First Identification of a Gypsum-Based Preparatory Layer on Polychrome Wooden Figurines from the Mawangdui Han Tomb No. 1 (2nd Century BCE), Changsha, China

Abstract

This study presents the first scientific characterization of the white preparatory layer and polychrome pigments on painted wooden figurines excavated from the Mawangdui Tomb No. 1, dating to the Han dynasty. A combination of analytical techniques, including XRF mapping, SEM, ATR-FTIR, XRD, and Raman spectroscopy, was used to investigate the composition, structure, and potential additives in the white layer. The results reveal that the preparatory layer is primarily composed of gypsum (CaSO₄·2H₂O) and calcite (CaCO₃), with minor phases such as anhydrite and larnite. SEM observations show a porous microstructure of needle-like crystals, while spectroscopic data suggest possible traces of organic binders. The preparatory layer was likely applied to smooth surface irregularities and support polychrome decoration, such as cinnabar and carbon-based pigments, and may have also functioned as a putty in localized areas. This represents the first confirmed use of gypsum-based plaster in ancient Chinese woodcarving, showing unexpected parallels with surface preparation techniques used in New Kingdom Egypt. However, the presence of organic additives and the internal structure of the figurines remain unresolved due to equipment limitations. These findings provide new insights into ancient material practices and highlight the importance of environmental control and material-specific conservation strategies for fragile gypsum-based heritage objects.

1. Introduction

Polychrome painting has long served as a vital decorative technique in ancient art, applied across a wide range of materials, including pottery, stone, wood, murals, and bronzes [1,2,3]. Over the past three decades, scientific investigations into polychrome surfaces in Chinese cultural heritage, particularly those dating from the Warring States period (475 BCE to 221 BCE) to the Han dynasty (202 BCE to 220 CE), have significantly enhanced our understanding of ancient pigments, painting materials, and artistic practices. However, painted wooden artifacts, which constitute an important category within this domain, are rarely preserved in a complete state. This is primarily due to the degradation of cellulose in archaeological wood over time, which results in severe structural weakening. Additionally, unstable environmental conditions during excavation and storage, including fluctuations in temperature and humidity, frequently cause drying, cracking, warping, and surface damage. These factors often lead to the detachment of polychrome paint layers, which poses serious challenges to the study of original materials and the reconstruction of historical production techniques. While considerable research has focused on pigment identification, such as cinnabar, minium, azurite, malachite, Han blue, and carbon black [4,5,6], studies on the detailed painting processes remain relatively limited, particularly in relation to the preparation of underlayers and the techniques used in wood carving and surface modeling.

Among wooden artifacts, funerary figurines hold particular significance because they were specifically created to accompany the deceased in burials. Archaeological evidence suggests that such wooden figures originated in the Chu state, a major state in southern China during the Warring States period, and have been primarily unearthed from tombs in Hunan, Hubei, Anhui, and Jiangsu provinces, dating from the Warring States to the Han dynasty [7,8,9]. Since these figurines were interred immediately after production and not subjected to daily use, many retain intricate carving and polychrome decoration in a relatively original condition. Such wooden figurines vary in size and form; they display a wide range of decorative techniques, including carving, surface modeling, polychrome painting, and simulated clothing. These finely crafted objects provide valuable insights into the social customs, costume styles, and technological practices of early imperial China.

A large number of wooden human figurines have been excavated from tombs in the Changsha region, ranging in height from a few centimeters to life-size [10,11,12]. Typically, these figurines were carved from whole blocks of wood to form the basic shape. The surfaces were further refined through techniques including direct polychrome painting, the addition of separate elements, and the application of miniature garments. Facial features and clothing details such as sleeves and collars were often modeled with precision to enhance lifelike expression. Based on their posture and attributes, these figurines have been systematically classified into categories such as attendants, musicians, craftsmen, guards, dancers, and stewards. They reflect the social status of the tomb owner and represent a belief system in which idealized servants continue to assist the deceased in the afterlife.

Notably, standing figurines from Wulipai Tomb M406 and the Mawangdui (MWD) Han Tombs Nos. 1 and 3 were entirely coated with a white preparatory layer (putty layer) prior to the application of polychrome paint. This treatment created a distinctive appearance characterized by white faces and white bodies [13]. Similar white preparatory layers have also been observed in other funerary objects, such as mica-coated wooden bi disks from the MWD Tomb No. 1 [14], wooden horse-shaped figurines from a Han tomb in Mozuizi, Wuwei, Gansu [15], and terracotta figurines from royal Han dynasty tombs in Qingzhou County, Shandong [1]. These cases suggest that the use of white ground layers was a common practice in early imperial mortuary art. However, across all these examples, the material composition, preparation techniques, and functional roles of the white layers remain poorly understood due to the lack of systematic scientific investigation.

