- Email: [email protected]
Ion beam analysis of golden threads from Romanian medieval textiles
Nuclear Instruments and Methods in Physics Research B xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research B journal homepage: www.elsevier.com/locate/nimb
Ion beam analysis of golden threads from Romanian medieval textiles Z.I. Balta a,⇑, L. Csedreki b, E. Furu b, I. Cretu c, R. Huszánk b, M. Lupu c, Z. Török b, Z. Kertész b, Z. Szikszai b a
National History Museum of Romania, Calea Victoriei 12, Sector 3, Bucharest, Romania Institute for Nuclear Research, Hungarian Academy of Sciences, H-4001 Debrecen, P.O. Box 51, Hungary c National Art Museum of Romania, Calea Victoriei 49-53, Sector 1, Bucharest, Romania b
a r t i c l e
i n f o
Article history: Received 11 July 2014 Received in revised form 27 December 2014 Accepted 13 January 2015 Available online xxxx Keywords: Ion beam analysis Golden threads Medieval textiles
a b s t r a c t In this study, metal threads from Romanian religious embroideries and precious velvet brocades dated from 15th to 18th century were analyzed by using IBA methods (PIXE and RBS) which, in comparison to the traditional analytical techniques (XRF, EDS), allowed the detection of their structures and accurate identification of the trace elements (detection limits of few tens of ppm). PIXE results confirmed that both types of the metal threads studied – wires and strips – have layered structures being made of fine silver, refined by cupellation, and gilded most probably with pure gold, and not of Au–Ag alloy, or gilded Ag–Cu alloy or Au–Ag–Cu alloy, as resulted from the previously performed SEM-EDS analysis. Trace elements of historical interest like lead, mercury and bismuth have been also possible to be detected by PIXE. The resulting elemental maps allowed us to identify the areas from which the metal thread structure and quantitative composition could be accurately determined. RBS measurements revealed that the gilding layer is separated from the silver bulk by an interface layer resulting through atomic diffusion of silver into the gold, which lead to the conclusion that the methods used for gilding were probably either the diffusion bonding or the fire gilding. The gilding layers thicknesses were estimated by PIXE with the GUPIX software and also determined from RBS measurements. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Technical studies by classical techniques routinely used for analysis of cultural heritage materials (XRF, SEM-EDS) have been done previously in order to determine the chemical nature and morphology of the metal threads used in ancient textiles [1–4]. Some studies revealed that XRF and EDS could not distinguish between alloys and layered structured materials for the small and extremely thin gilded metal threads. Besides, EDS may sometimes give erroneous results. It was observed that the gold concentrations resulted for the gold coating and reported in previous publications are often too low depending on the high accelerating voltage used (usually 20–30 keV in practice), and thus incorrect [5,6]. Recently, in order to overcome some of these difficulties, more sensitive analytical techniques like AES, XPS, SIMS and laser-ICP/ MS have been used, and to our knowledge, few characterization studies by using IBA methods have been done so far [5–8]. Elemental analysis either on surface or in a depth profile by IBA methods are non-destructive and could provide unique information at the
⇑ Corresponding author. E-mail address: [email protected] (Z.I. Balta).
trace level sensitivity which make them ideal for the study of the very small and thin, possibly multilayered metal threads. Aim of our study was to demonstrate the necessity of integrating the advanced IBA methods with the classical analysis techniques frequently used in museums, for an in-depth applied interdisciplinary research that brought new developments and rich accurate information on golden threads constituent materials, especially the trace elements, elemental depth distributions and layers thicknesses, the ancient production technologies, metal threads structure, their provenance and origin. IBA measurements were carried out within the EU FP7 CHARISMA FIXLAB Transnational Access programme, in two phases: the projects IBATEX 1 and IBATEX 2.
2. Selection of textiles Metal threads selected for ion beam analysis were taken from medieval gold brocaded velvets and religious embroideries preserved in the textiles collections of the Putna Monastery Museum and the National Art Museum of Romania. For the PIXE and RBS analysis 23 textiles were chosen, most of them with known provenience (Moldavia, Italy, Near East and Greece) and consistent dating as appeared in the art historical literature [9,10]: brocades with
http://dx.doi.org/10.1016/j.nimb.2015.01.027 0168-583X/Ó 2015 Elsevier B.V. All rights reserved.
Please cite this article in press as: Z.I. Balta et al., Ion beam analysis of golden threads from Romanian medieval textiles, Nucl. Instr. Meth. B (2015), http:// dx.doi.org/10.1016/j.nimb.2015.01.027
2
Z.I. Balta et al. / Nuclear Instruments and Methods in Physics Research B xxx (2015) xxx–xxx
the inscription of donation embroidered, embroideries typologically and stylistically similar and brocades with no inscription of donation. Embroideries were made in most cases for liturgical purposes, worn by priests as church vestments or used during the religious services (epitrachelion, nabedernita, altar door curtain, aer, epimankia, etc.), most of them being produced locally in the Putna Monastery’s Embroidery School. They were worked in the traditional Byzantine technique of pattern couching, using the cartoon models painted on religious themes similar to those in the mural paintings dating from the 15th to 17th centuries. According to the sources, Byzantine embroidering technique consisted in laying golden threads on the surface of the background material and attaching them with stitches made of silk threads. The stitches were passed over the golden thread at regular intervals, creating different patterns [10]. Brocaded velvets were imported and produced abroad in the Italian workshops from Venice or Florence, or in the Ottoman Empire workshops, as mentioned in the art historical literature [10]. Brocades were sumptuous materials of silk velvet fabrics richly decorated with golden metal threads that were used in the articles of clothes of Italian and Oriental influence by the Moldavian and Wallachian princes, princesses and boyars (ceremony costumes, court vestments, caftans, granatza, etc.). They were sometimes donated to churches and monasteries after being worn. With a simple cut, those clothes could be dismantled, the resulting fabric fragments being used during the religious services as temple veils, covers for lecterns and the Communion Table or as tomb covers. 3. Experimental IBA techniques were applied at the Oxford-type nuclear microprobe facility in the MTA Atomki, Debrecen, Hungary [11–13]. The analyses were performed in vacuum and the size of the beam was typically about 2 lm. The length of the threads was approximately 3–10 mm. PIXE measurements were performed on 50 samples taken from 18 medieval textiles (IBATEX 1) and on 23 cross-sections of threads prelevated from 8 textiles (IBATEX 2) by using a proton beam of 3 MeV energy and 100–500 pA intensity. PIXE analysis modes performed were: full area elemental mapping of the sample surface, selected raster, and point analysis. The resulting PIXE spectra were evaluated with the GUPIX code [14]. In IBATEX 1, three different types of metal threads were analyzed: wires wrapped around a dyed silk yarn, strips wrapped around a dyed or undyed silk yarn and wires with no core yarns. In order to obtain more precise and accurate results regarding the metal threads composition and chemical structure, our study continued with analysis on cross-sections within the IBATEX 2 project. Cross-sections were
obtained by embedding the wires and strips in epoxy resin, the resulted specimens being cut in transversal sections which were then wet grounded and polished down to grit 8000. RBS analysis were carried out on 10 samples taken from 6 textiles (IBATEX 1) and 26 samples prelevated from 12 textiles (IBATEX 2) using a He+ beam of 2 MeV energy and a few hundred pA intensity. During the first phase (IBATEX 1), for the evaluation of the PIXE data, it was assumed that the bulk silver includes less than 1% gold, and the gilding layer contains only pure gold, all the trace elements being present in the silver substrate only. The gilding layers thicknesses were also estimated with the GUPIX software. For the second measurements phase (IBATEX 2), carefully cut and polished cross-sections were prepared to assess the silver bulk composition by PIXE without any possible contamination from the gilding layer or the silk yarn. The thicknesses of the gilding layers were measured by dedicated RBS analysis on metal threads using an ORTEC-type surface barrier silicon detector (‘‘ULTRA’’ IonImplanted, 50 mm2 sensitive area and 25 keV system energy resolution). The detector was placed at a scattering angle of 165° at Cornell geometry [15,16]. The scan size was set to 500 500 lm2. RBS spectra were evaluated with the SimNRA computer code version 6.06 [17]. With the second approach (IBATEX 2), we also could check the validity of the assumptions made during the first phase (IBATEX 1). Preliminary examinations in reflected and polarized light, at different magnifications, was performed in order to determine the metal threads technological and morphological characteristics, also the wires diameters (in cross-sections), with a Nikon Eclipse LV100D microscope equipped with a D90 digital camera, a Camera Control Pro 2.0 imaging software and a NIS Elements-BR3.0 image analysis software. 4. Results and discussion Preliminary optical microscopy measurements showed that the strips had a total width of 0.2–0.6 mm and a thickness of 0.01– 0.05 mm, while the wires diameters were of approximately 0.1– 0.3 mm. Optical microscopy and PIXE maps revealed that wires, and some of the strips (especially strips from the brocades), present striations on the surface caused by drawing in the manufacturing process. In Fig. 1, the parallel longitudinal lines on the surface of sample K3 could be visible. Microscopy observations and PIXE lead to further information, two different types of strips – cut and flattened – being identified. According to written sources [5,18,19], the cut strips are early times threads dated from the 13th century onwards, uneven in size and quality and of variable thickness, obtained by cutting a gilt metal sheet. The flattened strips, dated from the 14th century
Fig. 1. The K3 – gilded silver wire microscopy image in reflected light and its corresponding AgL and Au PIXE maps. In the Au map, could be seen that the gold distribution is uneven showing some parallel lines.
