Introduction

The major subjects of research related to the pigment analysis, reproduction, and restoration of colored cultural properties reported in South Korea over the past 30 years are pigments mixed in murals, Buddhist paintings, portraits, decorative paintings, documentary paintings, dancheong, and lacquerware. Analytical research on the pigments of temple murals makes up most of these studies, particularly focusing on artworks from the seventeenth to nineteenth centuries during the late Joseon Dynasty (1392 to 1910) in Korea [1].

The paintings are found not only on the walls, ceilings, and pillars of the buildings, but there are also called of Taenghwa, Gyunghwa, and folding screen paintings. Despite the different time periods, research on Buddhist paintings has also been conducted; for instance, research on pigments [2], manufacturing techniques [3], pigment discoloration characteristics based on the environment [4, 5], and reproduction and restoration of traditional natural pigments [6].

Among these studies, identification of pigments for the Buddhist paintings are of fundamental importance for studying their manufacturing technology and residential environments, as well as for their conservation and restoration. Large scale banner paintings (Taenghwa) became popular in Korea during the seventeenth and eighteen centuries. The size and iconography of this type of painting suggest that they were originally important installations in monasteries. These paintings most likely hung behind a statue of Buddha Shakyamuni and would have been displayed in a hall’ dedicated to the Lotus Sutra or in some other major building within a temple complex [7, 8].

The White-Robed Water-Moon Avalokiteshvara is an artwork that represents the Buddhist paintings of Magoksa temple located on the rear wall on the west side of Daegwangbojeon (main hall), which is the dharma hall of Magoksa temple. A frame is made up of wood structures between high pillars, and several layers of mulberry fibers are applied to create a rear wall, which appears as if it were painted on a typical canvas. However, it is difficult to discover the exact date of creation because there is no manufacture date. In Magoksa temple, although this painting has been said to be painted by the Buddhist monk māyā Yakhyo (1846 to 1928), this claim is questionable because the expressions and descriptions of Amitabha about the crown of the white-robed Gwaneum Bosal (Buddhist Goddess of Mercy) are slightly different.

The lines are overall smooth, and the techniques of landscape paintings in the late Joseon Dynasty (seventeenth to nineteenth century) of Korea can be seen in the expression of the rocks. Nevertheless, considering the formal description technique observed in the face of Buddha and the use of dark red, the creation period is estimated to be in the late nineteenth century. Although the contours and colored surface of this White-Robed Water-Moon Avalokiteshvara remain undamaged, there is blistering, exfoliation, and discoloration of some pigment layers; contamination and stains; cracks and separations of the pigment layers; and cracks and separations of the wall. Thus, the painting urgently requires stable conservation treatment.

This painting was safely dismantled after performing a basic investigation and emergency treatment at the site. It was then transferred in a vibration-free vehicle for precise conservation treatment, and it has recently been placed at its original location. Prior to the conservation treatment of the painting, this study recorded the degree and condition of the overall damage in detail and conducted a scientific investigation on the colored layer to analyze the components and characteristics of the pigments used for coloring. Through such research, we have studied the characteristics by analyzing colored pigments used in ancient Korean art works [9, 10]. These results are important data that can reveal the pigments and coloring techniques used in Buddhist paintings at that time, which also can be used for research on long-term preservation and repair.

Object and method

Object

Magoksa Temple was founded in 643 by the ancient Korean monk Jajang (590 to 658) with the support of Queen Seondeok during the Joseon Dynasty (fourteenth to nineteenth centuries). This temple gained widespread fame as one of the leading temples in the country under support from the royal family and the Provincial Office of Chungcheong Province. Moreover, Magoksa Temple was listed as a UNESCO World Heritage site in 2018 under the category of “Sansa, Buddhist Mountain Monasteries in Korea” in recognition of its historical and cultural value as a temple that has sustained the tradition and culture of the Buddhist community.

Magoksa Temple, surrounded by Mount Taehwa as a folding screen, is a space where monks’ and believers’ practices, beliefs, and lives blend together. It has served as a driving force for social integration and the preservation of traditions throughout Korean history. Daegwangbojeon Hall is Magoksa Temple’s dharma hall, rebuilt in 1788 and designated as Treasure No. 802. Inside Daegwangbojeon Hall, Vairocana Buddha is uniquely enshrined, facing towards the east from the building’s west wall. On the back wall of Vairocana Buddha, White-Robed Water-Moon Avalokiteshvara is also enshrined (Fig. 1).

Fig. 1
figure 1

Location map and status showing the temple in Republic of Korea. A, B Map showing the location of research area. C Aerial photograph of Magoksa temple. D, E Frontal view and Cross-sectional diagram of Daegwangbojeon (main hall) in Magoksa temple

Full size image

This study was conducted prior to the conservation treatment of the White-Robed Water-Moon Avalokiteshvara, a Buddhist painting on the rear wall of main hall in Magoksa temple. It aimed to secure basic data can be used for conservation treatment through the precise analysis of coloring pigments by each color. Thus the surface of the painting as well as the integrity and damage status of the pigment layer were reflected in a drawing, which was intended to be preserved as a record of the repair and treatment areas of the colored layer.