During routine conservation of the MWD figurines, we conducted a comprehensive scientific analysis to characterize the white preparatory layer and overlying pigments. Using X-ray fluorescence (XRF) imaging, SEM, Raman spectroscopy, ATR-FTIR, and XRD, this study aims to clarify the material composition and application techniques of the preparatory layer and contribute to future conservation strategies and technical understanding of early Chinese polychrome woodcraft.

2. Materials and Methods

2.1. Description of the Wooden Figurines and Sampling Strategy

A total of 162 wooden attendant figurines (站立侍从俑) were recovered from the side chambers during the 1972 excavation of the Mawangdui Tomb No. 1 (马王堆一号汉墓) in Changsha. Among them, 101 polychrome standing figures from the eastern and southern chambers are particularly notable (Figure 1). These were found in a semi-upright or fallen state, with some partially submerged in approximately 40 cm of stagnant water at the bottom of the chambers. Associated burial objects included bamboo-woven containers filled with silk fabrics, along with terracotta items, pottery, and lacquerware.

The standing figures typically measure 42 to 51 cm in height, with head lengths of 7 to 8 cm and shoulder widths of approximately 13 cm. They were most likely carved from single blocks of wood, as there is no evidence of jointed or assembled components. A uniform white preparatory layer also referred to as a preparatory layer, was applied across the surface. Red and black pigments were used to render facial features, while the garments were decorated with additional patterns in red, black, and dark green. The figures wear cross-collar long robes with flat necklines, and their arms hang naturally, with hands enclosed within the sleeves at the abdomen. At the time of excavation, most figurines retained vivid coloration and appeared relatively well preserved. However, extended storage and environmental fluctuations have resulted in significant deterioration, including fading, dehydration, warping, cracking, and the delamination of painted layers.

One figurine, in particular (ID: 5035), exhibited extensive damage, with severe cracking visible on the face and throughout the body (Figure 2), especially in the lower half. In this area, both the painted layer and its underlying preparatory layer had detached from the wooden substrate. The detached fragment measures approximately 56 mm in length, with a maximum width of 28 mm and a thickness of 16 mm at its base (Figure 3). The front surface is decorated with red and black pigments, while the reverse shows the impression of the underlying wood. The fragment consists entirely of a gray–white solid material, with no visible traces of wood or other fillers. Based on its morphology and detachment characteristics, it is inferred that the wooden substrate had surface irregularities that were compensated by a thick application of preparatory, which served to fill defects prior to painting. Small samples of the preparatory and pigment layers from this fragment were collected for scientific analysis.

2.2. Elemental Mapping by MA-XRF and µXRF

In the investigation of the painted surface and an underlying preparatory layer of these wooden figurines, XRF mapping has proven to be an indispensable analytical method. This non-invasive technique enables in situ acquisition of elemental distribution maps across the entire surface rather than limited point measurements. It generates detailed visualizations of elemental concentrations, which reveal potential hidden structural features, clarify the spatial arrangement of pigments, and help distinguish between the chemical compositions of the painted layer and the underlying substrate. XRF is particularly effective for detecting complex pigment layering and subtle material transitions. Its relatively fast scanning speed and portability make it especially suitable for on-site heritage analysis [16,17,18].

Elemental mapping of the sample was carried out using two complementary XRF systems: the Bruker CRONO macro-X-ray fluorescence (MA-XRF) scanner and the Bruker M4 micro-X-ray fluorescence (µXRF) scanner. The MA-XRF system was employed for rapid, large-area scanning of the outer surface of the detached fragment [19]. It operated at 50 kV and 200 µA, achieving a spatial resolution of 0.5 mm and an energy resolution below 140 eV (Mn Kα). In contrast, the µXRF system [20] was used for high-resolution mapping of a selected area on the lower front side of figurine 5035. It operated at the same voltage but at a higher current of 600 µA, achieving a spatial resolution of 30 µm and an energy resolution below 145 eV. The µXRF scanning covered an area of 34.9 × 30.5 mm in this study. Together, these two systems provided a complementary analytical approach, combining rapid overview mapping with detailed analysis of localized elemental distributions. This dual-system strategy enabled both the identification of general compositional patterns and the detection of fine-scale material variations across different regions of the figurine.

2.3. Scanning Electron Microscopy (SEM) for Microstructural Observation

The microstructure of the gray–white material from the inner surface of the preparatory layer was examined using a JEOL JCM-6000Plus (JEOL Ltd., Tokyo, Japan) scanning electron microscope. Prior to imaging, the sample was coated with a thin layer of gold under vacuum conditions using a sputter coater in order to enhance conductivity and minimize surface charging. The SEM system was operated at an accelerating voltage of 10 kV, with a working distance between 7 and 53 mm. The filament current and probe current were maintained at standard settings. These parameters ensured optimal resolution and contrast in the acquired images. The analysis focused on identifying crystal morphology, porosity, and surface features of the preparatory material.

