Weijuan Zhao

To understand the firing technology and the level of production of glazed tiles in the Ming Dynasty, an ultraviolet-visible-near-infrared (UV-Vis-NIR) spectrophotometer and an energy-dispersive X-ray fluorescence (XRF) spectrometer were used to systematically analyze the glaze color and chemical composition of glazed tiles unearthed from the southern area of the Bao’ensi site. The results revealed that little difference exists in the main wavelengths of the same glazed tile samples, but that the colored elements of glaze with different colors are different. Iron (Fe) is the main element that affects the color of yellow glaze; copper (Cu) is the main element that affects the color of green glaze; and the main colorants that affect the color of black glaze are Fe, Cu, and manganese (Mn). The yellow and black glaze unearthed from the southern area of the Bao’ensi site were relatively mature in terms of raw material composition and technological formula, whereas the raw material and technological formula of green glaze were not stable enough. To meet temple construction needs, workers were skilled in firing glazed tiles with different glaze color requirements, indicating that the glaze color of the glazed tiles was relatively mature at that time.

Architectural glazed tile, a type of low-temperature lead-glazed pottery, is a common building material used in traditional Chinese architecture with good decorative and waterproof functions. The process of fabricating glass components in Chinese architecture has a long history. The earliest historical records can be found in the Book of Wei, compiled by the Wei Dynasty during the Northern Qi Dynasty. Since that time, with the development of glazed-tile firing technology, many fine architectural glass components have been developed.

In recent years, ample research of the manufacturing technology of ancient glazed tile in China has been conducted (1–3). The study of ancient glazed tile in Nanjing and Beijing has been a particular area of focus, given that Nanjing and Beijing were the capital cities of the Ming and Qing Dynasties, respectively. The ancient glazed tile has been examined for the chemical composition of yellow glaze, the mechanism of peeling glaze, and the degradation of green glaze (4–7). The Ming Dynasty was a key stage in the transition from the mature period of the Song and Yuan Dynasties to the heyday of Ming and Qing Dynasties on the basis of inheriting the technology of the Song and Yuan Dynasties. Glazed-tile firing technology during this period developed a variety of glaze colors, such as yellow, green, red, purple, blue, black, and white, which met the needs of all types of buildings and laid a foundation for the appearance of glazed tiles in the Ming and Qing Dynasties’ official buildings. However, there is relatively little research on the technology of glaze making at this stage, and even less research on the world-famous buildings of the Ming Dynasty, such as the great Bao’ensi site and the Ming Palace. Yinzhong and others studied the chemical composition of glazes, the melting temperature range of yellow glazes, and the matching of the thermal expansion coefficients of tire glazes of six architectural glazed components unearthed from the northern part of the Bao’ensi site in Nanjing (8).

In the Ming Dynasty, the Nanjing Bao’ensi was built in the tenth year of Yongle (AD 1412). Its temple was built in accordance with the standards of the imperial palace, and it was resplendent and magnificent in scale (9). From its completion to its decay, the pagoda has always been the highest building in China. It ranked among the Seven Wonders of the World in the Middle Ages, and it was regarded by Westerners as a landmark building representing China (9). The Ming Dynasty Bao’ensi site is generally divided into two areas: the south and the north. To cooperate with the construction of the “Bao’ensi Temple Site Park,” the Nanjing Institute of Archaeology conducted a comprehensive and systematic archaeological excavation in the northern area from 2007 to 2010 and the southern area in 2017.

Glazed tiles unearthed from this site represent the highest level of technology of glazed tile-making in the Ming Dynasty. For the purposes of studying the firing technology and production level of glazed tiles in the Ming Dynasty, 31 colored glaze samples were selected at the southern district and excavated archaeologically in the gratitude temple site, using an ultraviolet-visible-near-infrared (UV-Vis-NIR) spectrophotometer and energy-dispersive X-ray fluorescence (XRF) spectrometer. The glaze color and chemical composition of glazed tile unearthed from the great gratitude temple were measured, and the results were analyzed by using a multivariate statistical analysis method.

A total of 31 samples were selected from the southern area of the Bao’ensi site, including bobbin and slab tiles. Some samples were well preserved, with yellow, black, green, and peeling glazes. The body color was mainly gray and white with delicate quality. The samples are shown in Figure 1.

FIGURE 1: Images of several glazed tile examples from Bao’ensi in Nanjing.

For chromaticity analysis, a UV-Vis-NIR spectrometer was used to measure the reflection spectrum of the sample, with a 10° field of view during the spectral range of 380–800 nm and a slit width of 5 nm.

A D65 light source was used for testing, and daylight illumination with a color temperature of 6500 K was simulated.

The energy-dispersive XRF (EDXRF) experiments were carried out at the Laboratory of Science and Technology Archaeology at Peking University. An XGT-7000 X-ray fluorescence microscope was used for the experiments. The diameter of the X-ray incident spot was 1.2 mm, the voltage of the X-ray tube was 30 kV, the current of the tube was 0.029 mA, and the data acquisition time was 120 s.

