Developing efficient cleaning agents for textile relics affected by mold stains

Abstract

Mold causes significant damage to textile-relics, yet specific cleaning-agents are underdeveloped. This study systematically evaluated surfactant-bio-enzyme-bacteriostat ternary systems for cleaning textiles infected with Trichoderma SM4, Aspergillus SM12, and Rhizopus BF3. The optimal formulation was highly dependent on the mold-fiber combination. Rhamnolipid achieved 93.32% removal rate for Rhizopus-infected textiles, while C₈E₉ showed over 94.35% efficacy on cellulose fibers infected with Aspergillus SM12 or Trichoderma SM4. Alkaline protease was broadly effective on protein fibers (>73.17%), whereas cellulase demonstrated wide-spectrum efficacy on cellulose (>80.77%). Meanwhile, studies also demonstrate that surfactants alone removed surface mycelia, while the ternary system eliminated mycelia, mucus, and pigments, restoring the fiber’s microscopic morphology. The optimization system was successfully applied to the Qing Dynasty Miao-ethnic embroidered pieces, achieving near-complete removal of mold without causing any damage to embroidery. Therefore, this study presents a rationally designed, effective, and safe cleaning methodology for mold-affected textile relics, offering significant potential for conservation.

Introduction

Textile relics, as material carriers of ancient Chinese civilization, embody the wisdom and creativity of the Chinese nation^1,2,3. These relics crystallize the wisdom and creativity of the Chinese nation, documenting the evolution of ancient aesthetic arts, textile technology, and manufacturing processes, while also recording cultural distinctions among various ethnic groups, thereby highlighting the diversity of Chinese national culture^4,5,6. Simultaneously, textile relics are organic artifacts composed of carbohydrate materials such as cellulose fibers (cotton/linen) and protein fibers (silk/wool)^5,6,7. Moreover, molds, a category of microorganisms capable of decomposing these cellulose/protein fibers for nutrients, readily colonize such materials^8,9,10. Additionally, molds also have the characteristics of biodiversity, rapid proliferation rates, broad distribution, and strong adaptability^5,10,11. And during the growth and reproduction metabolic process, molds secrete organic acids, adhesive substances, and pigments (yellow, red, green, cyan, brown, black, etc.) that penetrate into the interior of fibers (see Fig. 1)^11,12,13. Consequently, when textile relics are stored in environments with poor ventilation, high humidity, excessive dust, or elevated temperatures, textile collections are highly susceptible to mold infestation^12,14,15. Especially in hot-humid conditions, infected molds thrive and multiply rapidly, causing irreversible damage within short periods through fiber decomposition, oxidation of reductive groups, reduction in degree of polymerization, destruction of chemical components, decline in mechanical strength, obscuration of color patterns and textile structures, and shortened lifespan, and thus leading to irreparable losses to the historical, scientific, and artistic values carried by these relics^14,15,16. Therefore, removing mold stains without damaging surface-related information remains a critical focus for conservation professionals in their efforts to preserve and process these invaluable relics.

Fig. 1: Examples of textile relics infected with various mold stains.

a Textile artifacts unearthed from the tomb of Qing Dynasty military official in Shijingshan. b A stone-blue round-collared magua (riding jacket) with four medallions of multi-colored clouds, bats, and golden dragons in leno weave. c A lightblue long-sleeved silk gown for women from the Han Dynasty, excavated at the Niya Ruins and the ancient city of Loulan in Xinjiang. d A yellow satin dragon robe for women embroidered with multicolored five bats and flat gold Fo (Buddha) characters (Provided by the China National Silk Museum). e A half-sleeve short jacket worn in the Yuan Dynasty (1271–1368).

However, contemporary research predominantly focuses on addressing issues related to the preservation and aging of ancient textiles, microbial species identification in textile artifact stains, optimization of mold cultivation and physiological characterization, as well as anti-mold measures for storage environments like repositories^5,16,17. Studies on cleaning technologies specifically for mold-stained textile relics are scarce. For instance, J. Y. Tian et al. investigated molds on silk textiles, identifying Rhizopus, Penicillium, and Aspergillus niger as primary genera responsible for biodeterioration¹⁷. J. He et al. isolated and identified molds on leather, revealing Aspergillus and Penicillium as dominant genera¹⁸. Y. Y. Xiao et al. extracted, isolated, and purified molds from mildewed clothing, identifying causative agents including Rhizopus stolonifer, Penicillium corylophilum, Aspergillus flavus, Cladosporium cladosporioides, Penicillium citrinum, Aspergillus versicolor, and Phoma spp., with Aspergillus and Penicillium being predominant¹⁹. Furthermore, existing studies also indicate that molds on textiles are closely related to fungal species in their storage environments, suggesting that textile artifacts may be susceptible to a wide variety of molds, which vary with differences in preservation conditions^4,20,21. Additionally, a survey on ‘susceptible’ microorganisms in Chinese textile relics preservation identified Trichoderma SM4 (T. SM4), Aspergillus SM12(A. SM12), and Rhizopus BF3 (R. BF3), as common microbial diseases in textile relics ²².

Meanwhile, according to stain cleaning theory, stain removal involves separating stains from substrates through the wetting, emulsification, penetration, solubilization, dispersion, and foaming effects of cleaning agents^23,24,25. Clearly, cleaning agents are central to this process, serving two primary functions: removing surface contaminants and dispersing/suspending stains to prevent redeposition^25,26,27. The composition of cleaning agents (including types of surfactants, bio-enzymes, antimicrobial agents, etc.) and variations in their formulations cause significant differences in surface tension, interfacial properties, solubility, wettability, emulsification, and dispersion capabilities, as well as their compatibility, thereby affecting the final cleaning efficacy and performance on substrates^28,29,30. Moreover, studies have shown that different types of molds, depending on their taxonomic classification, produce distinct metabolic by-products (such as viscous secretions/pigments)^23,31,32. Additionally, due to the preciousness and rarity of textile artifacts, the criteria for their cleaning differ significantly from those for ordinary textiles. That is, the principle of “cleaning to retain the aged appearance” must be followed: ensuring the removal of harmful substances while maximizing the preservation of the historical authenticity and long-term stability of the artifacts. Furthermore, existing research indicates that inappropriate cleaning agent formulations or cleaning treatments can have comprehensive impacts on the core preservation states of textile artifacts, including their appearance, color, fiber structure, and physical strength^33,34,35. Hence, the development of cleaning system for textile artifacts must comprehensively consider the precious nature of the artifacts, the specificity of fiber substrates (which often exhibit varying degrees of deterioration or even carbonization), the complexity of mold metabolic products, the safety and efficiency of cleaning agents (including the wetting and penetration capabilities of surfactants, the specific degradation effects of bio-enzymes, and the antimicrobial and regeneration properties of antimicrobial agents), as well as the combined interactions among these factors. And, the evaluation of cleaning effectiveness must encompass multidimensional indicators such as macroscopic morphology, color stability, microstructure, and mechanical integrity ^25,26,36.