This White-Robed Water-Moon Avalokiteshvara painting is a colossal mural measuring 515 cm in length and 296 cm in width, with Gwaneum Bosal (Avalokiteshvara) in a white robe sitting on a lotus pedestal above a rippling wave and a strange rocky stone, with her left leg hanging down. She is at the center of the painting surface and facing the front in a half-lotus position, her right leg placed on her left (Fig. 1a). Gwaneum Bosal wears a white robe from her head to below knees and red bottoms having a green bottom with red hem and tied with a pink belt inside the white robe (Fig. 2a).

Fig. 2
figure 2

The present status of the White-Robed Water-Moon Avalokiteshvara painting in this study. A Foreground of the painting. B Face part of the painting. C Nirmāna-Buddha on crown. D Left hand of the painting. E Bamboo in the upper right corner. F Seonjae Dongja (priestling) in the center right. G Kundika and willow branches in the middle left

Full size image

In addition, Gwaneum Bosal wears a crown on her head with a standing Amitabha statue in the middle to indicate identity as Avalokiteshvara (Fig. 2b and c). Her right hand rests comfortably on right leg, and her right ankle is held with left hand. Her hands and feet are large and realistically depicted (Fig. 2d). Behind Gwaneum Bosal, four bamboo trees are rising from the rough rocks on the left side (Fig. 2e); Seonjae Dongja (priestling) stands thereunder, leaning toward Gwaneum Bosal (Fig. 2f); and a kundika, in which a willow branch is arranged, is placed on the opposite rocks (Fig. 2g).

Method

In the preservation status investigations, the damage types of the colored layer were subdivided to create a map, and empirical data were presented to increase the reliability. This study applied a method with recently proven reliability in the damage assessment in the field of cultural heritage conservation science, including stone artifacts [11, 12].

Damage map creation and non-destructive investigation for the painting as a research object were conducted with special care to prevent any impact on the painting by using an installed scaffold. First, the preservation status was precisely recorded by checking the damage status at the site, and the analysis target was further selected after selecting the colors, such as blue, green, yellow, red, black, and white, by visual observation to confirm the color status.

As the analysis point for each color, a site was selected, whose measurement area of approximately 5 mm in diameter was sufficient among the colors in which the difference in color was distinguished by visual observation, and where the pigment in the base layer with the target color was not disturbed (Fig. 3). Non-destructive chromaticity and composition measurements were performed for the selected analysis points, and a trace amount of pigment that had flaked off was collected by using a sample for precision analysis. The names of the analysis points are indicated in Table 1.

Fig. 3
figure 3

Photograph showing the detailed measurement points of P-XRF and chromaticity analysis in this study

Full size image
Table 1 Summary on colors and designations for analysis points in this study
Full size table

In the detail investigation, chromaticity measurement, observation through a portable stereoscopic microscope, and elemental analysis through a portable X-ray fluorescence analyzer were performed. All were conducted through non-destructive analysis, and the same area was analyzed for each measurement item. Furthermore, the field analysis was carefully conducted by minimizing the applied pressure by selecting only the part of the color pigment that was most closely adhered onto the base layer to avoid any loss from the split colored layer due to the contact-based analysis.

A total of 55 analysis points were used to identify the components of the color pigments in this painting for measurement under classification into blue, green, yellow, red, black, and white series. This study secured quantitative data based on the results of a color component analysis that had already been reported for this study, and the types of pigments used in this painting were identified in comparison to the data on previously studied traditional pigments [13,14,15,16,17,18,19,20].

A CM-2500d spectrophotometer (KONICA MINOLTA) was used to measure the chromaticity of the pigments, and three measurements were obtained at the same point with a measurement area of 3 mm through the light source D65. The chromaticity values of the analyzed pigments were expressed by the CIELab system defined by the International Commission on Illumination (CIE), and the color coordinates were represented by luminosity (L*), red to green value (a*), and blue to yellow value (b*). In addition, the color components can be more intuitively understood by indicating chromaticity (C*ab) and hue (hab) values on the polar coordinates. Chromaticity (Eq. 1) is expressed as a position between the center of the sphere (black) and the outer surface (white). Hue (Eq. 2) is expressed in angle, with red = 0°, yellow = 90°, green = 180°, and blue = 270° [21, 22].

$${\text{C}}^{*}_{{{\text{ab}}}} { = }\left( {{\text{a}}^{{*2}} {\text{ + b}}^{*} } \right)^{{1/2}}$$
(1)
$${\text{h}}_{{{\text{ab}}}} = {\text{ arctan}}\left( {{\text{b}}^{*} /{\text{a}}^{*} } \right)$$
(2)

The measurement range of L* is from 0 to 100, and 100 refers to nearly white, which is close to white light, and 0 refers to black. The measurement range of a* and b* for the colorimeter used in this study is − 100 to + 100. The a* indicates a redder color in a positive direction and a greener color in a negative direction; b* indicates a yellower color in a positive direction and a bluer color in a negative direction.