Chromaticity Analysis of Glazed Tiles of Bao’ensi Site

Glaze color is an important expression of glazed tile appearance, and the brightness, saturation, and other related information on glaze color also reflects the level of glazed-tile firing technology. Human eyes have limitations in color discrimination, whereas a UV-Vis-NIR spectrometer is used to measure the reflection spectrum of glaze, and can distinguish the differences and changes in glaze color. To express the glaze color of glazed tiles scientifically and objectively, a color test of the glazed tiles of Bao’ensi was carried out.

The main wavelength of the glaze color corresponds to the peak of the reflectivity curve. Color saturation and brightness are important optical properties of glazes. The main wavelength determines the hue of the glaze, whereas the saturation depends on the selectivity of the glaze to light reflection, and brightness depends on the relative lightness of the glaze (10). The reflectivity of glaze is one of the factors influencing glaze gloss, which affects the specular reflection ability of glaze to visible light. When the reflectivity is high, light is mainly reflected by the mirror, and the glossiness is good. When the reflectivity is low, light is given priority over diffuse reflection, and the glossiness is poor. L* represents brightness or lightness; the larger the value, the lighter the color. The a* represents the red and green phase axes, with positive and negative directions indicated in red and green, respectively; the b* represents the yellow and blue phase axes, with positive direction indicated in yellow and negative direction indicated in blue. The c* represents the chroma, purity, or saturation; the higher the value, the higher the spectral color content; finally, the h* represents the tone angle value (11). The CIE L*a*b* color space values of the green, yellow, and black glaze colors of the glazed tiles of Bao’ensi that were tested and analyzed are shown in Table I.

Comparing the spectral curves of the yellow and green glaze from Figure 2, the main wavelength of the yellow glaze is greater than 800 nm and is located in the NIR region, while the spectral curve rises gently in the shortwave region (12). The wavelength of the green glaze is approximately 540–570 nm, rising sharply in the shortwave within the wavelength range of green light, and the reflection peak is narrow. Black glaze has low reflectivity and no main wavelength.

FIGURE 2: The Vis-NIR reflection spectral curve of glazed tiles of Bao’ensi.

Table I and Figure 3 show the chromaticity values of the samples, and the corresponding scatter plots of L*, c*, and h*. In Figure 3a, the L* value of the green glaze is between 40 and 70. In Figure 3b, the glaze color of the sample is distributed in the yellow–green region, close to the yellow axis, and there is little difference between the hue angle h* and saturation c*. The L* value of yellow glaze is approximately between 40 and 80, and there are samples with similar brightness of glaze color. The glaze color is close to the yellow axis, hue angle h* is not the same, glaze saturation c* is quite different, and some samples are on the high side while others are on the low side. However, there are also samples with the same glaze saturation, among the three glaze colors, with the yellow glaze having the highest saturation. The black glaze L* value (Figure 3a, y-axis) was between 30 to 50. There are two black samples of higher lightness (L*) than other black samples (be70, be81). Note in Figure 3b, that the black sample with higher brightness (be70) also has higher saturation.

FIGURE 3: (a) Scatter plot of L* of sample glaze; (b) Scatter plot of c* and h* of sample glaze.

The chromaticity analysis of glazed tiles at the Bao’ensi site shows that the wavelengths of glazed tile and the samples of the same glazed color were not different. The brightness and saturation of the glazed color samples were high or low, which may be because of the different application positions of the buildings. Therefore, glazed tiles with different glaze color requirements are needed. It can be inferred that in order to meet the needs of the construction of the temple, the workers were skilled in firing glazed tiles with different glaze color requirements, indicating that the glaze color of glazed tiles used was well-established (mature).

Chemical Composition Analysis of Glazed Tile Layer in Bao’ensi Site

EDXRF was used to determine the chemical composition of Al

, CuO, and PbO in the sample glaze. The partial test results are shown in Table II.

The main components of the glazed tile glaze in the Nanjing Bao’ensi were PbO, SiO

O, and CaO, and the main coloring elements were Ti, Mn, Fe, and Cu. Al

 and alkaline oxide to prevent glaze cracking.

To reveal the characteristics and differences in the chemical composition of the glazed tiles at the Bao’ensi site, a distribution analysis of the chemical composition of the experimental results was carried out. Using the content of the chemical components of one element as the abscissa and the content of the chemical components of another element as the ordinate, a two-dimensional (2D) dispersion analysis diagram can be constructed.

Figure 4a shows that the yellow glaze sample has a high content of Fe

, indicating that Fe is the main element affecting the color of the yellow glaze. The Fe

 content in most of the green glaze sample was low, but the CuO content was high, and the Fe

 content in a few samples (be37, be29) was high. Clearly, Cu is the main element affecting the color of the green glaze. The contents of Fe

 and CuO in the black glaze samples were relatively high, so Fe and Cu jointly affect the color of the black glaze. According to the content of Fe

 in Figure 4a, the glaze of the Bao’ensi glazed tile can be divided into two parts: one is green glaze, in which the content of Fe

 is less than 1%, and the other is yellow glaze and black glaze, in which the content of Fe

 is more than 2%. Therefore, it can also be shown that Fe is the coloring element of yellow, black, and green glaze. However, there are certain differences in the chemical composition of the three, which may indicate that the color of the glaze is related to the coloring elements and their content.

