Experimental Testing of the Efficiency, Stability, and Compatibility of Fillers in the Conservation and Restoration of Water-Gilded Wooden Heritage

Abstract

The conservation and restoration of water-gilded wooden cultural heritage, such as polychrome sculptures, frames, panels, altarpieces, etc., requires the use of fillers that guarantee structural stability, physicochemical and mechanical compatibility with the original support, and the ability to adapt to dimensional movements induced by thermo-hygrometric variations. This study, conducted as part of the DorART Project, analyzed the behavior of nine formulations, both commercial and non-commercial, selected through a review of the state-of-the-art specialized literature, along with the use of participatory science, which focused on the practices and materials most commonly used by professionals in the field. The experimental design was based on three types of specimens: two with wooden supports, selected for evaluating their interaction with the original material and with the traditional water gilding technique, and a third type for analyzing the individual behavior of the tested materials. Analyses of adhesion, tensile strength, Shore C hardness, gloss, abrasion test results, wettability, pH changes, and chemical composition were performed using ATR-FTIR spectroscopy. The results showed significant differences depending on the type of curing used and the composition and aging behavior of the specimen. Some of the fillers demonstrated improved compatibility with water-based gilding, facilitating workability and providing structural strength. M3 and M9 demonstrated an optimal balance of workability and aging stability. The results of this study can help restorers select materials based on their specific needs, considering the requirements of mechanical adaptation to the substrate, compatibility, and durability.

1. Introduction

The purpose of this research, conducted within the framework of the DorART Project at the Universitat Politècnica de València (Spain), was to study, through experimental testing, the mechanical properties, efficiency, and stability of the most commonly used gap fillers in the restoration of wooden artworks, with particular attention paid to their compatibility with traditional water gilding techniques.

Given the structural fragility and artistic complexity of gilded wooden heritage objects—such as polychrome sculptures and altarpieces—this study aims to identify filler formulations that not only perform effectively under mechanical and environmental stress but also interact appropriately with water-based gilding layers. The significance of this work lies in its contribution to the development of restoration materials that ensure both structural reinforcement and esthetic coherence while respecting the physical and chemical behavior of the original materials.

The deterioration of wooden works of art is a common phenomenon, as wood is a hygroscopic organic material that is susceptible to thermo-hygrometric fluctuations, anthropogenic factors, fire, and biodeterioration, including wood-decaying fungi, all of which contribute significantly to its degradation [1]. Consequently, the conservation of such objects depends on multiple factors, especially the specific properties of the wood [2], a thorough understanding of which is essential for selecting the most appropriate intervention methods to ensure its long-term preservation [3].

The current state of research on the behavior of filler materials used in the conservation and restoration of cultural heritage has given rise to several experimental studies aimed at analyzing the fillers’ behavior under artificial aging conditions in different contexts, although these studies have not specifically focused on wood. In the field of heritage restoration, experimental examples can be found for the treatment of works of art on canvas [4], bone [5], and metal [6], as well as wax sculptures [7]. These cases reflect the diversity of materials addressed in conservation practices and the ongoing search for appropriate restoration solutions.

A paradigmatic and pioneering example of material testing on artworks with wooden supports is provided in the study of Grattan and Barclay (1988), who evaluated 33 different filling materials under conditions of high relative humidity [8]. Building on this seminal work, Craft and Solz (1998) analyzed commercial fillers based on acrylic and vinyl resins, which contributed to the evaluation of more accessible formulations for use in conservation [9]. Other advances include the study by Ellis et al. (2004), who evaluated combinations of epoxy adhesives for application on wood, highlighting the role of intervening layers in improving reversibility [10].

Williams, in his research on flexible adhesives for gap filling, emphasizes the need for filler materials to replicate the mechanical behavior of the substrate and remain stable over time. Based on his practical experience, he proposes a hierarchy of materials, with ethylene vinyl acetate (EVA) being the most suitable choice [11]. Fulcher (2017), focusing on structural fillers for wood artwork, confirmed the hygroscopic compatibility and dimensional stability of cellulose-based materials [12].

More recently, Kryg et al. (2020) examined exterior adhesives combined with inert fillers, such as wood dust and glass microspheres, and showed that low moisture absorption and adequate drying are key to preventing shrinkage on wood substrates [13]. Complementarily, Cura D’Arts de Figueiredo Junior et al. found that expanded vermiculite-based composites offered excellent mechanical and hygroscopic stability for indoor heritage works, as well as chemical compatibility with wood [14]. Finally, in 2023, Chen et al. confirmed that hydrophobic wood fiber/epoxy composites exhibit structural and hygroscopic compatibility with wood, reinforcing their potential as durable restoration materials [15].

In 1979, Green identified the types of material losses that are typically found on water-gilded surfaces [16], a framework that continues to be relevant in conservation diagnostics today. Frinta (1963) had previously suggested wax as an alternative material for use on wooden statues [17], anticipating later concerns about material compatibility and aging. More recent studies on the behavior of fillers and their interaction with water gilding highlight the importance of avoiding watery materials to prevent moisture from compromising organic structures. For example, Wilson’s 1998 study, building on previous research on water-sensitive surfaces [18], proposed Paraloid^® B72 combined with colored fillers as a safe and reversible binder for filler putties. This formulation allows polishing to achieve gloss and ensures good aging stability [19].

To address the challenge of identifying fillers in water-gilded works, Salimnejad (2005) introduced the use of bismuth oxide as a radiopaque marker, detectable by X-ray, in combination with traditional plaster made with rabbit glue and fillers such as ZnO and CaCO₃ [20]. Similarly, Thornton recommended barium sulfate for its high X-ray opacity and, in 1991, suggested various materials for filling ornamental compositions, including plaster and thermosetting resins such as epoxy and polyester [21].

