Artificial coloration of ancient agate beads: a mineralogical study

The process of staining was frequently employed to enhance or alter the color of agate beads in ancient times. One of the key challenges in studying ancient beads is comprehending the intricate techniques employed to color agate stones. An understanding of the staining mechanism from a mineralogical standpoint offers insights into the level of technological advancement in different civilizations. In this study, the mineral structure of eight ancient agate beads from Xinjiang Uygur Autonomous Region, NW China, was analyzed using Micro X-ray fluorescence (µXRF), Raman spectroscopy, Scanning Electron Microscope (SEM), and Fourier Transform Infrared (FTIR) techniques. The color, transparency, mineral phase, and surface roughness of the beads were examined, revealing variations ranging from colorless to light violet to dark violet. Raman and FTIR spectroscopy were employed to determine the SiO2 phase and the changes in optical characteristics of agate beads after artificial staining. The black color of the beads was formed by carbon penetration, while the red color was produced by heating. The coexistence of α-quartz and moganite phases in the red, the dark red, the black, the idiochromatic white and the part translucent zones of the ancient beads was confirmed by the 464 cm−1 peak of α-quartz and the 502 cm−1 peak of moganite phase. The analyzed red, the dark red, the black, the idiochromatic white and the part translucent zones exhibited remarkably similar FTIR spectral features, with two prominent bands at ~ 1097 and ~ 1187 cm−1, as well as two weak bands at 798 and 778 cm−1, indicating the presence of moganite and α-quartz in the unetched ancient beads. In contrast to the idiochromatic white appearance of natural agate, the scattered white coloration in etched beads was generated by an etching reaction. Both Raman and FTIR spectroscopy indicated the absence of moganite in etched beads, indicating that the scattered white color was produced by the loss of moganite and a portion of α-quartz, resulting in a rough surface.

Previous studies decades ago [2,3,8,[21][22][23][24][25][26] have illustrated the shape, design, drilling technique, and color of the ancient beads.The beads chosen as ancient ornaments obviously underwent a series of different handicraft staining, which may indicate a symbolic and economic value.Previous studies have also found that the designed pattern changes over time and excavated regions, although there may be commercial activity conducted among the ancient civilizations at different locations [5,10,[27][28][29][30].Hence, excavated beads from different historical periods reflect craftsmanship advances and indicate craft skills development.
To a certain extent, the color styles of the ancient beads affect the appreciation of the people of the spreading destination through cultural infiltration.With the development of the urban society and technical expertise, natural pattern with simple surface polishing cannot express the motif of the elite communities [29].The beads have changed dramatically during their spreading period, which covers the Neolithic Period around 7000 B.C. to the Early Historic Period [16,29,31].One of the essential characteristics is the color, which can be correlated to the diverse line patterns of the beads.With the development of manufacturing, advances in agate beads-making technology resulted in new colors and various color patterns, such as white, red, and black [1,4,32,33].Pioneering works have divided colors of ancient beads into three color types: type I white-on-red, type II black-on-white, and type III black-on-red.Type I is the most common type, while black-on-white was less discovered.Type III was extremely rare.Artificial color patterns of the ancient beads can reflect the source area, production age, and handicraft development level.Differentiating the artificial origin color from natural color is the priority in the coloration research of the ancient beads.Therefore, interpreting the mineral character and the methods of manufacturing color is crucial for further research.
Agate mainly consists of microcrystalline quartz with a chemical composition of SiO 2 and Mohs hardness of 6.5-7.Naturally produced agate can show various colors with waxy lustre, generally colorless or translucent due to the difference of various trace elements [34][35][36].It is usually a difficult task to differentiate the manufactured color and idiochromatic color by optical observation.For the black-on-white type, it was considered that the beads were whitened entirely at the first step and then drawn with a metallic solution [4,16,37].Humancontrolled heat treatment plays a significant role in the red coloration manufacturing process [22].Experimental research showed that the white line of type I comes from alkali solution etching [4,16,31,37].However, SiO 2 was believed to have low reactivity.The terms etching, bleaching, and painting were used when describing the manufacturing mechanism of the white color [2,4,16,38].How do different mineral phases and colors respond differently to artificial staining?Illustrating the staining mechanism allows us to analyze humans' manufacture and intellectual level at that time [22,24,30,33].
Raman and FTIR spectroscopy are non-destructive and have been employed to identify mineral polymorphs and analyze mineral structure [17,[39][40][41][42][43][44].The µXRF allowed us to in-situ identify the element composition of the mineral.SEM helps to investigate the coarseness of the beads' surface.Here, we studied eight ancient beads with different color types by µXRF, FTIR, Raman, and SEM techniques at a micrometer scale.The staining mechanism was revealed from a mineralogical perspective.