FIGURE 4: Distribution analysis diagram of glazed tile glaze of Bao’ensi: (a) is the dispersion analysis figure of CuO-Fe

, (b) is the dispersion analysis figure of MnO-TiO

, and (c) is the dispersion analysis figure of SiO

In Figure 4b, the MnO content of the black glaze sample is relatively high. Based on the analysis of the Fe

 and CuO contents of the black glaze sample in Figure 3a, the main colorants that affect the color of the black glaze are Fe, Cu, and Mn.

In Figure 4c, the PbO content in the green, yellow, and black glaze increases gradually with a decrease in the SiO

 content, and there is a negative correlation overall. SiO

 can improve the melting temperature and viscosity of the glaze; provide a glaze with high mechanical properties (such as hardness and wear resistance); improve the whiteness, transparency, and chemical stability; and reduce the thermal expansion coefficient of the glaze. As a flux, PbO can reduce the thermal expansion coefficient of the glaze, improve the thermal stability, and reduce the melt viscosity, such that the glaze has good fluidity. This can increase the melting temperature range of the glaze, improve the glaze elasticity and gloss, and increase the tensile strength. PbO has the strongest flux and can easily react with SiO

 to form lead silicate with a low melting point. Owing to the high refractive index of lead silicate, it can form a high-gloss glaze (13).

 is 2.15:1 for the green glaze, 2.72:1 for the yellow glaze, and 2.81:1 for the black glaze. In Figure 4c, the ratio (5,9) of PbO to SiO

 in the yellow-glazed tile is slightly higher than that in the Ming glazed tile of the Beijing Imperial Palace; there are also differences in chemical composition, which may be related to the glaze materials used and the production system at that time (5,9).

The Song Dynasty work (14) by Li Jie “Ying Zao Fa Shi (State Building Standards)” has a detailed record. It was also recorded during the Wanli period of the Ming Dynasty: “306 Jin of Huangdan, 102 Jin of Maya stone, 8 Jin of ochre” (15). Huangdan is rich in lead, whereas Luohe stone and Maya stone are rich in silicon. The ratio of Huangdan (PbO) to Luohe stone:Maya stone (SiO

) in the two books is approximately 3:1, indicating that the glaze formula from the Song dynasty to the Ming Dynasty tended to be stable (16).

 of the yellow glaze with the highest grade of glazed tiles in the Bao’ensi site is relatively close. However, it is similar to that recorded in “Ying Zao Fa Shi” and “Gong Bu Chang Ku Xu Zhi.” The 3:1 ratio is somewhat different, which is the main reason for this phenomenon (5). On the one hand, as the initial formula recorded in the literature, Pb volatilizes to a certain extent in the actual firing process, resulting in a decrease in the content of PbO in the ingredients, so that the ratio cannot reach 3:1. On the other hand, as the glazed tiles of the Bao’ensi site had been buried underground for a long time, the Pb content would be continuously dissolved owing to the humidity and abundance of bacteria and microorganisms, resulting in a decrease in Pb content.

FIGURE 5: Lead:silicon ratio of the three glazes.

The comprehensive analysis shows that the firing technology of glazed tiles at the Bao’ensi site has matured, and artisans can flexibly make glazed tiles of different glaze colors by adjusting the proportion of raw materials and adding appropriate flux to meet the needs of the Bao’ensi site buildings.

Factor analysis is a statistical technique used in the analysis of EDXRF data for tile analysis. It allows for the identification of underlying factors (components) that contribute to the observed variations in the elemental composition of the glazed tiles. By analyzing the covariance matrix of the EDXRF data, factor analysis can extract the common sources of variation and determine the relationships between the measured elements in the different glaze types (17).

To directly observe whether there is any difference in the chemical composition of glazes of different colors, 31 samples from Bao’ensi site were analyzed using the factor analysis method.

Figure 6 shows the load diagram of the yellow glaze, green glaze, and black glaze of the Bao’ensi site. The seven colored elements were selected, such as Al

, and CuO. The primary and secondary quantitative element composition factors extracts two factors, F1 and F2, with a cumulative contribution rate of 77%, as shown in equations [1] and [2].

FIGURE 6: The factor analysis diagram of glaze.

As can be seen from the factor analysis diagram, the distribution of most data points of the green, black, and yellow glazes was concentrated and separated from each other, indicating that the glaze formulas of the three glazes were significantly different according to the coloring elements. In terms of the dispersion of data points, the yellow and black glaze were relatively stable in terms of raw material composition and process formula, whereas the data points of the green glaze were relatively dispersed, indicating that the glaze formula may be different.