In 2015, Pierce conducted an experimental study that compared traditional water-based putties with various synthetic alternatives, evaluating their application properties, surface finish, and reversibility. The findings demonstrated that several synthetic materials were able to faithfully replicate the aesthetic and functional characteristics of gilded surfaces, offering viable options for conservation treatments [22]. Addressing long-term behavior, Sawiki (2017) examined Paraloid^® B72, Plextol^® B500, and PVAC AYAF—polyvinyl acetate resin—after more than 10 years of natural aging under diurnal fluctuation in relative humidity (RH) and temperature through window glass, concluding that the synthetic adhesive Paraloid^® B72 exhibited greater surface fragility, despite showing minimal molecular changes according to FTIR analysis [23]. However, there is still limited comparative data on the performance of these synthetic formulations when integrated into typical multilayer water-gilded surface systems, particularly with regard to their long-term behavior under fluctuating environmental conditions.

In this context, previous studies highlight the use of traditional aqueous materials, such as rabbit glue gesso and inert fillers, while also emphasizing the need to investigate the compatibility of these materials with the water gilding technique. Additionally, they underscore the importance of evaluating the mechanical properties of non-aqueous synthetic adhesives in order to gather data on their adaptability to dimensional alterations in wood caused by thermo-hygrometric changes, which is a key requirement for preparatory layers in water-gilded surfaces.

2. Materials and Methods

2.1. Material Selection

For the selection of the materials to be tested, data from casting trials and bibliographic research were cross-referenced and analyzed. Participatory science was employed, involving citizen contribution at Level 2 of distributed intelligence [24], through initial crowdsourcing in which participants played an active role in the study design and decision making, following a mixed-methods approach. A relevant precedent is found in Fulcher’s 2014 social study on the most commonly used filler putties in woodwork, based on surveys of conservators and restorers, which focuses on the use of hydroxypropyl cellulose and paper [25]. This study led to subsequent experimentation with cellulose in 2017 (op. cit.).

The objective of our national survey, which included both qualitative and quantitative analyses, was to analyze the most commonly used fillers in gilded woodwork and to determine the materials to be tested in the experimental phase. This approach allowed for the systematic collection of data on current research needs and the selection of relevant materials. To assess the reliability of the questionnaire and validate it before distribution, the qualitative think-aloud technique was employed with three professional restorers, following Willis’s cognitive interviewing methodological model [26] and a mixed-methods approach [27,28].

The results were anonymized by removing keys and identifiers, in compliance with ethical standards. The processing and handling of participants’ personal data adhered to Regulation (EU) 2016/679 of the European Parliament and of the Council [29], as well as to the Spanish Organic Law 3/2018 on the Protection of Personal Data and Guarantee of Digital Rights [30]. The survey and its distribution were approved by the Ethics Committee of the Polytechnic University of Valencia, under the ethical approval code P02_28-06-2023.

The participants, who gave their informed and voluntary consent, were Conservation–Restoration professionals working with wood-based materials and gilded surfaces, who represented five sectors within the field of cultural heritage conservation: museums, institutions, freelancers, companies, and universities. The national scope of the survey allows for a focused understanding of current conservation practices in Spain, a country with a rich tradition in polychrome wood and water gilding. Moreover, the materials analyzed in this research complement those discussed in the literature review section, providing a contextual perspective that enriches and broadens the conclusions drawn from previous international studies.

The DorART survey, designed to select the materials to be tested in the experimental phase, examined both the criteria used by professionals for selecting putties and the most commonly used materials, taking into account their nature [31]. The materials were labeled as M1 to M9 in the order indicated by the respondents (

Table 1. Material selection for experimental evaluation.

Label	Material	Binder Family
M1	Araldite® SV 427-2/HV 427-1. RenPaste™ SV 427-2 /Ren® HV 427-1 (Huntsman Advanced Materials, GmbH, Basel, Switzerland)	Synthetic; epoxy resin (thermosetting)
M2	Balsite® (CTS Conservation SRL, Altavilla Vicentina, Italy)	Synthetic; epoxy resin (thermosetting)
M3	Epo® 127 (CTS Conservation SRL, Altavilla Vicentina, Italy)	Synthetic; epoxy resin (thermosetting)
M4	Vinavil® 59 (Vinavil® S.p.A., Milan, Italy) + sawdust (0.4–0.7 mm)	Synthetic organic polymer; Polyvinyl Acetate; PVAc
M5	Beva® Artist Gesso -p (Gustav Berger Original Formula, Kremer Pigmente GmbH & Co., Aichstetten, Germany) + calcium sulfate, CaSO4	Synthetic thermoplastic polymer
M6	Beva® 371 (Gustav Berger Original Formula, CTS Conservation SRL, Altavilla Vicentina, Italy) + vermiculite	Synthetic thermoplastic polymer
M7	Paraloid® B-72 (The Dow Chemical Company, Midland, TX, USA) + glass microballoons	Synthetic organic polymer; acrylic
M8	Gypsum (CaSO4)/deionized water	Inorganic mineral
M9	Rabbit-skin glue/deionized water + 50% calcium carbonate (CaCO3)/50% calcium sulfate (CaSO4)	Organic hot-melt adhesive

Source: Authors’ own elaboration, DorART Project, 2025, with the collaboration of Sofía Lacueva Gil.

).

2.2. Specimen Preparation for Aging Tests

Three types of samples were prepared to analyze different aspects of the materials’ behavior. Test Specimen 1 focused specifically on evaluating the interaction of the fillers with the wood, particularly in response to its thermo-hygrometric movements. This included analyzing the contact angle to assess surface hydrophobicity and monitoring the pH evolution during aging cycles to detect possible acidification or alkalization processes. Mechanical tests were also conducted, including tensile adhesion tests on wood substrates to evaluate the coating’s resistance to detachment, and Shore C hardness measurements to assess the surface’s resistance to penetration.

