Applied Physics A

, Volume 116, Issue 3, pp 1053–1058 | Cite as

Comparison of gold leaf thickness in Namban folding screens using X-ray fluorescence

  • Sofia Pessanha
  • Teresa I. Madeira
  • Marta Manso
  • Mauro Guerra
  • Agnès Le Gac
  • Maria Luisa Carvalho
Article

Abstract

In this work, the thickness of the gold leaf applied in six Japanese folding screens is compared using a nondestructive approach. Four screens belonging to the Momoyama period (~1573–1603) and two screens belonging to the early Edo period (~1603–1868) were analyzed in situ using energy dispersive X-ray fluorescence, and the thickness of the applied gold leaf was evaluated using a methodology based on the attenuation of the different characteristic lines of gold in the gold leaf layer. Considering that the leaf may well not be made of pure gold, we established that, for the purpose of comparing the intensity ratios of the Au lines, layers made with gold leaf of high grade can be considered identical. The gold leaf applied in one of the screens from the Edo period was found to be thinner than the gold leaf applied in the other ones. This is consistent with the development of the beating technology to obtain ever more thin gold leafs.

1 Introduction

Gold was one of the first metals manipulated by men because of its remarkable malleability, which makes it easily beaten to form the desirable shape [1]. Its noble metal stability and resistance to tarnish gives durability to the objects on which it is applied to. Its suitability to adorn small or larger areas renders it an excellent material and the finest among metals for use in the decoration of artworks. Two different terms have been employed to define metal sheets of varying thickness: foil and leaf. The former is generally used to describe a sufficiently thick sheet that can support its own weight while the latter, which mainly applies to gold, refers to a very thin sheet that has almost no weight and that can only be handled by specially designed tools.

With modern technology, the gold leaf can be obtained as thin as 0.1 μm [2]; however, the thickness of the leaf used in ancient artworks could be as thick as the 5–10 μm applied in an Egyptian 50 BC ceramic masque, or as thin as 0.2–0.5 μm used in the gilding of medieval Islamic glazed ceramics [3]. According to Koyano [4], the Japanese gold leaf is said to be the thinnest in the world.

In Japanese traditional painting, decorating with gold leaf is named Kin-haku, and the finest examples of this craft are the Namban folding screens, or byobu, in paper support, belonging to the late Momoyama (~1573–1603) and early Edo (~1603–1868) periods. Traditionally, a grid was marked slightly across the paper as a guide for laying the hand-beaten gold leaves (~10 cm × 10 cm) in horizontal rows, starting from the top left-hand corner [5].

Scanning electron microscopy (SEM) imaging [6, 7], and proton induced X-ray emission [8, 9] are often used to determine the thickness of coatings in art and Cultural Heritage; however, sampling is required in the first technique, while the second only can be applied in objects that can be moved to the laboratory. X-ray fluorescence techniques are more suitable for this task because they are nondestructive and portable equipments are available. Fiorini et al. [10] and Trojek et al. [11] studied the substrate under different angles in order to determine the thickness of the coating layer. By tilting the object around the X-axis, the path length increases the intensities of the Au characteristic lines, while the characteristic lines of the elements of the under layer are more attenuated. Cesareo et al. [12, 13, 14] evaluated the gilding of several artworks by determining the intensity ratio of Lα/Lβ lines of the substrate and Pessanha et al. [15] used a similar methodology for determining the thickness of the gold leaf applied in a sixteenth century illuminated manuscript, making sure that the substrate layer could be considered an infinitely thick sample. However, these studies always assumed a layer of pure gold and the effect of the presence of other metals in the alloy was not considered. In fact, by mixing gold with silver and copper, the goldsmith could produce a large number of alloys with different mechanical properties (strength, hardness and ductility) as well as different colors, meaning that each alloy is tailored to a particular application. In the studied screens, Cu appeared in very low, almost negligible concentrations (Fig. 1 and [16, 17]) and according to Guerra [18] an amount of Cu up to 1 % may be considered as being part of the natural gold alloy. This way, in this work, we investigated the influence of silver in the attenuation of the characteristic lines of gold in the gold leaf layer. Using this methodology, we were able to compare the thickness of the applied gold leaf layers in six Namban folding screens.
Fig. 1

EDXRF spectrum obtained for the gilding of the screen belonging to the private collector

2 Methodology

As demonstrated by Cesareo [14], the different lines of an element are attenuated differently as the thickness of the layer increases so the intensity ratio of Lα/Lβ (and Mα/Lα) varies until we can consider the sample infinitely thick and it remains constant.