The be54 sample in the yellow glaze has a distant relationship with other yellow glaze samples, but it is close to the green glaze samples. According to EDXRF and Figure 4a, it can be found that not only are the contents of Fe, Cu, and other elements in be54 glaze not very different from those of other yellow glazes, but that the contents of PbO are lower than those of other yellow glazes. It was also observed that the contents of SiO

 were higher than those of other yellow glazes. The sum of the SiO

 and PbO contents in the glaze accounted for more than 90% of the glaze formula. It was speculated that the be54 yellow glaze sample may have been caused by a lack of strict formula control.

The be30 and be43 samples in the green glaze are far from the other green glaze samples. According to EDXRF and Figure 4c, it can be found that the PbO content of be30 and be43 samples is low, and the SiO

 content was high, but the total was over 90%. Therefore, it was inferred that glaze color modulation technology has been relatively mature to meet the different requirements of glazed tiles.

The chroma and chemical composition of the glazed tile glaze of the Bao’ensi site were analyzed, and the following conclusions were drawn:

The firing technology of glazed tiles in the great Bao’ensi is relatively mature, and craftsmen could already skillfully fire glazed tiles with different glaze color requirements according to their needs. There are no differences in the main components and flux content of various glaze formulations. Different glaze colors are caused by different types and contents of coloring elements. Cu and Fe are the main elements that affect the color of yellow glaze, Cu is the main element that affects the color of green glaze, and the main coloring elements that affect the color of black glaze are Fe, Cu, and Mn. Yellow and black glaze are relatively mature in terms of raw material composition and process formula, whereas green glaze is not stable enough in terms of raw materials and process. The glaze formulations of the green glaze of the great Bao’ensi have both similarities and certain differences, indicating that the glazed tiles of the great Bao’ensi may have come from the same kiln, using different glaze formulations, or they may have been the result of later repairs.

This work was supported by National Natural Science Foundation of China (11975210), and the National Social Science Fund of China (17CKG016).

(1) Jiang, W. Y.; Liu, Y. L.; Sun, W. On the application of glazed tiles in Forbidden City buildings. 

Glaze Technology of Chinese Ancient Architecture

; China Architecture & Building Press, 1987.

(3) Yang, G.; Gao, S.; Wang, R. Z.; et al. The glaze color of ancient Chinese architecture. 

Studies in the History of Natural Sciences

(4) Kang, B. Q.; Li, H.; Miao, J. M. Research on the composition and related problems of the glazed tile glaze of the Imperial Palace in Qing Dynasty. 

(5) Li, H.; Ding, Y. Z.; Chen, T. M.; Miao, J. M. An Analysis on the Yellow Lead-glaze of Architectural Tiles of the Ming and Qing Dynasty in Beijing. 

China Cultural Heritage Scientific Research

(6) Miao, J.M.; Wang, S.W., Duan, H.Y.; et al. Research on the Elements in the Vitreous Glazed Components Used in Ancient Chinese Buildings. 

(7) Duan, H. Y.; Miao, J. M.; Li, Y.; Kang, B. Q.; Li, H. Research on Diseases & Damages to Green-Glazed Tiles on Chinese Traditional Architecture. 

(8) Ding, Y. Z.; Li, H.; Duan, H. Y.; et al. Science and technology research on glazed tiles of Bao’ensi Pagoda in Nanjing. 

(9) Leng, M. Analysis on the glaze pagoda of the Temple of Bao’ensi in Ming Dynasty. 

Identification and Appreciation to Cultural Relics

(11) Wang, C. W.; Zhao, W. J.; Wu, B.; Zhang, B. Scientific study on the celadon glaze from the Donggou kiln, China. 

Nucl. Instrum. Methods Phys. Res., Sect. B

(12) Li, Y.; Zhang, B.; Cheng, H. S.; Zheng, J. M. Revealing the coloration mechanism in the earliest Chinese celadon glaze. 

; Bibliographic Literature Publishing House, 2000.

(16) Hou, J. Y.; Li, H.; Kang, B. Q.; et al. The origins of imperial yellow glazed porcelain in the Ming Dynasty (1,368 to 1,644) China: Technical comparison to low-fired tile with yellow glaze. 

SPSS Statistical Analysis Method and Application

; Publishing House of Electronics Industry, 2013.

 are with the School of Physics and Microelectronics at Zhengzhou University, in Zhengzhou, China. 

 is with the Nanjing Research Institute of Archaeology, in Nanjing, China. 

 is with the School of Archaecology and Museology at Peking University, in Beijing, China. 

 is with the School of History, Zhengzhou University, Zhengzhou, China. Direct correspondence to: 

Articles

UV-Vis-NIR can be used to understand how ancient buildings were constructed. Here, a UV-Vis-NIR and EDXRF spectrophotometer were used to analyze glazed tiles that comprised a historical site built in Ancient China.

