Research on the Correlation Between the Microscopic Structure of Cultural Relics Faded Painted Layers and Surface Color Characteristics

Highlights

What are the main findings?

• Fading of painted relics is mainly due to light scattering by sub-micron pores.
• Ionic liquid or glycerol pore filling effectively suppresses scattering and restores color.
• Pore filling reduces color difference by 30%–50% and reflectivity by 20%–40%.
• The mechanism was validated on authentic painted bricks from the Western Qing Tombs.

What are the implications of the main findings?

• This study provides a reversible, non-chemical restoration strategy for faded relics.
• The approach offers a nondestructive alternative to traditional retouching methods.
• It has potential applications for preserving other porous cultural heritage materials.

Abstract

The fading of painted relics is a widespread deterioration phenomenon in ancient painted cultural relics, yet its underlying mechanism has long been attributed solely to pigment oxidation. Directed at colored drawings with complex surface microstructures, such as pottery paintings, wall murals and architectural paintings, here we challenge this view by demonstrating that light scattering induced by sub-micron pores within the paint layer plays a dominant role, especially Mie scattering when pore sizes approach visible light wavelengths (400–700 nm). In order to minimize the damage to the genuine painted relics, a large number of simulated experiments were conducted first. Using porous polyacrylamide (PAM) membranes and nylon 6 filter membranes as model systems, we show that pore-induced scattering reduces the optical path length for light absorption, leading to a significant decrease in color saturation and brightness. By filling the pores with non-volatile colorless ionic liquids ([BMIM]PF₆) (n = 1.41) or glycerol (n = 1.47)—both possessing refractive indices close to those of the pigments—the scattering is effectively suppressed, and the original color is restored. The filling treatment reduces the color difference (ΔE*ab) by 30%–50% and the surface reflectivity by 20%–40%. Mercury intrusion porosimetry and fluorescence spectroscopy confirm that pore elimination and optical path lengthening are responsible for the color recovery. The proposed mechanism and restoration strategy were successfully validated on authentic painted brick fragments from the Western Qing Tombs (Hebei, China), where severely faded green and red patterns reappeared after ionic liquid treatment. This study provides a new interface-regulation paradigm for the conservation of painted cultural heritage, shifting the focus from irreversible chemical remediation to reversible physical restoration and offers a generalizable platform for controlling light scattering in porous optical materials.

1. Introduction

Fading of painted relics is a common deterioration phenomenon in cultural relics. Fading causes the loss of original information, greatly reducing the historical value of the relics. Archeologists and conservators often observe that unearthed painted relics initially exhibit brilliant colors, but these colors soon disappear. For a long time, this phenomenon was widely attributed to the oxidation and aging of binding materials and photo-degradation of the pigments [1,2]. However, most ancient Chinese pigments are inorganic minerals, which are less prone to oxidative decomposition. The real cause may be related to light scattering [3,4].

Unearthed relics contain substantial moisture, which fills surface voids or particle gaps. When the relics are exposed to air, this moisture evaporates quickly, creating interfaces that cause strong light scattering [3]. This scattering reduces the absorption of light by the pigments, eventually leading to color fading [5]. In this study, by selecting colored film materials with different pore size distributions, we aim to reveal the mechanism of fading based on light scattering through correlations between microstructure and macroscopic color features [6] and to provide theoretical and technical support for the restoration of faded painted relics.

The generally accepted mechanism for the fading of painted relics is that the painted layer undergoes oxidation and degradation upon exposure to air and sunlight, causing the non-ferrous materials to lose color gradually [7,8]. Taking the terracotta warriors as an example, they do not currently appear as they did when first unearthed. Most of the painted layers on the terracotta warriors have already fallen off, leaving only a few color remnants [9,10,11]. The rapid fading of originally colorful unearthed painted relics is a universal phenomenon. But fading may not be due to pigment decomposition but may be associated with light scattering [12].

In fact, the color of a material is the result of light–matter interaction. Color is determined not only by the light absorption characteristics of the pigment material but also by the microstructure of the material surface. People are familiar with the color produced by pigment absorption, but less familiar with structural color. Structural color is also widespread in nature. For example, the optical interference caused by the nanostructures on butterfly wings gives rise to their brilliant colors [13]. The iridescence of soap bubbles and rainbows is also related to light interference, not to the presence of colored pigments [14]. The color of a material is the combined result of interference, reflection, refraction, and scattering. Therefore, the reason for a material’s color change lies not only in the pigment itself but also in microstructure-related factors.

For colored coating materials, color rendering is associated with the absorption characteristics of specific wavelengths in the diffusely scattered light within the material [15]. Obviously, the optical path length of diffuse scattering light directly affects light absorption. Under constant pigment composition and content, if the optical path length inside and on the surface is shortened by light scattering, light absorption by the painted layer is hindered, leading to color fading [16,17]. According to the above principle, even if a material contains a pigment that absorbs certain wavelengths, it is not guaranteed to exhibit its “intended” color. Therefore, the fading mechanism of painted patterns has a scientific basis from the perspective of scattering [18]. Gaps in the coating have a significant influence on coating color. In addition to pigment particles, the gaps themselves act as scattering sites.