Therefore, this study selects dominant microbial species commonly found as diseases in textile relics- Trichoderma SM4, Aspergillus SM12, and Rhizopus BF3-as targets to simulate the mildew infestation of textile relics and prepare substitute samples of mildewed textile. Building on this, surfactants such as tea saponin, rhamnolipid, and isotridecyl alcohol ethoxylate, bio-enzymes including alkaline protease, cellulase, and lipase, as well as antimicrobial agents such as sodium benzoate, zinc sulfate, and potassium sorbate, were selected for research on blended systems. And cleaning efficacy was systematically evaluated through methods including appearance imaging, color difference analysis, mechanical testing, and scanning electron microscopy observations. The findings are expected to investigate the individual and combined effects of different surfactants, bio-enzymes, and antimicrobial agents on cleaning performance, elucidate the mechanisms of key components in removing mold stains and their metabolites, and provide theoretical foundation and technical solutions for the scientific cleaning and conservation of textile relics.

Methods

Research framework and content

The research framework and content of this study were divided into the following three parts (see Fig.2). The first part focuses on the construction of the research foundation, specifically the preparation of simulated samples of mildewed textile relics. By selecting four typical types of textiles—cotton, linen, silk, and wool—and subjecting them to artificial hydrolysis aging and contamination with specific molds (Trichoderma SM4, Aspergillus SM12, and Rhizopus BF3), alternative samples of mildewed textile relics for cleaning studies are prepared. Basic data, including macroscopic morphology, colorimetric measurements, and microstructural characteristics, are systematically collected. The second part constitutes the core experiments and mechanistic analysis, focusing on the screening and optimization of high-efficiency cleaning agents. Based on a four-level formulation design involving single surfactants, additive incorporation, enzyme compounding, and the introduction of antibacterial agents, ultrasonic cleaning technology is employed to treat the simulated mildewed samples. By comparing color difference changes (Reflectance Color Difference, RCD value), macroscopic and microscopic morphological features before and after cleaning, the study systematically investigates the relationship between the components of the cleaning agents (surfactants, enzyme preparations, antibacterial agents) and the cleaning efficiency of various mold stains, as well as the associated damage to the substrate fibers. This process ultimately identifies the optimal combination of cleaning agents for different fiber materials. The third part involves physical validation. The optimized cleaning formulations are applied to authentic ethnic relics (e.g., Miao embroidery) for empirical research. By comprehensively utilizing scanning electron microscopy to observe the microscopic morphology of fibers and residual mold before and after cleaning, the safety of the developed cleaning agents for the cultural relic substrates is evaluated. This confirms their potential as cleaning agents for mold-contaminated textile relics.

Fig. 2: Research framework and content of this study.

a Preparation of simulated mildewed textile relics through artificial aging and mold contamination. b Screening and optimization of cleaning agents using ultrasonic technology. c Physical validation on authentic ethnic relics (Miao embroidery).

Experimental materials and equipment

To achieve the objectives outlined above, and considering the precious value of textile artifacts^6,17,26, cotton, linen, silk and wool fabrics from Anhui Tuoyan Experimental Materials Co., Ltd. (Hefei, Anhui) were selected as the primary experimental materials, as actual textile relics are prohibited from use in experimental work. Detailed specifications of these textiles are provided in Table 1. Additional materials used in the experimental study included: hydrochloric acid solution, solid sodium hydroxide particles (Sinopharm Chemical Reagent Co., Ltd.), potato infusion powder (Yeasen Biotechnology), agar powder (Shanghai Hushi Laboratory Equipment Co., Ltd.), glucose (Macklin Biochemical Co., Ltd.), ethanol solution (Sinopharm Chemical Reagent Co., Ltd.), deionized water (laboratory-prepared), Trichoderma SM4, Aspergillus SM12, and Rhizopus BF3 (self-cultured), sodium citrate (Hangzhou Gaojing Fine Chemical Co., Ltd.), tea saponin, rhamnolipid (Shanghai Yuanye Bio-Technology Co., Ltd.), isooctyl alcohol polyoxyethylene ether (C₈E9; Tokyo Chemical Industry Co., Ltd.), sodium oxalate (Aladdin Biochemical Technology Co., Ltd.), alkaline protease (XiYa Reagent), cellulase (Macklin Biochemical Co., Ltd.), lipase (Bide Pharmatech), sodium benzoate, zinc sulfate, and potassium sorbate (Sinopharm Chemical Reagent Co., Ltd.). In addition, the following equipment was utilized during the investigation: a biochemical incubator (Shanghai Yiheng, LRH-250F), laminar flow cabinet (Suzhou Jinghua, SW-CJ-1D), high-pressure steam sterilizer (STIK, IMJ-85A), digital camera (Sony, ZV-1F), fabric density meter (Y511B), fabric thickness gauge (YG141), electronic balance (YH-C30001), air-drying oven (DG-F75366BCX), fabric strength tester (YG026HC), high-definition video microscope (SCD-TZ500G5), water bath shaker (SHA-C), and a scanning electron microscope (SEM; Guoyi Quantum, SEM3200).

Table 1 Specification of textiles used in this work

Experimental process and design

To achieve the aforementioned research objectives, the research process was divided into two phases: the preparation of simulated samples of mold-stained textile relics and the development of high-efficiency cleaning agents, with the detailed procedures outlined as follows: In the preparation phase of substitute samples for mold-contaminated textile relics, the process mainly comprises textile aging and fungal contamination of the aged textiles. The steps for the aging treatment are as follows: First, commercially purchased cotton, linen, silk, and wool textiles were cut into 5 × 5 cm pieces. Then were cleaned three times in a 20% (w/v) ethanol aqueous solution (500 mL) to completely remove surface impurities. Subsequently, cotton, linen, and wool textiles were immersed in 1 mol/L hydrochloric acid solution (500 mL), while silk textiles were placed in pH 14 sodium hydroxide solution (500 mL). Additionally, to prevent solution spillage, the beakers containing the aging solution and fabric were sealed with aluminum foil during the experiment. Cotton and linen samples underwent hydrolytic aging in a 70 °C constant-temperature water bath for 6 hand 4 h, respectively, while wool and silk samples were treated at 65 °C for 50 min and 48 h respectively. After hydrolytic aging, textiles were rinsed with deionized water until neutral pH and dried in a 35 °C forced-air drying oven to obtain second-level aged textiles samples. The steps for the fungal contamination of aged textiles, the process included: Preparing PDA solid medium (2% glucose [8 g], 6 g/L potato infusion powder [2.4 g], 1.5–2% agar [1.7 g] in 200 mL total volume); Material Sterilization (Seal conical flasks with aluminum foil and sterilize in a high-pressure steam autoclave at 121 °C for 20 min); Preparing colony suspensions (culturing Trichoderma SM4, Aspergillus SM12, and Rhizopus BF3 strains on PDA plates at 30 °C for 5 days, then harvesting spores with 30 mL sterile water, and prepared spore suspension with a concentration of ${10}^{5}$ CFU/mL, where CFU stands for colony-forming unit.); Inoculating aged textiles (transferring 5 mL dense fungal hyphae and spores onto samples using inoculation loops, and culturing at 30 °C for 5 days with 5 mL deionized water sprayed every 12 h to maintain moisture, ultimately yielding substitutes of mildew-infected textile relics). The detailed preparation workflow was illustrated in Fig. 3.

