance of the materials, and on their alteration mechanisms. In the frame of project NewAGLAE (EQUIPEX n°2011-EQ1-22, French Ministry of Research), a new detection system has been implemented on the extracted beam line [25]. By using several detectors with high active surfaces, the total solid angle of X-ray detection was increased tenfold, and a new scanning system was developed to produce fast elemental maps with high spatial resolution (down to 10  10 mm²). This is achieved by scanning a rectangular area of selected dimension with a vertical magnetic deflection of the beam coupled to a horizontal/vertical mechanical movement. The fast data acquisition through a list mode is compatible with high sensitivity PIXE measurements, even for trace elements.

The acquired raw data is transformed into EDF (ESRF data format) and then visualised using the homemade software AGLAEMap [25]. This software has been developed to handle data provided by a large number of detectors and series of scanning acquisitions. Its interface allows switching easily the visualisation on a screen from the analytical data acquired by a detector to another for a set of analyses. Elemental mapping is achieved by selecting in the spectra the regions of interest that correspond to X-ray lines chosen for the chemical elements to be studied. The elemental distribution can be saved in the form of a map in intensity range. This software has other resources such as the selection of data in a specific area for further calculation. The areas can be drawn on the screen directly on the elemental distribution map. After selection of the pixels, data is summed and saved in a spectrum file in GUPIX format. Quantitative processing of the spectra is carried out with GUPIXWIN software [26], which is coupled to the homemade TRAUPIXE software. When necessary, quantitative mapping can be achieved by calculating every pixel using the software TRAUPIXE_EDF also developed at the AGLAE facility [27]. For the analysis of the gold objects, PIXE was carried out with four 50 mm² SDD detectors covered with two types of filters [28,29], and one SDD detector with no filter but with a deflecting magnet for protection against backscattered particles. Data acquired by the detectors covered with the same filter was summed. Therefore, the system was shifted to two “super detectors”: the first with an Al filter of 200 mm for the determination of the major elements of the gold alloys and the second with a Cu filter of 75 mm for the determination of the trace elements. The fifth detector is usually devoted to low energy X-rays and in this case has not been considered. Analyses were performed with a 3 MeV proton beam of 10 nA intensity and 50 mm diameter and a total acquisition time of 5 min. The mapped area was adapted regarding the area of interest from 240  640 mm² to 1960  1280 mm² with a pixel size of respectively 20  20 mm² and 40  40 mm².The small size of the beam combined Table 1 Average composition of the gold alloys determined by PIXE for the nine Egyptian objects analysed in this work. The two gold standards analysed to check the reproducibility of the analysis are also presented. Cu wt%

Ag wt%

Au wt%

Standards CLAL 6905 certified Measured CLAL 6917 certified Measured

19 18.2 7 6.8

6 6.0 18 17.6

75 75.9 75 75.7

Earrings N1855B N2084 E14435B AF2444 E14335 D

0.8 3.2 5.2 2.5 1.7

25.5 41.0 21.9 42.4 23.7

73.7 55.8 72.9 99.2 74.5

Bracelet E7168 Cartouche Lion

2.8 1.2

31.3 26.9

97.9 98.5

Necklace 22658 Bead with hard-solder Average on 6 beads

1.2 1.17 0.42

15.7 11.6 7 5.8

82.3 87.3 7 6.2

Rings E3297 N747

1.6 0.9

27.8 7.8

69.0 91.3

Q. Lemasson et al. / Talanta 143 (2015) 279–286

with the capacities of the system to acquire mappings with small pixel sizes allowed high spatial resolution acquisitions thus making possible the fine analysis of small size areas of interest down to beam size such as inclusions or joining regions.