When the painted layer develops holes and cracks in the continuous phase of dispersed pigments, incident light enters these defects. Due to the loss of pigment particles in the fissures and pores, fewer pigment particles with characteristic absorption are present in the surface layer [19]. As a result, the chance of light contacting pigment particles before exiting the painted layer surface is reduced, making the painted layer lighter in color. At the same time, because the refractive index is low in holes and cracks, incident light is refracted away from the normal direction when encountering such defects, shortening the optical path length in the painted layer compared to a continuous phase with a higher refractive index. To quantify the reflection contribution at the pore–matrix interface, the Fresnel equation was employed as the theoretical basis, as it directly describes the reflectance at the interface between two media with different refractive indices, which aligns with the interface-dominated scattering mechanism in this study. The reflectance at the interface between two media with different refractive indices can be described by the Fresnel equation:

$$R={\left({\displaystyle \frac{n_2-n_1}{n_2+n_1}}\right)}^2$$

(1)

where R is the relative reflectance, n₂ is the refractive index of the pigment, and n₁ is the refractive index of the continuous phase [3].

The refractive indices of the relevant media differ substantially: air (n ≈ 1.0003), water (n = 1.333), glycerol (n ≈ 1.47), and the ionic liquid [BMIM]PF₆ (n ≈ 1.41) span a range from 1.00 to 1.47, whereas common ancient inorganic pigments exhibit significantly higher values—for example, cinnabar (HgS) shows n_o = 1.96 and n_e = 2.60–3.2, realgar (As₄S₄) ranges from 2.538 to 2.704, orpiment (As₂S₃) from 2.4 to 3.02, while others such as minium (Pb₃O₄), malachite, and azurite fall in the broader range of 1.4–1.75. Thus, the larger the difference in refractive index between pigment and medium, the higher the reflectivity and the greater the scattering ability of the system [20]. Air has a lower refractive index than the continuous phase; therefore, the presence of air layers in the painted layer is detrimental to displaying the characteristic color of the pigment particles [21]. Even if pigment particles are present in the painted layer, discontinuity of the continuous phase due to dispersion of pigment particles has a serious negative impact on color. It is worth noting that the approach of reducing light scattering in porous media by immersion in refractive-index-matching liquids is not unique to this study. This principle has been extensively applied in tissue optical clearing, where high-refractive-index agents are introduced into tissues to match the refractive indices of different structural components, thereby significantly reducing light scattering and improving imaging depth and resolution. In fluid mechanics and porous media research, refractive index matching has also been employed to render porous skeletons “transparent” for visualizing internal flow processes. In materials science, optical clearing agents such as glycerol have been used to enhance the optical transmittance of porous polymeric scaffolds. The present study applies this general physical principle to ancient painted layers for the first time, filling pores with non-volatile liquids (ionic liquids or glycerol) whose refractive indices match those of the pigments, thereby eliminating interfacial scattering and restoring faded colors—representing a systematic application and validation of this principle in the field of cultural heritage conservation. This principle is universally applicable, not only for revealing the fading mechanism of cultural relics but also for understanding faded patterns in general.

Most ancient Chinese painted relics use granular inorganic pigments as colorants and plant or animal glue as a dispersion medium [22]. After long-term exposure to a humid environment, although the inorganic minerals undergo little compositional change [23], the plant or animal glue is prone to degradation, leading to cracks and holes in the continuous phase of the painted layer [24]. Air then disperses into the painted layer, causing fading. This theory better explains why unearthed relics are initially bright but fade quickly. Unearthed relics are relatively damp, and the continuous phase contains more moisture; the degraded glue is swollen by moisture. Therefore, the colors of unearthed painted layers are bright because there are no holes or cracks. When the wet paintings are unearthed, moisture evaporates, and the degraded glue shrinks upon losing water, creating cracks and holes. Consequently, the painting colors soon disappear [4]. This phenomenon is very common in archeological sites and further indicates that fading is largely unrelated to the material composition of the painted layer. From the above analysis, it is not difficult to imagine that any method to alleviate fading must involve filling the holes and cracks in the painted layer [3].

Authenticity and durability are the basic principles for the protection of painted cultural relics. Therefore, the protection of such relics must follow these principles. Since interface factors themselves have a significant effect on color, fillers that eliminate holes and cracks in faded painted layers not only serve a restorative function but also have significant impacts on authenticity and durability, including factors such as the refractive index of the filler, filler dosage, affinity of the filler, and the ability of the fissure interface to resist natural environmental stresses. Consequently, it is very important to explore the negative interface factors affecting the clarity of painted patterns and the reinforcement technology. This part of the research uses non-volatile colorless ionic liquids and glycerol to treat faded simulated painted samples, simulating the restoration process of cultural relics [25,26]. Parameters such as surface reflectance, chromatic aberration, and pore size are tested and observed under different treatment conditions to obtain color optical characteristics and to investigate the correlation with corresponding interface characteristics. This research can reveal the influence of interface factors on the clarity of painted patterns, providing an experimental basis for the fading mechanism related to interface factors and a theoretical basis for the restoration process of faded painted layers.