Samples
The Jirzankal Cemetery, which is approximately 2500 years old, takes its name from the tableland on the eastern Pamir Plateau where it is located.Situated in SW Xinjiang, China, the cemetery shares borders with Tajikistan, Afghanistan, Pakistan, and India to the west, providing significant potential for western connections.During the extensive excavations carried out between 2013 and 2015, a range of exotic objects were discovered in the cemetery.For example, musical instruments such as harps found in burials M14 and M16 likely originated from the Near East, while silk fabrics discovered in burials M1 and M2 may have been transported from the eastern part of China [45].Additionally, a variety of bone, glass, and agate beads discovered in the cemetery suggest broader connections with various cultures.Of these, 81 agate beads marked the cemetery as the largest bead-buried cemetery of its time in China.These beads exhibit various colors, both natural and artificial, and display intricate designs, providing valuable insights into the coloration of ancient agate beads.

Instrumentation and operating conditions
Raman spectroscopy was performed at room temperature using a Witec alpha300R confocal Raman microscope at the Institute of Geology and Geophysics, Chinese Academy of Sciences (IGGCAS).The laser beam was focused on the sample surface using a 50 × Zeiss microscope (NA = 0.75).The Raman data has been calibrated with a silicon peak of 520.7 cm −1 .The spectra were collected with a 488 nm laser with a spectral acquisition time of 1 s and 60 accumulations for each measurement.The investigated spectra range from 100 to 4500 cm −1 using 300 grooves/mm optical grating.At least ten measurements were operated for one sample.
FTIR spectra were collected using a Bruker Vertex 70 V spectrometer at IGGCAS.A Hyperion 2000 IR microscope with an MCT detector cooled with liquid nitrogen connected to the Bruker Vertex 70 V spectrometer was used to obtain the microscopy IR spectra.Spectra were acquired from a single 100 × 100 μm spot on each analysed point.Background spectra were collected from a gold mirror.The FTIR spectra were recorded using reflectance mode with the co-addition of 64 scans between 650 and 4000 cm −1 at a resolution of 1 cm −1 .
A Zeiss Gemini 450 electron microscope instrument at IGGCAS was used to study the morphology of the surface of the beads.The SE images were collected from the original surface.Samples for SEM measurement were not coated with carbon or any metal.The SE images were acquired at 2 kV accelerating voltage and beam current of 200 pA and working distance of 8.5 mm.The SEM was carried out under 3.01 × 10 −6 mbar vacuum conditions.
MicroXRF experiments were performed on the Bruker M4 TORNADO plus µXRF spectrometer at IGGCAS, which enables the detection and analysis of the entire element range from carbon to americium.A dual silicon drift detector (30 mm 2 ) measures X-rays with a resolution < 145 eV measured on Mn-Kα at a spatial resolution of∼20 µm [46].In this study, all the scanning µXRF element mapping experiments were performed with an X-ray tube energy of 50 kV and a current of 600 µA, with 8 ms per pixel spectrum acquisition time and a pixel step-size of 18 µm.Data analyses, including obtaining elemental maps on all objects and quantitative XRF analysis from zone of interest, were undertaken with the characterization software provided by Bruker Micro Analytics.