The intensity ratio of Mα/Lα lines of Au is dependent of the thickness of the layer. For a detector placed at an angle of 45° relatively to the sample this ratio reads:
$$ \left( {\frac{{M_{\alpha } }}{{L_{\alpha } }}} \right) = \left( {\frac{{M_{\alpha } }}{{L_{\alpha } }}} \right)_{0} \left( {\frac{{\varepsilon (M_{\alpha } )}}{{\varepsilon \left( {L_{\alpha } } \right)}}} \right)\left[ {\frac{{\left( {\mu_{0} + \mu_{2} } \right)}}{{\left( {\mu_{0} + \mu_{1} } \right)}}} \right]\left[ {\frac{{1 - e^{{ - \left( {\mu_{0} + \mu_{1} } \right) d}} }}{{1 - e^{{ - \left( {\mu_{0} + \mu_{2} } \right)d}} }}} \right] $$
(1)
where \( \left( {\frac{{M_{\alpha } }}{{L_{\alpha } }}} \right)_{0} \) the intensity ratio for an infinitely thin sample; ε(E) is the detector efficiency for a given energy; µ0 is the linear attenuation coefficient of Au at incident energy E0; µ1 is the linear attenuation coefficient of Au at energy of Mα emission line (2.12 keV); µ2 is the linear attenuation coefficient of Au at energy of Lα emission line (9.71 keV); d is path length of the characteristic radiation and \( d = \frac{x}{cos\theta } \) where x is the thickness of the layer.

Similar equations can be written for the Au (Lα/Lβ) intensity ratio.

The X-ray tube used in the analysis of the Namban screens has a silver anode, so the evaluation of the presence of Ag in low concentrations is challenging. This way, we need to investigate the influence of the presence of silver in the gold alloy in the Au (Mα/Lα) and Au (Lα/Lβ) intensity ratios considering different alloy compositions: a pure Au layer, an alloy of 99 % Au and 1 % Ag; an alloy of 97 % Au and 3 % Ag and an alloy of 95 % Au and 5 % Ag. Figures 2 and 3 represent the plot of these dependences with the thickness of the layer, and we consider these curves to be coincident (with a maximum deviation from a mean curve of 0.6 %). This way, and taking into account that the uncertainties related with the determination of the peak areas are in the 1 % order of magnitude, we conclude that layers can be considered identical concerning the attenuation of the Au-Mα Au-Lα and Au-Lβ lines.
Fig. 2

Plot of the intensity ratio Mα/Lα for a layer of pure Au, Au99 %–Ag1 %, Au97 %–Ag3 % and Au95 %Ag5 %

Fig. 3

Plot of the intensity ratio Lα/Lβ for a layer of pure Au, Au99 %–Ag1 %, Au97 %–Ag3 % and Au95 %Ag5 %

Equation 1 implies the use of monochromatic incident radiation and curves in Figs. 2 and 3 were plotted considering a 15 keV incident radiation. However, the measurements on the screens were taken using the bremsstrahlung from the X-ray tube, so we cannot use these equations to effectively determine the thickness in our case studies. Nevertheless, using this methodology, we can assess the comparative thicknesses of the gold leafs applied in artworks, as will be demonstrated for the Namban folding screens.

3 Artworks description

Material characterization of five of the studied screens was already presented [16, 17], and the use of a high-grade gold leaf, most likely applied using an iron-based bole layer, was confirmed. The folding screens we compared are as follows: One pair of screens signed by kano Naizen (1570–1616) kept at Museu Nacional de Arte Antiga (MNAA) (Fig. 4a); one pair of screens also kept at MNAA attributed to kano Domi (1568–1600) [17, 19] (Fig. 4b); one screen belonging to the collection of Museu Oriente (MO) attributed to the early Edo Period (Fig. 5a); and finally another screen attributed to the early Edo period belonging to a private collector (Private) [20] (Fig. 5b).
Fig. 4

Photographs of the two pairs of screens belonging to the Momoyama period a MNAA-Naizen; b MNAA-Domi