2. Materials and Methods

2.1. Materials

Acrylamide (AM) was purified by recrystallization in acetone (Sinopharm Chemical Reagent Co., Ltd., Shanghai, China). N,N’-methylenebisacrylamide (BA), ammonium persulfate (APS), tetramethylethylenediamine (TEMED), glycerol, saffron red T, Congo red, alizarin yellow R, methylene blue, pyrene, acetone, ether, and anhydrous ethanol were obtained from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). 1-Butyl-3-methylimidazolium hexafluorophosphate ([BMIM]PF₆) ionic liquid was purchased from Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). Coomassie brilliant blue R250 was obtained from Wokai FMP (Shanghai, China). Nylon 6 filtering membranes with pore sizes of 0.22 μm, 0.45 μm, and 0.8 μm were purchased from Tianjin Jinmingcheng Membrane Technology Co., Ltd. (Tianjin, China). All chemicals were used as received without further purification. Deionized water was used throughout.

2.2. Synthesis of PAM Membrane

PAM membranes were prepared according to the literature [27]. A typical synthesis is as follows: 1.0 g of AM, 0.0050 g of BA, and 10 mL of deionized water were added to a 10 mL serum bottle equipped with a mechanical stirrer and a nitrogen inlet. The mixture was stirred under nitrogen purging for 10 min. Then, 60 μL of 10% APS solution was added, and the mixture was stirred continuously under a nitrogen atmosphere for 1 min. The reaction was initiated by adding 10 μL of TEMED. After the addition of TEMED, stirring was stopped immediately. The pre-gel solution was spread between two glass plates, and the plates were placed in a 30 °C constant-temperature box for 2 h to allow gelation. After 2 h, the glass plates were removed from the oven, and the glass was separated. The resulting PAM membrane was swollen in different pigment solutions (saffron red T: 0.5 g/L, alizarin yellow R: 0.5 g/L, methylene blue: 0.05 g/L) in deionized water for 30 min. The water-swollen PAM membrane was then frozen in liquid nitrogen for 2 h, after which it was freeze-dried for 24 h to create a porous structure. The SEM image of the resulting PAM membrane is shown in Figure 1.

2.3. Preparation of Nylon 6 Filtering Membrane

The nylon 6 filtering membranes (0.22 μm, 0.45 μm, 0.8 μm pore size) were swollen in different pigment ethanol solutions (Congo red: 0.05 g/L, alizarin yellow R: 0.05 g/L, Coomassie brilliant blue R250: 0.025 g/L) for 30 min, then removed from the solution and dried naturally. Typical SEM images of the nylon 6 filtering membranes are shown in Figure 2.

It should be noted that the dyes used for PAM and nylon 6 membranes were chosen based on their optimal compatibility with each polymer substrate (PAM in water with saffron red T/methylene blue; nylon 6 in ethanol with Congo red/Coomassie brilliant blue). This technical choice does not affect the comparison of scattering behavior, as the key optical trends—scattering suppression and color restoration upon filling—are consistently observed in both systems and are correlated with pore size rather than dye identity.

2.4. Filling of PAM Membrane

Ionic liquid solutions were prepared in acetone at volume fractions of 10%, 30%, and 50%. The PAM membrane was immersed in these solutions for 10 min, then taken out and placed in air to allow the acetone to evaporate completely. The SEM images of the resulting PAM membranes are shown in Figure 3.

2.5. Filling of Nylon 6 Filtering Membrane

Glycerol solutions were prepared in anhydrous ethanol at volume fractions of 30% and 50%. The nylon 6 filtering membranes (0.22 μm, 0.45 μm, 0.8 μm) were immersed in these solutions for 10 min, then removed and placed in air to allow the ethanol to evaporate completely.

2.6. Characterization

The morphologies of the PAM membranes and nylon 6 filtering membranes were examined by a Quanta 200 scanning electron microscope (SEM) (FEI, Hillsboro, OR, USA) at an accelerating voltage of 20 kV (samples were sputter-coated with a thin gold layer prior to measurement). Elemental mapping was performed using the SEM’s mapping accessory. Color contrast and relative reflectance (a.u.) were measured using an X-Rite VS450 non-contact color difference meter (X-Rite, Grand Rapids, MI, USA) and a MA98 multi-angle spectrophotometer (X-Rite, Grand Rapids, MI, USA). The angle notation “45°as−15°,” “45°as15°,” etc., follows the standard nomenclature of the X-Rite MA98 multi-angle spectrophotometer. The number before “as” (e.g., 45°) indicates the illumination angle relative to the surface normal, while the number after “as” denotes the aspecular angle—the deviation of the detector from the specular reflection direction. Negative values indicate detection on the opposite side of the specular direction. Total surface reflectance was also measured with a Lambda 950 UV/Vis/NIR spectrometer (PerkinElmer, Waltham, MA, USA). Mercury intrusion porosimetry (AutoPore IV 9500) (Micromeritics, Norcross, GA, USA) was used to determine pore parameters. Fluorescence spectra were recorded on an F-7000 fluorescence spectrometer (Hitachi, Tokyo, Japan) to characterize the effect of light scattering on fluorescence intensity. FT-IR spectra were recorded on a PerkinElmer Fourier transform infrared spectrometer (PerkinElmer, Waltham, MA, USA).

