analysis of pigments used in scroll paintings of a ... · 1. introduction x-ray fluorescent...

ANALYSIS OF PIGMENTS USED IN SCROLL PAINTINGS OFA NATIONAL TREASURE "TALE OF GENJI" USING

A PORTABLE X-RAY FLUORESCENCE SPECTROMETER

Keiichi SUGIHARA1, Koichi TAMURA1, Masao SATOH1,Yasuhiro HAYAKAWA2, Yoshimitsu HIRAO2, Sadatoshi MIURA2,

Hideki YOTSUTSUJI3 and Yoshitaka TOKUGAWA3

1Seiko Instruments Inc., Chiba, Japan2Tokyo National Research Institute of Cultural Properties, Tokyo, Japan

3Tokugawa Art Museum, Aichi, Japan

AbstractX-ray fluorescence spectrometry (XRF) is frequency applied for the

measurement of the chemical composition of cultural properties because of its rapidityand non-destructiveness. However, it is hard to analyze a sample at the site byconventional XRF. In order to solve this problem, a portable XRF has been developed.

The pigments used in the Scroll Paintings of the Tale of Genji in the possessionof the Tokugawa Art Museum have been analyzed by portable XRF. This instrument wasbrought into the museum to avoid various risks that may be caused by moving thepaintings. The Scroll Paintings of the Tale of Genji are considered as most important inunderstanding the materials and techniques of painting in the 12th century. The pigmentsused in the fifteen illustrated sheets in the museum were once estimated by anexamination using X-ray radiography and ultra-violet fluorescence spectroscopy in 1950-1954. However, the elements and the compounds of the pigments have not beenidentified yet. In this study, the elements of the pigments were measured by using an X-raybeam with Φ2mm diameter. As a results of measurement of about 20-30 points per eachillustrated sheet, composition of pigments corresponding to different colors wereidentified. For example, 4 kinds of white pigment were detected; major component are(1) Pb (2) Ca (3)Hg (4) not detectable. It was also made clear that the white pigmentsconsisting mainly of Pb used for drawing a human face contains a trace of Hg.Furthermore, the color used for drawing the human face was slightly different betweenillustrated sheets. This study might give some understanding the material and techniquesof painting in the 12th century.

In conclusion, this portable XRF analyzer is a powerful tool for non-destructiveinvestigation of cultural properties.

Copyright(c)JCPDS-International Centre for Diffraction Data 2001,Advances in X-ray Analysis,Vol.44 1Copyright(c)JCPDS-International Centre for Diffraction Data 2001,Advances in X-ray Analysis,Vol.44 431Copyright(c)JCPDS-International Centre for Diffraction Data 2001,Advances in X-ray Analysis,Vol.44 431Copyright(c)JCPDS-International Centre for Diffraction Data 2001,Advances in X-ray Analysis,Vol.44 432ISSN 1097-0002

http://www.icdd.com/

This document was presented at the Denver X-ray Conference (DXC) on Applications of X-ray Analysis. Sponsored by the International Centre for Diffraction Data (ICDD). This document is provided by ICDD in cooperation with the authors and presenters of the DXC for the express purpose of educating the scientific community. All copyrights for the document are retained by ICDD. Usage is restricted for the purposes of education and scientific research. DXC Website – www.dxcicdd.com

ICDD Website - www.icdd.com

ISSN 1097-0002

http://www.dxcicdd.com/

http://www.icdd.com/

1. IntroductionX-ray fluorescent spectrometry is widely applied for the qualitative and quantitative

analysis in various fields because of its rapidity and non-destructiveness1),2). This methodhas been used in the field of cultural artifacts for a long time and still today is the oneanalysis method used most frequently in performing composition analysis of material.

In recent years, many new functions, including X-ray incidence capability at an orderof µm, elements distribution mapping function, and the ability to measure samples thatexceed 1 meter in size, have developed in the field of fluorescent X-ray spectroscopy 3).However analytical samples had to be brought to and placed in a sample chamberbecause the conventional instruments are large and immovable. In the field of culturalartifacts there are many instances in which the sample cannot be transported, and even ifthe sample can be transported, it can be damaged during transporting or by changes intemperature and humidity. Besides, there have been vigorous requests for a tool that canimmediately analyze unearthed artifacts at the excavation site.

In order to accomplish these demands, a portable fluorescent X-ray analyzer hasrecently been developed that can do the elemental analysis at the site4), 5).