Optical appearance
Table .1 shows the optical characteristics of the beads.Our visual observations of all the studied beads showed roughly faceted oval shapes with rounded edges.The most noticeable characteristic of the beads is the color and pattern.The dark red and black beads were transparent and showed a dull lustre, while the dark red and black beads were part translucent and showed a greasy lustre.The B and C beads feature a translucent zone located at the head of the beads(yellow rectangle schematic in Figs. 1, 2).Of all the color types, the whiteon-red variety was the most common.According to Beck's division of the prevailing era of the beads type, their pattern characteristics show the producing age of  approximately 2300 years, which is consistent with the dating results of the cemetery [2,12,14,47,48].The carnelian beads E, F, G, and H were translucent and showed a vitreous lustre.They varied from white ('scattered white' or 'idiochromatic white' , see Fig. 1 and Section "Idiochromatic white and scattered white") to light red to dark red to black (Table 1).
Figure 2 shows that the scattered white zones of samples A, B, and G are pits.The light red, the dark red, the black, and the idiochromatic white zones are flat and smooth compared with the scattered white zone.The scattered white line covers the flat surface with a distance range from approximately 0.5 mm (Sample G) to 6.0 mm (Sample B).The idiochromatic white gel layers and the cavity of the sample D are seen without artificial alteration [49,50].The extension trend of the white layers with different degrees of whiteness was cut by the edge of the black zone (Fig. 2-b).µXRF All samples contained Si, O, Na, K, Ca, and Fe.Si and O account for 98wt.% of the elements in weight percent (Table 2), indicative of the chemical composition of the α-quartz mineral [51].The scattered white zone contains significantly more Na and Ca than other color zones.This artificial staining process must have used some Na-containing and Ca-containing liquid-for example, plants and calcium carbonate minerals.K is evenly distributed across samples at a lower content.Based on the µXRF data (Table .2) and optical microscope observation (Figs.1,2), Fe is uniformly distributed in samples A and B, while the Fe measured by the other four carnelians (E, F, G, and H) show irregular distribution on its surface.Therefore, higher Fe content (0.04%) was observed (Table 2).

Raman spectra
All beads exhibited the characteristic Raman peaks of α-quartz at 464 cm −1 and weak peaks at 128 and 207 cm −1 .The 464 cm −1 is the strongest Raman peak assigned to the vibration of the SiO 4 six-membered ring, and the 128 and 207 cm −1 peaks were assigned to Si-O-Si bending vibrations of α-quartz microcrystalline [52].The shoulder at 502 cm −1 corresponds to symmetric stretching vibrations of the four-membered Si-O-Si ring, which was proved to be moganite by previous Raman studies [52][53][54][55].All the Raman spectra in Fig. 3 and the spectra of the idiochromatic white zone of B-b, C, and D (Fig. 4) confirmed the presence of moganite.This indicates that the moganite and microcrystalline quartz coexists in the original agate stone materials of beads.According to the integral ratio of the 502 cm −1 and 465 cm −1 bands [56], the moganite content of all the studied samples is in the range 0 to 50wt%.
The distinct Raman peaks demonstrate that the black zone of samples A and D is formed due to artificial alteration without breaking the main chemical bond (Figs. 2, Fig. 2 a, c, d  3, 4).Also, two relatively strong bands were distinctly observed at 1380 and 1580 cm −1 , which are absent in the spectra of most of the red and white zones.These features can be interpreted as the D and the G peaks of disordered carbon [57,58].The minor carbon content could account for the dark red and black color of A and D samples, respectively.For the spectra of the white zone, no carbon was detected except sample B.
In the part translucent zones of B and C (Dashed rectangle symbol shown in Figs. 1, 2), the intensity of the Raman peaks is much stronger than any other zone (Fig. 4), indicating a low-defect quartz structure with less intercrystalline porosity [56].As a result, a zone with a higher crystallinity degree could accommodate less exotic elements.This indicates that the crystallinity degree of the translucent zones is much greater, which in turn prevents the staining process.
Selected Raman spectra collected from the scattered white zone of the beads are shown in Fig. 4. It can be seen that the moganite peak is absent.However, the moganite peak was still observed from spectra collected from the idiochromatic white zone of samples C and D.
Raman spectra of the irregularly distributed red dots at the surface of the sample E (Fig. 5) confirmed the presence of hematite, with peaks centered at 229, 299, and 409 cm −1 [59,60].