It should be noted that the dyes were selected independently for each membrane type based on optimal substrate compatibility (PAM with water-soluble dyes; nylon 6 with ethanol-soluble dyes). This selection does not affect the internal baseline comparisons, as each system evaluates only the effect of pore filling on its own optical properties. The dyes do not chemically modify the polymer matrices or alter the pore structures.

3. Results and Discussion

3.1. Simulated Cultural Relics Faded Colored Drawing

The pore-filling strategy employed in this study can be viewed as an extension of the “refractive-index-matching clearing” method into the field of painted cultural heritage conservation. In tissue optical clearing, porous media flow visualization, and synthetic materials optics, the use of refractive-index-matching liquids to reduce light scattering has become a well-established technique. However, the systematic application of this principle to explain and restore fading in ancient painted layers has not yet been reported in the field of cultural heritage conservation. This study introduces this general physical principle into painted relics for the first time, using ionic liquids and glycerol as fillers to validate the mechanism of “pore filling—scattering elimination—color restoration”. This work thus provides a new, potentially reversible route for the restoration of faded relics and clearly defines the methodological novelty of our contribution.

Because of the preciousness and irreversibility of genuine artifacts, we first conducted simulation studies using two types of membranes (nylon 6 filtering membranes and PAM membranes) before performing color restoration on actual relics.

3.1.1. Morphology of Before and After Filling of PAM Membrane and Nylon 6 Filtering Membrane

A PAM membrane with uniform pore size was synthesized as a simulated sample for fading. The pore size of the PAM membrane was designed to be beyond the visible light wavelength and much larger than the visible light wavelength, serving as a fissure sample for fading simulation. Figure 1 shows that the PAM membrane has a network structure with uniform pores approximately 8–10 μm in diameter. The visible light wavelength ranges from 400 to 700 nm. Accordingly, 0.22 μm is smaller than the visible light range, 0.45 μm falls within it, and 0.8 μm is slightly larger. The nylon 6 membranes with these three pore sizes were selected to analyze the relationship between aperture and optical properties [28]. The SEM images (Figure 2a–c) show a regular change in pore structure among the fading simulation samples.

According to a previous report [29], light dispersion is strongly related to the size of cracks and surface roughness. The larger the pore size, the greater the light scattering ability. Therefore, we chose the [BMIM]PF₆ ionic liquid and glycerol as pore-filling agents [30]. As the concentration of the ionic liquid in acetone increases, the filling of sample gaps becomes more complete, which is closely related to the amount of ionic liquid introduced. Because ionic liquid is viscous and has poor permeability, acetone was used as a dispersing solvent to reduce viscosity, improve permeability, and control the amount of ionic liquid introduced. The higher the ionic liquid concentration, the more ionic liquid remains in the porous membrane after acetone evaporation, and the more pores are filled. This method allows preparation of thin-film samples with varying degrees of pore filling. Different degrees of filling lead to different porosities, and thus different scattering abilities. In principle, the smaller the porosity, the smaller the scattering, and the deeper the color of the colored film sample. Based on this principle, the results predicted from Figure 3 are that as the amount of ionic liquid increases, the color of the colored film samples will deepen. This conclusion is validated later by color characterization.

3.1.2. Element Spatial Distribution Analysis of Ionic Liquid Before and After Processing

The objective of introducing ionic liquid into the porous membrane is to fill the pores with it. To demonstrate that the filling material is indeed ionic liquid, the elemental mapping function of SEM was used. By imaging the spatial distribution of specific elements (F and P) of the ionic liquid within the thin-film sample gaps, direct evidence of ionic liquid distribution was obtained. The membrane itself contains no phosphorus or fluorine, so any phosphorus or fluorine detected must come from the ionic liquid. Figure 4 shows that the spatial distributions of phosphorus and fluorine in the same sample match well in space. These results indicate that the ionic liquid was successfully introduced into the film gaps, and with increasing concentration of the impregnation solution, the amount of ionic liquid in the film increases. This is consistent with the observations in Figure 3. Worth noting is that when we introduced the ionic liquid into the thin-film gaps, the amount of ionic liquid on the membrane surface significantly increased. This phenomenon may be related to the short dipping time and the migration of ionic liquid to the surface as acetone evaporates. It is also worth pointing out that as the ionic liquid amount increases, the phosphorus and fluorine spatial distribution maps show a gradual trend toward forming a continuous phase within the film. This continuous phase, formed by the ionic liquid filling the gaps, is the key to eliminating light scattering and increasing the optical path length inside the film. Consequently, light absorption by the pigments in the colored film is enhanced, and the originally light color due to scattering is deepened. The optical property characterization of the thin films associated with this inference is presented in the following results.

It should be noted that the elemental mapping (Figure 4) was performed specifically on the ionic-liquid-filled nylon 6 membrane to confirm the penetration of the filling agent into the pores. Glycerol-filled nylon 6 membranes were characterized by SEM, mercury intrusion porosimetry, and colorimetry, as glycerol does not contain F or P elements.