Test Specimen 2 was designed to evaluate the compatibility of the fillers with traditional water-gilded wood surfaces, focusing specifically on the aging of the materials upon contact with gilding performed with an inert base and rabbit glue. The gilding was applied half-burnished and half-unburnished to observe any differential behavior. The interaction between the wood, fillers, and gilt layer was assessed through physical and chemical analyses, including gloss measurements—as an optical indicator of surface texture changes and material loss—and mechanical abrasion tests to determine the fillers’ resistance to wear and the impact of this on the stability of the gilded finish.

Finally, Test Specimen 3 was specifically designed for the mechanical and spectroscopic analysis of the materials, independent of their interaction with the wood. Mechanical tests were performed using a Universal Testing Machine with a load capacity of 100 kN, with controlled displacements applied, following a standardized procedure, to determine the mechanical strength of the system. In addition, chemical characterization was performed using ATR-FTIR spectroscopy, which allowed the identification of functional groups on each material’s surface and the analysis of possible chemical changes induced by aging or external agents.

To ensure experimental reproducibility and statistical reliability, each material was tested in triplicate, following a standardized protocol that included photographic and dimensional documentation. The test specimens were evaluated for performance, stability, and compatibility through physicochemical and mechanical tests and analyses, including surface examination by optical microscopy. This approach allowed for the identification of consistent trends, minimized the influence of outliers or anomalies, and facilitated a more robust comparative analysis between materials. Due to the exploratory nature of this study and the small sample size (n = 3), statistical significance tests (e.g., ANOVA) were not applied. Instead, standard deviation and coefficient of variation were used to describe dispersion and detect trends or anomalies in material behavior.

2.3. Experimental Design

2.3.1. Description of the Types of Samples

• Type 1 and 2 test specimens: wood-supported specimens

The selection of the wood for use in this study was based on prior studies of wood substrates, taking into account the characteristics of density, hardness, cut orientation, and grain direction. Pinus sylvestris was selected based on its historical use in gilded cultural heritage objects, density of 500–540 kg/m², mechanical strength (static bending: 90–110 N/mm²; elastic modulus: 8600–13,000 N/mm²), and dimensional stability. With a hardness of 2.0 and a straight grain, it is easy to work with and highly suitable for use as a substrate. To enhance dimensional stability, a tangential cut was chosen, and samples were dimensioned to fit within the climatic chamber. To prevent warping, each support was divided into two pieces, each 2 cm thick.

• Type 1 test specimens (Figure 1 and Figure 2):

For each material, three circular replicates were prepared for each of the five test conditions, resulting in 15 samples per material and a total of 135 test samples.

Figure 1. Type 1 test specimens M1–M8 Source: DorART Project (photo by Juan Valcárcel Andrés).

Figure 1. Type 1 test specimens M1–M8 Source: DorART Project (photo by Juan Valcárcel Andrés).

Figure 2. Type 1 test specimens M9. Source: DorART Project (photo by Juan Valcárcel Andrés).

Figure 2. Type 1 test specimens M9. Source: DorART Project (photo by Juan Valcárcel Andrés).

• Type 2 test specimens (Figure 3):

The Type 2 test specimens were prepared in a quadrangular format, with one sample of each material used for the control condition and four additional samples used for aging cycles (totaling 45 samples). Half of each sample was water-gilded using 22 kt gold, with traditional preparation involving M9 and bole, and divided into matte and gloss sections by burnishing half of the surface with an agate stone.

Figure 3. Type 2 test specimens. Source: DorART Project (photo by Juan Valcárcel Andrés).

Figure 3. Type 2 test specimens. Source: DorART Project (photo by Juan Valcárcel Andrés).

2.

Type 3 test specimens: freestanding samples

The Type 3 test specimens were used for tensile strength testing in order to evaluate mechanical stability. A total of 168 test samples were prepared for tensile testing, which was carried out using a SHIMADZU^® Universal Testing Machine (Shimadzu Corporation, Kyoto, Japan) following the UNE-EN ISO 527-2 standard [32], type 1BA (Figure 4). To avoid internal air pockets and potential fractures, a vacuum degassing process was applied to the samples. This eliminated gaseous inclusions, improving the samples’ homogeneity and mechanical performance.

2.3.2. Environmental Test Chamber

The design of the thermo-hygrometric variable parameters for the cycles. Four specimens were submitted to accelerated aging cycles, while a control specimen was not subjected to this process. The established parameters were applied weekly, aiming to reproduce the progressive aging of each material. A specimen of each material was extracted from the climatic chamber every 7 days after each successive cycle for evaluation. The stipulated parameters were replicated weekly, in an attempt to mimic the natural progressive aging process of each material.

In the climatic chamber, the parameters for the aging cycles were adjusted according to ISO 9142:2003 [33], with a temperature of 55 °C and an RH of 30% applied for a total of five cycles. Following the recommendations of ISO 9142:2003, which provides guidance on the selection of standard laboratory aging conditions for adhesive bond testing, these accelerated aging cycles were selected to simulate thermo-hygrometric stress under controlled and repeatable laboratory conditions. The selected cycle corresponded to variable D1—the heat and humidity cycle—with an exposure period of one week at a temperature of 55 °C and a relative humidity of 30%.

While these tests cannot fully replicate the complexity of natural aging processes, they provide a structured methodology for evaluating material responses over time. As Artioli et al. [34] point out, “we always face the difficulty of replicating at a laboratory scale all the processes that take place during natural aging,” and they emphasize the importance of interpreting experimental results with caution. Their perspective highlights the need for documentation, standardized measurements, and the identification of reliable markers of deterioration in order to improve our understanding of long-term material changes.

3. Results

First, the factors evaluated in relation to the preparation, application, and hardening of the tested materials—including the type of substrate, the coating application methodology, and the curing or drying conditions—were analyzed. Once the accelerated aging conditions—exposure to humidity and temperature cycles—had been defined, the control specimen was analyzed and the experimental data were processed, grouped by test sample type. Each specimen was subjected to various tests to evaluate its behavior in the presence of deteriorating agents, specifically to assess the efficiency, stability, and compatibility of fillers.