Fig. 3: Preparation flow chart of simulated samples of textile relics infected with mold.

a Textile aging process including cutting, cleaning, and hydrolytic aging. b Fungal contamination process including PDA medium preparation, sterilization, colony suspension preparation, and inoculation.

In the development stage of efficient cleaning agents for mold-infected textile relics, the primary work involved testing the macroscopic morphology, microscopic morphology, and color difference of mildew-stained textiles before and after cleaning with different formulations. The detailed cleaning formulations are provided in Table 2. Additionally, to ensure that any differences in cleaning efficacy among the various formulations could be attributed solely to the formulation itself, ultrasonic cleaning technology was selected as the experimental methodology due to its gentle yet effective cleaning performance and excellent compatibility with fragile historical textiles^35,36. The specific cleaning parameters were set as follows: ultrasound frequency (40 kHz), temperature of washing water (40°C), main wash bath ratio (1:20), main wash cycle (1), dosage of cleaning agents (1 g/L), rinse bath ratio (1:30), rinse cycles (2), rinse time (20 min/each time). After cleaning, the samples were laid flat to dry under controlled indoor environmental conditions (temperature 20 °C ± 2 °C, relative humidity 65 ± 2%) to ensure that any observed changes in the test samples’ performance were attributable solely to the washing treatment. Furthermore, to validate the effectiveness of the developed cleaning formulation, this study conducted physical verification experiments by applying the optimized cleaning formulation to the overall style of the elegant and simple natural-colored silk Miao embroidery pieces. The embroidery featured a pink floral motif symbolizing happiness, prosperity, and other auspicious meanings, complemented by brown branches and green leaves, demonstrating the exceptional craftsmanship of folk artisans while reflecting the folk culture and aesthetic values of a specific historical period. The specific cleaning formulation of the cleaning process was as follows: C₈E₉ (200 g/L), Water (750 g/L), Sodium oxalate (50 g/L), Alkaline protease (1000μ/g), Potassium sorbate (0.12 g/mL), Sodium citrate (adjusted to Ph≈7). The cleaning procedure was executed as follows: First, the cultural relic-embroidery piece was placed on a clean operating platform underlaid with sterilized cotton gauze. Next, the prepared cleaning agent was dissolved in 100 ml of deionized water to create a cleaning solution, and placed the artifact (natural-colored silk Miao embroidery pieces) into the obtained-cleaning solution to form a mixture of relics and cleaning solution. Subsequently, the mixture was subjected to ultrasonic cleaning under controlled parameters: vibration frequency (40 Hz), temperature (40°C), bath ratio (1:20), and duration (20 minutes). Upon completion, the artifact was rinsed twice with deionized water until the reinstate reached visual clarity, marking the conclusion of the cleaning experiment. Finally, the cleaned artifact was laid flat to air-dry under controlled indoor environmental conditions maintained at 20 ± 2 °C and 65 ± 2% relative humidity.

Table 2 Detail cleaning formulations

Experimental testing indicators and methods

To validate the aforementioned research objectives and comprehensively characterize the cleaning efficacy of the developed agent, indicators such as appearance, color difference, and mechanical properties were evaluated. The detailed testing procedures were performed as outlined below.

In order to investigate the relationship between cleaning agent formulations and cleaning effect of textile relics, the macro-morphology of simulated sample of textile relics before and after the treatment of different cleaning formulations was individually measured with the help of digital camera (SONY ZV-1F) and automatic optical microscope (GP-300C).

In order to quantitatively evaluate the effect of different cleaning formulations on the washing efficiency of textiles affected by mold stains, the color difference of simulated samples of the textile relics infected with mildew before and after washing with different cleaning formulations was tested at 5 spots using whiteness meter (Data color 650). To ensure the stability and repeatability of the results, the average color difference of sample tested for 5 times was taken as the final value of tested sample for the calculation of cleaning efficiency. Additionally, to quantify the cleaning effect, the relative change in color difference (RCD) was calculated using Eq. (1):

$${RCD}=\frac{{\Delta E}_{b}-{\Delta E}_{a}}{{\Delta E}_{b}}\times 100 \%$$

(1)

$${\triangle {\rm{E}}}_{b}=\sqrt{{\left({L}_{b}^{* }-{L}_{0}^{* }\right)}^{2}+{\left({a}_{b}^{* }-{a}_{0}^{* }\right)}^{2}+{({b}_{b}^{* }-{b}_{0}^{* })}^{2}}$$

(2)

$${\triangle {\rm{E}}}_{a}=\sqrt{{\left({L}_{a}^{* }-{L}_{0}^{* }\right)}^{2}+{\left({a}_{a}^{* }-{a}_{0}^{* }\right)}^{2}+{({b}_{a}^{* }-{b}_{0}^{* })}^{2}}$$

(3)

Where ${RCD}$ is relative change in color difference (%); ${\triangle E}_{b}$ is the color difference between the mildew infected textiles (the simulated samples of textile relics infecting mildew) before cleaning and the aged textile (the simulated samples of textile relics); ${\triangle E}_{a}$ is the color difference between the mildew infected textiles (the simulated samples of textile relics infecting mildew) after cleaning and the aged textile (the simulated samples of textile relics); L is lightness – ranging from 0 to 100, where 0 represents black and 100 represents white; ${L}_{b}^{* }$ is lightness value at the measurement point before cleaning; ${L}_{a}^{* }$ is lightness value at the same point after cleaning; ${L}_{0}^{* }$ is lightness value of the aged textile (the simulated samples of textile relics) without infecting mildew; a is red-green axis – positive values indicate a shift toward red, negative values indicate a shift toward green; ${a}_{b}^{* }$ is red/green value at the measurement point before cleaning; ${a}_{a}^{* }$ is red/green value at the same point after cleaning; ${a}_{0}^{* }$ is red/green value of the aged textile (the simulated samples of textile relics) without infecting mildew; b is yellow-blue axis – positive values indicate a shift toward yellow, negative values indicate a shift toward blue; ${b}_{b}^{* }$ is yellow/blue value at the measurement point before cleaning; ${b}_{a}^{* }$ is yellow/blue value at the same point after cleaning; ${b}_{0}^{* }$ is yellow/blue value of the aged textile (the simulated samples of textile relics) without infecting mildew.