Meanwhile, molecular spectroscopic techniques are frequently applied in cultural heritage research. In particular, X-ray fluorescence (XRF) spectroscopy is extensively and successfully used for investigating objects of historical and cultural value, since it is non-destructive and, if necessary, measurements can be performed in situ, using portable equipment. The XRF provide valuable information regarding the material composition of artworks, contributing with substantial information about the time and geographical origin of an artifact or about production technologies [23].

Although XRF is an excellent technique for painting analysis, and it is widely used in literature, it does have some limitations. For example, because it is an elementary technique, pigments must to be identified indirectly by means of key elements associated with the colors observed in the painting. In this context, techniques that allow molecular analyses can bring information in order to complement the data obtained by XRF. In order to overcome these limitations of non-destructive analysis, a number of studies have been conducted to supplement it with analysis equipments [24,25,26,27,28,29]. However, all analysis results have their advantages and disadvantages, revealing some of the limited applications in this study where only non-destructive analysis was permitted without obtaining samples.

In this study, the component analysis for the color pigment at the site, an X-ray fluorescence spectrometer (P-XRF; Oxford Instruments, X-MET7500) was used for the part selected as the analysis location. The operating conditions of this spectrometer were an X-ray tube voltage of 15 to 40 kV (40 kV for heavy elements and 15 kV for light elements) and a current of 10 to 50μA. Measurement was performed in the ore mineral mode, the analysis area was approximately 5 mm in diameter, and the measurement time was 20 s each in conducting the elemental analysis on an atomic number of 15 or higher.

The pigments were identified based on the color associated with the presence of key-elements in the XRF spectra. In addition, the name of each pigment, even with the same components, may vary depending on the period, region, material, supply route, particle size, and areas of use. This study mainly recorded the names used based on previous studies regarding the color pigments of paintings. Furthermore, the estimation of black pigments and raw materials was performed through a comprehensive review of the results of scanning electron microscope-energy dispersive spectroscopy (SEM–EDS) analysis and X-ray fluorescence analysis performed in the field investigation by collecting a trace amount of color pigment samples that had flaked off during the stripping process.

Results and interpretation

Preservation status

The types of damage observed on the painting surface of White-Robed Water-Moon Avalokiteshvara are typically divided into microcracks (Fig. 4a), separation (Fig. 4b), fragmentation (Fig. 4c), contamination (Fig. 4d), and blistering and exfoliation of the pigment layer (Fig. 4e and f). Overall, although the contour lines and colored surfaces of this painting remain undamaged, damage appears on the painting surface along with blistering and exfoliation of some pigment layers, cracks and separation of the pigment layer, and cracks and separation of the wall (Fig. 4).

Fig. 4
figure 4

Photographs showing the representative damages of the studied painting. A Microcracks. B Separation. C Fragmentation. D Black contamination. E Blistering and exfoliation of the pigment layer. F Stereoscopic photographs of the exfoliation part on the pigment layers. G Deterioration of wooden frame after stripping the painting. H Residual pigments and cracks in the wooden frame

Full size image

The blistering and exfoliation of the pigment layer are mainly observed around the face and on the red bottoms of the Gwaneum Bosal, which is believed to be accompanied by contraction, expansion, and cracking due to moisture. Moreover, after completely stripping the painting, the preservation condition of the wood materials on the rear wall were checked. As a result, it was found that the deterioration of the front lining paper and the wooden frame had severely progressed due to decomposition by wood-decay fungi and insect damage (Fig. 4g and h).

The structural condition of the wooden frame constituting the painting wall was found to be relatively stable, and while there were cracks and separation along the edge of the wall, the preservation condition of the mural painting as a whole was fair. In addition, because the blistering, exfoliation of the colored layer, contamination, and fragmentation were found to be scattered throughout the painting surface, it was recommended that appropriate measures for long-term preservation be taken.

Chromaticity measurements

The pigments used to color this painting mainly include blue, green, yellow, red, black, and white, and violet and pink appear in smaller amounts (Fig. 5). This study measured the chromaticity of a total of 55 points for representative color pigments: six points for blue, eleven points for green, seven points for yellow, eight points for red, six points for black, eleven points for white, and ten points for other colors (Fig. 6 and Table 2).

