Technical Study of Polychrome Arhat Figures Dated From the Song Dynasty(960-1279 CE) From the Lingyan Temple, Changqing, Shandong, China

Scientic analysis revealed the materials and techniques used in the process of making the polychrome sculptures providing a solid foundation for the protection and restoration of the painted statues. In addition analysis revealed changes in colour schemes applied to the sculptures can provide the basis for the virtual restoration of the painted statues. In order carry out scientically-informed protection and restoration of the Bodhidharma statue from the Lingyan Temple, Changqinq, Shandong, several analytical methods such as optical microscope (OM), Micro-Raman spectroscopy (μ-RS), scanning electron microscopy coupled with energy dispersive X-ray analysis (SEM-EDS) and Fourier transform infrared spectroscopy (FTIR) were employed. Analysis clearly reveal the information including the stratigraphic structure and the composition of pigment. The use of silver foils and golden yellow pyrophyllite mineral to replace gold foils were found in the gilding paint layer in the later repainting after the Song Dynasty. This work reports the coexistence of emerald green (Cu(C 2 H 3 O 2 ) 2 ·3Cu(AsO 2 ) 2 ) and the degradation product lavendulan (NaCaCu 5 (AsO 4 ) 4 Cl·5H 2 O) in large areas of the paint stratigraphy and on the surface conrming that the degradation of emerald green is related to the thickness of the paint layer; in thinner paint layers emerald green is transformed in lavendulan, while thicker layers of contain both lavendulan and emerald green, suggesting an environmental source of chlorides. red lead (Pb 3 O 4 ). EDS analysis results show that the content of Pb in the pigment layers containing red lead are above 90%, which

In May 1981, Jinan cultural relics management committee, Jinan museum and the Changqing county LingyanTemple cultural relics management institute jointly initiated the protection and repair of the arhat statues of LingyanTemple [1]. In the process of maintenance, C-14 dating of the wooden structures within some of the arhat sculptures determined that of the 40 arhat statues, 27 were from the Song Dynasty 13 date to the Ming Dynasty [1][2]. The objective of this paper is the technical study of the polychromy of Bodhidharma, which is a typical Song Dynasty work. This statue is colorful and rich, the expression of the characters is solemn and serene, tolerant and introspective, the work is also relatively complete, has extremely high artistic value. Bodhidharma was a native of India who came to China during the Northern and Southern Dynasties (420-589 CE) and founded Chinese zen Buddhism ( ). Because of the spread and popularity of zen Buddhism in China, Bodhidharma was a household name in China. The statue of Bodhidharma was placed at the rst place on the right hand side of the entrance to the hall of Qianfo ( the rst place in the east), which meant the statue received more attention and favor from visitors than other statues. For a long time, natural aging and many environmental factors (such as dust, smoke, roof leakage, the presence of insects, pollution.) lead to degradation of pigments, e orescence and the aking and peeling of paint. It is of great signi cance to analyze and study the paint layers of arhat statues for the subsequent conservation of cultural relics and virtual restoration. In 2018, Wang Chuanchang and others analysed samples from the sixth and seventeenth arhat statues in the east of the hall of Qianfo, and determined the mineral composition of various mineral pigments [3]. Besides that, there was little scienti c analysis and research on the paint layers of arhat statues of Lingyan Temple. Complementary and multiple analytical techniques are widely used in the study of paint layers of painted cultural relics such as polychrome statues [4][5][6][7], mural paintings [8][9][10][11], painted pottery [12][13] and so on, which can successfully reveal the microscopic structure of paint layers, the composition of the pigment and the priming layers, the types of binding media found in paint layers, repair materials, and the paint stratigraphy. In this paper, multiple analytical techniques including ultra depth of eld 3D microscopy (OM), Micro-Raman spectroscopy (μ-RS), scanning electron microscopy coupled with energy dispersive X-ray analysis (SEM-EDS) and Fourier transform infrared spectroscopy (FTIR) were selected to analyze the paint layers from the Bodhidharma statue. The purpose of this study is to reveal the materials and techniques used in the making the painted statues, and the changes and to the composition of the paint layers as a consequence of aging, provide basic data for the protection and restoration of the painted statue and the virtual restoration of the color texture.