To identify whether occurring the drop in strength of textile relics after cleaning with different cleaning formulations or not, the elongation at break of simulated mildew-infected textile relic samples was measured before and after various cleaning treatments using fabric strength tester (YG026HC). The entire testing procedures and operations followed the standard GBT 3923.1-2013 “Textiles—Fabric tensile properties—Part 1: Determination of breaking strength and elongation at break (strip method).” Specifically, the sample dimensions and equipment parameters were scaled down proportionally by a factor of 5 according to the national standard^24,25,26,27. Each specimen had an effective width of 10 ± 0.1 mm (excluding selvage) and a length sufficient to accommodate a gauge length of 40 mm. The equipment settings included a pre-tension of 2 N, a clamping distance of 40 mm, a moving speed of 100 mm/min, a fixed force of 2000N, and a fixed elongation of 150%. Furthermore, to ensure data reliability, each sample was tested 10 times (five measurements each in the warp and weft directions), and the average of these 10 measurements was taken as the tensile strength of the sample. The effect of cleaning treatment on the mechanical properties of textile relics was characterized by the retention rate of breaking strength, calculated according to the following formula.

$$\triangle T=\frac{{T}_{a}}{{T}_{0}}\times 100 \%$$

(4)

where $\triangle T$ is the retention rate of breaking strength (%), ${T}_{a}$ is the average breaking strength of the mildew infected textiles after cleaning; ${T}_{0}$ is the average breaking strength of aged textiles before mold infection.

To evaluate the applicability of the optimal cleaning formulation in practical conservation of precious textile relics and to elucidate its cleaning mechanism, micro-morphology of the simulated sample of textile relics cleaned with the optimal cleaning formulations was systematically tested with the help of scanning electron microscopy (SEM3200, National Instrument Quantum). Observation was conducted under high vacuum mode with an accelerating voltage of 10 kV and a working distance of approximately 10 mm. The samples were sputter-coated with a thin layer of gold prior to imaging to improve conductivity and image quality.

Results

In order to develop efficient cleaning agents for textile relics affected by mold stains, appearance morphology, washing efficiency and mechanical strength of mildew-stained textiles (the simulated samples of textile relics infecting with mildew) before and after cleaning with different cleaning formations, were analyzed and compared. The specific results were presented as follows:

Appearance morphology analysis of textile relics

Figure 4 clearly showed the macro-morphological alterations of textiles infecting mildew before and after cleaning with different cleaning agents. As illustrated in Fig. 4, regardless of textile’s type, textiles infected with Trichoderma SM4, Aspergillus SM12, or Rhizopus BF3 exhibited dense, visible hyphal networks (mycelial layers) on their surfaces, obscuring the textile structure, with colony edges displaying wavy, serrated, or filamentous morphologies. However, the color and appearance of mildew stains varied depending on the fungal species. Textiles infected with Trichoderma SM4 showed dark green surfaces; those infected with Aspergillus SM12 displayed white surfaces with short, velvety, cotton-like hyphae; while Rhizopus BF3 -infected textiles exhibited black surfaces. After cleaning, regardless of the cleaning formulation, the surface mycelium, colonies and mucous-like substances, were visibly reduced on textiles infected with Trichoderma SM4 /Aspergillus SM12 /Rhizopus BF3, suggesting that the cleaning treatment was capable of removing visible mold-stains from textiles.

Fig. 4: Macroscopic morphology of textile infected with mold before and after treatment with different cleaning formulations.

The figure shows digital photographs (7× magnification) of cotton, linen, silk, and wool textiles contaminated with Trichoderma SM4, Aspergillus SM12, and Rhizopus BF3, both before cleaning and after treatment with ten different cleaning formulations (1# to 10#). The images illustrate the varying degrees of stain removal and fiber surface restoration achieved by each formulation, highlighting the influence of surfactant type, enzyme addition, and antimicrobial agents on cleaning efficacy across different mold-fiber combinations.

In addition, comparing different cleaning formulations, it was found that the composition and unit system design combination of the cleaning agent significantly affected the post-cleaning appearance. Surfactants alone removed only surface-adhered hyphae/colonies, leaving residual mucous or pigmented deposits with persistent dark stains. However, with the addition of auxiliary agents, bio-enzymes, and antibacterial agents, the cleaning effect showed a gradual upward trend, particularly notable when bio-enzymes were added. This observation suggests that mildew stains on textile relics may be adhered by metabolic secretions, and these metabolites are often considered to be biological macromolecules that are difficult to remove by surfactants alone^22,30,31. Therefore, the results indicate that surfactants alone had limited effectiveness in completely removing mildew stains from textile relics. In contrast, bio-enzymes, primarily hydrolases, can degrade insoluble macromolecules into soluble peptides/amino acids while lowering activation energy, and thus their combination with surfactants likely enhances removal efficiency. Furthermore, the optimal surfactant and bio-enzyme types varied significantly depending on the mold species and textile substrate. Specifically, after cleaning with tea saponin, small amounts of mucous membrane or green/reddish-brown pigment remained on the surfaces of cotton, linen, silk, and wool textiles infected with Trichoderma SM4 /Aspergillus SM12 /Rhizopus BF3, suggesting that tea saponin has limited cleaning capability under these conditions. This may be due to the weak penetrative and biofilm-degrading capacities of tea saponin compared to isomerol ethers and rhamnolipid, especially for silk/wool textiles infected with Trichoderma SM4 /Aspergillus SM12. The mucosa/pigment on surface of cotton and linen textiles infected with Trichoderma SM4 /Aspergillus SM12 after cleaning with C₈E₉, was completely removed, indicating C₈E₉ had exceptional adaptability to the cleaning of cotton and linen textiles infected with Trichoderma SM4 /Aspergillus SM12. This is because C₈E9 (EO = 9) are branched-chain iso-octanol ethoxylates. The microporous structure of cellulosic fibers (e.g., cotton and linen) facilitates the rapid adsorption of these branched surfactants. In contrast, the penetration into the cuticle (scale layer) of proteinaceous wool fibers is slower for their linear-structure counterparts. Therefore, C₈E₉ had a better cleaning effect on cotton and linen (cellulose) textiles infected with Trichoderma SM4 /Aspergillus SM12. There was no pigment residue on the surface of cotton, linen, silk and wool textiles infected with Rhizopus BF3 after being cleaned with rhamnose, indicating rhamnolipid effectively removed Rhizopus BF3-induced pigments across all textile types due to its high surface activity, biofilm degradation capability, and specific recognition of β-1,3-glucan secreted by Rhizopus BF3. Alkaline protease showed relatively ideal removal effect without remaining mycelium/mucosa on the surfaces of all mildew-infected textiles, as mold and its metabolites were predominantly proteinaceous, aligning with enzymatic specificity. Therefore, among the three biological enzymes, it had the best cleaning effect on all mildew-infected textiles in this study. Cellulase shows wide applicability in cotton and linen fabrics. This is consistent with the fact the matrix of cotton and linen textiles was cellulose, and cellulase was helpful to loosen mildew through cellulose matrix decomposition^25,30,31. Therefore, the removal of mildew on surface of cotton and linen textiles was relatively superior compared to silk and wool textiles. Lipase demonstrated poor cleaning performance across textiles, likely due to its limited action on phospholipids (primary components of mold cell membranes) compared to triglycerides. This indicated that the cleaning capability of enzymes was not only directly related to the type of extracellular enzymes secreted by mold (such as protease deproteination), but also had a certain relationship with the type of substrate (such as cellulase having a broad-spectrum advantage in cleaning moldy cotton and linen textiles). Zinc sulfate displayed broad-spectrum superiority against cotton, linen, silk and wool textiles infected with Trichoderma SM4. This may be because, compared with potassium sorbate and sodium benzoate, Trichoderma SM4 was more sensitive to zinc sulfate. Moreover, the zinc ions in zinc sulfate could block enzymatic reactions by chelating metal ions, interfering with the germination of Trichoderma SM4 spores and the growth of mycelium. Therefore, the inhibitory effect of zinc sulfate on textiles with mold spots was the most significant among the tested antimicrobials. Potassium sorbate showed broad-spectrum antibacterial advantages in cotton, linen, silk and wool textiles infected with Aspergillus SM12 /Rhizopus BF3. This is likely because potassium sorbate could dissociate into sorbic acid, which destroyed the cell membrane of microorganisms, interfered with the energy metabolism of molds, inhibited ATP synthesis, and then inhibited the activity of molds.