3. Application to the study of Egyptian jewellery We selected for this study nine gold jewellery items from the collection of the Louvre museum with apparent hard-solder joining and PGE inclusions of different sizes. This group of objects includes five 18th Dynasty penannular earrings of three distinct typologies, King Ahmose's I bracelet (first King of the 18th Dynasty) E7168, King Horemheb's signet ring (last King of the 18th Dynasty) N747, one of Queen Ahhotep's ring (end of the 2nd Intermediate Period) E3297 and one necklace containing gold tubular spacers 22358. The average compositions of these objects are summarised in Table 1. The majority is made of electrum alloys with Ag contents higher than 20 wt%. These results are consistent with the few data available for jewellery from equivalent periods [2,3,8]. The Cu contents range from 0.1 wt% to 5.2 wt%. We remind that objects made from electrum alloys may show as much as 2–3 wt% Cu [4] and objects made from aurian silver may show higher amounts of this element [5]. In order to estimate the information depth at our experimental conditions, we calculated for four alloys with compositions close to the data obtained for the studied objects the penetration depth of the incident proton beam and of the emitted X-rays for the three major elements of the gold alloys. Table 2 shows that the proton beam penetration is about 30 mm for all the four alloys whilst the information depth is 10–15 mm for the Au L-lines and the Cu K-lines. The depth of analysis for the Ag K-lines is 30–55 mm, which means limited by the depth of penetration of the protons. The data obtained in this work should not be affected by the surface preservation of the objects. Atmospheric corrosion of gold alloys is generally contained in the first mm [30, 31], only

281

induced surface depletion gilding on very particular alloys produces surface layers which may attain 5–10 mm [32]. The dimensions of both the PGE inclusions and the joining areas for the nine Egyptian objects analysed in this work go from 50 to 200 mm diameter. Fig. 1(a) and (b) shows a region with several easily detectable inclusions of different sizes on the head of the lion belonging to bracelet E7168 and one small inclusion visualised on ring E3269. The identification and localisation of small inclusions under a stereomicroscope before analysis by SEM–EDS are time consuming and only achievable for visible inclusions and objects entering in the SEM chamber (some SEM have however large chambers). An equivalent analytical difficulty is faced when identifying the brazing alloys, particularly for high quality joining (Fig. 1(c)), which are difficult to localise [4]. We could carry out at AGLAE PIXE point analysis with good spatial resolution by reducing the beam width. However, when increasing the spatial resolution the counting rates are lowered, which becomes a real analytical challenge for gold alloys as the data obtained may be non representative of heterogeneous PGE inclusions and brazes. With the new AGLAE detection system all these difficulties are overcome by mapping the area of interest. Scanning a whole area containing PGE inclusions or the joining areas makes possible to construct an elemental distribution map of the selected chemical elements. The data obtained can be exploited by choosing the areas of interest within the map (i.e. PGE inclusions or joining) and the concentrations can be calculated by using the extracted spectra. In addition to this, by mapping a whole region, it is possible to complement the information obtained for the brazes and the PGE inclusions with the composition of the surrounding regions (for example the base gold alloy) since several areas of interest can be defined in a single map.

Table 2 Depth of analysis calculated for four typical Egyptian gold alloys at our experimental conditions: an incident proton beam of 3 MeV and the emitted X-rays used for quantitative analysis. Alloy Cu–Ag–Au (wt%)

1–6 – 93 4–43 – 53 4–27 – 69 1–19 – 80

3 MeV proton penetration depth (mm)

28.1 32.5 31 29.7

X-ray penetration depth (mm) Cu Kαline

Ag Kαline

Au Lαline

8.2 10.9 10.0 8.9

29.1 56.6 43.1 35.6

13.0 16.7 15.2 14.2

Fig. 2. Elemental distribution by PIXE of Ir (Lα-line) on a selected region of the cartouche belonging to bracelet E7138 (640  640 mm², pixel size 40  40 mm²).

Fig. 1. SEM BSE images of PGE inclusions on the surface of (a) bracelet E7138 (scale 200 mm); (b) ring E3297 (scale 20 mm); and (c) one joining area of ring E3297 (scale 400 mm).