Materials
A total of 19 samples were collected from different colored paint on the Bodhidharma statue. The schematic diagram of sampling position was shown in Fig. 2. Except for D1-3, D1-10, D1-19, other samples were embedded in the epoxy resin and polished.

Optical microscopy (OM)
A KEYENCE VHX-6000 ultra depth of eld 3D video microscope was used to observe and document the cross sections, at magni cations from 20 to 2000×. The Leica DM2700 polarizing microscope is used for the polarizing microscopic analysis of individual pigments.
Scanning electron microscopy coupled with energy dispersive X-ray analysis (SEM-EDS) A Tescan vega3 XMU scanning electron microscope equipped with a Bruker XFlash 610M X-ray energy spectrometer was used to analyze the microstructural characteristics of paint layers and semi-quantitatively analyze the relative contents of major elements in pigment minerals in different layers. Analyses were carried out in a high vacuum environment, with a scanning voltage of 20 kv, 90 s of acquisition time at a 15 mm working distance.

Micro-Raman spectroscopy (μ-RS)
A HORIBA XploRAPLUS Raman spectrometer con gured with an Olympus microscope and an integrated motorized stage were used to qualitatively analyze the mineral composition of different layers under 50× and 100× objective lenses. The blue and green pigments were analysed using a 532 nm laser, and the other pigment were analysed using the 785 nm laser. The laser energy ranged from 1%-50%, the spectral range from 50-2000 cm -1 , the collection time was 15-25 s with 2 acculumations. The instrument was calibrated using the 520 cm -1 silicon Raman band.

Fourier transform infrared microscopy (FTIR)
A Thermo Nicolet iN10 MX Fourier transform infrared spectroscopy was used to study the priming materials of paint layers based on the preparation of dispersions in KBr in the spectral range from 4000-675 cm -1 , at spectral resolution of 4 cm -1 .

Page 4/29
The EDS results of relative elemental composition in each layer of the painting and the attribution from μ-RS analysis are shown in Table 2. The elemental composition is given as the mass percentage after normalization treatment. In the processing of element relative content analysis, the free selection mode was used to select areas of interest within each layer. The layer sequence was numbered in the micrographs of cross sections in Fig. 3, 5 and 8 correspond to the same numbers used in EDS analysis. Organic layers that did not contain metal ions are: D1-2 L0; D1-9 L2, L5, L8, L10, L12, L14; and D1-16 L3. The elemental composition of these layers are not listed in Table 2.

Analysis of stratigraphy
The cross sections observed under the optical and scanning electron microscope clearly show that the paint layers contains multiple layers of pigments and priming layers. D1-1, D1-7and D1-9 are samples with the most layers. As shown in Fig. 3 a and b, D1-1 has 14 layers from the statue surface to the unpainted clay ground, among which the pigment layers have 7 layers, and the layer structure distribution are as follows: L1 reddish-brown pigment layer→L2 white priming layer→L3 red pigment layer→L4 white priming layer→L5 orange pigment layer→L6 white priming layer→L7 yellow pigment layer→L8 white priming layer (mixed with red pigment particles)→L9 orange pigment layer→L10 white priming layer→L11 red pigment layer→L12 white priming layer→L13 white pigment layer→L14 gray priming layer.
The structural relationship between the layers of D1-9 gilding sample can be shown more clearly from the backscatter electron image (Fig.3  The layer structure is basically the same as the layer structure of D1-9, but there is an extra priming layer under the 12th layer of adhesive. According to our analysis the D1-7 and D1-9 contain six gilding layers and two pigment layers demonstrating that the Bodhidharma statue has experienced at least 8 repainting treatments. In other samples of paint there are only two layers. As seen from Fig. 5 a, sample D1-2 located on the face of the statue has only one reddish-brown pigment layer and one priming layer suggesting that the color of the face has always remained the same in the previous polychrome process. The reddish-brown pigment layer was covered with a dark translucent organic material (this translucent material is almost invisible in backscatter electron images).
The particle size distribution using the statistical analysis functions of the built-in software of the KEYENCE VHX-6000 ultradeptho eld 3D microscope were used to measure the thickness of the paint layers, the rst pigment layers, the priming layers and the particles sizes of the pigment. The measurement results are shown in Table1  The pigment layers and the priming layers usually appear alternatively, indicating that during repainting the statue was generally primed with a white layer that was applied over the last pigment layer. For the gilding layers very few priming layers are found. Instead new gold foil layers were applied directly over the last gold foil using an adhesive. In some places there is no priming layers between the old and new pigment layers. For example, the green pigment layer L1 of the D1-15 was applied directly over the red pigment layer L2 (Fig. 4 d).
Paint layers represented by D1-8 and D1-19 are clearly different from other paint layers in terms of layer structure. The whole paint layers were painted on a layer of white paper ( gold foils with high gold content. The thickness measured by the electron microscope are only 2μm at the thickest point.
The 7 th and 9 th layers of sample D1-7 can hardly be observed under the optical microscope, while they are clearly evidenced in the backscattered electron images ( Fig. 3