Technological Analysis of Glazed Tiles Unearthed from Bao’ensi in Nanjing

“Ru-type ware” and unglazed firing bodies were unearthed from the Qingliang Temple Kiln in Baofeng County in the Henan province of China in 2014; the excavation site was close to the central firing area of the ancient Ru Kuan Kiln. Ru-type ware has several similar characteristics to Ru Kuan ware. Here, four samples of unglazed firing bodies, 10 samples of sky green Ru-type wares, and seven samples of sky green Ru Kuan wares were selected. The chemical compositions of 16 elements in body and glaze were measured by energy-dispersive X-ray fluorescence (EDXRF), and multivariate statistical analysis was conducted to determine the classification relations between Ru-type and Ru Kuan wares. Results indicate that the source of raw material and glaze formula of sky green Ru-type are different from Ru Kuan wares. For the sky green Ru-type ware, the raw materials for body is concentrated, and the glaze formula is stable, which conforms to the characteristics of official celadon.

The Ru Kuan kiln is one of the top five famous kilns of the Song Dynasty, and it was reserved for the Imperial Court. The Ru kiln produced an extremely rare type of pottery and porcelain, called Ru Kuan ware, that was only created for the Imperial Court in the Northern Song Dynasty (1). Ru Kuan ware pottery was famous for its delicate body quality, its exquisite workmanship, and its bright and smooth glaze color. The Ru Kuan kiln existed only approximately 20 years, prospering as a result of the continuous war between the Song and Jin Dynasty at the end of the Northern Song Dynasty. The result of this war led to the rarity of products handed down from ancient times and made this material even more precious.

Qingliang Temple, located in Baofeng County in the Henan Province of China, was identified as the historical site of the official Ru kiln of the Song Dynasty in 2000 (2). From March 2014 to September 2014, the Henan Provincial Institute of Cultural Heritage and Archaeology conducted archaeological excavations in the central firing area and surrounding areas of the Ru Kuan kiln. A kiln furnace for unglazed firing and a small amount of unglazed firing shards were unearthed in the stratum of the nearest exploration pit (Figure 1) to the central firing area of Ru Kuan kiln (3). The stratigraphic relationship showed that the kiln furnace was built later than the mature period of the Ru Kuan kiln, which belonged to the late Northern Song or Jin Dynasty. In the same layer, another celadon-glazed porcelain was also unearthed; its body had fine and solid quality, with a greyish-white color and a little bit of the incense ash color normally associated with Ru Kuan ware. The celadon-glazed porcelain also contained the similar glaze color of sky green and grass green that was historically associated with Ru Kuan ware porcelain. Although the glaze color was bright and the glass texture is obvious, its jade texture was not as good as that of Ru glaze. This material was temporarily named “Ru- type ware” to distinguish it from Ru porcelain and also to acknowledge the several similarities that the two types of pottery possess (4). Ru-type ware is concentrated in the pit under the Yuan Dynasty layer (Layer 3) and the accumulation of the Jin Dynasty layer (Layer 4), and the stratigraphic age is also later than the mature period of the Ru Kuan kiln. Moreover, the Ru-type ware is large in size, with a regular shape, solid body quality, and a pure glaze color. Ru-type ware is also produced by full-glaze and pad firing. According to the standards for judging official and civil use, Ru-type ware should be for official use (5).

FIGURE 1: Stratigraphic profile of T85 exploration pit. Note: (1) Modern living and cushion soil layer, (2,3) Yuan and Ming Dynasties layer, (4) Jin Dynasty layer, and (5) late Northern Song Dynasty layer.

Amongst the celadon glaze color of Ru Kuan kiln, the sky green color is the most precious, as it represents the highest firing level of Ru Kuan ware. In this paper, the energy-dispersive X-ray fluorescence (EDXRF) technique is used to measure the chemical compositions of body and glaze samples of unglazed firing, sky green Ru-type ware, and Ru Kuan ware (Figure 2). The experimental results were subjected to multivariate statistical analysis to obtain the relationship between the raw material sources and glaze formulas of unglazed firing, sky green Ru-type ware and sky green Ru Kuan ware. Both Ru Kuan ware and Ru-type ware are very precious archaeological samples and there was no relevant research on Ru-type ware until now. This study lays the foundation for the preservation of samples and subsequent research studies. Furthermore, it provides a scientific basis for the firing age of Ru-type ware, identification of Ru Kuan and Ru-type wares, and research on the Ru kiln as well as the Northern and Southern Song Dynasties.

FIGURE 2: The figures of the typical samples. (a, b) Sky green Ru Kuan wares. (c, d) Sky green Ru-type wares.

This experiment was conducted with the X-ray fluorescence spectroscopy elemental analysis system in the super-large sample room (Eagle III XLL μ-Probe) of the Institute of High Energy Physics Chinese Academy of Sciences. The spectrometer is equipped with a molybdenum tube and a 125 μm beryllium window that has an incident beam angle of 65° and an emergence angle of 60°. A silicon–lithium X-ray detector with a beam spot diameter of Ф = 1 mm, operating voltage of 40 kV, operating current of 250 mA, and the energy resolution of 160.3 eV, was used in this study because it can analyze the elements of 

U. To reduce the absorption of the low-energy secondary X-rays produced by the low-Z elements, the spectrometer was equipped with a vacuum chamber.