A newly developed portable fluorescent X-ray analyzer was applied for theidentification of pigments used in a Japanese national treasure scroll paintings "Tale ofGenji". The paintings occupy an important position in art history as representations ofHeian Period (9-12th century) art. The paintings were illustrated on papers about thetypical scenes of the romantic novel of "Tale of Genji", which written toward thebeginning of the 11th century by a court-lady known as Murasaki-shikibu.

Up until recently, there has been much discussion in terms of art history, but only oneexamination of scroll paintings using the scientific method. From 1949 to 1953, thepredecessor of Tokyo National Research Institute of Cultural Properties performed asystematic examination using optical methods to appraise old art objects6). Detection ofglue, paint-over, and peeling-off of paint was performed by observing enlargedphotographs and photographing with infrared/ultraviolet lighting. X-ray radiography wasalso applied for the estimation of material and thickness of the pigments.

This examination used the leading edge inspection and analysis technology availableat that time. However, adequate conclusions could not be made for the pigments used, orthe reason of color change or slightly different tints.

Our analysis was able to directly detect elements composition within the pigment."Tale of Genji" scroll paintings is now separated and mounted on a folding mat with thesize of 60cm x 35cm maximum respectively for conservating. Although conventionaltools can analyze the painting, a portable X-ray fluorescent spectrometer was taken to theArt Museum to avoid any risk by transporting the paintings outside the museum.


2. Experimental2.1. Equipment2.1.1. Analyzer

Seiko Instruments Inc, SEA200 energy dispersive X-ray spectrometer, the PortableXRF Element Monitor was used. A rhodium X-ray tube (5kv,15kV or 50kV 3-stepswitching) and maximum current of 1000 µA was equipped. A schematic diagram isgiven in Fig.1. and the scene where this instruments is used is given in Fig.2.

X-ray Tube

Detector

Linearamplifire

MCA

Lap-top PC

CCD

High-voltage

Sample

Measurement Head

Power Unit

Preamplifier

The X-rays irradiated from the top. The size of the X-ray beam was adjusted in therange of Φ2-10mm by a collimator. The analysis spot can be observed by the CCD image,resulting that the primary X-rays precisely irradiated the analysis spot. The comparativetable is given in Table 1 between the portable EDXRF and the conventional EDXRF.

Table 1 The Comparative Table

Portable EDXRF Conventional EDXRF

SizeHead: 185(W)×210(H)×320(D)mm

Power Unit : 400(W)×200(H)×300(D)mmLap-Top PC

Head(include Power Unit)1000mm×1000m× 1000mm

Desk-Top PC

WeightHead: 5kg

Power Unit : 15kg >100kg

Power Supply AC100V/Car Battery AC100V

X-ray TubeVoltage: 50kV(max)Current; 1mA(max)

Voltage: 50kV(max)Current: 1mA(max)

DetectorSi-PIN

FWHM: 250eV(@Mn-Kα)Si(Li)

FWHM: 150eV(@Mn-Kα)X-ray Beam Size Φ2-10mm Φ0.1-20mm

Sample View CCD image CCD image

Fig. 1 Schematic diagram Fig. 2 Measurement Head


2.1.2. Sensitivity Energy dependency of sensitivity was investigated previous to the analysis of the scroll

paintings of “Tale of Genji”. 6 kinds of reference materials, which were deposited thin film on Mylar, were measured. The spectrum for each reference material is given in Fig.2. Al, SiO and CuS were analyzed under He gas purge condition. Each peak of Al, Si and S is clearly observed.

S

/ /’ .^K

2

Rh-scatter + A

Ti n

5o.grgkm2 as cus Energy(keV)

L- 45.8pgkti as so 51.7pgkd as CaF, -

42.lpgkd as Al 38.5p g/cd as Ti

49.5 g/cd as L--l

Mn

Fig. 3 Energy dependency of sensitivity

2.1.3. The measurement setup The measurement setup is shown in Fig. 4.

Tale of Genj i

Wooden Plate

4- Measurement Head

- Aluminum Stand

Fig. 4 The Measurement Setup

Copyright(c)JCPDS-International Centre for Diffraction Data 2001,Advances in X-ray Analysis,Vol.44 434Copyright(c)JCPDS-International Centre for Diffraction Data 2001,Advances in X-ray Analysis,Vol.44 435ISSN 1097-0002

The scroll paintings were placed on a wooden board and the wooden board was thenplaced below a 10 mm aluminum stand combined with the analyzer stand. The portablefluorescent X-ray analyzer was set at the center on top of the stand with the X-ray beamdirection facing down. The scroll painting was arranged 10 mm from the X-ray unit.The X-ray beam was 2 mm in diameter. With the main part of the instrument secure ontop of the stand, the analysis position can be monitored by the CCD camera imagedisplayed on the monitor. The analysis position was then set by manually moving thewooden board on which the scroll paintings are placed.