Fig. 5
figure 5

Photographs showing the measuring points and magnified images for pigment analysis in this study. Numbers are the same as those of Fig. 2

Full size image
Fig. 6
figure 6

Diagrams showing the chromaticity analysis for coloring pigments in this study. A Luminosity. B Position of the analysis points projected in the a*-b* plane. The abbreviations are shown in Table 1 and the symbols are shown in Fig. 6

Full size image
Table 2 Descriptive values statistics (mean, standard deviation, maximum and minimum) for the CIELab colour parameters of the painting in this study
Full size table

The chromaticity values for all measurement points represent various luminosity values, which are distributed between red to yellow (hab: 0 to 90°) and yellow to green (hab: − 90 to 0°) as a result of projection on the CIE Lab quadrant. Among the values, yellow and red pigments have relatively higher chromaticity and hue variability than blue and green pigments (Table 2, Fig. 6).

Regarding the chromaticity range of the blue-1 pigment, L*, a*, and b* range from 65.5 to 65.62 (mean 65.56), − 1.77 to − 0.8 (mean − 1.285), and 10.25 to 12.87 (mean 11.56), respectively. Regarding the chromaticity of the blue-2 pigment, L* on means is 70.71, a* is − 5.07, and b* is 10.52, respectively. Blueness increases as b* increases in the negative direction, while the measured blue-2 pigments all have positive b* values, showing that the blueness of the blue-2 pigments is lower than that of the blue-1 pigments (Table 2 and Fig. 6).

According to the result of measuring the chromaticity for the green pigments, L*, a*, and b* range from 42.18 to 54.82 (mean 48.00), − 15.7 to − 3.87 (mean − 10.16), and  3.75 to 6.57 (mean 2.71), respectively. As a* increases in the negative direction, greenness increases, and the a* values of all analysis points are negative. BG-7 shows the highest green level with a L* of 53.88, a* of − 15.7, and b* of 11.75, and BG-10 shows the lowest green level with a L* of 44.14, a* of − 3.87, and b* of 4.81 (Table 2 and Fig. 6).

Regarding the chromaticity of the yellow-1 pigments, L*, a*, and b* range from 54.53 to 60.56 (mean 57.41), 9.36 to 11.75 (mean 10.67), and 22.62 to 28.75 (mean 26.39), respectively. In the case of the yellow-2 pigments, L*, a*, and b* range from 56.64 to 61.02 (mean 57.41), 0.52 to 5.75 (mean 3.38), and 26.61 to 37.08 (mean 30.54), respectively. Yellowness increases as b* increases in the positive direction. Y-3 shows the highest yellowness, and OY-3 shows the lowest yellowness (Table 2 and Fig. 6).

The chromaticity of red pigments is widely distributed in the + a* and + b* coordinate planes. L*, a*, and b* range from 32.96 to 6.22 (mean 39.92), 12.71 to 30.39 (mean 20.67), and 14.45 to 19.68 (mean 17.28), respectively. Redness increases as a* increases in the positive direction. R-2 shows the highest redness with a L* of 38.38, a* of 30.39, and b* of 17.43, and R-8 shows the lowest redness with a L* of 46.22, a* of 12.71, and b* of 16.47 (Table 2 and Fig. 6).

The chromaticity range of black pigments is concentrated near the origin. L*, a*, and b* range from 25.42 to 58.65 (mean 34.51), 0.25 to 8.95 (mean 1.80), and 1.27 to 25.9 (mean 6.59), respectively. The analysis point closest to the origin is BL-1, where L* is 25.43, a* is 0.25, and b* is 1.27. The chromaticity range of white pigments is distributed vertically along the + b* axis. L*, a*, and b* range from 53.87 to 80.72 (mean 73.99), − 2.55 to 2.89 (mean 1.02), and 8.36 to 21.78 (mean 16.09), respectively. The analysis point with the lightest color corresponds to W-8, and the analysis point with the darkest color corresponds to W-9 (Table 2 and Fig. 6).

Meanwhile, regarding the chromaticity range of violet pigments, L*, a*, and b* range from 51.42 to 53.28 (mean 52.35), − 0.74 to − 0.44 (mean − 0.57), and − 1.74 to 0.74 (mean − 1.24), respectively. Because a* and b* show negative values at both measurement points, the redness and yellowness are not high, which does not deviate significantly from the negative direction. Furthermore, regarding the chromaticity range of pink pigments, P-1 shows higher redness and yellowness than P-4 (Table 2 and Fig. 6).

Chemical compositions by P-XRF

P-XRF was utilized to analyze the chemical composition of pigments in White-Robed Water-Moon Avalokiteshvara painting. The analysis locations were the same as the points for the chromaticity measurement and observation using the stereoscopic microscope. The analysis location for each colored pigment is shown in Fig. 6.

In analyzing a pigment layer through P-XRF, it is difficult to accurately measure only the pigment compositions because X-rays penetrate through the thin pigment layer and further analyze the underlying background. To overcome these shortcomings, studies have aimed to unravel only the components of the pigment from which the background component was removed. This study also detected the compositions of the pigment layer according to the method provided by Lee et al. [2] and Kim and Lee [10].