Please cite this article in press as: Z.I. Balta et al., Ion beam analysis of golden threads from Romanian medieval textiles, Nucl. Instr. Meth. B (2015), http:// dx.doi.org/10.1016/j.nimb.2015.01.027
Z.I. Balta et al. / Nuclear Instruments and Methods in Physics Research B xxx (2015) xxx–xxx
3
Fig. 2. Optical microscopy images in reflected light of samples J2 - gilded silver wire wrapped around a red silk yarn and K2 - gilded silver strip wrapped around a yellow silk yarn ‘‘drappi d’oro’’ and their cross-sections PIXE maps (IBATEX 2). For wire J2, iron seems to be present in the red dyed silk yarn (Fe map), respectively, for strip K2, copper could be seen to be present not only in silver, but also in the yellow silk yarn (Cu map).
onwards, were obtained from cast, drawn and rolled wires. They have even width all along their length and show parallel lines on their surface resulted from the wire drawing process. PIXE maps obtained in IBATEX 2 on cross-sections, in completion to maps resulted in IBATEX 1, revealed the thread structure and distribution of the elements in the layers. Fig. 2 shows the distribution maps of some of the samples’ J2 and K2 constituent elements. In Table 1 are presented the resulting concentrations of the elements: silver, gold, copper, lead, calcium, mercury, iron and bismuth, that were detected in almost all samples (wires and strips) by PIXE bulk analysis (full selected areas). Table 1 also lists brief descriptions and dating of the selected textiles from where samples were taken. Silver concentration in the wires and strips measured by PIXE (IBATEX 1) varies between 94.3% (sample B4) and 99.5% (sample Z2), and for the wires in cross-sections (IBATEX 2) between 97.3% (B3 cs.) and 99.4% (D3 cs.) indicating that a ‘‘pure’’ silver (Ag concentration > 95%), typical of a silver obtained by cupellation was employed. According to Biringuccio [20], a cast fine cupelled silver was used by the medieval smiths to make the so called spun silver or gold which is actually the gilded silver strip wrapped around a silk yarn employed in embroideries and velvet brocades. The resulting lead concentration is in accordance with the lead content mentioned in the literature for medieval silver [21,22]. The majority of the samples, wires and strips, contain less than 0.4% Pb, which strongly indicates cupellation. No lead, for wires A1 and A7, or very small content of lead, for wire X2, was detected, suggesting that they were obtained from native silver (argentite) or from rich silver ores (cerussite, cerargyrite or jarosite). Ancient silver originated from native silver or from rich silver ores that could have been smelted without recourse to lead to absorb the silver, may contain very little or no lead at all. In the strips P3 and P4, high amounts of lead was measured, which indicated that different silver ore sources and different technologies of production were used. Based on the lead content results could be assumed that some samples may belong to different periods or have a different provenance. Lead can be a good indicator of the technological level of purifying processes. Low concentrations of lead could indicate an advanced technology for the refining process. Similar contents of
lead indicate similar technologies of silver extraction and refining, while less lead suggests very advanced technologies of production. Gold content in silver is considered a good indicator of silver sources and one of the most important trace elements for silver authenticity [22]. Gold concentrations as resulted in IBATEX vary between 0.01% (Z1) and 1.8% (E4), and between 0.03% (X2 cs.) and 1.4% (B3 cs.) in IBATEX 2, which is in accordance with the values mentioned in the literature for gold contents in the medieval and post-medieval silver [22]. When high content of gold with low content of lead is detected it is assumed that re-melting of ancient old silver could occur (possibly sample U2). In medieval production of precious metals important sources were recycled silver or gold. Metal smiths used to re-melt ancient coins, jewelry or other gold or silver objects as their raw materials [23,24]. Another element determined by PIXE (IBATEX 1) was calcium which according with the elemental maps seems to have a homogeneous distribution and to be originated from the silk yarn, being probably present as a constituent of the silk thread, or resulting from the silk washing waters or silk treating solutions. PIXE analysis studies on silk composition have shown that silk, besides C, H, O and N, contains the elements Ca, Cu, P, S, Fe, and Zn [25]. No calcium was observed in the specifically prepared, presumably contamination free cross-sections (IBATEX 2). The presence or absence of copper (<3%) may be significant in indicating the purity of the silver employed. Concentration of copper for all samples was less than 2%, indicating that it was not deliberately added. In some cases, copper was observed to be present not only in silver, but also in the silk yarn as confirmed by PIXE cross-sections maps (Fig. 2, Cu map of the sample K2). Presence of mercury revealed by PIXE quantitative results (IBATEX 1) may indicate that a gold amalgamation technique was employed for gilding (fire gilding). Mercury may have resulted also, from an amalgamation process used for gold extraction, or from gold refining or parting using mercury, methods employed long ago before the 14th century only in the Middle East, and later on toward the end of the 15th century also in Europe. Biringuccio [20] refers to a mercury amalgamation technique that may have been used in silver extraction from lead ores containing large amounts of silver, but this technique seemed to start expanding
Please cite this article in press as: Z.I. Balta et al., Ion beam analysis of golden threads from Romanian medieval textiles, Nucl. Instr. Meth. B (2015), http:// dx.doi.org/10.1016/j.nimb.2015.01.027
4
Z.I. Balta et al. / Nuclear Instruments and Methods in Physics Research B xxx (2015) xxx–xxx
Table 1 Elemental concentrations in wt% (Ag, Cu, Au, Pb, Ca) and in ppm (Fe, Hg, Bi) resulted by PIXE bulk analysis (full selected areas) within IBATEX 1and IBATEX 2 (cs = cross-section, W = wire, S + S = strip wrapped around a silk yarn, W + S = wire wrapped around a silk yarn). The analytical uncertainty is 2–5 rel.% for the major elements and 5–20 rel.% for the minor and trace elements. Selected textiles
Sample
wt%
ppm
Ag
Au
Cu
A1/W A2 cs./W A5 cs./W + S A5/W + S A5b/W + S A6 cs./W A7 cs./W A7/W A12 cs./W A12/W
97.5 98.1 98.6 95.8 98.6 98.7 98.6 98.2 98.5 97.9
0.30 0.49 0.74 0.83 0.55 0.36 0.23 0.23 0.74 0.83
1.54 1.16 0.34 0.36 0.21 0.67 1.12 1.03 0.35 0.38
B2 cs./W + S B2/W + S B3 cs./W B4 cs./W + S B4/W + S
98.6 96.0 97.3 97.7 94.3
0.22 0.27 1.44 0.65 0.71
1.12 1.41 0.95 1.27 1.34
D1cs./W D1/W D2 cs./W D2/W D3 cs./W D3/W
98.7 98.4 98.7 98.2 99.4 99.3
0.87 0.94 0.87 0.89 0.15 0.24
0.14 0.15 0.14 0.13 0.22 0.24
0.16 0.15 0.15 0.13 0.18 0.18
E1/S + S E2/S + S E3/S + S E4/S + S
98.1 97.6 98.2 96.7
0.24 1.30 0.64 1.82
0.50 0.30 0.27 0.24
0.04 0.18 0.35 0.15
F1 cs./W F1/W F2 cs./W F2/W F3 cs./W F3/W
98.9 98.6 99.1 99.1 99.0 99.5
0.54 0.74 0.49 0.65 0.24 0.16
0.29 0.20 0.20 0.13 0.17 0.05
0.06 0.05 0.06 0.04 0.47 0.18
J. Nabedernita with pearls, religious embroidery, 15th cent.
J1/W J2 cs./W + S J3/S + S
97.2 98.3 97.2
1.24 0.76 0.11
0.50 0.05 1.56
0.26 0.74 0.14
K. Princess Maria Voichita tomb cover, velvet brocade ‘‘drappi d’oro’’ (Italy) with inscription of donation, 15th–16th cent.
