1Objects belonging to the Moche civilization (Vetter Parodi, 2006; Alva, 2006) were analyzed in the past using various destructive techniques (Andrade et al., 2005; Burger, 1992; Lechtman, 1998; Ruvalcada Sil, 2005; Saettone et al., 2003; Scott, 2000); fragments of gold and silver artefacts (and also of silvered gold) from Loma Negra, Peru, were accurately studied and analyzed by Schorsch (1998) by employing EDXRF attached to a scanning electron microscope, and wave length dispersive X-ray Spectrometry. The gold objects showed the following composition: Au ~ 80%, Ag = 10-20%, Cu = 5-15%. The silver objects showed a high Ag content, of about 97-99%. Copper represents the rest of the composition, to 100%. An interesting and unusual feature is the case of silvered gold, with a silver sheet measured to have a thickness of about 5 µm.
2Fragments from 17 Moche objects from the “Museo Tumbas Reales de Sipán” have been analyzed by Hörz and Kallfass (2000), using various techniques. These authors were able to identify: a) gilded copper objects: they are characterized by a thin gold film (2-6 µm). The coatings consist of a gold-copper alloy containing some silver; b) copper-silver alloys: fragments from several human head shaped beads have been analyzed, showing a mean composition of 79% Cu, 20% Ag, and 1% Au; c) copper-gold-silver alloys (tumbaga): fragments from a headdress, a chin ornament, an ornamental disc, and ornamental beads were analyzed; the average compositions were calculated to be varying between 20-60% Cu, 35-65% Au and 6-15% Ag. The alloy composition is strongly dependent on the distance from the surface.
Figure 1: Experimental setup, showing.
Figure 1 : Configuration expérimentale.
on the left, the Eclipse II X-ray tube and the Si-PIN detector (both collimated with a brass cylinder), measuring a Moche golden mask in the Museum “Tumbas Reales de Sipán”.
sur la gauche, le tube à rayons-X Eclipse II et le détecteur Si-PIN (les deux collimatés avec un cylindre en laiton) pendant l’analyse du masque d’or Moche au musée « Tumbas Reales de Sipán ».
3For the systematic analysis of Sipán and Sicán metal objects, the use of non-destructive techniques may be proposed, and, among them, energy-dispersive X-ray fluorescence analysis (EDXRF) appears to be the most suited, because it is non-destructive, multi elemental, reliable, rapid, and may be carried out in situ (Cesareo et al., 2004). This method is able to quantify the composition of a gold or silver alloy when standard samples are used. EDXRF analysis provides reliable results regarding the concentration of high carat gold alloys with a reduced quantity of copper, and also in the case of high concentration copper or silver alloys. In the case of tumbaga, of copper-rich gold alloys, of gilded copper, and of copper-rich silver alloys, the results provided by EDXRF are incomplete or erroneous.
4EDXRF is a surface analysis, and it is useful in this case because the respective thickness of the alloy is of the order of a few µm to a maximum of tens of µm; because of this peculiarity, EDXRF is able to distinguish a gold alloy from gilded copper or tumbaga, and a silver alloy from gilded silver, by using the internal ratio of Cu and Ag-lines. The method is also able to determine the gold thickness value in the case of gilded copper or tumbaga gold.
5The portable equipment employed for the analysis of Sipán and Sicán alloys is composed of an X-ray tube and a Si-PIN detector, both manufactured by AMPTEK. The X-ray tube has a weight of 300 g, length of 17 cm – including the collimator – and diameter of 3.7 cm. It has an Ag-anode, and works at 30 kV and 100 µA maximum voltage and current. X-rays from the tube irradiate areas of about 5 mm diameter, when the object is at a distance of ~2 cm. The X-ray beam intensity is too high when alloys are analyzed, because the Si-PIN detector is able to process only a few thousands of photons/sec without losing energy resolution. The X-ray beam is therefore collimated with a brass cylinder 2 cm long and with an internal hole of 2 mm in diameter. Furthermore, in order to excite silver in a more efficient manner, the X-ray tube output is also filtered with about 0.1 mm Ti.
6The X-ray detector is a thermoelectrically cooled Si-PIN, with 300 µm thickness, 7 mm2 area of the Si-crystal, and a thin Be-window. This detector has an energy resolution of about 180 eV at 5.9 keV, and an efficiency of 90%, 25% and 8% at 10, 20, and 30 keV, respectively. It has a weight of 150 g and a length of 14 cm. It is also collimated. The measuring time was of approximately 100 sec, according to sample composition and geometry.
7Standard gold and silver alloys, with certified Au, Ag and Cu concentrations, were employed for calibration and for the quantitative determination of the alloy composition.