Two types of blue pigments were used for the basic coloring of this painting. According to the results of chromaticity measurement and observation using the stereoscopic microscope, an analysis was conducted by classifying the results into Blue-1 and Blue-2. If grayish blue ((Co,Ni)As3) is used as a blue pigment, cobalt and nickel compositions are detected. However, these were not detected in this painting. On average, 9037 ppm (6363 to 11,710 ppm) and 33,763 ppm (12,844 to 59,991 ppm) of Cu were detected in Blue-1 and Blue-2 analysis points, respectively.

Thus, the results suggest that the color depth was raised by mixing a white pigment with azurite (Cu3(CO3)2(OH)2), which is one of the traditional blue pigments. The detection of the high concentration of Ca, Pb, and S compositions at all analysis points further suggests that the color shade could have been adjusted by oyster shell white (CaCO3) or white lead (2PbCO3·Pb(OH)2) (Table 3 and Fig. 7).

Table 3 Representative concentratios (ppm) by P-XRF of pigments on the studied painting
Full size table
Fig. 7
figure 7

Diagrams showing the measurement results (ppm) of coloring pigments by P-XRF in this study. Numbers are the same as those of Fig. 2

Full size image

According to the results of the P-XRF analysis regarding the green pigment layer measured at a total of 11 points, the major compositions detected were elements such as S, Ca, Ti, Fe, Cu, As, and Pb. However, the average content of Cu and As showed a significant difference:149,915 ppm (13,321 to 311,848 ppm) and 110,568 ppm (13,716 to 296,732 ppm). Typically, when Cu and As are simultaneously detected in the composition of a green pigment, it is considered emerald green (Cu(C2H3O2)2·3Cu(AsO2)2). However, in the relative content ratios of emerald green, the content of As is much higher than that of Cu. Although a high As content was detected in the green pigment used in this painting, it was difficult to determine it as emerald green because the ratio of As was mostly lower than that of Cu (Table 3 and Fig. 7).

The difference in the Cu and As content at each analysis point may have been affected by the thickness of the overlapped color. Furthermore, the pigments used for the face of Buddha and Keyura accessories remained relatively undamaged, whereas those used on the feet and the lower parts fell off, resulting in a relatively small amount. Thus, these results suggest that clinoatacamite (Cu2(OH)3Cl) made up of copper-based raw materials was mainly used for the green pigment layer, and it was likely painted over by modern pigments given the result that the Ca and Cl content is relatively high at some points (Table 3 and Fig. 7).

The points classified as yellow were analyzed by dividing them into Yellow-1 and Yellow-2 according to the results of chromaticity measurement and observation using the stereoscopic microscope. Among the main constituent elements detected, Al and Si appear to have been simultaneously detected as a dead color paint component using a white pigment, such as ūrnā (Table 3). The content of Pb was the highest at all points with an average of 288,094 ppm (246,801 to 323,689 ppm) and 252,850 ppm (187,988 to 289,484 ppm), in Yellow-1 and Yellow-2, respectively, followed by S with an average of 90,707 ppm (59,470 to 132,885 ppm) and 140,291 ppm (133,392 to 151,7544) (Table 3 and Fig. 7). Thus, the color yellow can be assumed to lower the chromaticity by mixing white pigments with massicot.

However, the detection of As (6014 ppm) only in Y-1 suggests that orpiment (As2S3) and massicot (PbO) were mixed with a white pigment for coloring. Regarding the white pigment used in the color combination, the Ti content was relatively high at 72,827 ppm (58,638 to 81,718 ppm) and 72,400 ppm (51,038 to 83,346 ppm) in Yellow-1 and Yellow-2, respectively, which can be assumed to be titanium dioxide (TiO2). In addition, compared to other color pigments, Ba is highly likely to be a modern pigment due to the detection of high levels of Ba with an average of 55,919 ppm (46,032 to 74,7499 ppm) and 67,742 ppm (56,966 to 82,836 ppm) in Yellow-1 and Yellow-2, respectively (Table 3 and Fig. 7).

According to the results of the component analysis on red pigments, the main composition were Pb, S, and Hg, and Si and Al were detected at all measured points. It can be assumed that the colored white clay under the red layer was detected together with the pigments. According to the results of the P-XRF analysis regarding a total of eight points, the patterns of constituent elements were classified into two, depending on the presence and content of Fe, Pb, and Hg.

Points R-1 to R-4 show a high Pb content with an average of 366,476 ppm (360,982 to 373,432 ppm). Points R-5 to R-8 show a relatively high content of Fe, with an average of 94,592 ppm (63,070 to 144,556 ppm). Thus, the red colors appear to be a mixture of minium (Pb3O4) or minium and hematite (Fe2O3), and it can be assumed that cinnabar (HgS) was used in smaller amounts together with these (Table 3 and Fig. 7).