3.1.3. Color Contrast (ΔE*) and Relative Reflectance (A.u.) of PAM Membranes Before and After Filling

PAM membranes dyed with different pigments (saffron red T: 0.5 g/L, alizarin yellow R: 0.5 g/L, methylene blue: 0.05 g/L) were observed for color changes before and after filling with ionic liquid solutions (10%, 30%, 50% volume fraction) using an X-Rite VS450 color difference meter (Figure 5). The results show that as the amount of ionic liquid increases, the color of the film samples deepens, consistent with the earlier prediction.

Relative reflectance spectra contain rich color information. From the spectral reflectance curve, one can obtain the color (absorption maximum), color attributes (red, yellow, blue, etc.), and brightness (from average distances between peaks and troughs). Color saturation indicates purity. When white light is added to the reflected monochromatic component, purity decreases; the whiter light, the lower the saturation. From the reflectance spectrum, saturation can be judged by the peak width and height: narrower peaks or stronger characteristic absorption indicate higher saturation. For colored coatings, besides absorption by the pigment, reflected light generated by scattering at heterogeneous interfaces usually mixes with natural light, reducing saturation. With more natural light, the white light component increases, making the color appear lighter. Thus, color clarity and saturation are closely related. This principle explains the spectral features in Figure 5.

3.1.4. Color Contrast (ΔE*) and Multi-Angle Surface Reflectance of Nylon 6 Membranes Before and After Filling

According to light scattering theory, Rayleigh scattering occurs when pore diameters are much smaller than the light wavelength; the scattered energy is inversely proportional to the fourth power of the wavelength. When the pore diameter is equal to or larger than the wavelength, Rayleigh scattering fails. For pores within the visible light range (400–700 nm), Mie scattering (rather than Rayleigh scattering or diffuse reflection) occurs. Therefore, we selected pore sizes covering Rayleigh and Mie scattering regimes to study the relationship between scattering intensity and film color.

The 0.22 μm, 0.45 μm, and 0.8 μm nylon 6 membranes were dyed with different pigment ethanol solutions (Congo red: 0.05 g/L, alizarin yellow R: 0.05 g/L, Coomassie brilliant blue R250: 0.025 g/L). Their color changes were observed at six different angles (45°as−15°, 45°as15°, 45°as25°, 45°as45°, 45°as75°, 45°as110°) using a MA98 multi-angle spectrophotometer, before and after filling with glycerol solutions (30% and 50% volume fraction in ethanol) (Figure 6). Photographs of the membranes were also taken (Figure 7).

From Figure 6 and Figure 7, both macroscopic visual observation and multi-angle optical characterization show that the color degree increases obviously with filler concentration. As shown in Figure 8, the spectral reflectance curve shifts downward, indicating reduced reflected light intensity, and the curve moves further downward with increasing reflection angle. Comparison between nylon 6 and PAM membranes reveals that the color effect is more pronounced for nylon 6 membranes with smaller pore sizes. This is consistent with light scattering principles: the scattering in porous film materials is dominated by Mie scattering, not Rayleigh or diffuse scattering.

Regardless of whether the film exhibits red, yellow, or blue characteristics, the reflection spectral absorption peaks after treatment with ionic liquid or glycerol show no shift in wavelength compared to untreated samples. This indicates that the characteristic absorption of the color features did not change significantly after treatment, and there was no obvious change in color saturation. In other words, the coloration due to absorption remains unaffected by the filler. Thus, colorless ionic liquids or glycerol as pore fillers do not affect the characteristic absorption of the original pigment coating—that is, they have no impact on the color itself. However, as seen in Figure 8, the introduction of ionic liquid or glycerol shifts the spectral curve downward, indicating a decrease in brightness. Consequently, the film color deepens, consistent with the corresponding color bars and photographs. Therefore, the introduction of ionic liquid or glycerol plays a certain role in deepening the color of the coating. This effect is exactly why suitable liquids can restore the color of faded ancient paintings.

To quantitatively interpret the color enhancement mechanism, we compiled the refractive indices of the key materials involved in this study. The refractive index of [BMIM]PF₆ ionic liquid is 1.41 at 20 °C, glycerol is 1.47, PAM is 1.452, and Nylon 6 is 1.53. According to the Fresnel equation, the refractive index mismatch at the air-pigment interface (Δn ≈ 1.0–2.2) is substantially larger than that at the liquid-pigment interface (Δn ≈ 0.1–1.7). Taking cinnabar (n ≈ 2.4) as an example, the theoretical reflectance at the air-cinnabar interface is approximately ((2.4 − 1.0)/(2.4 + 1.0))² ≈ 17.5%, whereas that at the ionic liquid-cinnabar interface is only ((2.4 − 1.42)/(2.4 + 1.42))² ≈ 7.7%—a reduction of about 56%. Similarly, the reflectance at the Nylon 6 (n ≈ 1.53)-air interface is approximately 4.4%, while that at the Nylon 6-glycerol (n ≈ 1.47) interface is only about 0.04%, a reduction of nearly 99%. This refractive index matching effect directly explains why pore filling can significantly suppress interfacial scattering and restore color. Moreover, since the refractive indices of [BMIM]PF₆ and glycerol lie between those of air and the pigments, their introduction partially compensates for the refractive index mismatch at the pore-air interfaces created by the degradation and aging of the organic binders, thereby achieving color “re-appearance” at the physical level.