K3/W K3 cs./W K4/S + S
98.5 99.2 98.4
0.67 0.24 0.18
0.40 0.40 0.37
0.12 0.12 0.11
P. Epitrachelion (Inv. 50), religious embroidery, 15th cent.
P1 cs./W P2 cs./W P3/S + S P4/S + S
99.3 99.0 97.2 97.1
0.51 0.62 0.62 0.21
0.02 0.03 0.31 0.32
T2/S + S T3/S + S T4/W T5/W
97.0 97.5 98.6 97.8
0.14 0.34 0.30 0.26
0.61 0.69 0.44 1.35
A. Epitrachelion (Inv.70), religious embroidery with inscription of donation, 15th cent.
B. Epitrachelion (Inv.36), religious embroidery with inscription of donation, 15th cent. (1469)
D. Epitrachelion (Inv. 40), religious embroidery, 15th cent.
E. Embroidery with eagles, religious embroidery with inscription of donation, 16th cent. (1536)
F. Altar door curtain of the Assumption of Holy Virgin, religious embroidery with inscription of donation, 15th cent. (1485)
T. Epitrachelion (Inv. 35), religious embroidery, 15th cent.
Pb 0.1 0.2 0.21 0.15
Ca
Fe
Hg
0.55
560
2.68 0.42
140 770 390 170
0.46
70
70
370 390 310
250 190 810
50
0.2 0.21
0.56
0.03
2.17
660 110
110
0.08 0.21
3.33
640
130
410 330
0.29
170
90
90
0.52
680
80
50
420
220 0.96 0.53 0.43 1.02
440 300 200 520
120 530 350 270 130
120
140 140 70 0.60
60
830
290
0.87
690
430
0.18
420
30
30
0.76
380
170
350
1.08 0.84
0.62 1.28
700 350
40 160
380 400 1660
0.12 0.08 0.07 0.02
1.83 1.25 0.53 0.53
2150 870 290 200
110 300 200 50
200
U. Sacos, religious embroidery, 17th cent., Near East
U2/W
98.6
1.06
0.06
0.03
0.20
X. Altar door curtain (Zavesa) of the Annunciation, religious embroidery with inscription of donation, 15th cent. (1484)
X1 cs./W X2 cs./W X2/W
98.2 99.4 99.5
1.19 0.04 0.38
0.23 0.15 0.04
0.31 0.09 0.004
0.04
100
Z. Nabedernita, religious embroidery, 18th cent. (1746), Greece
Z1/W Z2/W
99.2 99.5
0.01 0.04
0.13 0.08
0.05 0.052
0.53 0.26
320 120
in Europe only at the beginning of the 16th century. Low mercury concentrations combined with high content of lead observed in some samples (Table 1, samples P3 and P4) could exclude a silver production by amalgamation. Compared to all the threads investigated, only the samples P3 and P4 showed high concentrations of lead and low concentrations of mercury, which could be assumed that mercury enriched lead ores were used for producing the silver needed for their fabrication. Iron in the elemental maps could be seen on the metal thread surface, and sometimes as originating from silk. In most of the cross-sections no iron was measured, indicating that it is not a
Bi
200
200 190 230
100 250
constituent of the silver bulk. Presence of iron may also result from an iron-gall ink used for sketching out the preparatory underdrawings for the embroideries. Bismuth was also among the detected elements. Usually bismuth accompanies silver and may give information about its provenance [26,27], being determined in the medieval silver originating from Serbia, Transylvania, Carinthia and also from the Middle East. Absence of bismuth in some metal threads and its presence in other samples may be an indication of different silver sources. Less lead and no bismuth could also suggest a repeated cupellation which could have removed the bismuth.
Please cite this article in press as: Z.I. Balta et al., Ion beam analysis of golden threads from Romanian medieval textiles, Nucl. Instr. Meth. B (2015), http:// dx.doi.org/10.1016/j.nimb.2015.01.027
Z.I. Balta et al. / Nuclear Instruments and Methods in Physics Research B xxx (2015) xxx–xxx
5
Fig. 3. Relationship between copper and lead concentrations (in ppm) in wires ( ), wires wrapped around a dyed silk yarn ( ) and strips ( ) wrapped around a dyed or undyed silk yarn (IBATEX 1).
Copper and lead concentration comparison was done in order to determine if there are any qualitative differences between the types of metal threads analyzed in the IBATEX 1 project, and if there is any relation between the Cu/Pb ratio and the threads production processes. In Fig. 3, the samples showed to gather depending on the lead or copper content in three groups, the majority of them containing small lead and copper contents, suggesting that different provenience and metal processing methods for silver was employed. Samples P3 and P4 with high lead and small copper content, appear to gather together in a different group, and as revealed by optical microscopy, to be older, cut strips with no parallel striations on their surface. Also, the wire T5 seem to group with B2 wire wrapped around a silk yarn (high copper and small lead concentration), suggesting that T5 is similar with B2 and confirmed by optical microscopy. Fig. 3 showed also that the majority of the wires with no core yarn contain less than 2000 ppm copper, while the majority of the wires and strips wrapped around a dyed or undyed silk yarn more than 2000 ppm copper which is in accordance with the literature [18]. RBS measurements were also performed in order to determine the metal thread structure and the gold layer thickness. Due to the metal thread significant weathering, production technology and high surface roughness, RBS analyses were difficult to carry out in some cases. A small area was scanned over the thread, mapping the gold layer, and then point measurements were done on different locations. Evaluating the RBS spectra, results revealed that a significant atomic diffusion took place between the gilding layer and the silver substrate, making the fitting of the spectra challenging. The presence of this interface layer lead us to the conclusion that the possible methods used for gilding were either firegilding or diffusion bonding. In the diffusion bonding, an interface layer is formed between the two metal surfaces, brought in close contact by bounding the silver and gold bars together, by heating and soldering them to unite very well. The gilded silver bar is then hammered until it could be drawn into a wire or very thin, like a gold leaf, for being cut into strips. Biringuccio [20] described this method to have been used for producing both gilded silver strips and wires, and indicating that fine gold ducats served as raw material in the gilding process. In the fire-gilding method, gold and mercury are mixed into an amalgam paste which is evenly distributed on the silver surface. By heating, mercury evaporates and accelerates the diffusion between the silver substrate and gold coating
Fig. 4. RBS spectrum resulted for sample A6.
leaving very small amounts of mercury in the resulting gilding layer. Fig. 4 shows the diffusion zone in the RBS spectra obtained for sample A6 – gilded silver wire. PIXE measurements values of the gilding thicknesses estimated with the GUPIX software (IBATEX 1) were in good accordance with the RBS results (IBATEX 2): for wires, the thickness was varying between 0.1 and 0.8 lm, and for strips between 0.03 and 1.3 lm.
5. Conclusions Golden threads from Romanian medieval textiles have been analyzed for the first time by using multi-elemental non-destructive IBA methods. Compared to classical XRF and EDS techniques, IBA measurements, within IBATEX 1 and IBATEX 2 projects, allowed to accurately identify the elemental composition, distribution of the trace elements, the layered structures and thicknesses of the gilding layers. Useful information for characterization of the gilded silver threads due to elemental maps concerning the constituent elements spatial distribution in the sample was obtained. The cross-section elemental maps obtained in IBATEX 2 clearly showed the distribution of the constituent elements in the silver bulk, gilding layer and the silk core yarn.