282

Q. Lemasson et al. / Talanta 143 (2015) 279–286

3.1. PGE inclusions All the regions containing inclusions were scanned and for each region a map of the surface elemental distribution of the PGEs and the major elements of the alloys was constructed. A major concern for PGE inclusions is their homogeneity in composition and their dimensions, particularly their thickness. If their diameter is visible and their surface dimensions can easy be calculated with good precision, their thickness remains unknown. All the PGE inclusions on the Egyptian objects contain Ru, Os, and Ir, but Pt could also be found in some of them. Fig. 2 shows the intensity distribution map of Ir in colour scale for a small region (640  640 mm²) on the surface of the cartouche of bracelet E7168 where nine inclusions could be found. A variation on the number of counts within one inclusion clearly comes out. In order to assess the influence of the heterogeneous thickness of the inclusions on the calculated PGE compositions, we selected the biggest inclusion of Fig. 2 and calculated its composition in four different zones. Fig. 3 shows the areas selected from the centre to the rim and summarises the calculated compositions for each zone. As expected, the concentrations of Cu, Ag and Au increase from the centre to the rim, with the decrease of the inclusion's thickness, while the concentrations of Ru, Os, Ir and Pt decrease. However, when the PGE concentrations are normalised to 100 wt%, as shown on Table 3, similar compositions with 5–10% standard deviation (possibly caused by matrix effect) are obtained for all the inclusion areas. Thus the inclusions can be considered as homogeneous. A similar conclusion can be drawn for major elements when normalised to 100 wt%. The compositions of the rim zone in the inclusion and of the base gold alloys can be considered equivalent. Because of the low efficiency of the SDD detectors for high energy X-rays, the Ag K-line peak measured in the spectrum is frequently small. This is amplified by the presence of small quantities of Ag in the alloys and when the selected number of pixels is small. For these reasons, the error for the Ag contents may be above 10%. These facts explain the lower amount of Ag found at the centre of the inclusion. In order to estimate the depth of analysis of the PGE inclusions with our experimental conditions, we calculated for three alloys the penetration depth of the incident proton beam and of the emitted X-rays from both the PGEs and the major elements of the alloy. Table 4(a) shows that the beam penetration is of about 25 mm; the depth of penetration is lower than for the gold base alloy, particularly for high Ir contents. If we assume that the proton beam is always contained in 4  4 pixels (i.e. an area of 80  80 mm²) in Fig. 3 we can consider that area 1 only contains the inclusion which means that the surrounding gold alloy was not included in the analysis. Therefore, the Cu, Ag and Au contents calculated for area one might come from the underneath alloy. When we consider in Table 4(b) the depth of analysis calculated for the elements contained in the analysed layer, we can estimate that the biggest inclusion on the surface of the cartouche of bracelet E7168 is thinner than 10 mm.

1 2 3 4

Cu wt% Ag wt% 0.5 7.2 0.8 11.6 1.2 18.8 1.5 25.2

Table 5 summarises the average composition normalised to 100 wt% of all the analysed PGE inclusions by object. Although some grains are of iridosmine, ruthenosmiridium and osmiridium [24], many of them show in addition the presence of Pt that may reach 11.9 wt%. Similar contents were found in the base alloys of three gilding leaves of wood samples from the tomb of Tutankhamen [33]. We represented in Fig. 4 the composition of the whole PGE inclusions in a ternary diagram. In this diagram one axe represents the IrþOs, because these are the two common elements of the two ternary systems Os–Ir–Ru and Ir–Os–Pt defined by Harris and Cabri [24]. It is also interesting to remark the diversity of the compositions within one same object. Because of the presence of four elements, it is difficult to compare our data with the composition of PGE inclusions published for Egyptian jewellery, analysed by SEM–EDS. The PGE compositions published by Meeks and Tite [15], Troalen et al. [4], and Miniaci et al. [8] show that Pt is barely observed which could indicate either a lack of sensitivity of the analytical technique or, which might be our case, the use of gold from different sources. We must remark the difficulties connected to the measurement of Pt in an alloy containing high concentrations of Au [21,10]. Table 3 Composition of both the PGE inclusions and the base gold alloys when data in Fig. 3 is normalised to 100 wt% for each matrix and each selected area. The base alloy measured nearby the inclusion is also given.