2.4. ATR-FTIR Analysis

ATR-FTIR analysis was conducted using a Bruker ALPHA Fourier transform infrared spectrometer (Bruker Optik GmbH, Ettlingen, Germany) to identify the organic and inorganic components of the gray–white solid material from the preparatory layer. The sample was ground into a fine powder prior to analysis. An A220/D-01HP Platinum-ATR Universal sampling module (Bruker Optik GmbH, Ettlingen, Germany) was used, with a scanning range of 4000 to 400 cm⁻¹ and a spectral resolution of 4 cm⁻¹. Both background and sample spectra were acquired with 32 scans per measurement. The analysis was performed at room temperature and under normal atmospheric pressure in the cultural heritage conservation laboratory of the Hunan Museum.

For data processing, Python-based libraries were employed (Python 3.9). Matplotlib (version 3.5.1) was used to visualize the spectra, while Pandas handled data management. Numerical calculations were conducted using NumPy, and SciPy was used for spectral post-processing, including smoothing (Savitzky-Golay filter), peak detection (find_peaks), baseline correction (polyfit), and spectral fitting (curve_fit) to enhance the resolution and facilitate identification of weak absorption bands.

2.5. X-Ray Diffraction (XRD)

The phase composition of the gray–white solid material from the preparatory layer was analyzed using X-ray diffraction (XRD) with a Bruker D8 Advance Eco diffractometer (Bruker AXS GmbH, Karlsruhe, Germany). Prior to testing, the sample was finely ground and uniformly spread onto a monocrystalline silicon substrate. The instrument was equipped with a ceramic X-ray tube using a copper anode (Cu Kα₁, λ = 1.54060 Å), operated at 40 kV and 25 mA. The detector was configured in SSD160 one-dimensional mode with a divergence slit of 0.6 mm and a scatter slit of 2.5°. Data acquisition was performed over a 2θ range of 10° to 90°, with a step size of 0.008° and a counting time of 0.06 s per step. Phase identification was carried out using Jade software (version 6.0, Materials Data, Inc., Livermore, CA, USA).

Further analysis of the XRD data was conducted through Whole Pattern Fitting (WPF) and Rietveld Refinement. The peak shape function was modeled using the Pearson VII function with a fixed background. The refinement process converged after 26 iterations, yielding an R-factor of 19.05 percent, with E = 11.88 percent, R/E = 1.6, P = 38, and EPS = 0.5. These results ensured accurate phase quantification and structural refinement of the sample.

2.6. Raman Spectroscopy

Raman spectroscopy was used to analyze the red and black pigments applied to the figurine, as well as the gray–white solid material from the preparatory layer. The measurements were performed using a Bruker SENTERRA II confocal microscopic Raman spectrometer (Bruker Optik GmbH, Ettlingen, Germany), which incorporates a confocal optical system that enables high spatial resolution and depth discrimination. An integrated Olympus microscope (Olympus Corporation, Tokyo, Japan) was used for sample observation and precise laser focusing on targeted micro-regions, thereby minimizing interference from surrounding areas. All measurements were carried out at room temperature and under normal atmospheric pressure in the cultural heritage conservation laboratory of the Hunan Museum.

A 785 nm laser was used as the excitation source, with the laser power set to 1 mW to reduce fluorescence background and avoid thermal damage to the samples. The lateral spatial resolution was below 1 micrometer. Raman spectra were collected over the range of 50 to 3640 cm⁻¹, with a spectral resolution of approximately 4 cm⁻¹. Real-time calibration was performed using SureCAL™ technology (Bruker Optik GmbH, Ettlingen, Germany) to ensure both spectral accuracy and reproducibility. Exposure time and the number of accumulations were optimized to achieve an appropriate signal-to-noise ratio. Python-based data processing was employed for spectral visualization and peak annotation. The pandas and matplotlib libraries were used to import, organize, and plot the Raman data, allowing for clear identification of characteristic peaks.

3. Results

3.1. Microstructural Features

SEM observations reveal that the grayish–white substance consists of densely packed needle-like crystals, typically measuring 10–25 µm in length and 1–3 µm in width, accompanied by numerous finer crystals with dimensions below 3 µm. Although partial interlocking between some needle-like crystals is observed, the overall microstructure remains highly porous, with large voids and an uneven pore distribution. The crystals exhibit a random orientation, lacking a clear preferential direction, which contributes to a disordered aggregation.

In addition, film-like material was observed coating the surfaces of some needle-like crystals. These materials appear predominantly as thin films or fibrous structures, creating rough interfaces. The presence of this structureless phase appears to inhibit the growth of well-defined gypsum crystals, resulting in incomplete development and the formation of granular or aggregated morphologies with poorly defined boundaries. In some regions, a limited number of plate-like crystals were found embedded within the film-like matrix, further disrupting regular crystal growth (Figure 4).