3.1. Factors Evaluated in Relation to the Preparation, Application, and Hardening of the Tested Materials

This section details the mixed-methods approach employed to analyze factors related to the preparation and hardening of the tested materials, with a focus on assessing the compatibility and efficiency of the materials (Figure 5).

3.1.1. Preparation and Application Behavior

A multi-parameter evaluation was conducted, taking into consideration four criteria: preparation difficulty, complexity in application and modeling, textural adaptation, and workability. Each parameter was scored on a scale from 0 to 100. The resulting data were visualized using a Kiviat diagram with a radar chart, which allowed for a quantitative representation of multiple process variables and a comparative assessment of the materials’ performance, enabling the analysis of complex data through graphical representation [35,36].

Results. In terms of preparation difficulty, the highest scores were recorded for M6 and M9, indicating greater difficulty in their preparation, while M3 was the easiest to prepare. Regarding application and modeling complexity, M9 showed the highest complexity, followed by M8 and M1, whereas M3 and M4 had the lowest values, suggesting simpler application (Figure 5a).

For textural adaptation, the highest score was obtained for M3, indicating that it had the best fit for the required texture, while M9 and M7 had the lowest scores, reflecting a lower adaptation capacity. Finally, in terms of workability, M8 received the lowest score, indicating poor ease of handling, while M9 and M3 stood out for their high workability.

3.1.2. Post-Drying Properties

The analysis of different types of fillers revealed significant variations in their properties. As in the previous case, each parameter was scored on a scale from 0 to 100. The resulting data were visualized using a Kiviat diagram to enable comparative analysis between the tested materials (Figure 5b).

Results. As expected, dimensional shrinkage was more pronounced in the materials that were dried by evaporation (M4, M7, and M9), followed by M6, which was hardened through a combination of cooling and evaporation. No dimensional shrinkage was detected in the materials that were cured by chemical polymerization (M1, M2, M3). A more porous texture was observed in M4 and M5, while M7 and M3 exhibited a more compact and homogeneous appearance. In terms of ductility, M1, M3, and M4 showed the highest levels, with M8 presenting the lowest. Regarding peripheral color changes, M1 and M8 showed the most pronounced changes, followed by M9, while M4 presented the least chromatic impact. These results allow the suitability of each material to be evaluated according to the specific restoration requirements.

Figure 5. Kiviat diagrams showing comparative evaluation of the materials. Each axis represents a normalized score from 0 to 100, enabling a visual comparison of each material’s performance across different intervention stages. (a) Preparation and application behavior, evaluated through four parameters: preparation difficulty, complexity in application and modeling, textural adaptation, and workability. (b) Post-drying behavior, including dimensional stability, surface cohesion, adherence to the wooden substrate, and visual integration. Source: DorART Project (own data) 2025, visualized as Radar Chart [37].

Figure 5. Kiviat diagrams showing comparative evaluation of the materials. Each axis represents a normalized score from 0 to 100, enabling a visual comparison of each material’s performance across different intervention stages. (a) Preparation and application behavior, evaluated through four parameters: preparation difficulty, complexity in application and modeling, textural adaptation, and workability. (b) Post-drying behavior, including dimensional stability, surface cohesion, adherence to the wooden substrate, and visual integration. Source: DorART Project (own data) 2025, visualized as Radar Chart [37].

3.2. Control Specimen Analysis and Experimental Data Processing

A photographic record was created for the control specimen and for the test specimens after each of the aging cycles, in order to enable comparison and analyze the morphology of the initial specimens and any changes resulting from the thermo-hygrometric cycles, such as textural alterations, chromatic, formal, and dimensional modifications, adhesion to the surface, and material interaction. This provided an initial general overview for the subsequent analyses of efficiency, stability, and compatibility, which are detailed in the following sections.

3.2.1. Type 1 Test Specimens

• Physicochemical studies

• Measurement of the contact angle (CA) on the specimens’ surfaces. Contact angle measurements were performed in duplicate to evaluate the hydrophobic or hydrophilic behavior of the different fillers under various conditions. This parameter is relevant to the assessment of efficiency and compatibility, as it reflects the wettability of each putty. Furthermore, “contact angle (CA) measurements may provide quantitative information on the interfacial energy between a liquid drop and a solid surface” [38,39]. The procedure for measuring the contact angle [40,41] followed that established in UNE-EN 15802 [42], which specifies a method for determining the wettability of porous inorganic materials by measuring the contact angle of a water droplet. A 5 µL drop of distilled water was carefully deposited on the surface using a micropipette, and a photograph was taken 5 s after deposition to determine the contact angle. This value was used to assess the surface wettability of each sample.

Results. For the interpretation of the results, surface behavior must be taken into account, which can be classified according to the contact angle. A low contact angle (<90°) indicates that a surface is hydrophilic, favoring the absorption, adhesion, and penetration of the material. A high contact angle (≥90°) indicates that a material is hydrophobic. Materials M9, M8, M7, M6, and M5 presented contact angles of less than 90°, indicating that they have hydrophilic surfaces that favor the adhesion of aqueous materials such as clays with rabbit glue. The contact angle values obtained for each material are shown in Figure 6. M4 absorbed water immediately, and its content of wood chips allowed for very rapid drying; this sudden water absorption could lead to cracking. In general, the hydrophilic nature of these putties may enhance cohesion between layers and help to reduce delamination or separation issues. In contrast, materials M3, M2, and M1 exhibited low contact angles (≥90°), indicating that they have hydrophobic surfaces over which water does not spread well, making it difficult for aqueous materials to adhere. Their hydrophobicity means that they may require surface treatment, such as sanding or priming, to improve their interaction with water-based gilding.