Please cite this article in press as: Z.I. Balta et al., Ion beam analysis of golden threads from Romanian medieval textiles, Nucl. Instr. Meth. B (2015), http:// dx.doi.org/10.1016/j.nimb.2015.01.027
6
Z.I. Balta et al. / Nuclear Instruments and Methods in Physics Research B xxx (2015) xxx–xxx
Preliminary optical microscopy and PIXE maps revealed substantial differences in the construction of the metal threads. Parallel longitudinal striations were observed on wires and some of the strips, probably due to the drawing process, and two different types of strips – cut and flattened - were identified. PIXE and RBS results confirmed that both types of the metal threads studied – wires and strips – have a layered structure being produced from ‘‘pure’’ silver (Ag concentration > 95%) refined by cupellation and gilded probably by fire-gilding or diffusion gilding. In addition to silver, gold, copper, lead, calcium and traces of mercury, iron and bismuth were detected by PIXE. Based on the resulting PIXE maps, it was assumed that copper, lead, bismuth and probably mercury were components of the silver ores, while calcium and iron appeared to be mainly contaminations from the silk yarns. Variation of the gold content in the silver bulk, also the variable lead and copper contents, indicated that silver used for producing metal threads had different provenience. The silver ore sources used were probably native silver, rich silver ores, argentiferous galena, copper lead ores or mercury enriched lead ores. Furthermore, metal threads from the same textile seem to have different provenience and to be older than the period of time that the textile was dating. The assumed structure consisting in a layer of fine cupelled silver underneath a diffusion layer of silver into the gold and the gold layer on top was confirmed by RBS. Thicknesses of the gilding layers were estimated by PIXE with the GUPIX software and also determined from RBS quantitative measurements. For most of the samples, results revealed a non-uniform and variable thickness of the gilding layer that could be due to the non-homogeneity in the manufacturing processes, extensive wear, defects into the gold layer or contaminations. PIXE (IBATEX 1) and RBS measurements (IBATEX 2) showed that the gold layers are non-uniform and extremely thin: for strips, varying between 0.03 and 1.3 lm, and for wires between 0.1 and 0.8 lm. Data resulted on metal composition, especially the trace elements, metal thread structure, elemental depth distributions and layers thicknesses were used to clarify and complete the existing information concerning their techniques of production, gilding technologies and to identify if there are any qualitative differences between them. Considering that the selected historical textiles were consistently dated by art historians, the results obtained could contribute as reference and could constitute a preliminary database on historical metal threads to further studies. Since metal threads were easily transportable mediums of trade, it is possible that the ratios between their constituent elements to be indicative of the manufacturing and distribution sites. In regards with the implications for preservation and conservation practice, results obtained would help restorers and conservators to decide on appropriate treatments in order to bring the object closest to its originally intended appearance.
Acknowledgements Financial support by the Transnational Access to Research Infrastructures activity in the 7th Framework Programme of the EU
(CHARISMA Grant acknowledged.
Agreement
No.
228330)
is
gratefully
References [1] E. Hoke, I. Petrascheck-Heim, Mikroprobe analysis of gilded silver threads from medieval textiles, Stud. Conserv. 22 (1977) 49–62. [2] N. Indictor, R.J. Koestler, C. Blair, A. Wardwell, The evaluation of metal wrappings from medieval textiles using scanning electron microscopy – energy dispersive X-ray spectrometry, Text. Hist. 19 (1988) 3–22. [3] M. Jaro, Gold embroidery and fabrics in Europe: XI–XIV centuries, Gold Bull. 23 (1990) 40–57. [4] I.Z. Balta, I. Cretu, Scientific investigation of metal threads from Romanian medieval embroideries, in: ICOM-CC METAL 2001 (Chile) Proceedings, Western Australian Museum, 2004, pp. 90–96. [5] A.G. Nord, K. Tronner, A note on the analysis of gilded metal embroidery threads, Stud. Conserv. 45 (2000) 274–279. [6] K. Tronner, A.G. Nord, J. Sjostedt, H. Hydman, Extremely thin gold layers on gilded silver threads, Stud. Conserv. 47 (2002) 109–116. [7] A.M. Hacke, C.M. Carra, A. Brown, Characterization of metal threads in Renaissance tapestries, in: ICOM-CC Conference Metal 2004 (Canberra) Proceedings, National Museum of Australia, 2004, pp. 415–426. [8] O. Enguita, A. Climent-Font, G. Garcia, I. Montero, M.E. Fedi, M. Chiari, F. Lucarelli, Characterization of metal threads using differential PIXE analysis, Nucl. Instr. Meth. B 189 (2002) 328–333. [9] C. Paradais, Comori ale spiritualita˘ßt ii românesßti la Putna, Mitropoliei Moldovei sßi Sucevei, Iasßi, 1988, pp. 187–341. [10] C. Nicolescu, Costumul de Curte in Tarile Romane (Sec. XIV–XVIII), Intreprinderea Poligrafica Arta Grafica, Bucuresti, 1970. [11] I. Rajta, I. Borbély-Kiss, Gy. Morik, L. Bartha, E. Koltay, Á.Z. Kiss, The new ATOMKI scanning proton microprobe, Nucl. Instr. Meth. B 109–110 (1996) 148–153. [12] G.W. Grime, M. Dawson, Recent developments in data acquisition and processing on the Oxford scanning proton microprobe, Nucl. Instr. Meth. B 104 (1995) 107–113. [13] Zs. Kertész, Z. Szikszai, I. Uzonyi, A. Simon, Á.Z. Kiss, Development of a bio-PIXE setup at the Debrecen scanning proton microprobe, Nucl. Instr. Meth. B 231 (2005) 106–111. [14] J.L. Campbell, N.I. Boyd, N. Grassi, P. Bonnick, J.A. Maxwell, The Guelph PIXE software package IV, Nucl. Instr. Meth. B 268 (2010) 3356–3363. [15] A. Simon, R. Huszánk, M. Novák, Z. Pintye, Investigation of hydrogen depletion of organic materials upon ion beam irradiation by simultaneous micro-RBS and micro-ERDA techniques, Nucl. Instr. Meth. B 268 (2010) 2197–2201. [16] R. Huszánk, A. Simon, E. Szilágyi, K. Keresztessy, I. Kovács, Micro-ERDA, microRBS and micro-PIXE techniques in the investigation of fish otoliths, Nucl. Instr. Meth. B 267 (2009) 2132–2135. [17] M. Mayer, SIMNRA user’s guide, Tech. Rep. IPP 9/113, Max–Planck–Institut für Plasmaphysik, Garching, 1997. [18] A. Karatzani, T. Rehren, Clothes of gold: metal threads in Byzantine-Greek orthodox ecclesiastical textiles, in: J.F. Moreau, R. Auger, J. Chabot, A. Herzog (Eds.), Proceedings ISA 2006, Quebec, 2009, pp. 9–19. [19] A.D. Barker, A brief survey of the manufacture and use of precious metal wires and threads for embroidery and weaving, Northern Soc. Costume Text. (1980) 1–14. [20] C.S. Smith, M. Teach Gnudi, The Pirotechnia of Vannoccio Biringuccio. The Classic Sixteenth Century Treatise on Metals and Metallurgy, Dover Publications Inc, New York, 1990, pp. 377–383. [21] R.F. Tylecote, A History of Metallurgy, Maney Publishing for the Institute of Materials, London, 2002, pp. 89–118. [22] P. Craddock, Scientific Investigation of Copies, Fakes and Forgeries, Butterworth-Heinemann Elsevier Ltd., 2009, pp. 386–388. [23] Ch. Flament, P. Marchetti, Analysis of ancient silver coins, Nucl. Instr. Meth. B 226 (2004) 179–184. [24] P. Masjedi, F. Khandemi, M. Hajivaliei, S.M.M. Kouhpar, J. Neystani, Elemental analysis of silver coins of Seljuk’s’ of Rome by PIXE, Mediterranean Archaeol. Archaeom. 13 (2013) 181–187. [25] L. Yongcheng, Y. Tongyin, Y. Huiying, Y. Fujia, PIXE analysis of silk, J. Appl. Polym. Sci. 66 (1997) 405–408. [26] N. Kallithrakas-Kontos, A. Katsanos, A. Aravantinos, M. Oeconomid, I. Touratsog, Trace elements analysis of Alexander the Great’s silver tetradrachms minted in Macedonia, Nucl. Instr. Meth. B 171 (2000) 342–349. [27] Z. Smit, A. Semrov, Early medieval coinage in the territory of Slovenia, Nucl. Instr. Meth. B 252 (2006) 290–298.