3.2. Elemental Distribution and Material Zoning

Two XRF mapping systems were employed to analyze the spatial distribution of elements across the sample. Despite differences in resolution and scanning scale, both systems yielded consistent results regarding the distribution of key elements, including calcium (Ca), sulfur (S), silicon (Si), potassium (K), iron (Fe), and mercury (Hg). These elemental patterns correlate well with the optical image, particularly in relation to color variation across different regions (Figure 5 and Figure 6).

(1)

Calcium is widely distributed in the basal areas of the sample, showing a strong correlation with the gray–white regions observed in the optical image. This suggests that these areas are composed of Ca-rich materials. Silicon exhibits a similarly broad distribution, further supporting its role as a major component of the substrate. The µXRF system offers enhanced clarity, revealing minor variations in Ca and Si concentrations that are not readily apparent in the MA-XRF data.

(2)

Sulfur shows a distribution pattern closely aligned with that of calcium, although some regions exhibit strong Ca signals with weaker S signals. This suggests the presence of multiple Ca-bearing phases, with sulfur associated only with specific ones. The µXRF data provide finer detail, clearly illustrating small-scale heterogeneity in sulfur distribution and depletion.

(3)

The distribution relationship between Si and K is also noteworthy. Both elements show localized co-enrichment in certain areas, which may indicate the presence of a mixed Si-K-bearing phase. However, potassium is more sporadically distributed, suggesting that it exists in relatively minor amounts or is associated with surface deposits. In contrast, the broader and more consistent presence of silicon across the mapped area implies that Si may also exist in the form of a separate Si-rich phase independent of potassium. This interpretation is supported by the µXRF results, which revealed significantly lower concentrations of K than Si in most areas.

(4)

Iron is mainly concentrated in the dark-colored decorative motifs seen in the optical image but is also scattered across the entire sample surface. This implies that iron may be associated with black pigments as well as being a minor component of the underlying substrate.

(5)

Mercury appears as highly concentrated patches that precisely align with red motifs in the optical image. The sharply defined signal boundaries suggest that mercury was deliberately applied through painting rather than being naturally present in the substrate. In areas rich in Hg, the signals of Ca, S, and Si are significantly attenuated. This attenuation is likely due to the high atomic number of Hg, which can block or absorb X-ray fluorescence from underlying layers [21]. The µXRF system, due to its higher resolution, provides a more detailed view of this shielding effect, including areas where Ca and S signals are only partially reduced rather than completely suppressed.

In summary, the MA-XRF system is effective for rapid large-area mapping and provides an overall view of elemental distributions, which helps reveal indications of stratigraphic layering, suggesting a complex interplay between pigments and the underlying material. However, its relatively lower spatial resolution limits the detection of fine material transitions. The µXRF system complements this by offering high-resolution imaging of localized regions, enabling more detailed analysis of pigment application, material heterogeneity, and potential post-depositional alterations.

3.3. Mineralogical Composition and Spectroscopy Studies

3.3.1. XRD Phase Analysis

X-ray diffraction (XRD) analysis (Figure 7) identified gypsum (CaSO₄·2H₂O) as the dominant crystalline phase, accounting for 58.31 wt% of the total composition. Its characteristic peaks appeared at 11.72°, 20.81°, and 29.19° 2θ (PDF#00-033-0311). The full width at half maximum (FWHM) values at 11.72° and 20.81° were 0.1117 and 0.1366, corresponding to crystallite sizes of approximately 715.7 nm and 591.7 nm, indicating a well-developed crystalline structure. Anhydrite (CaSO₄, 1.9 wt%) was also detected (PDF#01-086-2270), with a broader peak at 29.19° (FWHM = 0.4674, crystallite size ~175.8 nm), suggesting smaller grains or higher microstrain. The overlap at 29.19° indicates contributions from both gypsum and anhydrite (Figure S1). Additional mineral phases identified include larnite (Ca₂SiO₄, 2.19 wt%) (PDF#00-033-0302), calcite (CaCO₃, 32.5 wt%) (PDF#00-005-0586), and quartz (SiO₂, 5.1 wt%) (PDF#03-065-0466), with diffraction peaks consistent with their respective reference patterns. The reported weight percentages are based on full-pattern Rietveld refinement, which produced a good fit between the experimental and calculated diffraction profiles. These results provide a reliable estimate of the relative phase composition of the analyzed material. Although Rietveld refinement does not yield absolute concentration values, it is widely recognized for providing accurate and reproducible semi-quantitative results in multi-phase systems when appropriate structural models and refinement parameters are applied.