Fig. 5

Photographs of the two screens belonging to the Edo period a MO; b private

4 Experimental setup

The EDXRF equipment used consists on a X-ray generator ECLIPSE II from Amptek (30 kV and 100 µA) with a Ag anode and an Amptek XR-100CR Si-PIN thermoelectrically cooled detector with a 7 mm2 detection area and 300 µm thick, and a 25 µm Be window. The energy resolution is 190 eV at 5.9 keV, and the acquisition system is Amptek PMCA [21]. For collimating the beam, an acrylic support with a 2 mm pinhole in Ta was used and a spot size of 0.5 cm on the sample is obtained. The components are placed in an aluminum structure in 90° geometry and mounted on a tripod with 1.5 m vertical amplitude. Spectrum deconvolution and evaluation was performed using PyMCA software package [22].

X-ray fluorescence analysis was executed in situ at the museums, and an average of 10 measurements was taken in the gilded area of each screen.

5 Results

The Au (Mα/Lα) and the Au (Lα/Lβ) intensity ratios are represented in the box-and-whisker plots of Figs. 6 and 7. These box plot charts compare the distribution of the values obtained for the gilded areas and divide them in quartiles. The small squares represent the mean value of this distribution, and the whiskers are the maximum and minimum.
Fig. 6

Box-and-whiskers plot for Au (Lα/Lβ) intensity ratio of the gilded areas in the studied screens

Fig. 7

Box-and-whiskers plot for Au (Mα/Lα) intensity ratio of the gilded areas in the studied screens

As could be expected, no significant differences were found in the Au (Mα/Lα) and Au (Lα/Lβ) intensity ratios for screens belonging to the same pair (MNAA-Naizen 1 and 2 and MNAA-Domi 1 and 2) as they were most likely executed with the same materials (Figs. 6, 7). Also when comparing these two pairs of screens, signed by different artisans, no significant differences were found in the intensity ratios, hence in the thickness of the leaf. Regarding the screens attributed to the early Edo Period, the obtained distributions for Au (Lα/Lβ) intensity ratios are very similar to the ones obtained for the aforementioned screens, which could mean that the thickness is too small for this different attenuation factors to be taken into account. However, the comparison of the Au (Mα/Lα) intensity ratios rendered more interesting results: while the intensity ratios obtained for the Private-boat screen are in the same range as the MNAA-Naizen and MNAA-Domi, the mean value and range distribution for the intensity ratios of the MO screen are significantly higher, meaning that the Au-Mα line is less attenuated and therefore the layer is thinner.

6 Conclusions

In this work, we demonstrated that the thickness of the gold leaf used for gilding different artworks can be assessed and compared, in a nondestructive way, by means of X-ray fluorescence spectroscopy. We also established that, for the purpose of comparing the intensity ratios of the Au lines, layers made with gold leaf of high grade (pure Au to 95 % Au 5 % Ag) can be considered identical. The gold leaf applied to six Namban screens was compared, and the leaf applied to the screen kept at Museu Oriente was found to be the thinnest. Actually, a thickness of 100 nm order of magnitude was obtained with a quasi-transversal view with SEM image of a sample collected from this screen [16]. Considering that the thickness of the gold leaf tends to decrease significantly with the advent of gold beating technological development [6], these results are indicative that this specimen from Museu Oriente is the most recent. This simple comparison allowed establishing a timeline between the manufacture of two pieces attributed to the same period—from 1603 to 1868, proving to be an important asset in the dating of artworks.

Notes

Acknowledgments

Authors would like to thank directors and staff at Museu Nacional de Arte Antiga and Museu Oriente for allowing the study, as well as Salvarte atelier for accommodating us and the screen belonging to the private collector.