2.2. Measurement conditionsWhen measuring light elements it is best to measure at a high sensitivity while

purging He gas. In this case we measured in air without a He gas purge to avoid He gasblowing on the samples during the measurement. These results performed on manypigment sample fragments using a 2 mm X-ray beam, tube voltage at 50 kV, tube currentat 50 uA, and a measurement time of 100 seconds, verified that measurements can beperformed on many elements at a statistical deviation of 3% to 5%. The deciding factorto analyze the scroll paintings at these measurement conditions was based on these results.

2.3. Sections MeasuredThe measurement conditions listed above were used to analyze the pigment on all 15

scroll paintings and two text scrolls of the "Tale of Genji." There are considerable colordifferences from area to area, and the number of measurement points differ depending onthe scene. However, we tried to be comprehensive in measuring areas that showcharacteristic colors in all scenes. We measured many points in areas of severediscoloration or where evaluating original color is difficult.

Each scene has roughly 20 to 30 measured points; however, scenes that had manytints had as many as 80 measured points. Figure 5 shows the measured points on part ofa scene named "Kashiwagi (Oak Tree)- 3 ". Our measurements gave emphasis to thehuman face. In this example we measured three points on the face of the man (nose,cheek and lips), and three points on the face of the child (forehead, cheek and lips). Wealso measured the parts of characteristic colors, for example a faded part and silverdesign of the man's kimono and areas showing like the green of the tatami mat.

3. Results and Discussion3.1. Measurement Results

Figure 6 shows an example of information displayed on the monitor during themeasurement. Figure 6 was obtained by measuring part of the nose of the man in figure 4.The fluorescent X-ray spectrum is displayed in real time and elements contained can bedirectly identified along with being able to verify measurement points by CCD cameraimage. Measurement conditions are displayed on the screen at the same time andimmediately after the measurement results can be analyzed quantitatively.

In this investigation we measured over 500 points in all paintings and texts. Thetypical colors and detection elements in this investigation along with the pigments listedin Table 2.


.

Fig. 5 The measured points on part of a scene named “Kashiwagi (Oak Tree)- 3 ”

Table 2 Typical pigments detected in “Tale of Genji”

Color Main Sub Estimated Pigments

I White

Pb Hg White Lead (+Cinnabar) Ca Shell White

2 Silver Pb,Ag Silver 3 Dark Red Hg,Pb Cu,Fe Cinnabar 4 Yellow Red Pb Red Lead 5 Light Red %Hg White Lead + Cinnabar

6 Brown Pb (C&Fe) White Lead + dyes 7 Yellow Pb (Cu) White Lead+dyes 8 Green cu Fe,Pb Malachite 9 Blue cu Fe,Pb Azurite IO Purple Pb (Cu> White Lead I1 Black

The main component in white pigment is thought to be white lead (Pb). Our measurements support this position, that is, that Pb is usually the main component. However, trace amounts of Hg were detected in white color of human faces. It is conceivable that trace amounts of cinnabar(HgS) was mixed in the white lead and its color showed white included a tinge of red. We could also observe that the white color included large amounts of Ca but without Pb. Other elements different than these were also detected in the white part, and will be described in the next section.


The large peaks detected at various measured points with various color were Pb. Thisis probably due to the white lead applied to the base on a paper. In other case, atmeasured points of yellow areas the only element detected was lead. This seems tosupport the idea that "Fuji yellow (Gamboge, organic) " was used as the color for yellow.That only white lead is detected in the bottom layer may be due to the fact that mostorganic material could not be detected in this analysis. Refer to Table 2 concerning othercolors and elements detected herein.

Fig. 6 Information displayed on the monitor

3.2. Results of measuring white pigmentsThe results of measuring 15 paintings from "Tale of Genji" scroll paintings show

that at least four types of materials were used to represent the color white. The typicalfluorescent X-ray spectrum of each material is shown in Figures 8 to 10.

3.2.1. Pb (lead) as a major component in white pigment (Figure 6)We thought that white lead would be the basic white pigment. Spectrum nearly the

same as the one in Figure 4 was obtained by measuring many points on white areas. Inmany cases, trace amounts of mercury were detected in the facial area, which color isthought to contain a small amount of cinnabar.