FTIR spectra
The red, the dark red, the black, the idiochromatic white and the part translucent zones of all of the samples show similar FTIR patterns with two prominent absorption   , 7), which can be assigned to the asymmetric stretching vibration of Si-O and symmetric stretching vibration of Si-O-Si, respectively [44,[60][61][62].FTIR bands indicate that the mineral composition of the red and black zones are characteristic of agate and contain a low amount of moganite (Fig. 6) [63].The band ratio of 1097 cm −1 peak and the trough at 1158 cm −1 is sensitive to the abundance of moganite [56].However, this parameter only works well for high moganite content samples (more than 50%) [56].
For the scattered white zone, the two prominent peaks show a redshift to the low wavenumber region.A sharp absorption band is observed at ~ 1084 cm −1 , and a subtle band at 1180 cm −1 .These bands are expected to be mainly associated with Si-O stretching vibrations of α-quartz [64].Moganite was considered to be an unstable precursor mineral of α-quartz in previous studies [44,[65][66][67], and there may be a tendency for a slow transition to quartz in natural environments.This is why moganite is easier to be etched at high temperatures.Unlike samples A, B-a, E, and G, the FTIR of the scattered white zone of F, the 1097 cm −1 peak remains unchanged while the 1187 cm −1 peak redshifts (Fig. 7) to 1180 cm −1 .This is proposed to be attributed to the grinding and polishing process after etching.The etched pit on F bead was eliminated by grinding, resulting in a slight redshift.

SEM
The SEM images reveal that most of the observed surfaces are flat, which means the beads had been produced by a grinding and polishing process (Fig. 8) [22,23].Figure 8i, j show the interface zone between the scattered white and the dark red zones with different magnifications.The scattered white zones always have coarse corrosion surfaces (Fig. 8f and j) compared with the red, the black, and the idiochromatic white zone.Scattered white zones are often lower in elevation than red zones (Fig. 8  i, j).This indicates that the white color was artificially added after polishing.The etching process weakened the  glaze slightly, which may result in a scattered white color.Furthermore, the chemical structure of the agate beads was partially damaged, resulting in a formation of the diffuse reflection surface (Fig. 9).

Red and dark red
Beads in this study can be divided into four color categories based on their character and staining mechanism.The first is red, the most common color of the excavated beads.
Natural occurring chalcedonies have different colors and intercrystalline porosities generated by the minor elements and mineral proportions [52,54].However, intentional heat treatment of natural agate produces the desired lustre, flaking properties, and the most significant result, red color [22,33].Since the structure of natural agate is not compact, the color of agate can be altered by exotic element infiltration into the micropores or subtle channels.Then the iron element in the pores is oxidized to ferric iron to form hematite after heat treatment, supporting the Raman data in this study together with previous literature [68,69].Hence, ferric iron is often uniformly present throughout the mineral pores and channels, causing the red color known as carnelian red.In some places with high concentrations, red dots found on the surface of the carnelian indicate the presence of hematite (Fig. 5).The irregular distribution of Fe on the surface of carnelian is caused by the oxidation of the Fe nuclei during the heating process in the bead manufacture or exogenous pollution during burial [6,60].Generally, this results in a color change throughout the particle.Therefore, carnelian is the most common material for bead making and is made into various shapes.
The evenly distributed Fe in A may be the source of its dark red color, indicating that these beads may not have undergone a heating process in the air, but are original Fe cations come from the infiltration of a specific iron-containing liquid.Fe and Mn cations are usually the colorforming elements of natural agate [69][70][71][72].The Fe cations penetrated the surface and spread out across the entire beads.
Although quartz has better crystallinity, it is unsuitable for staining due to its lack of channels.For zone with high crystallinity, the degree of staining will be significantly lower due to the compact crystal structure and smaller pores.This is the formation characteristics of the infiltration-induced red color.