1 2 3 4

Ru (wt%)

Os (wt%)

20.9 21.1 23.4 25.4

38.5 38.5 37.8 35.9

Pt (wt%)

36.8 37.2 36.3 33.9 Ag wt%

Cu wt%

1 2 3 4

Ir (wt%)

1.9 1.8 1.8 1.8 Base alloy 67.3

3.9 3.2 2.4 4.8 Au wt%

26.6 26.5 28.1 29.4 1.9

71.5 71.6 70.1 68.8 30.8

Table 4a Depth of analysis of the PGE elements calculated for three PGE inclusions typical of the analysed Egyptian objects at our experimental conditions: an incident proton beam of 3 MeV and the emitted X-rays used for quantitative analysis. Alloy Ru–Os–Ir (wt%)

47–35–18 25–40–35 2–38–60

Au wt% Ru wt% 19.4 14.3 31.2 11.4 7.1 46.8 2.9 59.0

3 MeV proton penetration depth (mm)

27 25.5 23.5

Os wt% 26.4 20.8 11.5 4.1

Ir wt% 25.2 20.0 11.1 3.8

X-ray penetration depth (mm) Ru Kαline

Os Lαline

Ir Lαline

Pt Lαline

37.7 25.6 17.4

13.5 11.5 9.5

14.6 12.4 10.2

15.7 13.4 11.0

Pt wt% 2.7 1.7 <0.8 <0.8

Fig. 3. Selected areas from the centre to the rim of the biggest PGE inclusion of Fig. 2 and the calculated compositions for each selected area.

Q. Lemasson et al. / Talanta 143 (2015) 279–286

283

Table 4b Depth of analysis of the major elements of the gold alloy calculated for three PGE inclusions typical of the analysed Egyptian objects at our experimental conditions: an incident proton beam of 3 MeV and the emitted X-rays used for quantitative analysis. Alloy Ru–Os–Ir (wt%)

47–35–18 25–40–35 2–38–60

3 MeV proton penetration depth (mm)

27 25.5 23.5

X-ray penetration depth (mm) Cu Kαline

Ag Kαline

Au Lαline

10.3 8.9 7.3

30.5 27.8 24.5

16.9 14.4 11.8

Table 5 Average composition normalised to 100 wt% of all the PGE inclusions analysed on the Egyptian objects. Ru wt%

Os wt%

Ir wt%

Pt wt%

2.1 22.3 8.3 24.3 26.3 12.1 13.2 22.5

18.0 38.0 50.3 36.9 36.2 43.8 51.5 40.3

69.6 36.5 41.4 35.4 34.1 41.7 30.0 34.6

10.3 3.1 o 0.2 3.3 3.4 2.3 5.2 2.6

Lion

16.9 0.9 29.8

41.0 33.8 36.6

39.2 58.7 28.3

2.8 6.6 5.3

Ring E3297

40.2 27.6 3.1 93.8 17.1

31.2 32.5 84.5 0.0 41.4

28.6 37.5 8.5 6.2 38.0

o 0.2 2.5 3.9 o 0.2 3.4

5.4

53.1

41.5

o 0.2

Earring 14435B

9.8 47.0 29.6 25.6 29.9 22.5 4.9 18.6 19.9 19.6 23.6 24.1

1.2 31.1 33.7 37.6 34.0 39.2 32.9 40.8 41.9 41.7 40.2 39.4

77.1 18.6 26.1 32.0 31.6 38.3 53.6 37.4 38.2 38.6 30.5 32.3

11.9 3.2 10.6 4.9 4.5 o 0.2 8.7 3.2 o 0.2 o 0.2 5.7 4.2

Earring 14435D

13.6

41.8

42.2

2.4

Earring 2084

5.0 5.3

56.6 53.1

38.4 41.6

o 0.2 o 0.2

Earring 2444

20.8 20.9

35.4 34.9

40.7 41.0

3.1 3.2

Bracelet E7168 Cartouche

Earring 1855B

3.2. Hard-soldering Hard-soldering or brazing is a joining technique with a final result on the object's morphology that very much depends on the

Fig. 4. Ternary diagram representation of the PGEs of Table 5 normalised to 100 wt%. Ir and Os are the common elements of the two ternary systems defined by Harris and Cabri [24].