Additionally, comparison of the post-cleaning appearance of different mold stains revealed that, regardless of textile type, the specimens contaminated with Aspergillus SM12 exhibited the cleanest surface (with minimal pigment deposition) under the same cleaning formulation, followed by Rhizopus BF3, while Trichoderma SM4 showed the poorest cleaning outcome. This is hypothesized to be due to the relatively weaker adhesiveness of the hydrophobics secreted by Aspergillus SM12, an Aspergillus species^29,32. Additionally, its predominantly white colony morphology and nearly transparent, colorless spores lack significant pigmentation, thereby avoiding visible pigment dispersion during cleaning and making it easier to remove. In contrast, Rhizopus BF3 and Trichoderma SM4 secrete black and green pigments, respectively, which penetrate deeply into the fibers and exhibit stronger adhesion. Particularly, Trichoderma SM4 possesses a notable ability to corrode and infiltrate the substrate, rendering its stains more difficult to eliminate.

Furthermore, by comparing the cleaning appearance of different textiles, this finding was discovered that regardless of the type of mold stains, the appearance of cotton/linen textiles infected with mildew after being washed with the same cleaning formulation was consistently cleaner (with minimal pigment deposition) than that of silk/wool textiles, suggesting that cellulose fibers may exhibit better cleanability compared to protein fibers under these tested conditions. This distinction arises because silk/wool (protein fibers) serve as ideal nutrient sources for mold growth, enabling vigorous colonization, while cotton/linen (cellulose fibers) exhibit relatively stronger mildew resistance. Additionally, mold colonies on protein-based textiles secrete more adherent fluids/pigments with deeper penetration, resulting in stubborn residues. Consequently, when textiles infected with mildew, protein textiles may be more difficult to remove compared to cellulose textiles. Moreover, the antibacterial advantage in textiles infected with molds were relatively significant. Sodium benzoate showed average performance in cotton, linen, silk and wool fabrics infected with Aspergillus SM12 /Rhizopus BF3. This is likely because sodium benzoate could exert its antibacterial effect only in a strongly acidic environment (pH < 4.5), and its effect drops sharply in other environments (neutral or alkaline environment). These findings underscored the synergistic interplay of surfactant specificity, enzymatic targeting, and antimicrobial synergy, while emphasizing the critical influences of substrate type, mold species, and formulation design on textile relic cleaning outcomes. Therefore, in the actual cleaning process of textile relics infected with mildew, it was necessary to tailor mildew cleaning formulations by integrating multidisciplinary considerations to achieve safe and efficient decontamination.

Color difference

Table 3 clearly presented the color difference (ΔE) and relative change in color difference (RCD, %) of mildew-stained textiles before and after cleaning with different formulations. As shown in Table 3, regardless of cleaning agent formulations, the color difference of textiles infected with Trichoderma SM4/Aspergillus SM12/ Rhizopus BF3 changed to varying degrees after cleaning treatment. And, the extent of change differed depending on the fungal species, fiber substrate, and cleaning formulation. The details are as follows:

Table. 3 Color difference and relative change in color difference (RCD) of mildew-stained textiles before and after cleaning with different formulations

Comparing the color difference (ΔE) of mildewed textiles before and after treatment with different cleaning agent components and their combined formulations revealed that treatments with single surfactants resulted in relatively large ΔE values. This was particularly evident for silk and wool textile contaminated with Trichoderma SM4 or Rhizopus BF3, where post-cleaning ΔE values were high (ranging from 8 to 37, corresponding to RCDs of 25% to 76%). However, with the addition of builders, biological enzymes, and antibacterial agents (especially biological enzymes), the ΔE decreased in a stepwise manner, ultimately reaching the range of 0–5 (with corresponding RCDs gradually increasing to above 90%). This phenomenon may be attributed to the fact that microbial metabolites are often proteins and polysaccharides (macromolecular biological substances not easily removed by surfactants), which can act as adhesives, binding stains to the textile surface^22,30,31. Consequently, relying solely on surfactants is insufficient for the complete removal of mold stains from textiles. In contrast, biological enzymes, being hydrolases, can lower the activation energy of reactions while simultaneously decomposing insoluble macromolecular mold stains into soluble peptides or amino acids, thereby enhancing the cleaning efficacy^31,33. Furthermore, the cleaning agent’s components also exhibited significant substrate specificity. For example, tea saponin showed relatively limited applicability across textiles infected with all three types of mold (RCD ranging from 22% to 89%). C₈E9 demonstrated high applicability for textiles infected with Trichoderma SM4/Aspergillus SM12/ Rhizopus BF3 (RCD ranging from 32% to 96%), with particularly high applicability for cotton and linen fabrics (RCD up to 95.19%). Rhamnolipids displayed exclusive applicability for BF3-infected fabrics, especially silk, achieving an RCD as high as 77.56%. By hydrolyzing peptide bonds, alkaline protease exhibited broad-spectrum and efficient cleaning capability across all fiber-mold combinations (significantly improving RCD). Cellulase, by decomposing the cellulose substrate, showed remarkable effectiveness on cotton and linen fabrics. In contrast, the effectiveness of lipase was relatively limited, possibly because the primary lipid components in mold cell membranes are not triglycerides.