8In order to measure the gilding thickness of gilded gold or silver, the Cu(Ka/Kb) or Ag(Ka/Kb) ratios, and the (Au-La/Cu-Ka) or (Au-La/Ag-Ka)-ratios were employed. Au leaves and Ag foils were employed for calibration (the foils were each 0.125 µm and 0.28 µm thick for Au and Ag, respectively). Thick sheets of pure Cu and Ag were also employed. Thickness values were tested by transmission measurements with mono-energetic X-rays.
9Artefacts of very different sizes, compositions and surfaces were analyzed. It is therefore very difficult to reproduce a fixed geometry. For this reason, determination of the alloy components on the basis of the fundamental parameters method is difficult, and an alternative approach was preferred, using the intensity ratio of two components (for example Cu/Au, assuming that Au(%)+Cu(%)+Ag(%) = 100), which is not dependent on the geometry. Other elements present are determined by using fundamental parameters.
10The Ka/Kb, La/Lb and La/Lg-ratios are tabulated (Cesareo, 2000; Markowicz, 1992). For example, for copper, Ka/Kb = 7.4, and for gold, La/Lb = 1. These values are valid for an infinitely thin sample, corresponding to a thickness < 1µm, < 0.5µm and < 0.2µm for Cu, Ag and Au, respectively. For larger thicknesses, self attenuation effects must be considered (Cesareo and Brunetti, 2008).
11When a sheet of metal (a), for example copper, of infinite thickness, is covered by a sheet of another metal (b), for example gold, then the ratio (Ka/Kb)a or (La/Lb)a is altered because of the attenuation of the covering sheet:
(Ka/Kb) = (Ka/Kb)s.a.exp-[µ1-µ2] d (1)
(Ka/Kb)s.a. is the ratio of metal (a) according to possible self-attenuation; µ1 and µ2 (cm-1) are the attenuation coefficients of element (b) at the energy of Ka and Kb rays of element (a); d is the thickness (in cm) of the sheet of element (b) (Cesareo and Brunetti, 2008; Cesareo et al., 2009). Similar equations may be calculated for L/Lb and La/Lg-ratios.
Figure 2: Gilded copper: attenuation ofCu(Ka/Kb) ratio by Au (or Au alloys) vs Au thickness.
Figure 2 : Cuivre doré: atténuation du rapport Cu(Ka/Kb) par l’Au (ou les alliages d’Au) versus l’épaisseur d’Au.
12Another way to experimentally determine, from the X-ray spectrum, the thickness of the second element (b) assuming that the first element (a) has an infinite thickness is the use of the X-ray ratio of the two elements, for example, the ratio (Au-La/Cu-Ka). This ratio, for two generic elements at fixed incident energy and geometrical arrangement, is provided by the following formula:
13-ma0 or mb0 (in cm2/g) are the total mass attenuation coefficients of elements (a) and (b) at incident energy; mab (in cm2/g) indicates the mass attenuation coefficient of element (a) at energy of the involved line of element (b); rb (in g/cm3) is the density of element b.
14In Equation (3), the proportionality should be quantitatively determined according to experimental data. Figure 3 represents theoretical values (from Eq. 3) and experimental measurements values for the (Au-La/Cu-Ka)-ratio versus Au thickness, assuming Cu with infinite thickness.
Figure 3: Theoretical (squares) and experimental values for (Au-La/Cu-Ka) ratio vs Au thickness, assuming Cu with infinite thickness.
Figure 3 : Valeurs théoriques (carrés) et valeurs expérimentales du rapport (Au-La/Cu-K) versus l’épaisseur d’Au, en assumant une épaisseur infinie pur le Cu.
15About 50 objects from the “Tumbas Reales de Sipán” were analyzed, made of gold, gilded copper, tumbaga, silver and copper alloys, the majority of them originating from the tomb “Señor de Sipán”.
16Several objects are made of gold (an example is shown in Fig. 4) and are characterized by Au, Cu and Ag as main components. In some samples, traces of Fe, Zn and Br are visible. The mean results of the EDXRF analysis carried out on gold objects are the following:
Au = 69.5 +/- 7%; Ag = 21 +/- 4%; Cu = 9.5 +/- 5%.
17Only a few objects are surely of gilded copper. They were identified by the exclusive presence of Cu in some analyzed areas, and by a deteriorated surface. In many cases, it was possible to clearly determine the Au leaf thickness from Cu (Ka/Kb) and (NAu-L/NCu-K) ratios.
18Several sheets of gilded copper were analyzed. They are characterized by a Cu(Ka/Kb) ratio of 6.1 +/- 0.1, corresponding to a gilding thickness of 1.2 +/- 0.5 µm. From the (Au-La/Cu-Ka) ratio of 0.1 results a value of ~ 0.5 µm, however. A beautiful mask of gilded copper was analyzed in detail, showing the following composition: Au ~97.5%, Ag ~2.5%. The gilding thickness was measured to be ~ 0.5 µm. Cu could not be determined, because it was present below the gilding.