expertise of the goldsmith. For this reason, it is possible to find on jewellery items thick visible joining areas with a typical morphology of brazing and in a very few cases so thin that they can be almost invisible even when observed on a SEM. [10], Diffusion bonding (or colloidal soldering, a mixture of copper salts with an organic flux), is a joining technique that produces thinner seams but it spreads in the Mediterranean basin in the 1st millennium BC and was never identified in Egyptian objects dated to the 2nd millennium BC. Fig. 5 shows the quantitative concentration distribution map in colour scale of Cu, Ag and Au on the joining region of one bead from necklace 22658. By mapping a surface of 640  440 mm² within the soldering area of this bead, it clearly comes out the variation in the alloy composition. To get more accurate results from the centre of the joining to the rim, the average composition has been calculated on 7 areas. Fig. 6 shows the selected areas and summarises the corresponding calculated compositions. As expected, the concentrations of Cu and Ag increase from the centre to the rim while the concentration of Au decreases. Only one layer has been considered when calculating the composition of the brazes as this is the normal situation for the objects studied in this work. In this case, either the joining thickness is superior to the depth of penetration of the beam and the results obtained represent the composition of the braze only, or the joining thickness is thinner and the results obtained represent the composition of the braze together with the substrate. The composition of the braze alone cannot be established. Moreover, the two layers being composed of the same elements it would thus be impossible to make an accurate calculation of a two layered distribution. The complexity of the joining region complicates the attribution of a gradient in concentrations to heterogeneities dues to brazing, to differences in thickness or to a likely combination of these two possibilities. An average concentration has been calculated for the braze in the different joining areas that could be found on the several objects of this study, by selecting the whole area in which the Cu or Ag contents showed different concentrations. The results obtained are summarised in Table 6. We remark the almost systematic increase of Cu in the joining areas. As expected for Egyptian jewellery [4], the same base alloy with different Cu contents was used to fabricate and

284

Q. Lemasson et al. / Talanta 143 (2015) 279–286

Fig. 5. Elemental distribution in colour scale of Au, Ag and Cu on a joining region of the selected bead from necklace 22658 (640  640 mm², pixel size 40  40 mm²).

3.3. Combining PGEs and gold brazes

Cu wt% 1 2 3 4 5 6 7

2.3 1.9 1.7 1.6 1.2 1.2 1.0

Ag wt% Au wt%

37.2 34.2 31.9 24.9 13.1 9.4 6.8

59.7 63.0 65.6 72.7 84.7 88.6 91.4

Fig. 6. Selected areas from the centre to the rim of the joining region of Fig. 5 and the calculated compositions for each selected area.

Table 6 Average composition normalised to 100 wt% of all the brazing alloys (brazes) analysed on the Egyptian objects. Cu (%)

Ag (%)

Au (%)

Ring E3297

Braze Braze Base alloy

5.2 8.6 1.6

27.3 27.5 27.8

66.0 62.4 69.0

Earring 2444

Braze Braze Base alloy

7.6 8.8 2.5

38.4 35.5 42.4

53.3 54.8 99.2

Earring 1855B

Braze Base alloy

1.8 0.8

24.0 25.5

73.6 73.7

Earring 14435D

Braze Braze Base alloy

3.2 3.4 1.7

24.4 24.6 23.7

71.8 71.3 74.5

Necklace 22658

Braze Base alloy

1.7 1.2

29.3 15.7

68.3 82.3

Finger-ring N747

Braze Base alloy

1.1 0.9

6.7 7.8

91.6 91.3

We tested the potential of our experimental system by considering in detail ring E3297 where one joining region could be analysed together with one PGE inclusion. A region of 240  640mm² has been mapped and the concentration distribution maps in colour scale for Cu, Ag, Au, Ir, Ru and Os are shown in Fig. 7. By using the mapping system developed at the AGLAE facility we could resolve the overlapping of the PGEs and the major elements Au, Ag and Cu that constitute both the base alloy and the braze. Clear differences in composition could be seen indicating the presence of a brazing alloy. The three PGE inclusions visible nearby the joining are placed either on the joining itself or on the base plate of the ring. Their compositions are different (two inclusions being however similar), but this fact is independent of their location on the object's surface. Table 7 summarizes the average composition of the three PGE inclusions and of the gold alloys used to make the ring plate and the braze, calculated by selecting the corresponding areas in the maps and normalising each alloy to 100 wt%. The Cu, Ag and Au ratios close to each PGE inclusion are similar to the ratios found either in the braze (for two PGE) or in the gold (third PGE) confirming the space delimitation considered for the PGE inclusions. Data clearly show that the brazing alloy is produced by simple addition of Cu to the alloy used in the fabrication of the parts to be soldered.