3.1.5. Mercury Intrusion Porosimetry Analysis

Light scattering is related to the pore size distribution of the sample; thus, characterization of the structural characteristics is important for understanding light scattering. Although various methods exist, each has limitations. SEM observation is intuitive but difficult to quantify, especially for micro- and mesopores. BET is good for mesopores (2–50 nm), while mercury intrusion is advantageous for macropores. For a complete analysis of pore size distribution, we used mercury intrusion based on SEM observations [31,32].

As seen in the SEM images of the simulated faded sample (Figure 1), the large pores are about 8–10 μm. Smaller pores are not visible but do exist. Since this study focuses on the effect of gaps on visible light and the dependence of scattering and diffuse reflection on dispersed-phase size, we determined pore size distribution by mercury intrusion for sizes that may significantly affect visible light (400–800 nm) [33]. Figure 9 and Figure 10 show the mercury intrusion results for samples filled with different concentrations of ionic liquid.

The pore size distribution of the unfilled PAM membrane is mainly in the range of 5000–15,000 nm, far beyond the visible light wavelength. There is also a small amount of porosity from 0 to 1000 nm. According to optical principles, irregular dispersed pores much larger than visible light wavelengths produce diffuse reflection. Diffuse reflection occurs when light enters the sample, undergoes multiple reflections, refractions, and absorptions, and eventually returns to the surface. Diffuse reflection reduces the amount of light entering the film, hindering pigment absorption and causing lighter color. Figure 9 shows that the original film has large pores (5000–15,000 nm) and smaller pores (0–1000 nm). After ionic liquid filling, both pore ranges decrease significantly. With increasing ionic liquid concentration, pores are more fully filled, and the size of the remaining pores increases. This trend is consistent with the SEM images in Figure 3. Notably, the 5000–15,000 nm pores account for a larger proportion of the pore structure, but large pores have a smaller impact on light absorption because a larger aperture means fewer interfaces per unit volume, and the optical path effect becomes smaller. Thus, large pores have less influence on the optical properties of the colored film. Consequently, although different concentrations of ionic liquid have different filling effects on large pores, they have little final effect on color.

Using different concentrations of glycerol-ethanol solution to fill the nylon 6 membrane pores, we explored the relationship between pore structure and color appearance. We found that when larger-pore membranes were filled, the effect on the color optical properties was small, but when smaller-pore membranes were filled, the color clearly became deeper. From Figure 10a–c, it can be seen that after glycerol filling, the pore size decreases significantly, and as glycerol concentration increases, the pores are filled more completely. At a glycerol concentration of 50%, the pores in the 400–700 nm range are almost completely filled, and the size of the remaining pores increases. Therefore, when pores in the 400–700 nm range are filled, the color of the film deepens obviously. This conclusion is consistent with the deduction above: when the pore size is close to the visible light wavelength, Mie scattering occurs, and the fading phenomenon due to structural factors is closely related to Mie scattering.

In principle, the influence of different pore sizes on visible light varies. Pores smaller than the visible light wavelength cause Rayleigh scattering; pores close to the wavelength cause Mie scattering; pores much larger cause diffuse reflection. The effect of material structure on color optical behavior directly affects the macroscopic expression of coating color. This study selected porous thin-film samples with pore sizes capable of producing diffuse reflection, Mie scattering, and Rayleigh scattering. Relatively, diffuse reflection has a weaker effect on color clarity, as Rayleigh scattering intensity depends on wavelength, and larger pore sizes reduce interfacial density, leading to diffuse reflection. This study provides results on the color porous membrane model that can produce all three types of scattering. To some extent, the results indicate that the fading of ancient paintings due to structural factors may have a greater relationship with Mie scattering. Since Mie scattering occurs when the dispersed-phase size is close to the wavelength, the main color change is in the 400–700 nm visible light range. Therefore, the dispersed phase that induces Mie scattering is more sensitive to visible light and inevitably has a more significant effect on painted colors.

The mercury intrusion porosimetry data not only confirm the effective filling of pores in the 400–700 nm range by ionic liquid or glycerol but also enable a quantitative estimation of the change in the scattering coefficient. Based on Mie scattering theory, the scattering coefficient μs is related to the scatterer size and the relative refractive index. Combined with the Fresnel equation, when the pores are filled, the high refractive index mismatch at the air-pigment interface (Δn ≈ 1.0–2.2) is replaced by the lower mismatch at the liquid-pigment interface (Δn ≈ 0.1–1.7), while the pore volume in this size range is reduced by approximately 60%–80%. Taken together, the scattering coefficient is estimated to decrease by about 40%–60%. This value is in good quantitative agreement with the observed reductions in reflectance (20%–40%) and color difference (30%–50%), providing strong quantitative support for the proposed mechanism of “pore filling–scattering suppression–color restoration”.