• Analysis of the average pH evolution on the material surfaces over the test cycles. This study reveals different trends for each material and provides an assessment of their chemical stability. Measurements were carried out using the Horiba^® scientific compact pH meter LAQUAtwin-pH-11 (Horiba^® Ltd., Kyoto, Japan) with Agarose Basic A8963,0100 PanReac AppliChem^® (AppliChem GmbH, Darmstadt, Germany)plugs, a polysaccharide that forms a thermoreversible gel: “(…) Gelation occurs at temperatures below 40 °C, whereas the melting temperature appears to be 90 °C” [43]. Agarose plugs, used to measure the pH of the surfaces, were prepared in a 4% solution in water, with double-boiling in a microwave, according to the method proposed by Chris Staroudis, from the Getty Conservation Institute [44].

Results. The results (

Table 2. Analysis of the average pH evolution over the test cycles.

Fillers	Control Sample	Average pH After 1 Cycle	Average pH After 2 Cycles	Average pH After 3 Cycles	Average pH After 4 Cycles
M1	8.6	7.2	7	6.9	6.4
M2	9	7.6	6.9	6.7	6.6
M3	9.4	8	7.3	6.4	6.5
M4	9.5	8.2	6.9	6.8	6
M5	7.9	7.3	7.2	7	6.7
M6	9	7.8	7.2	6.9	6.8
M7	7.9	7.8	7.2	6.5	5.9
M8	8.4	7.5	6.9	6.7	6
M9	8.4	6.9	6.8	6.7	5.8

Source: Authors’ own elaboration, DorART Project, 2025.

) correspond to our own experimental tests, which were adapted from Hughes’ method for the study of paper surfaces [45] and conducted under controlled laboratory conditions, and as such, they do not represent values certified by manufacturers. The pH behavior of the different filler materials (M1 to M9) was evaluated over four successive cycles, and in all cases, a clear downward trend in pH was observed, indicating a general acidification of the medium.

Although all the materials exhibited a decrease in pH over the course of the aging cycles, some of them—such as M5, M6, and M2—maintained relatively stable values, with the final values near or above pH 6.5. This suggests lower chemical reactivity and greater resistance to degradation under thermo-hygrometric stress. In contrast, other materials, such as M4, M7, and M9, showed more pronounced pH drops, with the final values close to or below pH 6.0, indicating higher susceptibility to chemical alteration over time. Despite the observed acidification, materials M5 and M6 maintained a stable physicochemical profile during the aging cycles. This behavior can be attributed to their formulation: both are based on low-reactivity thermoplastic polymers (Beva^®) combined with inorganic fillers—calcium sulfate (CaSO₄) in M5 and vermiculite in M6. These components are chemically stable and hygroscopically inert, which likely reduces their susceptibility to hydrolysis or oxidative degradation under thermo-hygrometric stress. Supporting this interpretation, both fillers also showed minimal variation in Shore C hardness and exhibited the lowest adhesion variability across all materials, with CV values of 0.00%.

2.

Mechanical studies

• Adhesion tests on wood substrates. Adhesion tests were used determine the coating’s resistance to detachment. The adhesion strength of the reintegration fillers was evaluated following the specifications of ISO 4624:2003 [46], which defines the pull-off test method for assessing the adhesion of coatings to rigid substrates. An electronic adhesion tester (KN-10, Neurtek) was used for this purpose. This device is specifically designed for measuring the adhesion strength of paint layers, coatings, and building materials to their respective substrates. For each material, three replicates were tested. The procedure involved bonding a 20 mm diameter dolly to the surface under study using Araldite@ epoxy adhesive. After the manufacturer’s specified curing time had elapsed, the instrument was then positioned on the dolly, and a progressive pull-off force was applied. At the point of failure, the device digitally recorded the maximum force exerted. The equipment operates within a range of 5 to 1000 kgf (0.05 to 10.00 kN).

Results. Regarding efficiency and compatibility (

Table 3. Analysis of the adhesion evolution over the test cycles.

Fillers	Control Sample	Cycle 1	Cycle 2	Cycle 3	Cycle 4	CV (%) *
M1	8.23 MPa (90% cohesive)	6.76 MPa (95% cohesive)	4.31 MPa (70% cohesive)	4.41 MPa (25% cohesive)	4.41 MPa (25% cohesive)	28.40
M2	>7.94 MPa (failure)	>7.84 MPa (failure)	6.66 MPa (40% cohesive)	2.94 MPa (40% cohesive)	10.19 MPa (60% cohesive)	33.44
M3	7.06 MPa (60% cohesive)	6.27 MPa (50% cohesive)	7.45 MPa (80% cohesive)	6.37 MPa (20% cohesive)	6.47 MPa (20% cohesive)	6.77
M4	>6.96 MPa	4.02 MPa	>7.06 MPa	>3.92 MPa	8.43 MPa	29.58
M5	<0.98 MPa	<0.98 MPa	<0.98 MPa	<0.98 MPa	<0.98 MPa	0
M6	<0.98 MPa	<0.98 MPa	<0.98 MPa	<0.98 MPa	<0.98 MPa	0
M7	1.27 MPa (35% cohesive)	<0.98 MPa	<0.98 MPa	<0.98 MPa	<0.98 MPa	11.18
M8	1.08 MPa (50% cohesive)	<0.98 MPa (30% cohesive)	<0.98 MPa (30% cohesive)	<0.98 MPa (30% cohesive)	<0.98 MPa (25% cohesive)	4.00
M9	5 MPa (95% cohesive)	5.39 MPa (90% cohesive)	5 MPa (75% cohesive)	4.51 MPa (65% cohesive)	3.14 MPa (100% cohesive)	17.04

* CV%: Coefficient of Variation calculated from the adhesion values across cycles. Values > 15% reflect heterogeneous behavior or sensitivity to degradation. Source: Authors’ own elaboration, DorART Project, 2025.