d, f). EDS results showed that the two layers mainly contain Cl and
Ag, with a mass ratio of about 1/3.7 to 1/3.2. The Raman spectrum of white particles in layer 7 (Fig. 6 a) showed characteristic bands at 147vs, 221m, 292m cm -1 similar to chlorargyrite (AgCl, Fig. 6b). Combined with the results of EDS, it can be inferred that the main phase of D1-7 L7 and L9 are chlorargyrite (AgCl). In terms of thickness and stratigraphic structure, D1-7 L7, L9 are only 1-2μm in thickness like other gold foil layers. Therefore, these two layers were silver foil that was pasted on the statue surface at two separate times. Corrosion later transformed the silver foils into chlorargyrite (AgCl).
The L3 and L6 of D1-7 and D1-9 paint layers appear golden yellow in cross-section with mineral granularity (Fig. 3 c, e).
The four layers of golden yellow material were painted over gold foil layers. From the color and its layer position, it can be con rmed that these layers were applied over gold foils. The results of EDS show that these four layers mainly contain Al, Si, K, Ca, Fe. In addition, D1-9 L3 contains 8% Au, which is due to the curious inclusion of gold particles in this layer ( In ancient China, especially since the Song Dynasty (960-1279 CE), the decorative technology of embossed painting and gilding ( ) on Buddha statues was very popular. As the saying goes, "Clothes to people, Gold to Buddha". However, the use of silver foil and gold mixed with golden yellow mineral to replace gold foil used in the decorating Buddha statues is rarely mentioned in the literature. The discovery of silver foil layers and golden yellow pyrophyllite mineral with gold in D1-7 suggests that there were other alternatives to gold that were employed to imitate gilding.