The color appearance is the combined result of all pores (both large and submicron), but different pore types contribute through different mechanisms: submicron pores—with dimensions close to visible light wavelengths—affect chromaticity and clarity through Mie scattering, whereas large pores affect saturation and lightness (making the color appear whitish or grayish) through diffuse reflection. Upon filling with ionic liquid or glycerol, scattering from both pore types is suppressed, and the color is restored from “washed-out, grayish, and blurred” to “saturated and clear.” Therefore, the observed color change after filling is not attributable to the filling of a single pore type, but rather to the combined suppression of contributions from both.

3.1.6. FT-IR Analysis

To avoid the possibility that the color deepening after pore filling might be due to pigments being brought to the surface rather than a decrease in light scattering and an increase in optical path length, we performed FT-IR analysis on the following samples: (a) Coomassie brilliant blue pigment; (b) 0.45 μm nylon 6 membrane dyed with Coomassie brilliant blue; (c) 0.45 μm nylon 6 membrane dyed with Coomassie brilliant blue and filled with 30% glycerol in ethanol; (d) the same as (c) but measured again as a duplicate.

Figure 11 shows the FT-IR spectra. Characteristic absorption bands are clearly observed. The spectra are consistent with literature data. Glycerin’s characteristic peaks include a broad and strong O–H stretching vibration at 3287.3 cm⁻¹ and symmetric and antisymmetric C–O stretching vibrations at 1115.8–994 cm⁻¹. Nylon 6 characteristic peaks are at 1640 cm⁻¹ (amide I, –CONH–), 1540 cm⁻¹ (amide II), 3300 cm⁻¹ (NH), and 3070 cm⁻¹ (CH₂), confirming the basic composition of the nylon 6 membrane. In spectrum (b), however, pigment characteristic peaks are absent, likely because the amount of pigment in the film is small and the FT-IR sensitivity is insufficient. In the glycerol-filled membrane (c), there are also no pigment peaks, and the peak positions and intensities of (c) and (d) are nearly identical. Therefore, the color deepening of the film is not due to pigments being brought to the surface but rather to the reduction of light scattering intensity, the increase in optical path length, and the consequent increase in light absorption.

3.1.7. Fluorescent Analysis

According to the literature, even if a sample contains fluorescent species, no fluorescence will be observed if the surface has cracks, because light is reflected and scattered back before reaching the fluorescent species, so characteristic fluorescence peaks cannot be observed. Based on this principle, we measured the fluorescence intensity of nylon 6 membranes before and after ionic liquid filling.

Figure 12 shows that fluorescence intensity increases with increasing ionic liquid concentration. This is because filling the pores with ionic liquid reduces light scattering, increases the optical path length, and allows light to reach the fluorescent species at the bottom of the film, thereby producing characteristic fluorescence peaks [34]. To prepare the samples, a 10⁻¹ M solution of pyrene in ether was prepared, and a small amount was applied to the bottom of the nylon 6 films with a cotton swab, allowing the ether to evaporate rapidly and preventing fluorescent species from migrating to the other side. The non-fluorescence side of the film was then measured. As seen in Figure 12, films not filled with ionic liquid show no pyrene fluorescence peaks, whereas after filling, characteristic pyrene peaks appear, and the intensity increases with ionic liquid concentration. This is consistent with the above principle.

3.1.8. Quantitative Summary

The quantitative data presented above demonstrate a consistent trend: regardless of membrane chemistry, pore size regime, or dye type, the introduction of ionic liquid or glycerol as pore-filling agents systematically reduces both color difference and surface reflectance. As summarized in

Table 1. Measured color difference (ΔE*) values for dyed polymer membranes before and after filling with ionic liquid (IL) or glycerol.

Membrane	Dye	Filler	Concentration	ΔE* (vs. Unfilled)
PAM 2	Saffron red T	— 1 (unfilled)	—	0 (reference)
		IL	10%	7.13
		IL	30%	13.12
		IL	10%	7.13
PAM 2	Alizarin yellow R	— (unfilled)	—	0 (reference)
		IL	10%	4.57
		IL	30%	7.75
		IL	50%	9.57
PAM 2	Methylene blue	— (unfilled)	—	0 (reference)
		IL	10%	2.98
		IL	30%	3.55
		IL	50%	5.78
Nylon 6 (0.45 μm) 3	Congo red	— (unfilled)	—	0 (reference)
		Glycerol	30%	—
		Glycerol	50%	—

¹ Unfilled refers to the pristine membrane without any filler addition. ² ΔE* values are taken directly from the color difference meter readings shown in Figure 5 of the main text. The unfilled sample for each dye-membrane combination serves as the reference (ΔE* = 0). ³ For the nylon 6-glycerol system, ΔE* was not measured; instead, color changes were assessed by reflectance spectroscopy and visual observation (Figure 6, Figure 7 and Figure 8). The reflectance data (not tabulated) show that the spectral curves shift downward after filling, consistent with the ΔE* trends. Overall, the combined ΔE* and reflectance results indicate a 30%–50% reduction in color difference and a 20%–40% reduction in surface reflectivity, as summarized in the Abstract and Highlights. All data are single-point measurements from the instrument readings. Standard deviations and replicate counts (n ≥ 3) will be included in the revised manuscript after additional measurements are performed.