Additionally, a comparison of color difference changes caused by different molds indicated that, regardless of fabric type, the ΔE values for textiles infected with Trichoderma SM4 or Rhizopus BF3 were consistently higher than those for textiles infected with Aspergillus SM12, when compared to aged, uninfected samples. This is because Aspergillus SM12 spores are nearly transparent and bright white, contributing minimally to color change, thus resulting in a smaller ΔE deviation. In contrast, Trichoderma SM4 and Rhizopus BF3 spores are green and black, respectively, containing pigments that contribute more significantly to color change, leading to larger ΔE deviations. After cleaning, ΔE values decreased to varying degrees for all formulations. For instance, on wool fabric using formulation 10#, the ΔE was 3.08 for SM12, 3.28 for Rhizopus BF3, and 4.07 for Trichoderma SM4. It was also observed that, regardless of the cleaning formulation, for most fabric-formulation combinations, the final post-cleaning ΔE values consistently followed the order: textiles infected with Aspergillus SM12 < textiles infected with Rhizopus BF3 < textiles infected with Trichoderma SM4. This suggests that, under the conditions of this experiment, the difficulty of cleaning follows the order: Trichoderma SM4 > Rhizopus BF3 > Aspergillus SM12. This is likely because Trichoderma SM4 mold can penetrate deeply into the fibers, corroding the substrate, making it the most difficult to clean. Rhizopus BF3 mold primarily forms a three-dimensional network on the fiber surface, potentially making it relatively easier to remove, hence ranking second in cleaning difficulty. The hydrophobins secreted by Aspergillus SM12 may have weaker adhesion to the substrate, and its spores are transparent and colorless, lacking pigments, potentially making it the least difficult to clean. Meanwhile, an interesting phenomenon was also noted: after cleaning, textiles infected with SM12(Aspergillus), despite having a small ΔE, often had many residual hyphae on the surface, indicating incomplete cleaning. Conversely, textiles infected with Trichoderma SM4 or Rhizopus BF3, despite having a larger ΔE, often appeared clean to the naked eye, with no residual hyphae. This is because the light-colored Aspergillus SM12, even with residual hyphae or metabolites, may cause only minor ΔE changes. In contrast, the pigmented spores of Trichoderma SM4 and Rhizopus BF3 can still cause a significant ΔE even when most surface hyphae have been removed and the textile appears clean to the naked eye. This finding indicates that relying solely on ΔE values can be misleading: a low ΔE does not necessarily signify thorough cleaning, and a high ΔE does not always correspond to poor cleanliness in terms of visible residues. Therefore, in practical applications, colorimetric data and visual assessment should be integrated to comprehensively evaluate cleaning effectiveness. Furthermore, previous color difference studies have shown that a ΔE between 2.1 and 3.5 for cleaned textiles compared to uninfected samples is imperceptible to the human eye but detectable by instruments. This further emphasizes the necessity of combining instrumental measurements with visual inspection for interpretation. Meanwhile, the above conclusions also demonstrate that the color difference (ΔE) of moldy textiles before and after cleaning varies depending on the mold species.

Moreover, comparing color differences among different textile types revealed that, for all mold species, cellulose-based textiles (cotton, linen) exhibited a greater reduction in ΔE after cleaning, with higher RCDs (73.6–98.33%). In contrast, protein-based textiles (silk, wool) showed lower levels of ΔE improvement (RCD 25.28–96.72%). This difference may be attributed to protein fibers themselves providing a richer nutrient source for mold growth, leading to deeper penetration of hyphae and metabolites and the formation of more stubborn stains³⁵. Consequently, under the conditions of this experiment, the change in apparent color difference after cleaning was smaller. Conversely, the microporous structure of cellulose fibers might facilitate the adsorption and penetration of cleaning agents, resulting in a larger change in apparent color difference after cleaning³⁵. Therefore, in the practical cleaning of mold-infected textile relics, it is essential to fully consider the type of infecting fungus, the characteristics of the artifact’s fiber substrate, and the compatibility of cleaning agent components to achieve efficient and safe mildew removal.

Mechanical strength analysis

Table 4 presented the mechanical strength of mildew-stained textiles before and after cleaning with various formulations. As shown in Table 4, the mechanical strength of mildew-infected textiles was slightly lower than that of uninfected textiles, indicating that even short-term mildew infestation reduced the mechanical properties of textiles. This occurred because mold growing on textiles’ surface penetrated deeply into the fibers, causing corrosion or decomposition of internal fibers, and then resulted in softening, dampness, stickiness, and diminishing mechanical strength of textiles. Additionally, the metabolic processes of mold generated heat, organic acids, and enzymes, which sharply increased acidity of textiles, damaged fibers, reduced mechanical strength, and accelerated material aging. These observations suggest that mildew removal may be beneficial for prolonging the lifespan of textile relics. In addition, by comparing the mechanical strength of mildewed textiles before and after cleaning, it was observed that regardless of textiles’ type or cleaning formulation, the mechanical strength did not continue to decline after cleaning; instead, it showed a slight increase. This suggests that the cleaning treatment did not lead to a degradation in the mechanical properties of mildew-stained textiles and may halt the ongoing enzymatic degradation caused by residual mold activity, thereby indicating the safety and efficacy of the developed cleaning strategy. The stability was attributed to the gentle, non-mechanical ultrasonic cleaning method employed in this study, combined with cleaning formulations maintaining a neutral pH (≈7). This mild, non-irritating environment minimized changes in mechanical strength during cleaning. Moreover, by analyzing mechanical strength variations before/after mildew infection and cleaning, it could be inferred that removing mold stains through cleaning is an effective method to inhibit the degradation of textile relics caused by mildew, thereby benefiting their long-term preservation. The slight increase in strength may be due to the removal of moisture and degradation products, as well as the cessation of enzymatic degradation by molds after cleaning^17,32.

Table 4 Mechanical strength of mildew-stained textiles before and after cleaning with different formulations

Microscopic morphology

Figure 5 clearly illustrated the alterations in the micromorphology of mold-infected textiles before and after cleaning. As depicted in Fig. 5, irrespective of textiles’ type or mold species, fungal hyphae and spores were found adhering to or embedded within the fibers following mold infection. This microbial colonization obscured the original fabric structure and fiber morphology, hindering both visual and instrumental identification of the inherent characteristics, thereby underscoring the importance of mildew removal for restoring the original appearance of textile relics. After treatment with the optimized cleaning formulation, microbial residues and mucus-like substances were effectively eliminated from all infected samples. The textile structures became clearly visible, and the surface morphology of the fibers was largely restored without observable damage. For example, cleaned cotton displayed the natural crimp of its fibers; cleaned hemp showed smooth surfaces with distinct longitudinal grooves; cleaned silk retained its characteristic smooth, cylindrical shape; and cleaned wool revealed overlapping cuticular scales. These results suggest that the appropriate cleaning formulation can efficiently remove mold contamination without causing physical damage to the fibers, supporting its potential application in heritage conservation.

Fig. 5: Microscopic morphology of textiles infected with mold before and after cleaning with the developed optimal cleaning formulation in this work.

The figure presents scanning electron microscopy (SEM) images at 500× magnification and optical macrographs at 50× magnification for cotton, linen, silk, and wool textiles infected with Trichoderma SM4, Aspergillus SM12, and Rhizopus BF3. The images compare the samples before cleaning, showing dense fungal hyphae and spores adhering to fibers, and after cleaning with the optimal formulation, demonstrating the effective removal of microbial residues and the restoration of the original fiber morphology without observable damage.