The relative position of the sample surface, the detector, and the light outlet were fixed, and the samples were positioned to the fixed focus of the analysis system accurately with the three-dimensional (3D) stepping motor, and high- and low- magnification charged-coupled device (CCD) cameras. The fundamental parameter (FP) method was used for nondestructive quantitative analysis of unglazed firing, sky green Ru-type and Ru Kuan ware body and glaze samples for 16 elements, such as sodium oxide (Na

O), magnesium oxide (MgO), aluminium oxide (Al

O), calcium oxide (CaO), titanium dioxide (TiO

), manganese(II) oxide (MnO), and iron(III) oxide (Fe

). To calibrate the matrix effects and improve the accuracy of the data, we used a set of ceramic reference materials for EDXRF by the Institute of High Energy Physics. The data were processed using the program version 32, which was provided with the equipment. Through discriminant analysis and hierarchical cluster analysis (HCA) in multivariate statistical analysis, the relationship between the chemical components was studied (6). The classification of unglazed firing, sky green Ru-type, and Ru Kuan ware body and glaze samples were obtained.

Four samples of unglazed firing, 10 samples of sky green Ru-type ware, and seven samples of sky green Ru Kuan ware shards were selected for this experiment. The shards were provided by the Henan Provincial Institute of Cultural Heritage and Archaeology, and those samples were from accurate and reliable sources. The selection of samples is shown in Table I. The glaze color of Ru-type and Ru Kuan wares mentioned in this paper is sky green and will not be repeated hereinafter.

The chemical composition of 16 elements of body and glaze samples of unglazed firing, Ru-type, and Ru Kuan ware was measured by EDXRF. The contents of seven elements, including Al

Body Raw Materials by Discriminant Analysis

The main raw material of celadon body in northern China is clay, which is comprised of many minerals. The oxide contents in clay minerals would vary in different geological formation conditions and include a small amount of alkaline-earth metal oxides, such as Fe

, in addition to hydrous aluminosilicate (7). Fe

 are usually dispersed in clay as fine grains, and their combinations combine and form compounds, such as FeO·TiO

 at high temperatures; thus, making the bodies show different shades of gray. The Fe

 content was taken as the horizontal coordinate and the TiO

 content as the vertical coordinate. A 2D dispersion analysis chart was made, as shown in Figure 3. Table II and Figure 3 show that the contents of Fe

: 1.2%~1.38 wt%), except for RG14B, were higher than those of Ru-type ware (Fe

: 1.11~1.16 wt%) bodies. This finding indicates that the source of bodies of Ru Kuan ware is different from that of Ru-type ware and unglazed firing, which may have used clay materials with higher hematite, ilmenite, or rutile minerals. The RG14B sample has a higher TiO

 content than unglazed firing and Ru-type ware, and the Fe

 content is only slightly lower than that of one Ru-type ware sample (QL15B). RG14B shows that the contents of Fe

 of Ru Kuan ware bodies are higher than that of the Ru-type ware and unglazed firing bodies. Ru-type ware and Ru Kuan ware can be classified by the iron and titanium contents of bodies.

 content range of Ru-type ware and unglazed firing body overlaps to some extent, the overall content of Fe

 of Ru-type ware was higher than that of unglazed firing. This result shows that the two may have differences in the choice of raw materials, but they may have the same sources of raw materials. However, the content of iron minerals is reduced in the raw material treatment processes, such as elutriation, in the preparation process of unglazed firing. The differences in Fe

 content imply that the level of raw material composition and origin result in that the body of Ru Kuan ware is incense ash color, Ru-type ware is greyish white, and unglazed firing is white.

Discriminant analysis can be used to train the discriminant function based on the different eigenvalues and the category to which the research object is known to belong; when a sample with an unknown category is entered, the sample is classified according to the similarity level between it and the discriminant function (8). In this experiment, Fisher discriminant method was used to analyze the content of all chemical components of 25 body samples. Figure 4 shows a two-dimensional (2D) visual classification scatter diagram. The body samples of unglazed firing, Ru-type ware and Ru Kuan ware are concentrated in three areas of A, B, and C, respectively. This condition is ideal and indicates that the raw material sources of the three kinds of body samples are completely different. Although unglazed firing and Ru-type ware belong to the same excavation layer, their sources of body materials differ, which is caused by differences in the origin of raw materials or firing batches. The two categories differ from those of Ru Kuan ware, which shows that the two differ in terms of body material selection for the porcelain made in the Ru Kiln. The Ru-type ware data points in area B and the Ru Kuan ware data points in area C are concentrated, indicating that sky green Ru-type ware body is strictly controlled in raw material selection and craftsmanship, which proves it was used for official occasions.

FIGURE 4: Discriminant analysis of bodies.