), M1 exhibited high initial adhesion strength (8.23 MPa) and showed increased internal cohesion after aging, maintaining cohesive failure throughout the cycles. Fillers M2, M3, and M9 demonstrated a progressive loss of internal cohesion across successive cycles. M2 initially showed high adhesion values (>7.94 MPa), but its adhesion evolved to exhibit partial cohesive failure (40–60%) and signs of structural degradation, including a sharp drop in adhesion during Cycle 3 (2.94 MPa). M3 maintained relatively stable adhesion values (ranging from 6 to 7.4 MPa), though with a shift toward adhesive–cohesive failure, reaching only 20% cohesive failure by Cycle 4. M9 presented stable mechanical behavior in the first cycles (5–5.39 MPa), with predominantly cohesive failures (75–100%), indicating a strong bond to the substrate, although a slight decline in adhesion and cohesion was observed in later cycles. The coefficients of variation (CV%) for adhesion values were calculated. Values > 15% were observed in some materials, notably M2 and M4, indicating heterogeneous internal cohesion or degradation trends under thermo-hygrometric cycles. These variations are consistent with their fluctuating performance across aging cycles and reflect intrinsic differences in formulation and response.

Fillers M5 and M6 consistently showed low adhesion values (<0.98 MPa) across all cycles, with adhesive failures, indicating weak bonding to the substrate. M7 began with low adhesion strength (1.27 MPa) and exhibited partial cohesive failure (35%), but its strength dropped below 0.98 MPa in subsequent cycles, confirming poor mechanical performance. M8 also showed low initial adhesion (1.08 MPa), with mostly cohesive failure (50%), and continued to degrade over time, with its adhesion strength stabilizing below 0.98 MPa with reduced cohesion (25% in Cycle 4). In contrast, M4 displayed more heterogeneous behavior, with fluctuating adhesion values throughout the cycles. Despite an initial drop (4.02 MPa in Cycle 1), its adhesion strength recovered in later stages, reaching 8.43 MPa in Cycle 4, suggesting improved adhesion to the substrate over time.

• Shore C hardness. Within the study of the stability and compatibility of materials, Shore hardness is a measure of the resistance of materials. The method for measuring this parameter consists of pressing the material with a hardened steel penetrator, designed with a specific shape and force according to the selected measurement scale [47]. Ten measurements were taken for each putty, and the average values obtained are presented in Figure 7.

Figure 7. Chart of Shore C hardness measurements of tested fillers. Source: Authors’ own elaboration, DorART Project, 2025.

Figure 7. Chart of Shore C hardness measurements of tested fillers. Source: Authors’ own elaboration, DorART Project, 2025.

Results. M1, M4, M5, M7, and M9 exhibited the highest Shore C hardness values (90–95 HC), whereas the M2 and M6 fillers showed the lowest values (80–82 HC). These results are directly related to the formulation of each material, particularly the type of binder/agglomerate, fillers, and plasticizers present. In this regard, the M4 and M8 fillers contain vermiculite and small wood fragments, respectively, bound with resin, which significantly influences their mechanical behavior. M3 and M8 presented intermediate hardness values of around 87 HC.

Regarding their behavior after the aging cycles (Figure 8), putties M2, M3, and M8 exhibited an increase in hardness, which may be attributed to crosslinking processes in the resins that compose these materials. In contrast, M4, M5, and M6 showed a significant decrease in hardness, suggesting the presence of degradation processes likely related to the breakdown of polymer chains. M1, M7, and M9 showed minor fluctuations (ranging between 0.7 and 4 Shore C units), indicating greater resistance to aging.

3.2.2. Type 2 Test Specimens

• Physicochemical studies

All samples were analyzed using optical microscopy, which revealed aspects related to efficiency and compatibility in the test specimens. Three surfaces were studied, putty, matte gold, and glossy gold, allowing for a comparison of the morphology of the different putties and their microscopic behavior after the aging cycles. Gloss measurements were also carried out on the gilded surfaces to assess changes in surface sheen.

• Gloss. Gloss measurements were also carried out on the gilded surfaces of Type 2 specimens, allowing for the evaluation of optical efficiency and compatibility with the burnished metal leaf. The variation in gloss was assessed using a KONICA MINOLTA^® Multi Gloss 268A portable glossmeter (Konica Minolta^®, Inc., Tokio, Japan), which features three reflectometer geometries (20°, 60°, and 85°) and complies with ISO 2813 [48]. The measurement window of the device was positioned to ensure full contact with the specimen surface, and five readings were taken per sample.

Results. Analysis of the data revealed notable differences in the ability of certain fillers to maintain or enhance surface gloss. Most materials exhibited gloss values between 83 and 97 gloss units (GU). However, two fillers displayed markedly distinct behavior: M3 showed the highest gloss level, exceeding 110 GU, suggesting that the surface of this filler interacted favorably with the gold leaf, possibly due to its texture and composition. In contrast, M4 stood out for its low gloss value (~60 GU), indicating that its surface may not be suitable for applications requiring a high-luster gold finish. This may be related to factors such as porosity, bole absorption, or the chemical composition of the filler. These results suggest that when using this base material, additional surface treatments may be necessary to improve optical performance.

2.

Mechanical studies

• Abrasion of the fine gold layer and stability of the water gilding on the tested materials. The efficiency and compatibility of the materials were assessed by evaluating their abrasion resistance using a Taber^® 5750 Linear Abraser (North Tonawanda, New York, US), in accordance with the UNE-EN ISO 7784-3 standard [49]. The tests were carried out under controlled conditions, with an abrasion path length of 1.27 cm and a speed of 60 cycles per minute. An H-10 hardness wearaser tip was used to simulate surface wear on the specimens.

Results. Burnished gilding demonstrated greater mechanical stability and better resistance to abrasion, whereas matte gilding proved more vulnerable under the same test conditions. These differences are clearly illustrated in

Table 4. Surface micrographs of M6 specimens after abrasion tests across aging cycles.

M6	INITIAL	Cycle_1	Cycle_2	Cycle_3	C_4
Burnished gilding
Matte gilding

Source: Authors’ own elaboration, DorART Project, 2025.

, which presents surface micrographs of M6 specimens across aging cycles after abrasion testing. The specimens that showed the best resistance to abrasion were M2, M4, M8, and M9. In these cases, the wear was mostly limited to the gold leaf, exposing the underlying bole layer. Occasional losses of bole were observed, but the base material (putty) remained intact, indicating that these fillers provided higher mechanical stability and stronger adhesion between layers.