Please cite this article in press as: Z.I. Balta et al., Ion beam analysis of golden threads from Romanian medieval textiles, Nucl. Instr. Meth. B (2015), http:// dx.doi.org/10.1016/j.nimb.2015.01.027
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research B journal homepage: www.elsevier.com/locate/nimb
Ion beam analysis of golden threads from Romanian medieval textiles Z.I. Balta a,⇑, L. Csedreki b, E. Furu b, I. Cretu c, R. Huszánk b, M. Lupu c, Z. Török b, Z. Kertész b, Z. Szikszai b a
National History Museum of Romania, Calea Victoriei 12, Sector 3, Bucharest, Romania Institute for Nuclear Research, Hungarian Academy of Sciences, H-4001 Debrecen, P.O. Box 51, Hungary c National Art Museum of Romania, Calea Victoriei 49-53, Sector 1, Bucharest, Romania b
a r t i c l e
i n f o
Article history: Received 11 July 2014 Received in revised form 27 December 2014 Accepted 13 January 2015 Available online xxxx Keywords: Ion beam analysis Golden threads Medieval textiles
a b s t r a c t In this study, metal threads from Romanian religious embroideries and precious velvet brocades dated from 15th to 18th century were analyzed by using IBA methods (PIXE and RBS) which, in comparison to the traditional analytical techniques (XRF, EDS), allowed the detection of their structures and accurate identification of the trace elements (detection limits of few tens of ppm). PIXE results confirmed that both types of the metal threads studied – wires and strips – have layered structures being made of fine silver, refined by cupellation, and gilded most probably with pure gold, and not of Au–Ag alloy, or gilded Ag–Cu alloy or Au–Ag–Cu alloy, as resulted from the previously performed SEM-EDS analysis. Trace elements of historical interest like lead, mercury and bismuth have been also possible to be detected by PIXE. The resulting elemental maps allowed us to identify the areas from which the metal thread structure and quantitative composition could be accurately determined. RBS measurements revealed that the gilding layer is separated from the silver bulk by an interface layer resulting through atomic diffusion of silver into the gold, which lead to the conclusion that the methods used for gilding were probably either the diffusion bonding or the fire gilding. The gilding layers thicknesses were estimated by PIXE with the GUPIX software and also determined from RBS measurements. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Technical studies by classical techniques routinely used for analysis of cultural heritage materials (XRF, SEM-EDS) have been done previously in order to determine the chemical nature and morphology of the metal threads used in ancient textiles [1–4]. Some studies revealed that XRF and EDS could not distinguish between alloys and layered structured materials for the small and extremely thin gilded metal threads. Besides, EDS may sometimes give erroneous results. It was observed that the gold concentrations resulted for the gold coating and reported in previous publications are often too low depending on the high accelerating voltage used (usually 20–30 keV in practice), and thus incorrect [5,6]. Recently, in order to overcome some of these difficulties, more sensitive analytical techniques like AES, XPS, SIMS and laser-ICP/ MS have been used, and to our knowledge, few characterization studies by using IBA methods have been done so far [5–8]. Elemental analysis either on surface or in a depth profile by IBA methods are non-destructive and could provide unique information at the
⇑ Corresponding author. E-mail address: [email protected] (Z.I. Balta).
trace level sensitivity which make them ideal for the study of the very small and thin, possibly multilayered metal threads. Aim of our study was to demonstrate the necessity of integrating the advanced IBA methods with the classical analysis techniques frequently used in museums, for an in-depth applied interdisciplinary research that brought new developments and rich accurate information on golden threads constituent materials, especially the trace elements, elemental depth distributions and layers thicknesses, the ancient production technologies, metal threads structure, their provenance and origin. IBA measurements were carried out within the EU FP7 CHARISMA FIXLAB Transnational Access programme, in two phases: the projects IBATEX 1 and IBATEX 2.
2. Selection of textiles Metal threads selected for ion beam analysis were taken from medieval gold brocaded velvets and religious embroideries preserved in the textiles collections of the Putna Monastery Museum and the National Art Museum of Romania. For the PIXE and RBS analysis 23 textiles were chosen, most of them with known provenience (Moldavia, Italy, Near East and Greece) and consistent dating as appeared in the art historical literature [9,10]: brocades with
http://dx.doi.org/10.1016/j.nimb.2015.01.027 0168-583X/Ó 2015 Elsevier B.V. All rights reserved.
Please cite this article in press as: Z.I. Balta et al., Ion beam analysis of golden threads from Romanian medieval textiles, Nucl. Instr. Meth. B (2015), http:// dx.doi.org/10.1016/j.nimb.2015.01.027
2
Z.I. Balta et al. / Nuclear Instruments and Methods in Physics Research B xxx (2015) xxx–xxx
the inscription of donation embroidered, embroideries typologically and stylistically similar and brocades with no inscription of donation. Embroideries were made in most cases for liturgical purposes, worn by priests as church vestments or used during the religious services (epitrachelion, nabedernita, altar door curtain, aer, epimankia, etc.), most of them being produced locally in the Putna Monastery’s Embroidery School. They were worked in the traditional Byzantine technique of pattern couching, using the cartoon models painted on religious themes similar to those in the mural paintings dating from the 15th to 17th centuries. According to the sources, Byzantine embroidering technique consisted in laying golden threads on the surface of the background material and attaching them with stitches made of silk threads. The stitches were passed over the golden thread at regular intervals, creating different patterns [10]. Brocaded velvets were imported and produced abroad in the Italian workshops from Venice or Florence, or in the Ottoman Empire workshops, as mentioned in the art historical literature [10]. Brocades were sumptuous materials of silk velvet fabrics richly decorated with golden metal threads that were used in the articles of clothes of Italian and Oriental influence by the Moldavian and Wallachian princes, princesses and boyars (ceremony costumes, court vestments, caftans, granatza, etc.). They were sometimes donated to churches and monasteries after being worn. With a simple cut, those clothes could be dismantled, the resulting fabric fragments being used during the religious services as temple veils, covers for lecterns and the Communion Table or as tomb covers. 3. Experimental IBA techniques were applied at the Oxford-type nuclear microprobe facility in the MTA Atomki, Debrecen, Hungary [11–13]. The analyses were performed in vacuum and the size of the beam was typically about 2 lm. The length of the threads was approximately 3–10 mm. PIXE measurements were performed on 50 samples taken from 18 medieval textiles (IBATEX 1) and on 23 cross-sections of threads prelevated from 8 textiles (IBATEX 2) by using a proton beam of 3 MeV energy and 100–500 pA intensity. PIXE analysis modes performed were: full area elemental mapping of the sample surface, selected raster, and point analysis. The resulting PIXE spectra were evaluated with the GUPIX code [14]. In IBATEX 1, three different types of metal threads were analyzed: wires wrapped around a dyed silk yarn, strips wrapped around a dyed or undyed silk yarn and wires with no core yarns. In order to obtain more precise and accurate results regarding the metal threads composition and chemical structure, our study continued with analysis on cross-sections within the IBATEX 2 project. Cross-sections were
obtained by embedding the wires and strips in epoxy resin, the resulted specimens being cut in transversal sections which were then wet grounded and polished down to grit 8000. RBS analysis were carried out on 10 samples taken from 6 textiles (IBATEX 1) and 26 samples prelevated from 12 textiles (IBATEX 2) using a He+ beam of 2 MeV energy and a few hundred pA intensity. During the first phase (IBATEX 1), for the evaluation of the PIXE data, it was assumed that the bulk silver includes less than 1% gold, and the gilding layer contains only pure gold, all the trace elements being present in the silver substrate only. The gilding layers thicknesses were also estimated with the GUPIX software. For the second measurements phase (IBATEX 2), carefully cut and polished cross-sections were prepared to assess the silver bulk composition by PIXE without any possible contamination from the gilding layer or the silk yarn. The thicknesses of the gilding layers were measured by dedicated RBS analysis on metal threads using an ORTEC-type surface barrier silicon detector (‘‘ULTRA’’ IonImplanted, 50 mm2 sensitive area and 25 keV system energy resolution). The detector was placed at a scattering angle of 165° at Cornell geometry [15,16]. The scan size was set to 500 500 lm2. RBS spectra were evaluated with the SimNRA computer code version 6.06 [17]. With the second approach (IBATEX 2), we also could check the validity of the assumptions made during the first phase (IBATEX 1). Preliminary examinations in reflected and polarized light, at different magnifications, was performed in order to determine the metal threads technological and morphological characteristics, also the wires diameters (in cross-sections), with a Nikon Eclipse LV100D microscope equipped with a D90 digital camera, a Camera Control Pro 2.0 imaging software and a NIS Elements-BR3.0 image analysis software. 4. Results and discussion Preliminary optical microscopy measurements showed that the strips had a total width of 0.2–0.6 mm and a thickness of 0.01– 0.05 mm, while the wires diameters were of approximately 0.1– 0.3 mm. Optical microscopy and PIXE maps revealed that wires, and some of the strips (especially strips from the brocades), present striations on the surface caused by drawing in the manufacturing process. In Fig. 1, the parallel longitudinal lines on the surface of sample K3 could be visible. Microscopy observations and PIXE lead to further information, two different types of strips – cut and flattened – being identified. According to written sources [5,18,19], the cut strips are early times threads dated from the 13th century onwards, uneven in size and quality and of variable thickness, obtained by cutting a gilt metal sheet. The flattened strips, dated from the 14th century
Fig. 1. The K3 – gilded silver wire microscopy image in reflected light and its corresponding AgL and Au PIXE maps. In the Au map, could be seen that the gold distribution is uneven showing some parallel lines.