Furthermore, comparative analysis of cleaned mold-infected textiles revealed that for textiles infected with Aspergillus SM12 and Trichoderma SM4, no significant contaminants remained on fiber surfaces or interstices, and fiber alignment returned to pre-infection states. In contrast, cotton textiles infected with Rhizopus BF3 exhibited minor flaky residues presumed to be mucous membranous substance remnants. Such residues were absent on other materials (linen, silk, wool). This indicates that post-cleaning morphological variations arise from differences in mold type, indirectly suggesting that cleaning efficacy varies among species. Notably, Rhizopus BF3 presented the greatest cleaning challenge compared to Aspergillus SM12 and Trichoderma SM4, particularly for cotton textiles. The results further demonstrated that the optimal detergent formulation effectively disrupted fungal hyphal networks and decomposed metabolic byproducts through synergistic action of surfactants and bio-enzymes, while preserving original fiber morphology. Furthermore, these conclusions not only confirmed the developed formulation’s effectiveness in microbial stain removal and textile fiber restoration, but also provided a reliable technical approach for treating mold-contaminated organic artifacts - particularly Trichoderma SM4/Aspergillus SM12/ Rhizopus BF3-infected textile relics - thereby supporting subsequent conservation and preservation efforts.

Practical application of optimal formulations in textile relics

Figure 6 clearly illustrated the appearance characteristics of the Qing original-color silk Miao embroidery piece before and after cleaning. As shown in Fig. 6a, before cleaning, the embroidery exhibited an overall dull beige hue with noticeable signs of wear. The edges showed fraying, and the textile structure appeared fragile. Multiple areas of yellowish tinges were observed on the surface, particularly around the embroidered patterns and along the edges, resulting from residual brown pigments due to mold contamination. After ultrasonic cleaning with the optimized formulation (C₈E9 (200 g/L), water (750 g/L), sodium oxalate (50 g/L), alkaline protease (1000μ/g), potassium sorbate (0.12 g/mL), sodium citrate (adjusted to Ph≈7)}, surface stains were significantly reduced, and the yellowish tinge was notably diminished compared to the pre-cleaned state (see Fig. 6b). However, due to the considerable age of the artifact, a certain degree of historical patina remained, and the frayed edges were unchanged. The experimental results demonstrate that ultrasonic cleaning with the optimized formulation can partially remove surface contaminants from textile relics and moderately improve its visual appearance, thereby partly restoring their aesthetic and artistic value. Nevertheless, given the historical age and inherent material degradation of the embroidery, the cleaning process could not fully restore it to its original condition. This indicates that cleaning treatment alone has limitations in preserving and transmitting the cultural heritage of mold-stained textile relics. Subsequent research should incorporate virtual simulation technologies to vividly reconstruct the craftsmanship and wisdom embodied in ancient textile relics³⁶.

Fig. 6: Practical application of optimal formulations in textile relics.

a Qing dynasty natural-color silk Miao embroidery piece before cleaning, showing overall dull beige hue, frayed edges, and yellowish tinges from mold contamination. b The same embroidery piece after ultrasonic cleaning with the optimized formulation {C₈E9 (200 g/L), water (750 g/L), sodium oxalate (50 g/L), alkaline protease (1000μ/g), potassium sorbate (0.12 g/mL), sodium citrate (adjusted to pH≈7)}, showing significantly reduced surface stains while preserving historical patina.

Discussion

Mildew is a significant and persistent threat to the conservation of textile cultural heritage. However, study on systematic, safe, and effective cleaning methods to address this issue remains limited. Therefore, this study systematically evaluated the cleaning efficacy of a ternary synergistic cleaning system composed of surfactants, biological enzymes, and antimicrobial agents on artificially contaminated samples (cotton, linen, silk, and wool fibers infected with Trichoderma SM4, Aspergillus SM12, and Rhizopus BF3, respectively), aiming to effectively solve the problem of mildew stain removal from textile relics. The results indicate that through rational design of the cleaning formulation, particularly employing the “surfactant-enzyme-antimicrobial agent” ternary synergistic system, effective removal of mildew stains (mycelia, slime, pigments) was achieved while maximally preserving the original morphology and mechanical strength of the fibers. Moreover, the cleaning efficacy exhibited significant formulation dependence, fungal species specificity, and fiber substrate selectivity. This demonstrates that the cleaning effect on stained textile relics results from the complex interaction among mold species, fiber substrate, and cleaning formulation, necessitating that conservation cleaning work must tailor formulations based on the properties of the artifact-contaminant composite system^6,25,35. Moreover, the successful cleaning of authentic mildew-stained silk artifacts from the Qing Dynasty using the optimized formulation not only demonstrated the practical application potential of the developed ternary synergistic cleaning system in cultural heritage conservation but also indicated that this multidisciplinary-based meticulous cleaning approach serves as an effective solution to complex challenges in cultural relic preservation.

The results of this study align with previous research findings on the microbial deterioration and cleaning of cultural heritage, while systematically extending upon them. Earlier studies on mildew stain removal from artifacts often focused on exploring single components, such as surfactants or enzymes alone, and it was widely recognized that single surfactants had limitations in removing microbial metabolites (e.g., proteinaceous and polysaccharide slime)^11,18,22,29. The present study corroborates this, finding that using surfactants alone (e.g., tea saponin, C₈E₉, rhamnolipid) could remove some surface mycelia but often failed to effectively decompose the extracellular polymeric substances (EPS) acting as “biological adhesives,” leaving behind pigments and slime, resulting in the color difference reduction rate (RCD) as low as 22%^11,30,31. A significant advancement of this study lies in the introduction of biological enzymes (protease, cellulase, lipase), which markedly improved cleaning efficiency, as evidenced by substantially decreased ΔE values, confirming the crucial role of enzymes in hydrolyzing macromolecular stain^31,33. This finding resonates with the research approaches of Ahmed et al. using α-amylase to remove starch adhesives and Miao et al. utilizing enzyme systems to remove iron-based stains^11,23,31. By systematically comparing different enzymes, this study further revealed specific relationships between enzymes, substrates, and fungal species: alkaline protease performed best due to its broad-spectrum hydrolysis of proteinaceous soils; cellulase exhibited a “loosening” effect on mildew stains on cellulose-based fibers (cotton, linen) by partially decomposing the cellulose substrate^25,30; whereas lipase was less effective, attributed to the molecular structural differences between fungal cell membrane phospholipids and the typical lipase substrates (triglycerides). Furthermore, incorporating antimicrobial agents (zinc sulfate, potassium sorbate) into the cleaning solution represents a novel contribution. The synergistic action observed with surfactants and enzymes not only enhanced immediate cleaning efficiency (by approximately 1–5%) but also potentially improves the longevity of the cleaning effect by inhibiting residual microbial activity^12,19. This “cleaning + inhibition” synergistic strategy is an important supplement to existing cleaning approaches. Additionally, this study systematically explored the differential cleaning efficacy among mold species (Aspergillus SM12 < Rhizopus BF3 < Trichoderma SM4) and fiber types (cellulosic fibers < proteinaceous fibers), deepening the understanding of fungal biofilm cleaning mechanisms. For instance, Trichoderma species, known for their potent cellulolytic capacity and hyphal penetration ability²¹, can deeply invade the fiber interior, and combined with their pigmented spores, make decontamination particularly challenging. In contrast, the white hydrophobic spores of Aspergillus SM12 adhere weakly and lack deep-penetrating pigments, making infected textiles easier to clean^29,32. Regarding substrate materials, compared to cellulosic fibers (cotton, linen), proteinaceous fibers (silk, wool) provide a more favorable nutritional environment (e.g., nitrogen sources, amino acids) for fungal proliferation, more readily inducing the formation of robust biofilms and allowing deeper penetration of metabolites, resulting in more tenacious stains³⁵. These findings underscore the necessity of integrating the biological characteristics of the contaminant, the material composition of the artifact, and the chemical properties of cleaning agents when formulating artifact cleaning strategies—a core principle increasingly emphasized in cultural heritage conservation science^6,16,25.

