Appl. Phys. A 90, 97–100 (2008)
Applied Physics A
DOI: 10.1007/s00339-007-4231-2
Materials Science & Processing
The gilded bronze panels of the Porta del Paradiso by Lorenzo Ghiberti: non-destructive analyses using X-ray fluorescence
m. ferretti1 s. siano2,u
1
Istituto per le Tecnologie Applicate ai Beni Culturali-CNR, Via Salaria km 29.300 – CP 10, 00016 Monterotondo St., Roma, Italy 2 Istituto di Fisica Applicata “Nello Carrara”-CNR, Via Madonna del Piano 10, 50019 Sesto Fiorentino, Italy
Received: 8 November 2006/Accepted: 17 July 2007 Published online: 23 August 2007 • © Springer-Verlag 2007 ABSTRACT We report the results of an analytical investigation on the bronze alloys of the Porta del Paradiso by Lorenzo Ghiberti using a portable X-ray fluorescence instrument. The study was carried out on five panels of the bronze artwork. It provided information on the artist’s selection of the sculptural alloys and on some manufacturing details of interest for understanding the development of the art foundry during the early Renaissance. PACS 81.05.Bx;
1
81.70.Jb; 82.80.Ej
Introduction
The Porta del Paradiso, crafted by Lorenzo Ghiberti between 1425 and 1452, is a fundamental masterpiece of the early Renaissance. It is composed by two single casting brass frameworks, which host 10 gilded bronze reliefs (80 × 80 cm2 panels) representing stories of the Old Testament and two gilded bronze friezes with 48 figurative elements (24 tondi and 24 niches). The door has undergone systematic investigations since 1966, the year of the flood. The studies were mainly aimed at assessing the deterioration phenomena and defining the conservation treatments and the strategy for its future preservation [1–3]. The increasing interest for the handicraft techniques of the Renaissance art foundry also stimulated some preliminary metallurgical analyses, which demonstrated the gilding was carried out through the mercury amalgam technique (fire gilding) and provided the first few data on the alloy composition of two panels [4, 5]. Recently, a dedicated technological investigation project has been promoted by the Opificio delle Pietre Dure of Florence, the institution responsible for the restoration work. Within this project we have carried out an overall recognition of the bronze alloys used to craft five of the ten sculptural reliefs of the door using a portable XRF spectrometer. The measurements allowed the identifying and grouping of the different alloy compositions of the main bodies of the panels and repairs. Furthermore, trace elements provided information about possible different material stocks, which represent the first step towards the temu Fax: +39-0555225305, E-mail:
[email protected]
poral collocation of the reliefs and the understanding of the casting and mounting sequences of Ghiberti’s masterpiece. 2
Materials and methods
The measurements were carried out using a homemade portable XRF spectrometer equipped with a watercooled X-ray tube, working at 60 kV, 1.6 mA, a 0.4 mm Cu filter, and a 5 mm2 Si-Drift detector. The acceleration potential of the tube is higher with respect to the typical one of the commercial instruments, in order to achieve an effective excitation of the K lines of elements of interest, such as Sn, Sb, Ag (22 – 30 keV), whose L lines (2.5 – 4.5 keV) present some resolution problems. The measured detection limits of the spectrometer were 50 ppm for Sn and 300 ppm for Pb with 95% confidence level in copper matrix and 120 s measuring time [6]. As mentioned above, the investigation was carried out on five panels of the Porta del Paradiso exhibited at the Santa Maria del Fiore Museum of Florence: Storie di Abramo (IV), Storie di Esaù e Giacobbe (V), Storie di Giuseppe ebreo (VI), Storie di Mosè (VII), Storie di David (IX). About 230 measurements were performed on the back side of these panels for achieving semi-quantitative information on the main casting and repair zones, through the comparisons of the corresponding clusters of data points in the space of the count-rates. The repairs were performed by means of re-castings and mechanically applied plugs. Among the latter the central plug that is present in all of the panels is particularly interesting. It closes a rather large intentional rectangular hole (about 1 × 2 cm2 ) whose function has not yet been understood. As an example, Fig. 1 displays the backside of the panel IV including a sketch of the front relief and white spots in correspondence of the repairs (left), along with details of the central plug of four panels (right). The XRF instrument was used in a strictly non-destructive way, which means without any preliminary preparation of the surface. In other cases, the removal of the outer corrosion layer significantly reduces the dispersion of the data and hence improves the quantitative evaluations, but on valuable artworks, such as the present one, this procedure is too invasive. Nevertheless, even in presence of strong line intensity variations, the portable XRF approach often allows the distinguishing of different alloy compositions since