Figure 7 shows comparative results about the amount of mercury used for drawinghuman faces or skins. Three representative scenes, which are “Kashiwagi (Oak Tree)- 3”as scene A, “Takekawa (Bomboo River)-2” as scene B and “Hashihime (BridgePrincess)” as scene C, with a lot of coloration were selected. The vertical axis in the


figure is the intensity ratio (Hg Lp / Pb Lp).. The ratio of Hg-LP/Pb-LP in scene A and B, which have some deviation, show nearly the same values. It is clear that scene C has a much larger value than the values of the Hg-LP/Pb-LP ratios obtained from scene A and B. If mercury (Hg) is a reflection of the red of cinnabar, then we can conclude that the red in the skin of the empress in scene C is much stronger than for humans drawn in scene A and B.

7.0

6.0

5.0

4.0

ocl 3.0

2 2.0

G 1.0 cl 2 0.5

.g * 0.4 e! .G 2 0.3 a2 2 0.2

0.1

q

q

nno- A A A A

Fig. 7 Intensity Ratio Hg-LP / Pb-LP

3.2.2. Ca (calcium) as a major component in white pigment (Figure 8)

component without Pb. Ca as the main component suggests that the white color is from shell white (created by grinding shells). We were unable to determine if this pigment was used as drawing at original or was an added later restoration.

There are only a few measurement points on the face of the court lady in figure 8. Ca is the main

17.72 cps A I ws.%o7.spc

3.2.3. Hg (mercury) detected in large amounts in white pigment (Figure 9)

pL-----~~ 3.12 keV 14.47 kaV

There are several measurement points on the face of the court lady and young nobleman. In the example

Fig. 8 Spectrum of the scene “Sawarabi”

shown in Figure 9, the amount of Hg is greater than the


amount of Pb. There are several possible reasons forthis including whether Hg existed in white pigment as amajor ingredient, or a complex paint-over was done.

3.2.4. No main component element detected in thewhite pigment (Figure 10)

Other areas were discovered in which no majorelement was detected in the white. As is clear fromfigure 10, the spectrum is almost at background leveland fluorescent X-ray peaks for the most part cannot bedetected. An earlier researcher concluded that this waspart of white ground paper where paint had not beenapplied. However, our measurements made clear thatsomething had been applied on a paper and was not justground paper. Detecting light elements cannot be doneunder these measurement conditions. If, for example,pure white clay containing as main components Al andSi were used, measurement results like those in figure10 would be obtained. It might be expected that theorganic dye with white color, which cannot be detectedby X-ray fluorescent spectrometry, was used.

4. Conclusionsi A portable X-ray fluorescent spectrometer that

can be carried almost anywhere was recentlydeveloped. This instrument can be taken to the sample location and used toperform rapid elemental analysis.

ii Detection characteristics of the portable X-ray fluorescent spectrometer wereextensively examined and its ability to measure cultural artifacts confirmed.

iii Paints used in scroll paintings of Japanese national treasure “the Tale of Genji” atthe Tokugawa Art Museum were analyzed using the portable X-ray fluorescentspectrometer. The Tokugawa Art Museum has in its possession 15 scroll paintings.On the average, 20 to 30 measurements (maximum of 80) were taken at differentpoints on each scene resulting in our acquiring a large amount of new information,such as the existence of four kinds of white pigment, and color differences attributedto scenes.

The authors wish to thank Ms. Miho Yoshikawa of Tokugawa Art Museum, Mr.Yonekura, Mr. Shimao and Mr. Tsuda of Tokyo National Research Institute of CulturalProperties for their cooperation on this work.

Fig.10 Spectrum of the scene“Yomogiu”

Fig. 9 Spectrum of the scene“Hashihime”


Reference[1] J.L.Ferrero, C.Roldan, M.Ardid, E.Navarro, Nuclear Instruments and Methods inPhysics Research, A422,868 (1999)[2] M.Mantler, M.Schreiner, F.Weber, R.Ebner, F.Mairinger:Adv. X-Ray Anal.35,987(1992)[3] Y.Hayakawa and Y.Hirao, Conservation Science, 37,137(1998)[4] Y.Hayakawa, Y.Hirao, S.Miura, K.Tamarura, K.Sugihara, M.Satoh, H.Yotsutsuji andY.Tokugawa, Proceeding of the 16th Annual Meeting of the Japanese Society forScientific Studies on Cultural Property, 54(1999)[5] Y.Hayakawa, Y.Hirao, S.Miura, K.Tamura, K.Sugihara, M.Satoh, H.Yotsutsuji andY.Tokugawa, Proceeding of the 48th Annual Conference of the Japanese Society forAnalytical Chemistry ,65(1999)[6] M.Akiyama, K.Yamazaki, S.Nakayama, K.Suzuki and Y.Nakamura,The Journal ofArt Study,174(1954)


analysis of pigments used in scroll paintings of a ... · 1. introduction x-ray fluorescent...

Documents