Figure 4: Peanut heads made of gold, belonging to a necklace.
Figure 4 : Perlesen forme de cacahuète exécutées en or et appartenant à un collier. D’après la Ref. (2).
The average composition is: Au = 58%, Ag = 26%, Cu = 16%
La composition moyenne est: Au = 58%, Ag = 26%, Cu = 16%
19The majority of the gold alloys were identified as tumbaga (an example is shown in Fig. 5), which behaves in a similar manner as gilded Cu for EDXRF analysis. The ‘gold-equivalent’ surface thickness can be determined from Cu-Ka/Kb and from (NAu-L/NCu-K) ratios (see Section 3). The mean Au-Cu-Ag concentration and Au thickness values are:
20Au = 60 +/- 10%; Cu = 30 +/- 9%; Ag = 10 ± 4%.
The (NAu-L/NCu-K) ratio = 1.44 +/-0.7, corresponding to an Au thickness of 3.10 +/-0.7 µm;
The (Cu-Ka/Kb) ratio = 5.35 +/-0.5, corresponding to an Au thickness of 2.5 +/- 1.4.
21The EDXRF analysis of silver objects shows that the silver content is relatively high, and that it systematically contains gold. The following mean concentration values were determined:
Ag = 92 +/- 4%; Cu = 5 +/- 2.5%; Au = 3 +/- 1.5%.
22Parts of a few objects are made of copper, composed of about 99% Cu, and traces of Fe and Ni.
23Many of the gold objects include turquoise, which is a hydrated phosphate of aluminium and copper. The turquoise shows an occasional presence of Zn, Fe and Cr as impurities, resulting in deviations from the blue colour. Results pertaining to Sipán turquoise show a systematic presence of Fe and Zn, at an average concentration of 10% and 8.5%, respectively.
Figure 5: Leg protector made of tumbaga.
Figure 5 : Protecteur de jambe en tumbaga.
The ratio Cu(Ka/Kb) is 4.8, corresponding to an Au thickness of 3.1 µm
Le rapport Cu(Ka/Kb) est 4,8, correspondant à une épaisseur d’Au de 3,1 µm
24About 20 objects from the Museum of Sicán (Shimada and Griffin, 1994) were analyzed, the majority of which were of gilded copper; others were made of gold, tumbaga, silver and copper alloys.
The following mean values could be determined:
Au = 62 ± 4% ; Ag = 32 ± 3% ; Cu = 6 ± 2%.
The following mean values could be determined for the gilding: Au = 67%; Ag = 33%.
25A few objects are of uncertain composition; they could either be made of gilded copper or of tumbaga. In fact, the gilded copper objects are identified because of the altered Cu(Ka/Kb) ratio, and because of the presence of highly corroded areas on almost pure copper. In other cases, the ratio Cu(Ka/Kb) is altered, but no corroded areas were detected.
One artefact, a beautiful mask, is certainly made of tumbaga, whose concentration and characteristic thickness parameters are the following:
26Au = 34 +/- 6%; Cu = 57 +/- 5%; Ag = 7 +/- 1.5%.
The Cu(Ka/Kb) ratio = 4.9 +/- 0.3, corresponding to an Au thickness of 3.8 +/- 0.9 µm.
The (Au-L/Cu-K) ratio = 1.05 +/- 0.02, corresponding to an Au thickness of 4.5 +/- 0.8 µm.
27Only one object is made of silver, i.e. a brooch in the form of a monkey, with the following composition: Ag = 94%, Cu = 3.7%, Au = 0.5%, Pb = 0.8%, Br = 1%.
The following mean values could be determined:
Cu = 98 ± 1%; Fe = 1.2 ± 0.5%; As = 0.8 ± 0.5%.
28EDXRF analyses of precious objects from Sipán and Sicán demonstrate their complexity and variety; they are made of the following alloys: gold, gilded copper, tumbaga, silver, silvered copper, silvered gold, copper, and so on, and their nature is not always easy to identify.
29From the metallurgical point of view, a comparison between the Moche (50-700 AD) and Sicán (750-1375) civilizations shows that:
Moche precious objects seem to be more sophisticated in terms of beauty and technology;
The average compositions of gold and silver are similar;
The Moche civilization largely used tumbaga-gold (representing more than 50% of the ‘gold’ artefacts), while Sicán used more gilded copper;
The Moche employed almost pure copper; the Sicán civilization had a knowledge of arsenical copper.
This work was partially carried out within the framework of the project IAEA-CRP (G4.20.02/1371) “Unification of nuclear spectrometry: integrated techniques as a new tool for material research”. J. Fabian expresses his gratitude to the International Centre for Theoretical Physics Abdus Salam for a 5 months grant at the University of Sassari.