A key contribution of this study is the experimental confirmation that superior cleaning efficacy arises from the multi-faceted synergistic action of surfactants, enzymes, and antimicrobial agents, and the initial elucidation of the underlying synergistic mechanism. Firstly, surfactants provide foundational physical detachment. For example, the superiority of C₈E₉ (branched-chain isomeric octyl ethoxylate) for Trichoderma/Aspergillus on cotton/linen may be attributed to favorable compatibility between its molecular structure and the microporous structure of cellulosic fibers, facilitating rapid adsorption and penetration of the cleaning agent, thereby effectively disrupting the surface mycelial layer and enhancing cleaning²⁵. The specific high efficacy of rhamnolipid against Rhizopus BF3 might stem from its outstanding biofilm penetration capability and potential recognition or affinity for β-1,3-glucan, a characteristic component of Rhizopus cell walls, enabling deep action within the colony. The broad-spectrum limitations of tea saponin may relate to its relatively lower interfacial activity and biofilm penetration power¹⁸. Secondly, the introduction of biological enzymes enables targeted degradation of recalcitrant metabolic products (the EPS secreted by molds, primarily composed of proteins, polysaccharides, and a small amount of lipids, forming a dense biofilm barrier)^22,30. The broad-spectrum suitability of alkaline protease likely arises from its efficient hydrolysis of protein components within the EPS and hyphal debris. The synergistic enhancement by cellulase on cotton/linen fabrics may result from a “chemical loosening” effect, where limited, controllable hydrolysis of the fiber surface cellulose detaches the intertwined mildew stains by removing their anchoring basis²⁵. The limited effectiveness of lipase may be because the primary lipid components in mold EPS are not triglycerides but other lipids like phospholipids, which are less accessible to conventional lipases. Finally, incorporating antimicrobial agents provides a dual “cleaning + inhibition” safeguard for mildew stain removal from textile artifacts. The specific efficacy of zinc sulfate against Trichoderma SM4 stems from Zn²⁺ ions acting as metal cofactor inhibitors for various enzymes, interfering with fungal metabolism and spore germination¹⁹. The broad-spectrum advantage of potassium sorbate against Aspergillus SM12 and Rhizopus BF3 is attributed to its undissociated molecules penetrating cell membranes and inhibiting or killing residual fungi by interfering with energy metabolism (e.g., inhibiting ATP synthesis)¹². This “ternary synergistic” cleaning strategy—combining surfactant-based physical detachment, enzymatic chemical decomposition, and antimicrobial biological inhibition—ensures comprehensive removal of mildew stains from macro-scale soiling to micro-scale residues and offers potential for preventing post-cleaning mold regrowth.

Despite the significant cleaning efficacy achieved on simulated samples of textile relics infected with mold and the preliminary evidence for application to real artifacts, this study has some limitations. Firstly, the inherent simplification of the experimental system. Although four representative fibers and three typical molds were selected for artificial contamination, mildew contamination on real-world textile artifacts is far more complex. Microbial communities on artifacts often consist of mixed bacterial and fungal populations, typically coexisting with dust/oils and so on; thus, relics’ stains are usually composite stain comprising dust, oils, and various metabolic by-products^6,13. Consequently, the efficacy of the optimal formulations derived here may require further validation when confronting such composite contamination. Secondly, the long-term safety of the cleaning agents has not been fully established. While mechanical testing and microscopic observation confirmed the immediate safety of the cleaning process for the fibers, the long-term effects, particularly whether enzymes and additives might remain within fibers and induce slow, long-term degradation (e.g., catalytic hydrolysis, metal ion-catalyzed oxidation), remain unknown. This risk is especially critical for already aged and fragile textile relics^5,33. Lastly, the depth of mechanistic understanding needs further exploration. For instance, the specific recognition mechanism between rhamnolipid and Rhizopus, the controllable degree of cellulase-induced “loosening” on cellulosic substrates and its impact on long-term fiber durability, and the long-term antimicrobial efficacy of the agents post-cleaning all require further investigation at the molecular level. Therefore, future research should proceed in the following directions: (1) Expanding contamination models to develop more realistic simulation systems incorporating mixed microbial consortia and composite soils, assessing the broad-spectrum adaptability and practical application effects of cleaning formulations. (2) Conducting long-term aging studies by introducing accelerated aging experiments on cleaned samples, employing multiple analytical techniques (e.g., X-ray diffraction, infrared spectroscopy, degree of polymerization measurement) to monitor the evolution of fiber microstructure, chemical composition, and mechanical properties, thereby comprehensively evaluating the safety of cleaning treatments^5,33. (3) Deepening mechanistic exploration using molecular biology, proteomics, and interfacial chemistry to investigate interaction mechanisms between enzymes and specific substrates, and between surfactants and different fibers, providing a more robust theoretical basis for the scientifically informed design of cleaning formulations.

In conclusion, by systematically investigating the efficacy of a ternary synergistic cleaning system comprising surfactants, enzymes, and antimicrobial agents for removing fungal stains from textile artifacts, this study establishes for the first time that effective cleaning of mildewed textile artifacts necessitates the targeted design of such a ternary synergistic system—integrating fiber- and mold-specific surfactants for physical removal, targeted enzymes for chemical decomposition of biological adhesives, and appropriate antimicrobial agents for biological inhibition—by holistically considering the biological characteristics of the contaminant (mold species), the material science of the artifact (fiber type), and the chemical properties of the cleaning agents (components and ratios) as an organic whole. Examples include the C₈E₉/alkaline protease system for Trichoderma/Aspergillus on cotton/linen, and the rhamnolipid/alkaline protease/potassium sorbate system for Rhizopus. This research not only provides conservators with practical, evidence-based solutions for mildew stain removal but also offers new design ideas and a theoretical framework for the future development of safer, more efficient, and controllable cleaning agents for cultural heritage conservation.