Black
Black-colored agate can form naturally by introducing carbon-bearing material into the porous silica matrix [35].Continuous black bands can be observed alternating with the white bands, indicating that the black bands were formed successively after or before the white band [73].Similar to the natural black agate, artificial blackcolored beads come from disseminated carbonaceous matter [29,36,60,74].However, black-colored beads usually appear in a regularly distributed black banded or overall black color type rather than alternating bands, regardless of their natural color bands.The transition boundary between the studied beads' black and white sections clearly cut the agate's concentric zoning bands.
The lack of a line pattern means that the methodology of making the black beads differs from the scattered white beads.This process may be performed by immersing the beads in some kind of liquid.In the iron age, diverse kinds of food, such as honey or sugar, could serve as the carbon source [31].The black staining process was most probably realized by the organic carbon penetration through the subtle channels in the agate [31,36].Because of the principle of penetration, the translucent part of the beads has a slightly different consistency and porosity.

Idiochromatic white and scattered white
Both idiochromatic and scattered white colors appear as common color types of the excavated beads.However, the bead craftsman did not use the same strategies to make idiochromatic white and scattered white beads.The idiochromatic white color comes from the mineral composition that was originally possessed after the quartz formation (Fig. 1, samples C and D).Researchers have found that carnelian beads with a scattered white pattern are clearly due to artificial processing technology rather than an idiochromatic white [1-3, 16, 21, 75].Some studies have proposed that the scattered white color was generated by the 'micro fractures' of the carnelian surface [8,31].Potash and washing soda were believed as an adhesive material for the white carnelian [4,16,75].Some scholars advocate using'bleaching' since they proposed that the beads are not actually etched but only in a way of white color painting on the red carnelians [1,38].According to the μXRF (Table .2), Raman (Figs. 3, 4), FTIR (Figs. 6, 7), and SEM (Fig. 8) data, we conclude that it is etched into red rather than painted with paint [38,68].The pores on the surface of the white zone of agate are irregularly distributed, indicating a corrosion surface.The incident light was scattered by the etched coarse surface of the carnelian, resulting in a white-looking and opaque color.
Generally speaking, the main components of the unearthed beads are α-quartz and moganite.However, the Raman spectrum shows no evidence of moganite peaks in the white zones (Fig. 4).This suggests that the formation of the white color is related to the reaction of moganite.Moganite belongs to the monoclinic crystal system and is generally considered to be an intermediate phase in the transformation of amorphous SiO 2 to crystalline α-quartz [54,65,66,76].It contains a large number of structural defects in the form of Si-OH, which has greater reactivity than the Si-O bond in α-quartz [77,78].Therefore, the chemical reaction should occur first in Si-OH rather than Si-O, although we cannot rule out the reaction of the Si-O bond.Alkali-silica reaction reactions can occur in quartzbearing rocks [79].Kirar plant juice was used when producing etched beads without water [4].However, the washing-soda was proposed to be the active ingredient of the etching liquid rather than the kirar plant juice based on the results of this study [68].Therefore, moganite is likely to be etched by the following reaction equation (Fig. 9): Therefore, our analysis considers that etching is more suitable and accurate for this kind of bead because the moganite was removed by chemical reactivity.Etching rather than bleaching was proposed to be used to accurately describe the carnelian beads because the moganite was removed by chemical reactivity.
In addition to Na, small amounts of K have been detected in μXRF, Na and K can come from etching liquid, and K may also come from plant ash which was made by burning the Amaranth species plant [31] or kirar tree [80].For the prehistoric period, carbonate stone could serve as a Ca source by heating (Eq.3).Therefore, relatively high levels of the element Ca are likely derived from the CaCO 3 in the etching liquid.It may have undergone the following reaction process (Fig. 9): (1) (2) In the presence of calcium hydroxide and sodium carbonate, the following reactions may also produce sodium hydroxide (Fig. 9): Regardless of the reaction pathway, the etching process is accomplished by breaking Si-O-Si or Si-OH of the moganite to form soluble silicate and carbon dioxide (Fig. 9).According to these results, moganite should be more reactive than α-quartz.In this sense, the temperature will affect the etching rate depending on the moganite proportion and different Si-OH content [51,54,81,82].Forming a corrosion surface with different depths eventually (Fig. 8f, i, j).A white color can be observed since the rough surface scatters incident light of different wavelengths in all directions.