3.3.2. ATR-FTIR and Raman Characteristics

ATR-FTIR analysis further confirmed the presence of gypsum (CaSO₄·2H₂O) in the examined samples [22]. The O–H stretching vibrations of structural water were observed as strong and broad absorption bands centered at 3523 cm⁻¹ and 3393 cm⁻¹. A narrow and intense band at 1619 cm⁻¹ was assigned to the H₂O bending vibration (ν2), while a weaker band at 1684 cm⁻¹ may also contribute to this vibrational mode. The sulfate group was identified by several diagnostic bands. The most intense absorption occurred at 1115 cm⁻¹, assigned to the asymmetric stretching vibration (ν3) of SO₄²⁻. A weaker band at 1009 cm⁻¹ corresponds to the symmetric stretching mode (ν1). Although this mode is typically Raman-active and IR-inactive under ideal Td symmetry, its presence here may reflect a slight symmetry distortion of the sulfate group, possibly due to hydrogen bonding or lattice effects [23]. Strong signals at 670 cm⁻¹ and 599 cm⁻¹ represent bending vibrations (ν4). Additionally, skeletal lattice vibrations such as Ca–O modes were observed near 460 cm⁻¹. These spectral features collectively confirm the presence of crystallized gypsum [24,25].

Absorptions at 1445 cm⁻¹ and 876 cm⁻¹ were attributed to the asymmetric stretching (ν3) and out-of-plane bending (ν2) modes of CO₃²⁻, indicating the presence of carbonate impurities, likely from calcite [26]. The spectral profile thus reflects a gypsum-dominated composition with a minor contribution from carbonate phases (Figure 8a). Since ATR-FTIR averages signals over the contact area, the data represent the bulk composition rather than localized surface variation.

In addition to the inorganic phases, weak absorption bands at 1740 cm⁻¹ and 1241 cm⁻¹ suggest the possible presence of organic components (Figure 8b). The 1740 cm⁻¹ band is likely due to C=O stretching vibrations, commonly found in esters, carboxylic acids, ketones, or aldehydes. These signals may originate from the oxidative degradation of proteinaceous binders such as animal glue, lipid or resin residues, or from the adsorption of environmental contaminants [27,28,29].

The weak signal at 1241 cm⁻¹ may correspond to the Amide III band. However, the characteristic Amide I (~1650 cm⁻¹) and Amide II (~1550 cm⁻¹) bands were not distinctly detected. This absence may result from the extremely low concentration of proteinaceous material (likely below 1 percent), leading to absorption signals below the noise threshold. Furthermore, overlapping signals from H₂O bending (1680–1620 cm⁻¹) and CO₃²⁻ stretching at 1445 cm⁻¹ may obscure Amide I and Amide II, respectively.

Notably, no CH₃ or CH₂ stretching vibrations were observed in the 2800–3000 cm⁻¹ range, suggesting the absence or only trace amounts of long-chain organic compounds. This further supports the hypothesis that the detected C=O signal may be attributed to degradation byproducts or surface contamination. The uneven distribution of organic residues may also lead to selective detection of certain peaks, thereby limiting the completeness of the spectral information.

Raman spectra indicate that the sample contains both sulfate (SO₄²⁻) and carbonate (CO₃²⁻) components. The symmetric stretching vibration of SO₄²⁻ (ν₁) is observed at 1008 or 1015 cm⁻¹, corresponding to gypsum and bassanite, respectively. Symmetric bending modes (ν₂) are found at 415 cm⁻¹ for gypsum and at 427 and 485 cm⁻¹ for bassanite, reflecting structural differences due to the hydration state. The simultaneous presence of these features may indicate a mixture of both phases [30,31]. Additionally, a symmetric stretching band of CO₃²⁻ appears at 1085 cm⁻¹, indicating the presence of calcium carbonate, likely in the form of calcite. A weak, broadband in the range of 3200 to 3500 cm⁻¹ corresponds to O–H stretching vibrations, suggesting hydrated sulfate phases such as gypsum (Figure 9a). Based on these features, the predominant phases in the sample are inferred to include gypsum, anhydrite, and minor calcite (Figure S2).

An additional set of Raman peaks is observed between 1200 and 1800 cm⁻¹, beyond the expected SO₄²⁻ region. Strong bands at 1260, 1358, 1443, and 1573 cm⁻¹, as well as weaker bands at 1674, 1762, and 1875 cm⁻¹ (Figure 9b), resemble Raman signatures associated with organic macromolecules such as proteins, collagen, elastin, nucleic acids, or their degradation products [32,33]. Specifically, the band at 1260 cm⁻¹ may correspond to Amide III (C–H deformation), the 1445 cm⁻¹ band to CH₂/CH₃ bending, the 1660–1670 cm⁻¹ band to Amide I (C=O stretching), and the 1760 cm⁻¹ band to additional C=O vibrations.