Analysis of the priming layers
In terms of element distribution, the priming under the pigment layers can be roughly divided into ve types. The rst type is the white mineral dominated by Al, Si (Al/Si ratio between 1/2-1/1.2) and containing a small amount of K, Fe, Ca, Pb and other elements, such as D1-1 with priming layers, D1-2 L2, D1-5 L2 L4 L6, D1-6 L2 L4 L6 L8, D1-11 L5. These white mixtures showed no obvious Raman scattering except quartz particles. In order to determine the mineral phase of the priming layers mineral, three samples with simple layer structure, D1-2, D1-5, D1-6, were selected for FTIR analysis. Fig. 7 A, B and C are the infrared spectra of the above three samples respectively, and their main infrared characteristic absorption bands are basically one-to-one corresponding to the infrared characteristic absorption bands of kaolin [14]. Si are only 2% and 3%. Micro Raman spectroscopy revealed two white minerals, chalk (CaCO 3 ) and lead white. D1-4 L4 is composed of chalk and lead white, mixed with a small amount of kaolin. The fourth type of priming layer mineral dominated by Pb, such as D1-11 L4 and D1-15 L4, containing chie y lead white with Pb content in the two layers as high as 85% and 60% respectively. μ-RS con rmed that the main mineral phase in these two layers is lead white, so this type of priming is lead white with a small amount of kaolin added. The fth type of priming layer mineral dominated by Si, Al, Ca, and containing a certain amount Fe. For example, under the pink pigment layer of D1-8 sample, there are two priming layers of white and brown, and the Al/Si ratio of their chemical composition are 1:3 respectively. Both of these two layers contain Fe (the white layer 13% and the brown layer 16%), which is different from other kaolin-based priming layers mineral in chemical composition. The priming of these two layers have no obvious Raman activity. According to the relative content of elements, it can be inferred that the white priming layer should be a low aluminum and high calcium clay, while the brown priming layer is a low aluminum and high iron clay.
In summary, the priming layers material used in the each previous polychrome of the statue is mostly white kaolin or a mixture of other white minerals based on kaolin as the matrix. The use of low aluminum and high iron priming in D1-8 is obviously different from other paint layers.

Analysis of pigment layers
The results of μ-RS analysis of mineral pigments in each pigment layers were shown in Table 2. Raman analysis were based on published literature [15][16]. Raman spectra of some pigments are shown in Fig. 9 and 10.

Reddish-brown pigment layers
The reddish-brown pigment layers have three layers: D1-1 L1, D1-2 L1 and D1-12 L1. μ-RS analysis show that the D1-1 L1 and D1-2 L1 pigment layers are composed of hematite (Fe 2 O 3 ) and red lead (Pb 3 O 4 ). EDS analysis show that the Fe content in the two layers are signi cantly different (17%, 31%, respectively), indicates that the mineral ratio of pigments was different due to the different color requirements of the pigment layers at different locations (D1-2 L1 is darker in color than D1-1 L1). The pedestal area represented by D1-12 L1 is reddish-brown ( Fig. 1), while the pigment layer presents a bright orange-red on the cross section (Fig. 4 a). Raman analysis shows that the orange-red mineral is red lead (Pb 3 O 4 ).
The change from bright orange-red color to reddish-brown color on the exposed surface of the pigment layer should be related to the oxidation and deterioration of red lead pigments, and it is well known that red lead can alter to brown-black lead dioxide (PbO 2 ) in high humidity environment [17][18][19].   Fig. 1). This is obviously different from the characteristics of paint layer painted over white paper in D1-8 L1, so it can be inferred that the pink area of hematite + anatase titan white in D1-3 L1 is the repair area after partial damage and shedding of the original hematite + chalk pink area.