According to the analysis results regarding the black layer, elements such as P, Fe, K, S, Ca, Ti, Cl, and Cu were detected. Most of these elements are not components of black pigments, which suggests the effect of dead color paints. However, no element indicating black color was detected (Table 3 and Fig. 7). The traditional black pigment is an Chinese ink, which is a carbon compound. Because Chinese ink is composed of light elements, it is difficult to detect with P-XRF. Thus, because only the same components as the white pigment used in the base layer were mainly detected at the measurement point of the black pigment, Chinese ink was likely used for black color development, requiring additional review.

In this black layer, the detection of slightly high levels of Pb and Ca, with an average of 59,164 ppm (0 to 331,475 ppm) and 85,953 ppm (from 16,859 to 194,497 ppm) through P-XRF, suggests that white lead and oyster shell white were mixed to lower the concentration of black pigment. Among the analysis points, the highest detection of Pb at 331,475 ppm in the right pupil of the Gwaneum Bosal statue suggests that the pigment was reapplied several times (Table 3 and Fig. 7).

The white color is mainly used on the Gwaneum Bosal. statue and cloth, in which elements such as Si, K, and Al were detected through the P-XRF analysis. Among these elements, the average contents of Si, K, and Al were 151,220 ppm (58,740 to 244,566 ppm), 11,979 ppm (5,722 to 24,628 ppm), and 46,856 ppm (7,846 to 89,153 ppm), respectively, and these high levels suggest the high likelihood of the use of kaolin (Al2Si2O5(OH)4). The detection of Ca and S with an average of 225,922 ppm (95,963 to 341,179 ppm) and 52,471 ppm (36,456 to 98,436 ppm) suggests that gypsum was likely mixed with feldspathic kaolin (Table 3 and Fig. 7).

The color violet was mainly used for coloring the daeui part of Gwaneum Bosal. According to the result of the P-XRF analysis, Cu and Fe were detected with an average of 27,799 ppm (22,263 to 33,334 ppm) and 26,454 ppm (23,255 to 29,653 ppm), respectively, as the main compositions. Because the composition is similar to that of the blue pigment, and high levels of Fe and Pb are detected, the mixture of azurite, hematite, and minium can be assumed (Table 3 and Fig. 7).

In addition, Ca, S, and Mg were detected here at 34,920 ppm (32,165 to 37,674 ppm), 72,511 ppm (66,265 to 78,756 ppm), and 72,223 ppm (71,575 to 74,871 ppm), respectively. This result suggests the addition of gypsum or talc (Mg3Si4O10(OH)2), which is a white pigment, or the detection of constituent minerals in the base layer to lower the chromaticity (Table 3 and Fig. 7).

The color pink is painted on the chest top part of Gwaneum Bosal as well as the ears of Buddha. According to the results of the P-XRF analysis, Si, Mg, Ca, and S were detected with an average of 177,186 ppm (43,800 to 230,1666), 128,059 ppm (104,302 to 158,080 ppm), 109,155 ppm (34,669 to 304,902 ppm), and 44,710 ppm (23,002 to 71,002 ppm), respectively. This result becomes the basis for assuming the use of kaolin or talc as a white pigment in combination with pink. Moreover, the detection of high levels of Pb and Fe with an average of 99,309 ppm (2228 to 277,324 ppm) and 11,374 ppm (7,0378 to 16,438 ppm) suggests that the color depth was raised by mixing kaolin or talc with minium and hematite in terms of components (Table 3 and Fig. 7).

SEM–EDS analysis

Although the locations of the micro-fragments of the flaked-off pigments collected during the stripping process for the repair of the Gwaneum Bosal were not clearly identified on this painting, a pale blue pigment was found underneath the thin black layer. Based on this finding, the colored layer was closely observed on a SEM (TESCAN, MIRA3), and the constituent elements on the pigment layer were analyzed using an EDS (Bruker, Quantax 200) mounted onto the SEM (Fig. 8).

Fig. 8
figure 8

Microphotograph images using optical microscope, SEM and EDS analytical spots for exfoliated black and cyan pigment samples in this study. A, B Occurrences under stereoscopic microscope at the analysis point. C SEM secondary electron image in Fig. 3A. D SEM Back scattered electron image of Fig. 3A. E Elemental mapping image using EDS analysis on the black and cyan layer. F SEM secondary electronic images of carbon particles in Fig. 3A. G Enlarged particle image of Fig. 3F. H SEM secondary electron image in Fig. 3B

Full size image

According to the observation at high magnification with the SEM, the microtextures of the pigment layer can be divided into two layers: a black layer and a cyan layer (Fig. 8c). The elemental mapping regarding the black pigment part confirmed that carbon was more distributed in the black part than in the light cyan part (Fig. 8e). An observation at 3000 magnification of this part showed that carbon particles with a diameter of 5 nm were relatively uniformly coated on the smooth surface (Fig. 8f).

According to the results of the EDS analysis regarding this matter, a significantly high carbon (C) content was detected with an average of 60.30 wt.% (Table 4). In particular, no compositions other than the base layer components underneath the black color were detected, suggesting that Chinese ink, which is composed mainly of carbon, was used for coloring (Fig. 8c to g, and Table 4). Furthermore, although the SEM analysis regarding the pigments attached to the background paper showed a large number of Blue-1 pigment particles that had exfoliated onto the tissue composed of mulberry fibers, no specific components were detected (Fig. 8b and h).