4. Conclusion

join parts of the items. All the Ag contents except in the bead from necklace 22658 remain constant. In the case of ring N747 we could not evidence a composition variation in the joining area.

We applied to the study of Egyptian gold objects the new detection system implemented on the extracted beam line of AGLAE. This setup is based on several detectors with high active surfaces and a new scanning system developed to produce fast elemental maps with high spatial resolution (down to 10  10 mm²). We focused on the application of this new setup to the analysis of PGE inclusions related to the alluvial origin of the gold and to the joining areas where a hard-solder (or braze) is applied. The thickness of the joining area and the form and dimensions of the PGE inclusions are very variable. We could show that when fast analysis must be performed in areas difficult to locate, the analytical difficulties can be surmounted and the experimental setup described in this work becomes a powerful system of analysis. We could also show that the system is suitable for the study of complex situations such as the overlapping of a joining area with PGE inclusions. Some difficulties could however be encountered when processing the data. These difficulties are connected to the selection of the areas to be processed for quantitative analyses. A compromise must be reached between selecting a high number of pixels, necessary to obtain more accurate measurements, and the precise

Q. Lemasson et al. / Talanta 143 (2015) 279–286

285

Fig. 7. Elemental distribution in colour scale of Cu, Ag, Au, Ir, Ru and Os on the region of ring E3297 where a PGE and a joining could be found together (240  640 mm², pixel size 40  40 mm²).

Table 7 Composition of the PGE inclusions, the brazing alloy, and the base alloy nearby one of the joining areas of ring E3297, when data in Fig. 7 is normalised to 100 wt% for each matrix and selected area.

PGE1 PGE2 PGE3

basealloy braze PGE1 PGE2 PGE3

Ru wt%

Os wt%

Ir wt%

39.5 26.1 2.9

31.2 33.1 83.2

28.5 37.5 10.0

Pt wt%

Cu wt%

Ag wt%

0.8 3.3 3.9% Au wt%

0.8 10.0 0.8 9.2 9.3

3.7 8.2 4.5 7.1 8.3

95.5 81.8 94.7 83.7 82.4

delimitation of the area of interest without adding extra information from the surrounding areas. The selection depends on the user’s choice. The biggest limitation to achieve good detection limits with fast measurements remains the study of very small areas of interest containing a very low number of pixels. The monitoring system of the mapping enables each pixel to correspond to a fixed charge. Therefore, for each map the detection limits are improved by expanding the selected area of interest. Data obtained for the joining areas analysed in this work show that they are obtained by brazing with gold alloys made by addition of Cu to the base alloy. These results match the recent data published on Egyptian jewellery attributed to the same periods. Concerning the PGE inclusions, we could show the variety of their compositions even within a same object, which could be associated to the re-melting for reuse of alluvial gold from different origins. Contrary to what could be expected from the few data on PGE inclusions published for Egyptian goldwork, some of our objects show in addition to the binary and ternary alloys belonging to the Os–Ir–Ru system the presence of a fourth element in the alloy: Pt. The presence of Pt can be either associated to the use of gold from a different origin or explained by the good detection limits of our technique (1500–6000 ppm according to the size of the inclusion, the acquisition time, and the intensity of the beam) compared to SEM–EDS, usually used to analyse PGE inclusions.

Acknowledgements The authors are grateful to Dr. Hélène Guichard curator at the Department of Egyptian Antiquities of the Louvre Museum for the access to the objects. This work falls under the scope of the Centre National de la Recherche Scientifique (CNRS) funded Project PICS 5995 “Analytical study of Bronze Age Egyptian gold jewellery”.