Glaze Formula by Dispersion Analysis and Hierarchical Cluster Analysis

The content of oxides of Ru-type and Ru Kuan ware glazes is shown in Table III. After the molar coefficient of alkaline oxide (RO+RO

) is adjusted to 1, the Seger formulas of Ru-type and Ru Kuan ware glazes are 1.1 and 1.2, respectively.

 are the colorants of ceramic glaze, with iron being the main coloring element of celadon glaze in northern China (9). Figure 5a shows the 1D dispersion analysis diagram of the Fe

 content of Ru-type and Ru Kuan ware glazes is at the range of 1.85~2.18 wt%, which shows that the iron mineral content in the glaze formula of the two is close. Manganese and titanium are not the main coloring elements, but they have certain influence on the hue of celadon. MnO and TiO

. Figure 5b shows the 1D dispersion analysis diagram of MnO and TiO

. The MnO contents of Ru-type and Ru Kuan ware glazes fluctuate around 0.11 wt%, and the content of each glaze sample is close to that of others. On the whole, the TiO

 content of Ru-type ware is higher than that of Ru Kuan ware glazes, which indicates that the mineral content of titanium of Ru-type ware glaze is higher, or some ilmenite is involved in the glaze ratio.

FIGURE 5: (a) 1D dispersion analysis of Fe

 accounts for 81~85.7 wt% and 78.9~80.1 wt% of the total oxides of Ru-type and Ru Kuan ware glaze samples, respectively. The two oxides with a great deal of fluxing agent chemically react to form glaze at high temperature (9), and the molar ratio of the two contents can intuitively characterize the changes in glaze formula and process technology (10). Figure 6 shows the dispersion analysis of the SiO

 contents and the silica-alumina ratio for Ru-type and Ru Kuan ware glazes. The molar silica-alumina ratios are listed in Table III. Except RG20G, Ru Kuan ware is concentrated in area C, and Ru-type ware is concentrated in area D. The comparison shows that Ru Kuan ware glaze has high aluminum and low silicon content, whereas Ru-type ware glaze has high silicon and low aluminum element content, indicating that Ru-type ware is different from Ru Kuan ware regarding its glaze formula. Ru Kuan ware is glazed with agate. A rich agate ore is observed near the site of the Qingliang Temple Kiln (11). Agate is a cryptocrystalline mineral of SiO

, which is added into the glaze to achieve the similar function to quartz. The SiO

 content of Ru-type ware glaze is higher than that of Ru Kuan ware glaze, which might result from the higher quartz content in the glaze formula, or the similar glaze technology with agate in Ru Kuan ware to some extent, but the amount used is higher than that of Ru Kuan ware glaze. Al

 is mainly brought by clay and feldspar, and the content of Al

 of Ru Kuan ware glaze is higher than that of Ru-type ware. Therefore, the Ru Kuan ware used high aluminous clay in the formula, and a larger amount of feldspar is found in the formula. Although RG20G is not in area C, its SiO

 contents are consistent with the characteristics of high aluminium and low silicon contents of Ru Kuan ware.

. (insert) 1D dispersion analysis of n(SiO

Table III and Figure 6 show that the molar radio n(SiO

) of Ru-type ware glaze ranges from 8.29 to 8.99, higher than that of Ru Kuan ware glaze with the range from 6.26 to 7.18, which also shows that the glaze formulas of the two differ. Moreover, the silicon:aluminium ratio fluctuation range of the two is relatively small, which implies that the glaze formula is relatively stable and also proves that Ru-type ware has been used for official occasions.

The traditional high-temperature glaze in China has always been glazed by limestone and plant ash, amongst which the fluxing agents, such as CaO, MgO, and K

O, can reduce the melting temperature of glaze. Their contents and types also have a decisive influence on the melting temperature and appearance characteristic of glaze. CaO in glaze mainly comes from limestone and calcite (main ingredient is CaCO

), whereas MgO mainly comes from dolomite (CaCO

). Figure 7a shows a 2D dispersion analysis of calcium and magnesium oxides. A positive correlation is observed between the CaO and MgO contents of Ru-type and Ru Kuan ware glazes, which may be associated with the dolomite composition of the initial mixed glazes. In addition, the contents of CaO and MgO of Ru Kuan ware glaze are higher than those of Ru-type ware glaze, which may be caused by higher content of anorthite, limestone, and dolomite in glaze.

FIGURE 7: (a) 2D dispersion analysis of CaO and MgO, (Insert) showing CaO and MgO content of various sample numbers. (b) Box plot of K

O as a fluxing agent is mainly from potassium feldspar, and ice cracks can be produced on the glaze when the content is high because of the larger coefficient of expansion. P

 as a kind of opacifier that can result in varying degrees of opacified texture on the glaze surface. P

 basically comes from plant ash, and the content of varies considerably from species to species. Figure 7b shows the box plot of K

 of Ru Kuan ware glaze are higher than those of Ru-type ware glaze on the whole. Therefore, Ru Kuan ware adopts the glaze formula with higher potassium feldspar ratio, while Ru-type ware is glazed with plant ash with less P

Hongjie Luo (12) proposed a reference standard for classifying Chinese ancient porcelain glaze based on the coefficient b of wood ash glaze formula, which is the molar ratio of alkali-earth metal oxide (RO) to the sum of alkali metal oxide (R

O) and alkali-earth metal oxide. This classification consists of three groups: calcium glaze (

 ≥ 0.76), calcium–alkali glaze (0.76 > 

 ≥ 0.5) and alkali–calcium glaze (b < 0.5). The calculated 

 values of each sample are listed in Table III, and the box plot is shown in Figure 8. The results show that the 

 values of Ru-type and Ru Kuan ware glazes are all above the dividing line 

 = 0.76, which indicates that both are high calcium glazes. The 

 values of the two are relatively close (the range of 

 values is 0.76~0.82 for Ru-type ware and 0.78~0.81 for Ru Kuan ware), which indicates that Ru-type ware may follow the glaze-making technology of Ru Kuan ware to some extent. With the small fluctuation range of the relatively stable glaze formula, Ru-type ware contains the characteristics of official celadon.

 values of Ru-type ware and Ru Kuan ware.