Table 4 Results of SEM–EDS analysis (wt.%) on samples after exfoliation from the painting in this study
Full size table

Discussions

This study secured basic data on the types, compositions, raw materials, and colors of the pigments used in this painting through a scientific investigation and precise analysis of White-Robed Water-Moon Avalokiteshvara of main hall in Magoksa temple and the pigments remaining on the wooden frame after its stripping. The Korean National Research Institute of Cultural Heritage has reported on traditional pigments and color-developing elements based on analysis of various periods and cultural heritages. Table 5 summarizes the pigments that are consistent with the manufacturing time and color of this study subject based on data from previous studies.

Table 5 A summary of representative main pigments used in Korean Buddhist paintings from the eighteenth to nineteenth centuries analyzed in previous studies on historical wall paintings
Full size table

Based on this data, in reviewing the analysis results regarding the target of this study, small amounts of modern synthetic pigments were identified at many points. Table 5 shows the results of the analysis using P-XRF and SEM–EDS, which are classified into the color series of pigments according to major color-developing elements.

Two types of blue pigments were assumed to be used for raising the color depth by mixing a white pigment with azurite (Cu3(CO3)2(OH)2). Regarding the average chromaticity, L*, a*, and b* are 11.6, 65.6, and − 1.3, respectively, for Blue-1 pigment and 70.7, − 5.1, and 10.5 for Blue-2. The higher L* value for Blue-2 suggests that Blue-2 was mixed with a larger amount of white pigment to lower the chromaticity. The white pigment used in this case is assumed to be white lead, kaolin, or titanium dioxide (Table 6).

Table 6 Summary on detected elements and estimated pigments for the painting in this study
Full size table

As pigments that are color development sources for green, copper chloride-based emerald green and clinoatacamite were identified, and white pigments such as white lead and kaolin were used as auxiliary materials for color combination, depending on the coloring position. The average chromaticity of the green series was determined as a L* of 48.0, a* of − 10.2, and b* of 4.1, and the chromaticity varied slightly depending on each analysis point.

Orpiment and massicot were identified as yellow pigments. These two types of yellow colors were mixed with white pigments to adjust their chromaticity for color combination. S and Pb were highly detected at most of the measurement points, suggesting that massicot (PbO) was likely used in combination with white lead or gypsum.

A trace amount of As was detected at the Yellow-2 point, which was difficult to determine as orpiment (As2S3) because the pigments colored on the background may have been analyzed together with it. The average chromaticity of Yellow-1 was determined as a L* of 57.4, a* of 10.7, and b* of 26.4, and that of Yellow-2 was determined as a L* of 59.4, a* of 3.4, and b* of 30.5. Thus, Yellow-1 was higher in luminosity, while Yellow-2 was higher in yellowness.

The compositions of Fe, Pb, S, and Hg were detected as color development sources for red pigments, which have compositions of hematite, cinnabar, and minium in common. This result suggests a mixture of hematite and minium as well as white pigments, such as lead and kaolin, for color combination. Mercury (Hg)-based cinnabar appears to have been used in trace amounts, or not used at all. According to the existing records, because cinnabar was highly expensive—with a value approximately three times that of Hg—Seokganju was used as a substitute it [30].

Among these analysis points, 56,922 ppm of Ti was detected in R-2, which suggests that titanium dioxide was likely used. The average chromaticity of red pigments including various color ranges was determined as a L* of 39.9, a* of 20.7, and b* of 17.0, showing a tendency of low chromaticity and redness, and there was no significant difference in the average values of chromaticity for each measurement point.

The black pigments were all assumed to be amorphous Chinese ink. Chinese ink and niram were mainly used as black pigments in Buddhist paintings or dancheong, making it difficult to accurately identify via P-XRF analysis alone. To overcome the limitations of this non-destructive analysis, a UV–Vis analysis was additionally performed for Gwaebultaeng in Boeun Beopjusa Temple (Treasure No. 1259). The results indicated that the reference spectra of the artifact’s black pigment was nearly identical to that of Chinese ink [10]. However, the application of this technique was not permitted in the present study.

The average chromaticity of the black series was determined as a L* of 59.4, a* of 3.4, and b* of 30.5, showing no significant difference at most of the analysis points, whereas the black pigment used for coloring the bamboo in the top right corner of Gwaneum Bosal showed the highest values with a L* of 49.92, a * of − 10.91, and b* of 1.75.

Kaolin and white lead were identified as pigments that become color development sources for the color white. Kaolin is composed of various minerals, which are subdivided into kaolinite with feldspar, and mica. The results suggest that white pigments were used alone, mixed with pigments of other colors, or used as extenders. The average chromaticity of the white pigments was determined as a L* of 74.0, a* of 1.0, and b* of 16.1.