References

  1. 1.
    C.J. Raub, Mater. Australas. 1–19, 7 (1986)Google Scholar
  2. 2.
    J. Winter, East Asian Paintings: Materials, Structures and Deterioration Mechanisms (Archetype publications, Washington, 2008)Google Scholar
  3. 3.
    E. Darque-Ceretti, M. Aucouturier, Dorure: décor et sublimation de la matière (Mines ParisTech, Paris, 2012)Google Scholar
  4. 4.
    M. Koyano, Gilding and Gilding Conservation in Japan. Gilded Wood: Conservation and History (Sound view press, Madison, 1991)Google Scholar
  5. 5.
    S. Grantham, Pap. Conserv. 9, 83 (1985)Google Scholar
  6. 6.
    A. Le Gac, A.I. Seruya, M. Lefftz, A. Alarcão, Revue d’Archéométrie—ArcheoSciences 33, 423 (2010)Google Scholar
  7. 7.
    N. Wang, L. He, E. Egel, S. Simon, B. Rong, Microchem. J. 114, 125 (2014)CrossRefGoogle Scholar
  8. 8.
    V. Corregidor, L.C. Alves, N.P. Barradas, M.A. Reis, M.T. Marques, J.A. Ribeiro, Nucl Instrum. Methods Phys. Res. B 269, 3049 (2011)ADSCrossRefGoogle Scholar
  9. 9.
    A. Migliori, N. Grassi, P.A. Mando, Nucl. Instrum. Methods Phys. Res. B 266, 2339 (2008)ADSCrossRefGoogle Scholar
  10. 10.
    C. Fiorini, A. Gianoncelli, A. Longoni, F. Zaraga, X-Ray Spectrom. 31, 92 (2002)CrossRefGoogle Scholar
  11. 11.
    T. Trojek, M. Hlozek, Appl. Radiat. Isot. 70, 1420 (2012)CrossRefGoogle Scholar
  12. 12.
    R. Cesareo, Nucl. Instrum. Methods Phys. Res. B 211, 133 (2003)ADSCrossRefGoogle Scholar
  13. 13.
    R. Cesareo, A. Brunetti, S. Ridolfi, X-Ray Spectrom. 37, 309 (2008)CrossRefGoogle Scholar
  14. 14.
    R. Cesareo, M.A. Rizzutto, A. Brunetti, D.V. Rao, Nucl. Instrum. Methods Phys. Res. B 267, 2890 (2009)ADSCrossRefGoogle Scholar
  15. 15.
    S. Pessanha, M. Guerra, S. Longelin, A. Le Gac, M. Manso, M.L. Carvalho, X-Ray Spectrom. 43, 79 (2014)CrossRefGoogle Scholar
  16. 16.
    S. Pessanha, A. LeGac, T.I. Madeira, A. Guilherme, M. Manso, M.L. Carvalho, X-Ray Spectrom. 42, 128 (2013)CrossRefGoogle Scholar
  17. 17.
    S. Pessanha, M.L. Carvalho, M.I. Cabaço, S. Valadas, J.-L. Bruneel, M. Besnard, M.I. Ribeiro, J. Raman Spectrosc. 41, 1220 (2010)ADSGoogle Scholar
  18. 18.
    M.F. Guerra, T. Calligaro, Meas. Sci. Technol. 14, 1527 (2003)CrossRefGoogle Scholar
  19. 19.
    Y. Lippit, Encompassing the Globe: Portugal and the World in the 16th and 17th Centuries (MNAA, Lisbon, 2009)Google Scholar
  20. 20.
    Art Namban - Catalogue of the Europália89 Exposition (Musées Royaux d’Art et d’Histoire, Brussels, 1989)Google Scholar
  21. 21.
    M. Guerra, M. Manso, S. Longelin, S. Pessanha, M.L. Carvalho, J. Instrum. 7, C10004 (2012)CrossRefGoogle Scholar
  22. 22.
    V.A. Solé, E. Papillon, M. Cotte, Ph Walter, J. Susini, Spectrochim. Acta B 62, 63 (2007)ADSCrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2014

Authors and Affiliations

  • Sofia Pessanha
    • 1
  • Teresa I. Madeira
    • 1
  • Marta Manso
    • 1
  • Mauro Guerra
    • 1
    • 2
  • Agnès Le Gac
    • 1
    • 3
  • Maria Luisa Carvalho
    • 1
    • 2
  1. 1.Centro de Física Atómica da Universidade de LisboaLisbonPortugal
  2. 2.Departamento de Física, Faculdade de Ciências e TecnologiaUniversidade Nova de LisboaCaparicaPortugal
  3. 3.Departamento de Conservação e Restauro, Faculdade de Ciências e TecnologiaUniversidade Nova de LisboaCaparicaPortugal

Personalised recommendations