References [1] D. Schorsch, J. Egypt. Archaeol. 87 (2001) 55–71. [2] C. Lilyquist, The Tomb of the Three Foreign Wives of Tuthmosis III, The Metropolitan Museum of Art, New York, 2003. [3] L. Troalen, M.F. Guerra, J. Tate, W.P. Manley, ArcheoSciences 33 (2009) 111–119. [4] L.G. Troalen, M.F. Guerra, J. Tate, J. Archaeol. Sci. 50 (2014) 219–226. [5] N. Gale, Z.A. Stos-Gale, J. Egypt. Archaeol. 67 (1981) 103–115. [6] P.T. Craddock, in: A. Ramage, P.T. Craddock (Eds.), King Croesus' Gold, British Museum Press, London, 2000, pp. 27–53. [7] P.T. Craddock, M.R. Cowell, M.F. Guerra, in: U. Yalcin (Ed.), Anatolian Metal III, 18, Der Anschnitt 18, 2005, pp. 67–77. [8] G. Miniaci, S. La Niece, M.F. Guerra, M. Hacke, Br. Mus. Tech. Res. Bull. 7 (2013) 53–60. [9] H. Maryon, I. Welding and Soldering, MAN 41-85 (1941) 118–124. [10] J. Tate, K. Eremin, L.G. Troalen, M.F. Guerra, E. Goring, W.P. Manley, ArcheoSciences 33 (2009) 121–128. [11] D. Schorsch, in: C. Brown, F. Macalister F, M. Wright M (Eds.), Conservation in Ancient Egyptian Collections, Archetype Publications, London, 1995, pp. 127–135. [12] P.M. Roberts, Gold Bull. 6 (4) (1973) 112–119. [13] Flinders Petrie, W M, Quibell, J W 1896. Naqada and Ballas, London, B Quaritch ed. [14] J.M. Ogden, J. Egypt. Archaeol. 62 (1976) 138–144. [15] N.D. Meeks, M.S. Tite, J. Archaeol. Sci. 7 (1980) 267–275. [16] M. Berthelot, Monum. Mémoires 7 (2) (1900) 121–141. [17] J.M. Ogden, J. Hist. Metall. Soc. 2 (1977) 53–72. [18] M.F. Guerra, J. Archaeol. Sci. 31 (9) (2004) 1225–1236. [19] M.F. Guerra, X-ray Spectrom. 37 (4) (2008) 317–327. [20] M.F. Guerra, Radiat. Phys. Chem. 95 (2014) 356–361. [21] M. Radtke, I. Reiche, U. Reinholz, H. Riesemeier, M.F. Guerra, Anal. Chem. 85 (3) (2013) 1650–1656. [22] J.F. Bouchard, M.F. Guerra, ArcheoSciences 33 (2009) 273–279. [23] A.H. Ahmed, Ore Geology Reviews 32 (2007) 1–19. [24] D.C. Harris, L.J. Cabri, Can. Mineral. 29 (1991) 231–237. [25] L. Pichon, B. Moignard, Q. Lemasson, C. Pacheco, P. Walter, Nucl. Instrum. Methods Phys. Res. B 318 (2014) 27–31. [26] J.L. Campbell, N.I. Boyd, N. Grassi, P. Bonnick, J.A. Maxwell, Nucl. Instrum. Methods Phys. Res. B 268 (2010) 3356–3363. [27] L. Pichon, L. Beck, P. Walter, B. Moignard, T. Guillou, Nucl. Instrum. Methods Phys. Res. B 268 (2010) 2028–2033. [28] M.F. Guerra, Nucl. Instrum. Methods Phys. Res. B 226 (2004) 185–198. [29] M.F. Guerra, T. Calligaro, J. Archaeol. Sci. 31 (2004) 1199–1208.

286

Q. Lemasson et al. / Talanta 143 (2015) 279–286

[30] M.F. Guerra, I. Tissot, Nucl. Instrum. Methods Phys. Res. B 306 (2013) 227–231. [31] M. Grimwade, Gold Technol. 26 (1999) 16–23. [32] I. Tissot, L.G. Troalen, M. Manso, M. Ponting, M. Radtke, U. Reinholz, M. A. Barreiros, I. Shaw, M.L. Carvalho, M.F. Guerra, Spectrochim. Acta Part B: At. Spectrosc. 108 (2015) 75–82, http://dx.doi.org/10.1016/j.sab.2015.03.012. [33] M.M. Rifai, N.M.N. El Hadidi, in: J. Dawson, C. Rozeik, M.M. Wright (Eds.), Decorated Surfaces on Ancient Egyptian Objects: Technology, Deterioration and Conservation, Archetype Publications, London, 2010, pp. 16–24.

Further reading [34] M.F. Guerra, M. Radtke, I. Reiche, H. Riesemeier, S. Strub, Nucl. Instrum. Methods Phys. Res. B 266 (2008) 2334–2338.