Conversely, specimens M1, M3, M5, M6, and M7 exhibited less favorable performance, with significant losses affecting not only the gold leaf but also the bole layer and, in some cases, the base material. M1 exhibited the most pronounced deterioration, suggesting a lower mechanical strength of both the gilded surface and the overall system. M5, although initially resistant to abrasion, began to show greater degradation from the third cycle onward, possibly due to a progressive loss of internal cohesion or reduced adhesion between layers. M3 and M1 showed intensified deterioration beginning in the fourth cycle, likely resulting from a combination of factors, such as reduced internal cohesion of the filler material and weakened interfacial bonding.

3.2.3. Type 3 Test Specimens

• Mechanical testing using Universal Testing Machine and TRAPEZIUM ^®X software (Version 1.3.0 2007-2012 Shimadzu© Corporation, Kyoto, Japan). Prior to testing, width and thickness measurements were taken at three different points on each specimen using a Mitutoyo^® 500-181-30 caliper, model ID- C150B (Mitutoyo Corporation, Takatsu-ku, Japan) (Figure 9), and an average value was then calculated (Table 5).

Figure 9. Measurements of all test specimens. Source: DorART Project, 2025.

Figure 9. Measurements of all test specimens. Source: DorART Project, 2025.

Table 5. Dimensions of test specimens.

Material	Sample	Width (mm)	Deviation (mm)	Thickness (mm)	Deviation (mm)
M1 control 1	M1MC-1	4.88	0.0	2.04	0.1
M1 control 2	M1MC-2	4.90	0.1	2.14	0.1
M1 control 3	M1MC-3	4.95	0.1	2.05	0.1

The data represent the mean values and standard deviations of the sample dimensions. Source: Authors’ own elaboration, DorART Project, 2025.

Table 5. Dimensions of test specimens.

Material	Sample	Width (mm)	Deviation (mm)	Thickness (mm)	Deviation (mm)
M1 control 1	M1MC-1	4.88	0.0	2.04	0.1
M1 control 2	M1MC-2	4.90	0.1	2.14	0.1
M1 control 3	M1MC-3	4.95	0.1	2.05	0.1

The data represent the mean values and standard deviations of the sample dimensions. Source: Authors’ own elaboration, DorART Project, 2025.

Furthermore, a gauge length of 25 mm was established for each of the 168 specimens; controlling this parameter was essential to ensure accurate evaluation, especially considering that the specimens did not have identical geometric dimensions (Figure 9).

For mechanical testing, the TRAPEZIUM X software was used. The testing machine was configured with a load capacity of 100 kN, and a standard procedure was followed, with a displacement rate of 1 mm/min applied to flat specimens until failure (Figure 10). Three valid measurements were obtained, from which an average value was calculated.

Based on these data, the software automatically generated a strain–displacement graphic. From this graphic, the engineering stress (a) and engineering strain (b) were calculated using the following equations:

$$(a) S={\displaystyle \frac{F}{A_0}}$$

$$(b) \epsilon ={\displaystyle \frac{\Delta l}{l_0\simeq 25 \mathrm{m}\mathrm{m}}}$$

where A₀ is the original cross-sectional area of the specimen before the test begins, l₀ is the original gauge length, and $\Delta l$ is the change in length after the force F is applied. With these parameters, stress–strain curves were created, recording the mechanical behavior of the materials under tension (Figure 11) [50].

In specimens M8 and M9, the ends were reinforced with aluminum sheets, as the grips tended to fracture the specimens due to the pressure exerted (Figure 12), thereby complicating the testing process.

The data enabled the calculation of the maximum stress and the elastic modulus, along with their respective standard deviations (

Table 6. Average and standard deviation of maximum stress and elastic modulus.

Cycle	Average (Max. Stress) (MPa)	Deviation (MPa)	Elastic Modulus (GPa)	Deviation (GPa)
M1MC	11.0	0.9	783	14
M1C1	10.7	0.7	780	95
M1C2	13.4	1.0	935	122
M1C3	12.3	1.8	1227	235
M1C4	11.5	1.4	1050	338

The data present the mean values and standard deviations of the maximum tensile strength and elastic modulus of the reference specimens and the specimens of putty M1 subjected to aging cycles. Source: Authors’ own elaboration, DorART Project, 2025.

), allowing for an assessment of the consistency and reliability of the results obtained throughout the testing cycles (Figure 13).

Results. Based on the analysis of these graphics and a comparison between the different filler formulations evaluated, it was determined that the formulations with the highest mechanical strength were M1 (13 MPa) and M4 (13 MPa), both of which achieved their peak values in the second cycle. On the contrary, formulations M8, M5, and M6 were classified as the least resistant, and they also showed high fragility during demolding. In general terms, the mechanical strength of most formulations reached its maximum in the second cycle, followed by a progressive decline in subsequent cycles, which may indicate a loss of structural capacity as a consequence of aging. However, some exceptions to this pattern were observed:

• M5 showed a progressive increase in strength throughout the cycles, although with very low stress values.
• The strength of M7 reached its maximum in the first cycle and then decreased.
• M3 presented high initial strength, which dropped in the second cycle and then increased progressively until the fourth cycle, suggesting that this specimen had more stable long-term mechanical behavior.

• ATR-FTIR Spectroscopic Characterization. A chemical–mineralogical analysis of the fillers was performed using ATR-FT-IR spectroscopy in order to characterize the molecular structure and chemical composition of the materials and to assess possible changes induced by the aging cycles. An ATR-FTIR spectroscope Vertex70 spectrometer (Bruker Optik GmbH 2012, Ettlingen, Germany) was used with MKII Golden Gate with an ATR accessory (Specac^®, Orpington, UK) and an FR-DTGS—a fast-recovery deuterated triglycine sulfate temperature-stabilized coated detector. A total of 32 scans were collected at a resolution of 4 cm⁻¹. The spectral range scanned was 500–4000 cm⁻¹. The spectra were processed using OPUS 7.2/IR software (Bruker Optik GmbH 2012, Ettlingen, Germany). The samples were analyzed directly.