Please cite this article in press as: Z.I. Balta et al., Ion beam analysis of golden threads from Romanian medieval textiles, Nucl. Instr. Meth. B (2015), http:// dx.doi.org/10.1016/j.nimb.2015.01.027
Z.I. Balta et al. / Nuclear Instruments and Methods in Physics Research B xxx (2015) xxx–xxx
3
Fig. 2. Optical microscopy images in reflected light of samples J2 - gilded silver wire wrapped around a red silk yarn and K2 - gilded silver strip wrapped around a yellow silk yarn ‘‘drappi d’oro’’ and their cross-sections PIXE maps (IBATEX 2). For wire J2, iron seems to be present in the red dyed silk yarn (Fe map), respectively, for strip K2, copper could be seen to be present not only in silver, but also in the yellow silk yarn (Cu map).
onwards, were obtained from cast, drawn and rolled wires. They have even width all along their length and show parallel lines on their surface resulted from the wire drawing process. PIXE maps obtained in IBATEX 2 on cross-sections, in completion to maps resulted in IBATEX 1, revealed the thread structure and distribution of the elements in the layers. Fig. 2 shows the distribution maps of some of the samples’ J2 and K2 constituent elements. In Table 1 are presented the resulting concentrations of the elements: silver, gold, copper, lead, calcium, mercury, iron and bismuth, that were detected in almost all samples (wires and strips) by PIXE bulk analysis (full selected areas). Table 1 also lists brief descriptions and dating of the selected textiles from where samples were taken. Silver concentration in the wires and strips measured by PIXE (IBATEX 1) varies between 94.3% (sample B4) and 99.5% (sample Z2), and for the wires in cross-sections (IBATEX 2) between 97.3% (B3 cs.) and 99.4% (D3 cs.) indicating that a ‘‘pure’’ silver (Ag concentration > 95%), typical of a silver obtained by cupellation was employed. According to Biringuccio [20], a cast fine cupelled silver was used by the medieval smiths to make the so called spun silver or gold which is actually the gilded silver strip wrapped around a silk yarn employed in embroideries and velvet brocades. The resulting lead concentration is in accordance with the lead content mentioned in the literature for medieval silver [21,22]. The majority of the samples, wires and strips, contain less than 0.4% Pb, which strongly indicates cupellation. No lead, for wires A1 and A7, or very small content of lead, for wire X2, was detected, suggesting that they were obtained from native silver (argentite) or from rich silver ores (cerussite, cerargyrite or jarosite). Ancient silver originated from native silver or from rich silver ores that could have been smelted without recourse to lead to absorb the silver, may contain very little or no lead at all. In the strips P3 and P4, high amounts of lead was measured, which indicated that different silver ore sources and different technologies of production were used. Based on the lead content results could be assumed that some samples may belong to different periods or have a different provenance. Lead can be a good indicator of the technological level of purifying processes. Low concentrations of lead could indicate an advanced technology for the refining process. Similar contents of
lead indicate similar technologies of silver extraction and refining, while less lead suggests very advanced technologies of production. Gold content in silver is considered a good indicator of silver sources and one of the most important trace elements for silver authenticity [22]. Gold concentrations as resulted in IBATEX vary between 0.01% (Z1) and 1.8% (E4), and between 0.03% (X2 cs.) and 1.4% (B3 cs.) in IBATEX 2, which is in accordance with the values mentioned in the literature for gold contents in the medieval and post-medieval silver [22]. When high content of gold with low content of lead is detected it is assumed that re-melting of ancient old silver could occur (possibly sample U2). In medieval production of precious metals important sources were recycled silver or gold. Metal smiths used to re-melt ancient coins, jewelry or other gold or silver objects as their raw materials [23,24]. Another element determined by PIXE (IBATEX 1) was calcium which according with the elemental maps seems to have a homogeneous distribution and to be originated from the silk yarn, being probably present as a constituent of the silk thread, or resulting from the silk washing waters or silk treating solutions. PIXE analysis studies on silk composition have shown that silk, besides C, H, O and N, contains the elements Ca, Cu, P, S, Fe, and Zn [25]. No calcium was observed in the specifically prepared, presumably contamination free cross-sections (IBATEX 2). The presence or absence of copper (<3%) may be significant in indicating the purity of the silver employed. Concentration of copper for all samples was less than 2%, indicating that it was not deliberately added. In some cases, copper was observed to be present not only in silver, but also in the silk yarn as confirmed by PIXE cross-sections maps (Fig. 2, Cu map of the sample K2). Presence of mercury revealed by PIXE quantitative results (IBATEX 1) may indicate that a gold amalgamation technique was employed for gilding (fire gilding). Mercury may have resulted also, from an amalgamation process used for gold extraction, or from gold refining or parting using mercury, methods employed long ago before the 14th century only in the Middle East, and later on toward the end of the 15th century also in Europe. Biringuccio [20] refers to a mercury amalgamation technique that may have been used in silver extraction from lead ores containing large amounts of silver, but this technique seemed to start expanding
Please cite this article in press as: Z.I. Balta et al., Ion beam analysis of golden threads from Romanian medieval textiles, Nucl. Instr. Meth. B (2015), http:// dx.doi.org/10.1016/j.nimb.2015.01.027
4
Z.I. Balta et al. / Nuclear Instruments and Methods in Physics Research B xxx (2015) xxx–xxx
Table 1 Elemental concentrations in wt% (Ag, Cu, Au, Pb, Ca) and in ppm (Fe, Hg, Bi) resulted by PIXE bulk analysis (full selected areas) within IBATEX 1and IBATEX 2 (cs = cross-section, W = wire, S + S = strip wrapped around a silk yarn, W + S = wire wrapped around a silk yarn). The analytical uncertainty is 2–5 rel.% for the major elements and 5–20 rel.% for the minor and trace elements. Selected textiles
Sample
wt%
ppm
Ag
Au
Cu
A1/W A2 cs./W A5 cs./W + S A5/W + S A5b/W + S A6 cs./W A7 cs./W A7/W A12 cs./W A12/W
97.5 98.1 98.6 95.8 98.6 98.7 98.6 98.2 98.5 97.9
0.30 0.49 0.74 0.83 0.55 0.36 0.23 0.23 0.74 0.83
1.54 1.16 0.34 0.36 0.21 0.67 1.12 1.03 0.35 0.38
B2 cs./W + S B2/W + S B3 cs./W B4 cs./W + S B4/W + S
98.6 96.0 97.3 97.7 94.3
0.22 0.27 1.44 0.65 0.71
1.12 1.41 0.95 1.27 1.34
D1cs./W D1/W D2 cs./W D2/W D3 cs./W D3/W
98.7 98.4 98.7 98.2 99.4 99.3
0.87 0.94 0.87 0.89 0.15 0.24
0.14 0.15 0.14 0.13 0.22 0.24
0.16 0.15 0.15 0.13 0.18 0.18
E1/S + S E2/S + S E3/S + S E4/S + S
98.1 97.6 98.2 96.7
0.24 1.30 0.64 1.82
0.50 0.30 0.27 0.24
0.04 0.18 0.35 0.15
F1 cs./W F1/W F2 cs./W F2/W F3 cs./W F3/W
98.9 98.6 99.1 99.1 99.0 99.5
0.54 0.74 0.49 0.65 0.24 0.16
0.29 0.20 0.20 0.13 0.17 0.05
0.06 0.05 0.06 0.04 0.47 0.18
J. Nabedernita with pearls, religious embroidery, 15th cent.
J1/W J2 cs./W + S J3/S + S
97.2 98.3 97.2
1.24 0.76 0.11
0.50 0.05 1.56
0.26 0.74 0.14
K. Princess Maria Voichita tomb cover, velvet brocade ‘‘drappi d’oro’’ (Italy) with inscription of donation, 15th–16th cent.