Conclusions
The identification of the artificial staining mechanisms used on ancient agate beads has been investigated using μXRF, Raman, and FTIR techniques.In this study, systematic mineralogical analyses were conducted on carnelian beads from an Iron Age cemetery on the Pamir Plateau and, for the first time, revealed the different coloration mechanisms of red, black, and white.
In unstained beads, Raman spectroscopic studies revealed a characteristic peak at 502 cm −1 , confirming the presence of moganite.All ancient beads with varying colors exhibited similar FTIR features, with variations in peak intensity.The position of the of the 1097 and 1187 cm −1 doublet shifted to 1084 and 1180 cm −1 with decreasing amounts of moganite.The white zones of samples A, B, E, and G were absent of moganite, which was confirmed by the FTIR data as two prominent peaks that shifted to the low-frequency region compared to the red zone.Higher amounts of Ca and Na were detected using μXRF analyses for the scattered white zone Through the application of heating, organic matter infiltration, and alkali etching techniques, beads could be regenerated with a novel appearance to cater to the varied preferences of people from different regions.The red color was produced from a heat treatment, while the black color came from carbonaceous matter infiltration.The scattered white appearance was assigned to the rough and irregular coarse mineral surface that reflected light, resulting from the chemical etching process.
In conclusion, samples C and D belonged to the blackon-idiochromatic white type.The black zones of B and D were manufactured through carbon infiltration, while the white-on-red type samples E, F, G, and H had undergone heat treatment that enhanced the red color and drove the etching reactivity.
The mineralogical study presented in this article is significant for analyzing the geographical spread of agate beads and studying ancient handicrafts.The unique mineral composition and excellent craftsmanship of these colorful ancient beads illustrate a developed pre-historic chemical technology.Further studies may be necessary to better understand the innovation and spread of the color styles and how they might have changed over time.

:
Fig. 2 a, c, d: Optical graph shows the pit of the scattered white zone, b Optical graph shows the cavity and the white layers were cut by the edge of the black zone

Fig. 3
Fig. 3 Raman spectra of A dark red zone, B black zone, C dark red zone, D black zone, E red zone, F red zone, G red zone, and H red zone

Fig. 4 Fig. 5
Fig. 4 Raman spectra of A scattered white zone, B-a scattered white zone, B-b part translucent zone, C-a idiochromatic white zone, C-b part translucent zone, D idiochromatic white zone, E scattered white zone, F scattered white zone, and G scattered white zone

Fig. 6
Fig. 6 FTIR spectra of A dark red zone, B black zone, C dark red zone, D black zone, E red zone, F red zone, G red zone, and H red zone

Fig. 7 Fig. 8
Fig. 7 FTIR spectra of A scattered white zone, B-a scattered white zone, E scattered white zone, F scattered white zone, G scattered white zone, C-a idiochromatic white zone, B-b part translucent zone, C-b part translucent zone, and D idiochromatic white zone

Fig. 9
Fig. 9 Schematic drawing of etching process observed in side view of the agate surface

Table 1
Optical characteristics of the eight studied beads

Table 2
Chemical compositions of the different color zones