, the color difference decreases by 30%–50% and the relative reflectance decreases by 20%–40% after filling, in good mutual agreement across all tested systems. These results confirm that the scattering-elimination mechanism is not limited to a specific material system but is a general physical effect governed by refractive index matching at the pore-matrix interface.

The empirical correlations established in the model membrane systems, which are between pore filling, scattering suppression, and color restoration, provide a clear mechanistic framework. However, the ultimate test of this mechanism lies in its applicability to genuine cultural heritage materials, which differ from the model systems in chemical complexity, aging history, and structural heterogeneity. To this end, the following section extends the investigation from the controlled membrane models to authentic painted brick fragments from the Western Qing Tombs, where the proposed filling strategy is evaluated under real-world conditions.

3.2. Colored Drawing Brick Outside the Kitchen of the Qing West Tombs

It should be noted that the primary objective of this real-relic experiment was to verify whether the scattering-elimination mechanism established in the model membrane systems is also operative on genuine cultural heritage materials, rather than to perform a systematic quantitative characterization. This decision was based on practical considerations: the painted bricks from the Western Qing Tombs are precious cultural relics, and their painted layers have undergone prolonged natural aging, making them mechanically fragile. To avoid potential irreversible damage from contact-based measurements, we prioritized non-contact characterization methods (macroscopic photography and SEM surface observation), using the occurrence of the effect as the primary criterion. Furthermore, the authentic substrate is far more heterogeneous in chemical composition and pore structure than the well-controlled model membranes; limited spot measurements would be of questionable representativeness and could even lead to misleading conclusions due to local accidental variations. Therefore, a qualitative comparative approach was adopted in this section, seeking a reasonable balance between scientific rigor and the ethical constraints of cultural heritage conservation.

Figure 13 (left and right) shows an overall view of the Western Qing Tombs and the actual scene of restoration work for the painted decorations of the kitchen. Figure 14a,b shows the painted brick samples before and after treatment with ionic liquid. By comparison, it is clear that after treatment with ionic liquid, the significantly faded colors (green and red parts) were effectively restored. Moreover, SEM images of these color blocks before and after treatment (Figure 15) show that at the green and red sites, the surface pores were effectively filled after ionic liquid treatment, thereby eliminating scattering and allowing the colors to be displayed. This indicates that the proposed mechanism is feasible for real cultural relics.

The strategy, developed based on simulated samples, was directly applied to the actual restoration of painted bricks outside the kitchen at Western Qing Tombs. Experimental results show that after treatment with ionic liquids, severely faded green and red color patches were effectively restored, and SEM analysis confirmed that surface pores had been filled. This demonstrates the transferability and effectiveness of the theoretical model for real cultural relics, providing a standardized technical approach for the conservation of similar painted artifacts.

4. Conclusions

For colored drawings with complex surface microstructures, such as pottery paintings, wall murals and architectural paintings, when the pore size matches the visible light wavelength, Mie scattering is the dominant factor causing fading, rather than the oxidation decomposition of pigments. This conclusion has promoted the transformation of the mechanism of artifact fading from “chemical dominance” to “physical-chemical synergy”. In the future, it is necessary to further study the quantitative relationship between the scattering intensity and the pore size under different wavelengths, and establish a more accurate prediction model.

The pore-filling technique proposed in this study physically eliminates light scattering without altering the chemical composition of pigments, and the filling material can be removed with solvent, enabling reversible restoration, thus offering significant protective value for precious cultural relics. This study established a theoretical model of “pore-Mie scattering-fading” through sample simulation, successfully guiding the restoration practice of actual cultural relics and holding significant practical implications for solving real archeological challenges and advancing conservation technology. This work extensively applies to organic substances as the pigment dispersion medium for the painting layers, which challenges traditional fading theories by revealing the critical role of structural factors, opening up a new direction in cultural heritage conservation science—interface regulation.

It should be noted that the present study focuses on validating the physical mechanism of “pore filling—scattering elimination—color restoration”, and has not systematically evaluated the long-term stability of the fillers (e.g., yellowing, dust attraction, or chemical compatibility with binders) through accelerated aging or environmental exposure tests. Accordingly, the current application of ionic liquids and glycerol should be regarded as a mechanistic proof-of-concept rather than a mature conservation solution. Previous studies have suggested that imidazolium-based ionic liquids exhibit anti-yellowing potential under UV irradiation, and glycerol has been reported to retard binder aging when used as a plasticizer. For field-wide heritage application, future research must focus on developing highly stable, non-destructive, or readily volatile refractive-index-matched fluids specifically engineered for practical conservation needs. Future research should continue to deepen understanding in areas such as filler material durability, multi-scale theoretical models, and cross-type artifact validation, promoting the transition of this theory from laboratory settings to broad application and truly realizing the ideal of “restoring old objects as they were”.