However, the absence of C–C stretching bands in the 800 to 940 cm⁻¹ range and protein cross-linking signals between 1100 and 1300 cm⁻¹ suggests that possible organic compounds may have undergone degradation. Similar spectral patterns have been reported in studies of fossilized bone, where peaks in this region were interpreted as resulting from altered carbonate minerals, secondary mineralization, or degraded biomolecular residues [34]. Raman analyses of plaster materials have also revealed comparable enhanced signals in this range. These signals have been tentatively attributed to structural impurities, including foreign anions, lattice substitutions, or vacancy defects, rather than to residual organic matter [35]. In the present study, the observed spectral features may be explained by the presence of degraded proteinaceous compounds, trace organics, or an increased degree of poorly crystalline disorder within the sulfate matrix.

The painted layer contains fine black particles. Raman spectral analysis reveals a D band between 1300 and 1400 cm⁻¹ and a G band near 1580 to 1620 cm⁻¹, which are characteristic features of disordered carbonaceous materials (Figure 10a). The G band corresponds to the E₂g vibrational mode of graphitic carbon, while the D band is associated with lattice defects, edge effects, and structural disorder. The broad profile of the G band and the prominent intensity of the D1 band (1300 to 1385 cm⁻¹) indicate a high degree of structural disorder. In addition, a weak band in the 1200 to 1290 cm⁻¹ range may correspond to the D4 band, which is commonly attributed to sp²–sp³ carbon hybridization, oxidized carbon precursors, or other structural irregularities. These spectral features suggest that the black pigment consists primarily of soot or highly disordered carbon black [36].

Red particles identified within the painted layer exhibit Raman signatures consistent with cinnabar (HgS). A strong band at 253 cm⁻¹ corresponds to the A₁ symmetric stretching vibration of the Hg–S bond. Additional peaks at 290 cm⁻¹ and 343 cm⁻¹ are attributed to E-mode shear vibrations and A₁ bending modes, respectively (Figure 10b). Some red–black grains display Raman peaks at 226, 293, 410, 498, and 610 cm⁻¹, which match the characteristic bands of hematite (α-Fe₂O₃) [37]. A weak band at 659 cm⁻¹ may indicate the presence of magnetite (Fe₃O₄) (Figure 10c), although potential interference from fluorescence cannot be ruled out [38].

4. Discussion

4.1. Technological Interpretation of the Preparatory Material

XRF mapping results indicate that calcium is the most abundant element in the analyzed sample, showing strong and widespread signals across the basal regions. This elemental distribution is closely associated with the gray–white areas observed in the optical image, suggesting that calcium-rich compounds constitute the dominant component of the underlying material. Further characterization by Raman spectroscopy and X-ray diffraction (XRD) confirmed that the main crystalline phases are gypsum (CaSO₄·2H₂O) and calcite (CaCO₃). SEM imaging reveals that the gypsum crystals display a needle-like morphology with a broad size distribution and random orientation. Crystallographic analysis shows that the gypsum crystals predominantly grow along the C-axis, with the {010} plane typically exposed on the crystal sides. The {120} plane is less developed, and the {001} plane is difficult to identify in SEM observations. As a result of this growth pattern and the irregular crystal alignment, the material exhibits a relatively loose and porous microstructure. These features closely resemble those observed in low-strength plaster formulations, particularly gypsum plaster produced through the hydration of thermally processed gypsum [39].

Gypsum plaster is typically prepared by calcining natural gypsum at temperatures between 130 and 160 °C to produce bassanite (CaSO₄·½H₂O), which rehydrates upon mixing with water to form gypsum again. When hydrated under a low water-to-plaster ratio, the material forms a highly porous matrix of interlocking needle-like crystals with limited strength. This results in low compressive and flexural resistance, making the material fragile and susceptible to mechanical damage.

Based on its microstructural features and phase composition, the grayish–white material constituting the preparatory layer in this study is best interpreted as gypsum plaster that has undergone thermal processing. XRD analysis confirmed minor amounts of anhydrite (CaSO₄) and larnite (Ca₂SiO₄), both of which are known to form under high-temperature conditions. Larnite formation typically occurs in systems containing silicates, sulfates, and calcium oxide at temperatures between 900 and 1100 °C, although it is theoretically possible that some larnite may originate as a natural contaminant, particularly in silicate-rich metamorphic deposits [40]. While anhydrite may begin to form at temperatures as low as 400 °C. Excessive calcination above 1100 °C can lead to the decomposition of calcium sulfate, resulting in irreversible loss of hydration capacity [41,42].