HCA classifies the samples into groups and subgroups according to the similarity and correlation of the samples (13). The HCA of Ru-type and Ru Kuan ware glazes is shown in Figure 9, and can be divided into two categories.

FIGURE 9: The HCA of Ru-type ware and Ru Kuan ware.

The first category includes all Ru Kuan ware glaze samples from RG14G to RG20G, indicating that the formula of Ru Kuan ware glaze in this category is relatively stable.

The second category includes all Ru-type ware glaze samples from QL11G to QL19G. The formula of Ru-type ware glaze in this category has also become mature and relatively stable, which conforms to the properties of the official glaze formula. Although sample QL11G is slightly far from the nine other samples of Ru-type ware, it can still be classified into one category. On the whole, the formulas of Ru-type and Ru Kuan ware glazes are completely different.

The raw material sources of unglazed firing, sky green Ru-type ware and sky green Ru Kuan ware bodies differ. Although unglazed firing and sky green Ru-type ware belong to the same excavation layer, their sources of body materials differ. The body of sky green Ru Kuan ware is made of clay raw materials with minerals, such as high hematite, ilmenite, or rutile.

Sky green Ru-type ware is different from sky green Ru Kuan ware in terms of glaze formula. The high content of quartz in sky green Ru-type ware glaze and the high content of feldspar in sky green Ru Kuan ware glaze show that the element composition characteristics of the former are a high ratio of silicon and low aluminium, while those of the latter are high ratio of aluminium and low silicon. Moreover, the sky green Ru Kuan ware glaze is glazed of plant ash with high P

 content, and the glaze contains higher limestone, dolomite and potassium feldspar. The content of coloring elements iron and manganese of the two glazes are close, whilst the content of TiO

 of sky green Ru Kuan ware is higher. Ilmenite is involved in the glaze ratio.

The glaze and body of sky green Ru-type ware are strictly controlled in raw material selection and craftsmanship, which proves that it has been used for official occasions. The sky green Ru-type and Ru Kuan ware glazes are high calcium glazes with the relatively close b values and small fluctuation. The glaze formulas of the two are stable and concentrated, which belong to the characteristics of official celadons.

The sky green Ru-type and Ru Kuan wares are classified by the Fe

This work was supported by National Natural Science Foundation of China 11275173 and 11975210.

(1) W. Li, J. Li, Z. Deng, J. Wu, and J. Guo, 

(2) W.J. Zhao, Z.J. Wu, G.X. Li, M. Guo, J.Z. Xie, X.K. Lu, X.M. Sun, S.L. Feng, and M.S. Guo, 

Archaeological Discoveries of Ru Kiln in Baofeng Qingliang Temple, Chinese Ru Kiln academic proceeding

 (Henan People’s Publishing House, Zhengzhou, China, 2017). 

(4) B. Wu, W.J. Zhao, S.L. Feng, H. Zhao, X.M. Sun, and M.S. Guo, 

(5) X.M. Sun, H. Zhao, J.L. Zhao, T.J. Wang, Z.J. Niu, W.B. Wang, and H.W. Lu, 

 (Materials Information, ASM International, Novelty, Ohio, 1996).

 (Shanghai People’s Fine Arts Publishing House, Shanghai, China, 2000).

(10) Y.Z. Ding, J.Y. Hou, H. Li, J.M. Wu, and X.M. Sun, 

History of Science and Technology of China (Ceramic Volume)

Chinese Ancient Ceramics and Mutivariate Statistical Analysis

 (China Light Industry Press, Beijing, China, 1997).

(13) R.A. Ikeoka, C.R. Appoloni, M.A. Rizzutto, and A.M. Bandeira, 

 are with the School of Physics and Microelectronics at Zhengzhou University in Zhengzhou, China. 

 are with the Institute of High Energy Physics Chinese Academy of Sciences in Beijing, China. 

 is with the Henan Provincial Institute of Cultural Heritage and Archaeology in Zhengzhou, China. Direct correspondence to Weijuan Zhao: 

As this study demonstrates, energy-dispersive X-ray fluorescence (EDXRF) and multivariate statistical analysis can be used to distinguish different classes of historical artifacts, such as ancient pottery—revealing insights about theirs origin and uses.

Weijuan Zhao

Articles

Technological Analysis of Glazed Tiles Unearthed from Bao’ensi in Nanjing

Studying the Source of Raw Material and Glaze Formula of Sky Green “Ru-type Ware” and Ru Kuan Ware by EDXRF