Results. In general, a reduction in the intensity of bands corresponding to hydroxyl groups [51] (–OH) located in the 3300–3400 cm⁻¹ region was observed, which can be interpreted as a consequence of the loss of structural or surface-bound water. This behavior was consistent across most spectra and is associated with dehydration processes triggered by the thermo-hygrometric conditions applied during accelerated aging in the climate chamber. These trends are illustrated in Figure 14, which shows the spectral evolution of two selected filler materials across the aging cycles.

However, in formulations M1 and M9, the initial moisture loss appears to be partially offset by the formation of new –OH groups, likely due to mild oxidation, bond cleavage, or exposure of previously hidden functional groups. The oxirane ring vibration band at ~910 cm⁻¹ is also consistently present and suggests curing reactions or the loss of unreacted epoxide groups, a phenomenon that is particularly evident for the samples containing epoxy components.

The carbonyl group (C=O) at ~1730 cm⁻¹ shows a slight increase in M1, M4, and M9, indicating the onset of oxidative processes or thermal degradation of the organic matrix.

In the case of inorganically filled materials, such as M5, M6, and M8, spectral variations are observed in the 1100–900 cm⁻¹ region, which are attributed to internal structural reorganizations that are likely related to sulfate- or silicate-based phases.

By contrast, M3, M7, and M9 show minimal chemical or mineralogical changes, suggesting that they had greater structural and chemical stability under conditions of moisture and temperature fluctuations.

In conclusion, thermo-hygrometric aging cycles affected all samples to varying degrees; however, the extent and nature of the observed changes clearly depended on the specific composition of each formulation.

4. Discussion

In terms of application efficiency, fillers M3 and M9 demonstrated excellent workability, ease of handling, and adequate surface adaptation, making them particularly suitable for on-site interventions. M3 also recorded the highest gloss level (>110 GU), indicative of a well-suited surface for burnished gilding. M9 combined this ease of application with stability during drying and good abrasion resistance. In contrast, M6 and M9 exhibited greater complexity in the preparation phase. Although inferential statistics were not applied due to the limited sample size, the data provide robust comparative trends that support material selection in conservation practice.

The evaluation of compatibility with the water gilding technique considered both optical performance and abrasion resistance. Formulations M2, M4, M8, and M9 performed remarkably well, limiting losses to the gold leaf while preserving the bole and base material, especially under burnished gilding conditions. M3 maintained acceptable performance even after the fourth cycle, whereas M1, despite its good internal cohesion, was among the most affected by abrasion, suggesting surface vulnerability under frictional stress. Regarding mechanical stability, tensile strength testing showed that M1 and M4 achieved their highest strength values (13 MPa) in the second cycle. Most formulations reached their peak strength during that cycle and then progressively declined, likely due to aging effects. M3, however, displayed different behavior: after an initial decrease, it recovered strength by the fourth cycle, indicating potential long-term stabilization. M5, although showing low strength values, displayed a slight progressive improvement.

In terms of adhesion to the wood support, M1 stood out with an initial value of 8.23 MPa and good performance over time. M2 showed high initial adhesion but suffered a progressive loss of internal cohesion.

Shore C hardness measurements revealed that M1, M4, M5, M7, and M9 presented the highest values (90–95 HC), while M2 and M6 had the lowest values. For some of the putties studied, such as M2, M3, and M8, their hardness increased after aging, suggesting internal crosslinking processes, while M4, M5, and M6 showed significant reductions in hardness, indicating possible degradation of their matrices.

Chemical–structural analysis using ATR-FTIR revealed a general reduction in –OH groups due to dehydration induced by thermo-hygrometric aging conditions. However, in M1 and M9, possible partial regeneration of these groups was observed, attributed to mild oxidation or exposure of previously inaccessible functional groups. The oxirane ring band (~910 cm⁻¹), present in materials with epoxy components, was suggestive of curing reactions or the loss of unreacted epoxide groups, particularly in M1. Carbonyl-related bands (~1730 cm⁻¹) increased slightly in M1, M4, and M9, suggesting the onset of oxidative processes. In materials with inorganic components (M5, M6, M8), variations were detected in the 1100–900 cm⁻¹ region, related to internal reorganization of sulfate or silicate phases. In contrast, M3, M7, and M9 showed minimal spectral changes, suggesting greater chemical stability during the aging process.

Overall, the data obtained indicate that formulations M3 and M9 offer a particularly favorable balance between application efficiency, compatibility with the water-based gilding technique, and mechanical and chemical stability on gilded wooden heritage objects.

5. Conclusions

This study provides restorers of wooden cultural assets with specific results that can support their decision making, especially when focusing on the restoration of water-based gilding on wood. The material selection was supported by citizen science, which enabled the identification of the most commonly used materials and was key to the design of the experimental study, ensuring that it would take end users into consideration, address their needs, and provide data applicable to diverse intervention contexts. The study showed that each filler formulation exhibited distinct behaviors, depending on its composition, curing type, and response to aging. The results can help to guide the selection of materials based on the specific technical and material characteristics of a particular item or the requirements of the intervention, in addition to the thermo-hygrometric conditions to which the work of art will be subjected. Without aiming to establish a classification or hierarchy among the volumetric reintegration materials analyzed, the data obtained allow us to objectively assess the materials’ strengths and limitations in three conditions: when presented on linear supports (test piece 1), when presented on wooden supports with water-gilded surfaces (test piece 2), and when existing as a freestanding material (test piece 3), providing a series of data that are intended to help guide conservation treatments. The findings also underscore the importance of selecting the most appropriate filler according to the specific intervention context, the decorative technique used, and the requirements for compatibility, stability, and adaptability in each case.