98
Applied Physics A – Materials Science & Processing FIGURE 1 Examples of the analysed surfaces: (left) backside of the panel IV (Storie di Abramo) with sketches of the front reliefs (lines) and repair sites (spots); (right) central repair plug of four panels
a number of measurements sufficiently high produces countrate clusters with different centroids. Often, bivariate scatter plots of representative elements are sufficient to spot the differences. The minimum observable difference depends on the surface conditions, number of measurements, and spectrometer sensitivity, as trace elements may have larger variations and hence provide better discrimination than major elements. Following the XRF campaign a microsample (around 100 mg each) was taken from the main body of each panel and analysed using atomic absorption spectroscopy (AAS). The achieved compositions were compared with the XRF data in order to support the conclusion derived from these latter with quantitative measurements and, conversely, to verify whether the AAS sampling was representative of the panel alloys. 3
satisfactory agreement with the results of the spectrochemical analysis. Regarding the impurities, the most effective discrimination was provided by Sb: The Ag-Sb scatter plot of Fig. 3 shows the panel VII is significantly lower in Sb with respect to the others. Two different supplies of cop-
Results and analysis
The detected elements were Cu, Sn, Zn, Pb, Fe, Ag, Sb and As. The main alloy elements Sn and Zn, along with Pb are considered meaningful of the founder’s choices, whereas the impurities Fe, Ag, As, and Sb are related to the different supplies of raw metals used over the years. Figure 2 displays the count-rate of Sn-Zn and Sn-Pb scatter plots of the main body of the panels where at least three different alloys can be clearly distinguished. The composition of the panels IV, V, and VI is different with respect to the ones of the panels VII and IX. Furthermore, the Sn-Zn scatter plot shows a Sn content increase when stepping from the panel VII to IX, IV and VI; whereas the panel V overlaps the last two. The Zn content provides a discrimination between the panels VII and IV (relatively high) with respect to IX, V, and VI (relatively low). A quantitative estimation of the principal element contents was achieved by means of a set of calibration curves. These were derived through fluorescence measurements on five quaternary alloy samples of known composition, which provided the following slopes. Zn, 53 (counts/s)/wt.%. Sn, 120 (counts/s)/wt.% Pb, 21 (counts/s)/wt.%. The results are compared with the ones provided by AAS in Table 1. Apart from the discrepancies in lead content, which are likely due to segregation phenomena both the Sn and Zn composition features pointed out by XRF are consistent and in
FIGURE 2 Principal element scatter plots showing composition differences among the five sculptural reliefs investigated here
FERRETTI et al.
Panel IV (Abramo), AAS IV (Abramo), XRF V (Giacobbe), AAS V (Giacobbe), XRF VII (Mosè), AAS VII (Mosè), XRF IX (David), AAS IX (David), XRF TABLE 1
The Porta del Paradiso by Lorenzo Ghiberti: non-destructive analyses using XRF
99
Cu
Sn
Pb
Zn
Fe
Ni
Ag
Sb
91.93 – 92.91 – 93.51 – 95.19 –
1.82 2 2.53 3 0.13 0.2 0.70 1
1.25 1 1.43 1 1.14 2 0.84 3
3.46 3 1.54 2 3.78 4 1.10 2
0.54 – 0.46 – 0.13 – 0.38 –
0.19 – 0.20 – 0.14 – 0.17 –
0.03 – 0.04 – 0.06 – 0.05 –
0.41 – 0.52 – 0.13 – 0.50 –
Composition (wt.%) of the main alloy of the panels, as measured by atomic absorption spectroscopy (AAS) and average XRF line intensity
analysis
per seem to have been used to cast the present set of panels. The situation becomes considerably more complex if one also considers cast-on repairs and plugs. From this standpoint the panels can be divided in two groups: a) “good” castings (VII and IX), which have only 1–2 visible repairs and b) “problematic” castings (IV, V and VI), which are more heavily repaired with alloys usually (not always) lower in Sn and higher in Zn. In particular, panel VI, with its 17 well-recognisable repairs, represents the worst situation. Alloys of very variable compositions and different trace patterns were used in this latter case, which suggests the repairs were carried out along a relatively long period of time by employing material from different stocks. The scatter plots achieved for the other panels were relatively more readable. In Fig. 4 the ones of the panel IV, V and VII are displayed. The Sn-Zn and Sb-Ag scatter plots show that some repairs of the panel IV and V are rather Zn-rich. The alloy used should actually be more properly called brass. Conversely, the rectangular central plugs of all the panels investigated here have a relatively high Sn content, which differentiates their composition with respect to the main alloys and other repairs. This difference is extended to the trace elements, thus suggesting this central plug was applied in a different time with respect to the main casting, as well as to the other repairs.