K3/W K3 cs./W K4/S + S
98.5 99.2 98.4
0.67 0.24 0.18
0.40 0.40 0.37
0.12 0.12 0.11
P. Epitrachelion (Inv. 50), religious embroidery, 15th cent.
P1 cs./W P2 cs./W P3/S + S P4/S + S
99.3 99.0 97.2 97.1
0.51 0.62 0.62 0.21
0.02 0.03 0.31 0.32
T2/S + S T3/S + S T4/W T5/W
97.0 97.5 98.6 97.8
0.14 0.34 0.30 0.26
0.61 0.69 0.44 1.35
A. Epitrachelion (Inv.70), religious embroidery with inscription of donation, 15th cent.
B. Epitrachelion (Inv.36), religious embroidery with inscription of donation, 15th cent. (1469)
D. Epitrachelion (Inv. 40), religious embroidery, 15th cent.
E. Embroidery with eagles, religious embroidery with inscription of donation, 16th cent. (1536)
F. Altar door curtain of the Assumption of Holy Virgin, religious embroidery with inscription of donation, 15th cent. (1485)
T. Epitrachelion (Inv. 35), religious embroidery, 15th cent.
Pb 0.1 0.2 0.21 0.15
Ca
Fe
Hg
0.55
560
2.68 0.42
140 770 390 170
0.46
70
70
370 390 310
250 190 810
50
0.2 0.21
0.56
0.03
2.17
660 110
110
0.08 0.21
3.33
640
130
410 330
0.29
170
90
90
0.52
680
80
50
420
220 0.96 0.53 0.43 1.02
440 300 200 520
120 530 350 270 130
120
140 140 70 0.60
60
830
290
0.87
690
430
0.18
420
30
30
0.76
380
170
350
1.08 0.84
0.62 1.28
700 350
40 160
380 400 1660
0.12 0.08 0.07 0.02
1.83 1.25 0.53 0.53
2150 870 290 200
110 300 200 50
200
U. Sacos, religious embroidery, 17th cent., Near East
U2/W
98.6
1.06
0.06
0.03
0.20
X. Altar door curtain (Zavesa) of the Annunciation, religious embroidery with inscription of donation, 15th cent. (1484)
X1 cs./W X2 cs./W X2/W
98.2 99.4 99.5
1.19 0.04 0.38
0.23 0.15 0.04
0.31 0.09 0.004
0.04
100
Z. Nabedernita, religious embroidery, 18th cent. (1746), Greece
Z1/W Z2/W
99.2 99.5
0.01 0.04
0.13 0.08
0.05 0.052
0.53 0.26
320 120
in Europe only at the beginning of the 16th century. Low mercury concentrations combined with high content of lead observed in some samples (Table 1, samples P3 and P4) could exclude a silver production by amalgamation. Compared to all the threads investigated, only the samples P3 and P4 showed high concentrations of lead and low concentrations of mercury, which could be assumed that mercury enriched lead ores were used for producing the silver needed for their fabrication. Iron in the elemental maps could be seen on the metal thread surface, and sometimes as originating from silk. In most of the cross-sections no iron was measured, indicating that it is not a
Bi
200
200 190 230
100 250
constituent of the silver bulk. Presence of iron may also result from an iron-gall ink used for sketching out the preparatory underdrawings for the embroideries. Bismuth was also among the detected elements. Usually bismuth accompanies silver and may give information about its provenance [26,27], being determined in the medieval silver originating from Serbia, Transylvania, Carinthia and also from the Middle East. Absence of bismuth in some metal threads and its presence in other samples may be an indication of different silver sources. Less lead and no bismuth could also suggest a repeated cupellation which could have removed the bismuth.
Please cite this article in press as: Z.I. Balta et al., Ion beam analysis of golden threads from Romanian medieval textiles, Nucl. Instr. Meth. B (2015), http:// dx.doi.org/10.1016/j.nimb.2015.01.027
Z.I. Balta et al. / Nuclear Instruments and Methods in Physics Research B xxx (2015) xxx–xxx
5
Fig. 3. Relationship between copper and lead concentrations (in ppm) in wires ( ), wires wrapped around a dyed silk yarn ( ) and strips ( ) wrapped around a dyed or undyed silk yarn (IBATEX 1).
Copper and lead concentration comparison was done in order to determine if there are any qualitative differences between the types of metal threads analyzed in the IBATEX 1 project, and if there is any relation between the Cu/Pb ratio and the threads production processes. In Fig. 3, the samples showed to gather depending on the lead or copper content in three groups, the majority of them containing small lead and copper contents, suggesting that different provenience and metal processing methods for silver was employed. Samples P3 and P4 with high lead and small copper content, appear to gather together in a different group, and as revealed by optical microscopy, to be older, cut strips with no parallel striations on their surface. Also, the wire T5 seem to group with B2 wire wrapped around a silk yarn (high copper and small lead concentration), suggesting that T5 is similar with B2 and confirmed by optical microscopy. Fig. 3 showed also that the majority of the wires with no core yarn contain less than 2000 ppm copper, while the majority of the wires and strips wrapped around a dyed or undyed silk yarn more than 2000 ppm copper which is in accordance with the literature [18]. RBS measurements were also performed in order to determine the metal thread structure and the gold layer thickness. Due to the metal thread significant weathering, production technology and high surface roughness, RBS analyses were difficult to carry out in some cases. A small area was scanned over the thread, mapping the gold layer, and then point measurements were done on different locations. Evaluating the RBS spectra, results revealed that a significant atomic diffusion took place between the gilding layer and the silver substrate, making the fitting of the spectra challenging. The presence of this interface layer lead us to the conclusion that the possible methods used for gilding were either firegilding or diffusion bonding. In the diffusion bonding, an interface layer is formed between the two metal surfaces, brought in close contact by bounding the silver and gold bars together, by heating and soldering them to unite very well. The gilded silver bar is then hammered until it could be drawn into a wire or very thin, like a gold leaf, for being cut into strips. Biringuccio [20] described this method to have been used for producing both gilded silver strips and wires, and indicating that fine gold ducats served as raw material in the gilding process. In the fire-gilding method, gold and mercury are mixed into an amalgam paste which is evenly distributed on the silver surface. By heating, mercury evaporates and accelerates the diffusion between the silver substrate and gold coating
Fig. 4. RBS spectrum resulted for sample A6.
leaving very small amounts of mercury in the resulting gilding layer. Fig. 4 shows the diffusion zone in the RBS spectra obtained for sample A6 – gilded silver wire. PIXE measurements values of the gilding thicknesses estimated with the GUPIX software (IBATEX 1) were in good accordance with the RBS results (IBATEX 2): for wires, the thickness was varying between 0.1 and 0.8 lm, and for strips between 0.03 and 1.3 lm.
5. Conclusions Golden threads from Romanian medieval textiles have been analyzed for the first time by using multi-elemental non-destructive IBA methods. Compared to classical XRF and EDS techniques, IBA measurements, within IBATEX 1 and IBATEX 2 projects, allowed to accurately identify the elemental composition, distribution of the trace elements, the layered structures and thicknesses of the gilding layers. Useful information for characterization of the gilded silver threads due to elemental maps concerning the constituent elements spatial distribution in the sample was obtained. The cross-section elemental maps obtained in IBATEX 2 clearly showed the distribution of the constituent elements in the silver bulk, gilding layer and the silk core yarn.