Violet and pink were used as auxiliary colors for coloring this painting. The analysis results of pigments showed that their composition was similar to that of the blue pigments, and the detection of Fe and Pb indicated that white lead was mixed with azurite or hematite. The average chromaticity of the color violet was determined as a L* of 52.4, a* of − 0.6, and b* of − 1.2, which has lower chromaticity and higher blueness than the blue pigments.

The pink pigments were estimated to be a mixture of minium and oyster shell white due to the detection of high levels of Ca and Pb at almost all measurement points. Furthermore, the results suggested that Mg, Al, and Fe were detected at the measurement points of P-2 to P-4 due to the influence of the base layer. The average chromaticity of the pink pigments was determined as a L* of 63.7, a* of 11.0, and b* of 13.8, which shows high luminosity and slight reddishness.

Elements such as Ti, Ba, and Hg were identified in most points on this painting, and thus, it was impossible to rule out the possibility that industrial synthetic pigments were used in the repair process. Research results reveal that pigments were imported from China or the Western Regions, suggesting a high likelihood that White-Robed Water-Moon Avalokiteshvara was repaired using modern pigments even though South Korea started manufacturing modern pigments in 1940.

Further research is required, such as more accurate component analysis and identification for each color of the pigment, to confirm the traditional pigments that were used prior to modern synthetic pigments. Table 6 summarizes the analysis results and estimated pigments for the aforementioned ten types of pigments.

Meanwhile, Ca, Fe, As, and Cu components were concurrently detected through the pigments’ analysis points. This finding could be attributed to dead coloring, which is one of the production techniques of traditional painting. Dead coloring is applied on the ground layer of paper, silk, or burlap used as materials in Buddhist paintings and folk tales. This technique impedes the absorbency of the ground layer, improves the coloring of the recoated pigments, and serves as a filler for pores or grain. Dead coloring was applied to nearly all ancient Korean paintings, and the same is true for modern Buddhist paintings and oil paintings [10].

There are some examples in which green pigment was used as a dead color paint in Buddhist paintings, and the ground layer of Dancheong was repainted using noerok. Thus, there is a likelihood that green pigments were also used as a dead color paint for White-Robed Water-Moon Avalokiteshvara. Concurrent detection of certain components—such as Ca, Fe, As, and Cu—could be attributed to kaolin and massicot containing Ca and Fe, which were often used as fillers for the ground layer. Moreover, Cu and As will also be detected assuming that green pigments were used as a dead color paint. To prove this assumption, additional cross-sectional observations of the applied pigments and/or studies focusing on other Buddhist paintings are necessary.

Amorphous carbon was observed in the black pigment remaining on the wooden frame after this painting was stripped. Although P, S, and Pb were also detected, this was assumed to be due to interference from the contaminants and the surrounding pigments. In addition, green, white, grayish white, and red pigments appeared, and the element content ratio of each color pigment was highly similar to that of this painting. Thus, the pigments remaining on the wooden frame are assumed to have been influenced by the pigments used in this painting.

Conclusions

The White-Robed Water-Moon Avalokiteshvara painting displayed on the rear wall of main hall in Magoksa temple, is painting representative of Buddhist paintings in the late nineteenth century of Korea, and a valuable resource for understanding the coloring techniques and characteristics of Buddhist paintings in terms of expression and description methods in landscape painting. The present study has applied a non-destructive method to analyze the eighteenth-century South Korean White-Robed Water-Moon Avalokiteshvara to determine which pigments were used for coloring.

Table 5 summarizes the estimated pigments. In total, 45 different colors and compositions were analyzed under the same conditions, and the pigments used for each color were examined based on the components detected via qualitative analysis. As a result, the types of pigments according to color, blue, green, yellow, red, black, and white were determined as azurite; emerald green or clinoatacamite; massicot; minium or hematite; Chinese ink; and kaolin, white lead, and gypsum, respectively. Violet and pink colors were assumed to have been prepared by mixing white with blue and red.

In most of these pigments, small amounts of synthetic pigment components from the modern era were detected at many points. In addition, a detailed further examination of the cross-section of the pigment layer is required because of the overlap between the traditional pigments and the modern pigments that were used for repair. In addition, according to the pigment analysis results, the components Ca, Fe, As, and Cu were detected in nearly all pigments. This indicates the presence of a filler or dead color paint for silk, which was used as the ground layer for many paintings.

The compositional analysis of the pigments in this study was limited to sampling, which allowed only non-destructive analysis and local analysis on exfoliated tiny samples that were recoverable. Additionally, the analysis results were compared to previously reported data. This preliminary study aims to secure the basic data necessary for preservation and restoration. The study’s findings have been utilized as important data in selecting pigments to repair White-Robed Water-Moon Avalokiteshvara in Magoksa Temple, as well as for the restoration of damaged areas and long-term preservation.