Blue pigment layers
The blue pigment layers are D1-5 L1 and D1-10 L1. Raman analysis shows that the blue mineral in D1-5 L1 is azurite (Cu 3 (CO 3 ) 2 (OH) 2 ), and there are white chalk (CaCO 3 ) particles mixed in L1. D1-10 L1 blue pigment's Raman characteristic bands at 258w, 548vs, 1091m cm -1 (Fig. 9 j) are completely consistent with the bands of ultramarine. Raman spectroscopy and chemical composition can not distinguish natural and synthetic ultramarine pigment, but polarizing microscope analysis is an effective and simple method. Natural ultramarine pigment is made of lapis lazuli mineral grinding, because lapis lazuli mineral is often accompanied by diopside, calcite, pyrite, so there are more impurities in the pigment; particle shape is irregular, angular, particle diameter in 10~20 μm. But arti cial ultramarine pigment is pure in texture, and the particle shape is regular, uniform and ne (average diameter is 5 μm) [20]. D1-10 L1 blue pigment is bright and colorful under plane polarized light (Fig. 8 a), with pure texture, smooth particle edge, particle diameter of 3-6 μm, completely extinction under orthogonal polarized light (Fig. 8 b), which corresponds to the characteristics of arti cial ultramarine pigment (Na 6-10 Al 6 Si 6 O 24 S 2-4 ).
Lavendulan(NaCaCu 5 (AsO 4 ) 4 Cl·5H 2 O) is a rare supergene arsenate mineral in oxidation zones of Cu-and As-bearing ore bodies [21]. It was rst described by Breithaupt in 1837 and named as a mineral containing As, Co, Ni; Vogl (1853) describes it as blue coating from Jáchymov; Foshag (1924a) published a detailed description of lavendulan and gave its refractive index [22]. Guillemin (1956a) rst used X-ray diffraction analysis to study lavendulan and proposed its original rhombic crystal structure. The mineral, which has a vivid blue-green color and a small spherical or wafer-like microstructure, is often found in arid climates, and in caves derived from copper sul des in cave walls. Also, it has been reported as an alteration products in ancient slag heaps. Further, it is certain that some of these minerals are of archaeological signi cance and it is apparent that they were used for cosmetics in ancient Egypt [23].
Lavendulan has also been found sporadically in Chinese polychromy such as the painted clay statue in Yungang Grottoes ( ), the colored drawing on timber structure in Kumbum ( ) [24], the wall painting in Anyue Grottoes ( ) [25], the . The distribution of lavendulan is extremely rare in nature. Due to its physical microscopic structure and the fact that the chemical composition is similar to the widely-used emerald green pigment from the 1830s to the early 20th century. Chinese scholars believe that the lavendulan found in polychromy is most likely an alteration product from emerald green pigment, and was not intentionally used as the blue-green pigment [24][25][26][27][28].
In literature [28], the mechanism of corrosion conversion of emerald green pigment into lavendulan has been studied in depth. It is suggested that emerald green (Cu(C 2 H 3 O 2 ) 2 ·3Cu(AsO 2 ) 2 ) was rstly dissociated copper ions (Cu 2+ ) and arsenite ions ((AsO 2 ) ) in a slightly acidic and humidity environment. The (AsO 2 ) was oxidized to arsenate ion ((AsO 4 ) 3 ), and nally Cu 2+ , (AsO 4 ) 3 reacts with soluble K + , Ca 2+ , Cl from the environment to produce lavendulan. There are no residual emerald green pigment particles were found in the blue-green pigment layer D1-18 L1. However, the simultaneous presence of layers of green and blue-green pigments can be seen in other blue-green pigment areas (Fig. 10 a), which shows the signs of green emerald green transformation into blue-green lavendulan. Therefore, it can be inferred that the blue-green pigment layer D1-18 L1 (the part of plant and ower pattern above the black pigment layer of the cassock of the Bodhidharma statue, as shown in Fig. 2) may have transformed from emerald green. The thickness of the pigment layer in D1-18 L1 is 80 μm, and its mineral phase was almost completely transformed into blue-green lavendulan, while the emerald green pigment layer D1-15 L1, with a thickness of 570 µm has no sign of alteration. This observation suggests that the transformation of emerald green pigment depends on the thickness of the pigment layer, with thinner pigment layers exhibiting lavendulan. This could be because the thinner the pigment layer, the easier it is to contact with the moisture and acid substances in the air, and the easier it is for the soluble salt ions to migrate in the pigment layer, accelerating the transformation of emerald green to lavendulan.

Black pigment layers
There are 4 layers of black pigment layers: D1-11 L1, D1-12 L2, D1-18 L2 and D1-19 L1. The Raman spectra of the black pigment shows two broad bands near 1325 cm -1 and 1580 cm -1 (Fig. 10 i), which is consistent with the standard Raman spectra of lamp black (C). The black pigment appears gelatinous under the microscope, and has no granular sensation ( Fig. 4 a). Therefore, it can be inferred that the black pigment is the Chinese ink, which the most commonly used black pigment in ancient China.