Please cite this article in press as: Z.I. Balta et al., Ion beam analysis of golden threads from Romanian medieval textiles, Nucl. Instr. Meth. B (2015), http:// dx.doi.org/10.1016/j.nimb.2015.01.027
6
Z.I. Balta et al. / Nuclear Instruments and Methods in Physics Research B xxx (2015) xxx–xxx
Preliminary optical microscopy and PIXE maps revealed substantial differences in the construction of the metal threads. Parallel longitudinal striations were observed on wires and some of the strips, probably due to the drawing process, and two different types of strips – cut and flattened - were identified. PIXE and RBS results confirmed that both types of the metal threads studied – wires and strips – have a layered structure being produced from ‘‘pure’’ silver (Ag concentration > 95%) refined by cupellation and gilded probably by fire-gilding or diffusion gilding. In addition to silver, gold, copper, lead, calcium and traces of mercury, iron and bismuth were detected by PIXE. Based on the resulting PIXE maps, it was assumed that copper, lead, bismuth and probably mercury were components of the silver ores, while calcium and iron appeared to be mainly contaminations from the silk yarns. Variation of the gold content in the silver bulk, also the variable lead and copper contents, indicated that silver used for producing metal threads had different provenience. The silver ore sources used were probably native silver, rich silver ores, argentiferous galena, copper lead ores or mercury enriched lead ores. Furthermore, metal threads from the same textile seem to have different provenience and to be older than the period of time that the textile was dating. The assumed structure consisting in a layer of fine cupelled silver underneath a diffusion layer of silver into the gold and the gold layer on top was confirmed by RBS. Thicknesses of the gilding layers were estimated by PIXE with the GUPIX software and also determined from RBS quantitative measurements. For most of the samples, results revealed a non-uniform and variable thickness of the gilding layer that could be due to the non-homogeneity in the manufacturing processes, extensive wear, defects into the gold layer or contaminations. PIXE (IBATEX 1) and RBS measurements (IBATEX 2) showed that the gold layers are non-uniform and extremely thin: for strips, varying between 0.03 and 1.3 lm, and for wires between 0.1 and 0.8 lm. Data resulted on metal composition, especially the trace elements, metal thread structure, elemental depth distributions and layers thicknesses were used to clarify and complete the existing information concerning their techniques of production, gilding technologies and to identify if there are any qualitative differences between them. Considering that the selected historical textiles were consistently dated by art historians, the results obtained could contribute as reference and could constitute a preliminary database on historical metal threads to further studies. Since metal threads were easily transportable mediums of trade, it is possible that the ratios between their constituent elements to be indicative of the manufacturing and distribution sites. In regards with the implications for preservation and conservation practice, results obtained would help restorers and conservators to decide on appropriate treatments in order to bring the object closest to its originally intended appearance.
Acknowledgements Financial support by the Transnational Access to Research Infrastructures activity in the 7th Framework Programme of the EU
(CHARISMA Grant acknowledged.
Agreement
No.
228330)
is
gratefully
References [1] E. Hoke, I. Petrascheck-Heim, Mikroprobe analysis of gilded silver threads from medieval textiles, Stud. Conserv. 22 (1977) 49–62. [2] N. Indictor, R.J. Koestler, C. Blair, A. Wardwell, The evaluation of metal wrappings from medieval textiles using scanning electron microscopy – energy dispersive X-ray spectrometry, Text. Hist. 19 (1988) 3–22. [3] M. Jaro, Gold embroidery and fabrics in Europe: XI–XIV centuries, Gold Bull. 23 (1990) 40–57. [4] I.Z. Balta, I. Cretu, Scientific investigation of metal threads from Romanian medieval embroideries, in: ICOM-CC METAL 2001 (Chile) Proceedings, Western Australian Museum, 2004, pp. 90–96. [5] A.G. Nord, K. Tronner, A note on the analysis of gilded metal embroidery threads, Stud. Conserv. 45 (2000) 274–279. [6] K. Tronner, A.G. Nord, J. Sjostedt, H. Hydman, Extremely thin gold layers on gilded silver threads, Stud. Conserv. 47 (2002) 109–116. [7] A.M. Hacke, C.M. Carra, A. Brown, Characterization of metal threads in Renaissance tapestries, in: ICOM-CC Conference Metal 2004 (Canberra) Proceedings, National Museum of Australia, 2004, pp. 415–426. [8] O. Enguita, A. Climent-Font, G. Garcia, I. Montero, M.E. Fedi, M. Chiari, F. Lucarelli, Characterization of metal threads using differential PIXE analysis, Nucl. Instr. Meth. B 189 (2002) 328–333. [9] C. Paradais, Comori ale spiritualita˘ßt ii românesßti la Putna, Mitropoliei Moldovei sßi Sucevei, Iasßi, 1988, pp. 187–341. [10] C. Nicolescu, Costumul de Curte in Tarile Romane (Sec. XIV–XVIII), Intreprinderea Poligrafica Arta Grafica, Bucuresti, 1970. [11] I. Rajta, I. Borbély-Kiss, Gy. Morik, L. Bartha, E. Koltay, Á.Z. Kiss, The new ATOMKI scanning proton microprobe, Nucl. Instr. Meth. B 109–110 (1996) 148–153. [12] G.W. Grime, M. Dawson, Recent developments in data acquisition and processing on the Oxford scanning proton microprobe, Nucl. Instr. Meth. B 104 (1995) 107–113. [13] Zs. Kertész, Z. Szikszai, I. Uzonyi, A. Simon, Á.Z. Kiss, Development of a bio-PIXE setup at the Debrecen scanning proton microprobe, Nucl. Instr. Meth. B 231 (2005) 106–111. [14] J.L. Campbell, N.I. Boyd, N. Grassi, P. Bonnick, J.A. Maxwell, The Guelph PIXE software package IV, Nucl. Instr. Meth. B 268 (2010) 3356–3363. [15] A. Simon, R. Huszánk, M. Novák, Z. Pintye, Investigation of hydrogen depletion of organic materials upon ion beam irradiation by simultaneous micro-RBS and micro-ERDA techniques, Nucl. Instr. Meth. B 268 (2010) 2197–2201. [16] R. Huszánk, A. Simon, E. Szilágyi, K. Keresztessy, I. Kovács, Micro-ERDA, microRBS and micro-PIXE techniques in the investigation of fish otoliths, Nucl. Instr. Meth. B 267 (2009) 2132–2135. [17] M. Mayer, SIMNRA user’s guide, Tech. Rep. IPP 9/113, Max–Planck–Institut für Plasmaphysik, Garching, 1997. [18] A. Karatzani, T. Rehren, Clothes of gold: metal threads in Byzantine-Greek orthodox ecclesiastical textiles, in: J.F. Moreau, R. Auger, J. Chabot, A. Herzog (Eds.), Proceedings ISA 2006, Quebec, 2009, pp. 9–19. [19] A.D. Barker, A brief survey of the manufacture and use of precious metal wires and threads for embroidery and weaving, Northern Soc. Costume Text. (1980) 1–14. [20] C.S. Smith, M. Teach Gnudi, The Pirotechnia of Vannoccio Biringuccio. The Classic Sixteenth Century Treatise on Metals and Metallurgy, Dover Publications Inc, New York, 1990, pp. 377–383. [21] R.F. Tylecote, A History of Metallurgy, Maney Publishing for the Institute of Materials, London, 2002, pp. 89–118. [22] P. Craddock, Scientific Investigation of Copies, Fakes and Forgeries, Butterworth-Heinemann Elsevier Ltd., 2009, pp. 386–388. [23] Ch. Flament, P. Marchetti, Analysis of ancient silver coins, Nucl. Instr. Meth. B 226 (2004) 179–184. [24] P. Masjedi, F. Khandemi, M. Hajivaliei, S.M.M. Kouhpar, J. Neystani, Elemental analysis of silver coins of Seljuk’s’ of Rome by PIXE, Mediterranean Archaeol. Archaeom. 13 (2013) 181–187. [25] L. Yongcheng, Y. Tongyin, Y. Huiying, Y. Fujia, PIXE analysis of silk, J. Appl. Polym. Sci. 66 (1997) 405–408. [26] N. Kallithrakas-Kontos, A. Katsanos, A. Aravantinos, M. Oeconomid, I. Touratsog, Trace elements analysis of Alexander the Great’s silver tetradrachms minted in Macedonia, Nucl. Instr. Meth. B 171 (2000) 342–349. [27] Z. Smit, A. Semrov, Early medieval coinage in the territory of Slovenia, Nucl. Instr. Meth. B 252 (2006) 290–298.
Please cite this article in press as: Z.I. Balta et al., Ion beam analysis of golden threads from Romanian medieval textiles, Nucl. Instr. Meth. B (2015), http:// dx.doi.org/10.1016/j.nimb.2015.01.027