FIGURE 3
panels
Trace elements (Sb-Ag) scatter plot of the main alloy of the
4
Discussion and conclusions
The non-destructive XRF investigation on Ghiberti’s sculptural reliefs demonstrated they were crafted using low-alloyed quaternary bronzes. As confirmed by AAS analyses, the total amount of white metals (Sn, Zn, Pb) resulted to be between about 3–8 wt. %. The relative changes are likely due to the difficulty in controlling the composition because of the low weight ratios of the single elements and/or to the optimisation changes during the casting work. If the latter hypothesis is assumed for the major changes, the panels VII and IX could have been cast before the others, but further insights are needed to support this conclusion. The alloy compositions of the present sculptural reliefs are significantly different from the ones of the first and second door of the Baptistery of Florence, crafted by Andrea Pisano (1330–1336) and Lorenzo Ghiberti (1403–1415), respectively. We believe that the need to change the material support is closely related to the stylistic differences between the Porta del Paradiso and the previous doors. Actually, in line with the new architectural taste that was growing during the second decade of the XV century, the 28 four lobe Gothic panels (about 50 × 50 cm2 ) of the former were replaced with ten larger square panels (about 80 × 80 cm2 ) and two friezes [7]. The complex figurative reliefs that the author decided to insert in this augmented space required a big amount of cold working (chiselling), which would have been impossible on the hard alloys used till then. The new alloy was also very suitable for applying the amalgam gilding. Regarding impurities, we related to the stocks of raw metals, it is worth nothing that the Sb content clearly distinguishes the panel VII (Storie di Mosè) from the others (Fig. 3). This feature, together with the ones of the main alloys just described, allows one to formulate the work hypothesis that panel VII could be the oldest one. Objective observations and 3D-scanning of the casting features support this conclusion. Finally, the compositional data achieved here also provide information about the repair procedure and sheds light on the role of the strange central plug, which is present in all the panels (Fig. 1 right). As pointed out above, the repairs have very variable composition and trace patterns. This suggests they were carried out using material remains from previous casting works without observing any specific rules. The frequency of the high Zn content is compatible with the composition of the alloys used before the crafting of the Porta del Paradiso. At the same time the relatively high Sn content and the different trace patterns of the central plugs with respect to the other
100
Applied Physics A – Materials Science & Processing FIGURE 4 Principal (Sn-Zn) and trace (Sb-Ag) element scatter plots of the main alloys (filled circles) and repairs (empty marks). C indicates the central plug
repairs lead us to conclude that the formers were applied during the final phase of the work. We think the central holes were drilled by chiselling after the completion of the panels in order to simplify their mounting into the framework. It probably required several insertion and extraction tests of the heavy panels (around 80 – 90 kg) to refine the fittings that were likely performed by suspending the panels with a chain through the hole. In conclusion, the non-destructive investigation presented here provided new data and presented new interpretations concerning the execution of the Porta del Paradiso. In the next future, the approach will be extended to other sculptural elements of the masterpiece and, more in general, to other bronze artworks of the early Renaissance.
ACKNOWLEDGEMENTS The authors wish to thank Dr. Anna Maria Giusti of the Opificio delle Pietre Dure of Florence for entrusting them with the present study.
REFERENCES 1 G. Alessandrini, G. Dassu, P. Pedeferri, G. Re, Stud. Conserv. 24, 108 (1979) 2 P. Fiorentino, M. Marabelli, M. Matteini, A. Moles, Stud. Conserv. 27, 145 (1982) 3 S. Siano, R. Salimbeni, Stud. Conserv. 46, 269 (2001) 4 P. Parrini, In: Metodo e Scienza operatività e ricerca nel restauro, ed. by U. Baldini (Sansoni Editore, Florence, 1983) 5 F. Garbassi, E. Mello, Stud. Conserv. 29, 172 (1984) 6 M. Ferretti, Nucl. Instrum. Methods B 226, 453 (2004) 7 R. Krautheimer, Lorenzo Ghiberti (Princeton University Press, Princeton, NJ, 1956)