Rose red painted layer
Page 14/29 The rose red painted layers are only distributed in the bottom step area of the statue pedestal in sample D1-16. From the preserved photos [29], this area was originally reddish brown before restoration and protection in 1981, and the painted layer was largely peeled off (see Fig. 11 a). According to the interview with the staff of Lingyan Temple, the rose red area was painted in the restoration and protection of the Arhat statue in 1981; in addition, a layer of varnish was also painted in the bottom step area of pedestal in order to isolate moisture [1].
EDS analysis shows that the mass percentage of C and O elements in D1-16 L1 is more than 75% consistent with an organic pigment. The characteristic bands in the Raman spectrum (Fig. 12 b) [30]. The white priming layer of D1-16 L2 mainly contains Ti, Ca and S. Raman analysis con rms that the layer contains anatase titan white (TiO 2 , Fig. 12 c). The infrared spectrum of the D1-16 was shown in Fig. 12  The L1 yellow pigment layer is the oral motif painted over the green pigment layer. It can be seen from the cross sections that there is no priming layer between the yellow pigment layer and the green pigment layer, indicating that the two layers of pigment were painted on the surface of the statue at the same time. Chrome yellow appeared in the early 19 th century [33,34], emerald green (chemical name "copper acetoarsenit"), was rst synthesized in Germany in 1814, mainly used in the 1830s and early 20 th century [35]. The use of lead chrome yellow and emerald green in the outermost pigment layer con rms that the statue was repainted later than 1814. Chinese scholars' textual research on the existing inscriptions and other materials in Lingyan Temple has unanimously concluded that the last polychrome of the arhat statues of Lingyan Temple was in the thirteenth year of the reign of Tongzhi (1874CE) [2,36] in the Qing Dynasty(1639-1912 CE).
Lead chrome yellow and emerald green were most likely painted on the surface of the statue in the 1874 redecoration. The use of anatase titan white pigment in D1-3 pink pigment layer and D1-16 L2 indicated that the statue had also been treated after 1874, because the commercial titanium dioxide pigment appeared in 1916 [37]. There is no mention of the polychrome repainting of the arhat statues in Lingyan Temple after 1874 in various Chinese documents. From the photographs taken before the restoration of the Bodhidharma statue in 1981 (Fig. 11 a), it can be seen that the state of the pink area from D1-3 (the inside of the hat) before 1981 is similar to its current condition. Based on the use time of anatase titan white, it can be inferred that the pink area containing anatase titan white should has been painted between 1920s to the 1980s. In addition, the Bodhidharma statue was painted with a rose red organic pigment on the previously reddishbrown step area. In summary, the most recent large-scale repainting of the statue took place in 1874, and there have been at least two partial repainting treatments since then.

Conclusions
Based on the scienti c analysis of the painted layers of Lingyan Temple's Bodhidharma Statue, there are some insights.
The statue has been painted at least eight times and there have been at least two partial repainting treatments in the 20 th C. Before each new polychrome white wash was applied over the last pigment layer. In the most recent polychrome, a new technique of pasting white paper on the clay gure surface and then painting the priming layer and the pigment layer was used in some areas.
The gold layer in the gilding decoration area mostly uses gold foil with a high content of gold, and its thickness is only 1-2 μm. In different periods of history, silver foil and gold mixed with golden yellow mineral pigments were used to replace the original gold foil layers in the gilding area. The priming layers material used in the each previous polychrome of the statue was mostly white kaolin or a mixture of other white minerals based on kaolin as the matrix. Sometimes lead white, chalk or a mineral mixture of the two white pigments were used as the priming layers material. In addition, the clay with a low aluminum and high iron content has been used as priming layers material in the most recent polychrome.
The pigment layers of the statue was mostly composed of a single inorganic mineral, and the individual pigment C. emerald green (Cu (C 2 H 3 O 2 ) 2 ·3Cu (AsO 2 ) 2 ) and copper-containing pigments in a glassy state; Black pigments were only lamp black (C). In addition, organic pigments and anatase titan white (TiO 2 ) has been used in partial repainting more recently. The blue-green lavendulan mineral on the surface of the statue is not intended to be used as a blue-green pigment, but rather is a degradation product of emerald green pigment. The transformation of emerald green pigment seems related to the thickness of the pigment layer. The thinner the pigment layer, the easier it is to transform.