The presence of both bassanite-derived gypsum and high-temperature phases suggests that the material may have experienced non-uniform calcination during ancient gypsum processing. In enclosed kiln-like environments, even when using wood or charcoal as fuel, it is plausible that localized thermal hotspots, albeit extremely limited in extent, could have reached temperatures as high as 1100 °C. Under such conditions, these localized hotspots may have facilitated the formation of anhydrite and even larnite while much of the material remained within the optimal temperature range for bassanite formation. This scenario would have allowed the preparatory layer to maintain its functionality as a workable gypsum-based material used for surface smoothing and support beneath the painted decoration.

XRD and Raman data also reveal calcite intermingled with the gypsum phase. In evaporitic gypsum deposits, calcite frequently coexists with gypsum, anhydrite, and dolomite, forming complex mineral assemblages shaped by depositional processes, water–rock interactions, and diagenetic or microbial activity [43,44]. The carbonate phases observed here may represent natural impurities introduced during raw material sourcing. Alternatively, the partial decomposition of gypsum and associated carbonates during uneven calcination, followed by atmospheric carbonation during hydration, may have led to the secondary formation of CaCO₃. The minor presence of quartz and silicate phases likely results from extraneous mineral contamination.

SEM analysis further revealed the presence of film-like or fibrous material on the surface of gypsum crystals. These structures appear to inhibit normal crystal growth, leading to granular or aggregated morphologies [45]. Such microstructural modifications are consistent with the known effects of incorporating animal glue, which alters crystal formation pathways and can significantly influence mechanical properties. ATR-FTIR and Raman spectroscopy support the possibility of organic binding materials within the sample, although the current spectral data are ambiguous and require cautious interpretation, such as Raman features in the 1200–1800 cm⁻¹ range deviate from standard reference spectra and allow for multiple interpretations.

The historical use of organic binding materials, such as animal glue in gypsum-based materials, is well documented [46,47,48]. In traditional plaster formulations, animal glue serves to slow hydration reactions, reduce moisture permeability, and enhance the dimensional stability and weathering resistance of the hardened material. It also improves mechanical performance by increasing flexural and compressive strength [45]. Despite these known effects, the presence of organic additives in Han dynasty gypsum-based materials remains to be definitively confirmed. Additional analyses using advanced techniques such as Py-GC-MS or proteomics will be necessary to determine the precise nature and role of organic substances in these ancient formulations.

4.2. Historical Context and Regional Utilization of Gypsum-Based Materials

Gypsum has long been valued for its ease of processing, low-temperature calcination, and excellent plasticity. These properties made it a widely used material in both architectural and artisanal contexts since the Neolithic period [49,50]. Early applications in the Near East included its use in construction, sculpture, and decorative artifacts [51]. In Egypt, during the Neo-Babylonian period and the 8th century BCE in Iberian fortresses, gypsum was employed as a key building material [52,53]. During the medieval period, its usage expanded across Europe, particularly in cities such as Paris, Chartres, and Lübeck, where it played important roles in both structural and decorative applications [54].

The production of ancient craft materials was closely constrained by the availability of natural resources, regional processing capabilities, and prevailing technological traditions. Different geographic areas developed distinct strategies for material selection and use, often shaped by their resource endowments. Since gypsum requires significantly less thermal energy for calcination than lime, it became the preferred material in fuel-scarce regions such as Mesopotamia and Egypt. In contrast, lime-based technologies were more prevalent in fuel-rich areas like the eastern Mediterranean and Anatolia [55]. In Egypt, abundant gypsum ore provided the foundation for large-scale and long-term use of the material in both architectural and decorative applications [56,57].

Similarly, the region surrounding Changsha in Hunan Province, China, is rich in gypsum resources. In Shaodong County, southwestern Hunan, for example, Cenozoic formations contain thick-bedded gypsum deposits that are often associated with calcite and other calcium carbonate minerals [58]. These gypsum layers are located at shallow depths and are suitable for manual extraction [59]. Such resource conditions supplied a material basis for local use of gypsum during ancient times.

Despite this geological abundance, archaeological evidence suggests that the widespread use of gypsum as a primary material for construction or sculptural modeling was extremely rare in China during and before the Han dynasty. In the Late Neolithic period, gypsum was primarily used as a white pigment for pottery decoration [60,61]. From the Han dynasty onward, it also appeared in traditional Chinese medicine, typically in its raw and ground form, rather than as a processed or calcined material [62]. Han dynasty wooden funerary objects, especially figurines, were usually carved from solid wood and directly painted without the use of preparatory layers.

This study provides the first evidence that a gypsum-based preparatory layer, serving as a plaster and as a putty to support surface modeling, was applied to wooden figurines that, to date, have only been identified in Hunan, as demonstrated by the findings of this study. This localized usage supports the view that ancient technologies were highly dependent on the availability and perception of regional materials and that material choices were shaped by both geological and cultural conditions.

Interestingly, wooden figurines from Han dynasty tombs in MWD and funerary statuettes from New Kingdom Egypt exhibit notable similarities in their surface preparation techniques. In both cases, a white preparatory layer was applied to the wooden substrate prior to polychrome decoration. Scientific analysis of an Egyptian shabti revealed that its ground layer consisted of a mixture of gypsum and calcite, likely used to smooth the surface and enhance visual brightness. In some instances, GC-MS analysis also suggested the presence of proteinaceous organic binders, which would have contributed to improved adhesion and cohesion of the overlying paint layers [63,64]. These parallels may reflect a shared functional response to the technical requirements of preparing wooden surfaces for painted decoration.

Nevertheless, the considerable geographic and temporal separation between East Asia and the Mediterranean–Egyptian world makes direct cultural transmission unlikely. The similarities observed in material application and surface preparation methods are more plausibly interpreted as independent innovations developed by artisans responding to similar technical needs. These findings highlight a broader phenomenon of convergent technological evolution, where distinct societies, working within their respective material contexts, arrived at comparable solutions through empirical engagement with available resources.

4.3. Deterioration Mechanisms and Conservation Implications

Gypsum-based materials are inherently susceptible to a range of physicochemical transformations when exposed to waterlogged burial environments or abrupt environmental changes following excavation. These transformations include mineralogical phase transitions, Ostwald ripening induced by cyclic wetting and drying, and partial dissolution followed by reprecipitation. Such processes can lead to the progressive coarsening of gypsum crystals, the formation of secondary crystal phases, and increased microstructural heterogeneity [50]. Collectively, these changes compromise the material’s structural cohesion and mechanical integrity, promoting degradation phenomena such as cracking, surface delamination, and powdering [54]. These effects present significant risks to the long-term preservation of gypsum-based cultural heritage. Similar observations have been reported in the conservation of Tang dynasty murals in Xinjiang, where spontaneous transformation of gypsum ground layers was identified as a key factor in mural deterioration [65].

The wooden figurines from the Han tombs at MWD were recovered from a waterlogged and anoxic burial environment, where they remained in a stable and saturated condition for over two millennia. The sudden environmental shift during excavation, from a moisture-rich to a dry and oxygenated atmosphere, likely initiated or accelerated the deterioration of the gypsum-based preparatory layers through the mechanisms described above. A clear understanding of the composition and behavior of both surface and substrate materials, especially those utilizing gypsum as a ground or fill layer, is essential for developing effective conservation strategies. Such knowledge not only informs decisions on material stabilization and environmental control but also enhances our understanding of historical technologies and the practical choices made by ancient artisans in response to material properties.

5. Conclusions

• This study presents the first scientific characterization of a white preparatory layer on polychrome wooden figurines excavated from the Mawangdui Tomb No. 1, dated to the Han dynasty. Multi-analytical techniques, including XRF mapping, SEM, ATR-FTIR, XRD, and Raman spectroscopy, revealed that the preparatory layer consists primarily of gypsum (CaSO₄·2H₂O) and calcite (CaCO₃), with minor high-temperature phases such as anhydrite and larnite. The preparatory layer exhibits a porous microstructure composed of interlocking needle-like gypsum crystals, consistent with the characteristics of low-strength plaster. In certain areas, it also functioned as a putty, filling surface irregularities and providing structural support for polychrome decoration. Spectroscopic analyses further indicate the possible presence of organic binders, although their composition remains inconclusive. The use of such a preparatory layer reflects a localized technological innovation specific to Hunan and represents the first confirmed application of gypsum-based materials in ancient Chinese woodcarving. While comparable surface preparation techniques involving gypsum and calcite have been observed in New Kingdom Egyptian artifacts, the considerable spatial and temporal distance between the two traditions suggests independent development. These similarities are better understood as examples of convergent technological responses, in which artisans in different cultural settings arrived at comparable solutions through practical engagement with similar materials.
• In addition to material identification, this study highlights several conservation challenges and methodological limitations. The gypsum-based preparatory layer showed deterioration after excavation, including cracking, delamination, and powdering, reflecting its sensitivity to environmental fluctuations such as humidity, temperature, and air exposure. These issues emphasize the need for environmental control during excavation, transport, and storage. However, due to equipment constraints and the immovability of the artifacts, advanced imaging techniques such as high-resolution X-ray radiography and computed tomography (CT) were not applied. As a result, the internal structure of the wooden cores and the extent to which the external morphology was shaped by wood versus gypsum remains unclear. Likewise, the absence of molecular analyses such as pyrolysis–gas chromatography–mass spectrometry (py-GC-MS) limited the identification of potential organic binders. Future research should address these gaps through the application of non-invasive or minimally invasive imaging and molecular techniques, which will be essential for clarifying the construction logic of these figurines and refining conservation strategies tailored to their complex material composition.