Early degradation mechanisms at the interface between acrylic grounds and oil paint films

This section presents a detailed discussion of the data obtained from the combined use of a multi-analytical approach of non-invasive techniques, such as multiband imaging and portable digital microscopy, with micro-invasive methods including HR-FESEM-EDX, ATR-FTIR, and GC–MS.

Observation of degradation phenomena by means of non-invasive techniques

This section focuses on the visual and diagnostic examination of degradation patterns using non-invasive methods such as multiband imaging and digital microscopy.

Multiband imaging (MBI) was carried out to observe the surface and the possible alterations affecting paintings. VIS, RL, IR, IRT, TL, UVL, and UVR were used to record the appearance of the paintings across different bands of the spectrum and under different modes.^{41,45,49,51,52,53}.

Figures 2–4 show the images obtained using the different multiband techniques applied to SET3, SET2, and SET1, respectively. The images shown are a selection of the most relevant results.

The following degradation phenomena were observed in the prepared mock-ups by reading cross-sectionally and comparing several images taken:

• Wrinkles: uneven (small) folds, ridges or grooves resulting in an undulating and wavy layer on the paint surfaces^54,55;

• Cracks: single cracks or networks of straight or slightly curved lines, known as craquelure^49,56,57.

• Protrusions: tiny, translucent, or opaque white globules that break through the surface of the paint layer (protruding masses). Typically ranging from 50 to 300 µm in size—though they can be larger—these protrusions create a cratered texture on the surface.^58,59,60,61.

Figure 2 collects the most interesting cases of SET3. The VIS and RL images show severe wrinkling. Considering that the samples kept at controlled room conditions also showed this alteration, it might be logical to think that the formation of wrinkles was not necessarily induced by the environment, but more likely related to the chemical composition of the paint films and/or the interaction between the paint and the preparatory layers. RL also allowed the cracks on the surfaces of the mock-ups to be better observed. Of particular interest are the mock-ups SET3-TiGD-T and SET3-ZnMM-T.

The observation of the surface morphology by optical microscopy (Fig. 3) revealed more pronounced and deeper cracks in the mock-ups with a zinc-based oil paint film. In such specimens, IR imaging showed darker cracks, suggesting that they also affected the inner layers⁶². In contrast, the mock-ups with a titanium-based oil paint film showed no light absorption, indicating that the cracks were more superficial and only affected the paint film (Fig. 4). The IR images were compared to IRT images, which provided additional information such as structural defects (voids, internal cracks, delamination or thermal changes caused by layers not visible in IR)^51,53. By comparing the cracks of the three types of mock-ups studied, it was possible to distinguish between surface cracks (affecting only the paint film) and cracks with cleavage (affecting several layers and showing a separation between the paint film and the preparatory layers, although the paint film was still attached to the underlying substrate). The results confirmed the previous hypothesis: only the specimens made with a zinc-oil paint film showed deep cracks with some degree of cleavage affecting the underlying layers.

This time-dependent process of crack propagation across the layers was determined by comparing the VIS, IR, and IRT images of the sample SET3-ZnMM-T. As shown in Fig. 5, neither the surface level captured in the VIS image nor the deeper level revealed in the IR image shows any cracks in the highlighted red area. However, the IRT technique highlights internal cracks in the layer closest to the canvas. The IRT image makes it possible to suppose that the cracks originate specifically in the acrylic primer before they become visible and propagate into the acrylic gesso (detected by IR) or the oil paint layer (detected by VIS).

Specific literature states that IRT images are sensitive to inhomogeneities, including trapped moisture or variations in material properties, which might alter the thermal behavior of painted surfaces⁶³. The acquired IRT images allowed to observe areas where heat was not uniformly absorbed, resulting in darker areas. This initial observation was confirmed by TL images, where a halogen light source is positioned behind the canvases at a distance of 80 cm, and a sensitive camera records the light uniformly transmitted through it⁶⁴. The quantity and quality of the light passing through depend on the thickness of the material. This technique allows the detection of defects or structural variations in the layers closest to the canvas, which would otherwise remain invisible on the surface.

In particular, in samples where darker areas were observed in IRT images, a similarly irregular surface is also visible in TL images. According to the literature, this phenomenon can be attributed to areas of different densities, aligning with previous observations⁶⁴.

IRT and TL images also suggested a significant alteration in the thermal and optical properties of some mock-ups, indicating localized changes in their composition and condition.

These alterations are observed not only in SET3 but also in SET2 and SET1. Specifically, the results observed in samples SET3-ZnGD-RH%, SET2-TiGD-RH% and SET1-ZnMM-RH% were particularly relevant. These mock-ups exhibited thermal and optical inhomogeneity spots areas: accumulation of moisture (which appeared darker in TL and exhibited different thermal behavior in IRT images), corresponding to protrusion areas observed in VIS. Moreover, the combination of these techniques evidenced structural defects and forms of degradation not yet evident on the surface in the visible range. The aggregation of material that had not yet migrated to the surface in the form of protrusions observed in SET3-ZnMM-NAT seems to originate in the preparatory layers (Fig. 8).

Two forms of three-dimensional failure were registered in Fig. 9:

• protrusions (as previously described);

• protrusions with medium separation: the separation and accumulation of the lipid medium in punctiform areas of the paint surface, forming a (locally) shiny, yellowish spot.

Such a degradation phenomenon is not always visible in the commonly reflected VIS photography. Instead, it can be better recorded by UVL imaging, which reveals features that were not evident under visible lighting conditions^65,66,67, In the studied examples, UVL allowed to clearly distinguish between protrusions and protrusions with medium separation. In the three mock-ups under examination (SET3-ZnGD-RH%, SET2-TiGD-RH% and SET1-ZnMM-RH%), it was observed that (Fig. 10):

• the protrusions exposed to UV radiation showed a reflection phenomenon in correspondence to all the areas with protrusions;

• only the protrusions with medium separation reflect and emit light in the VIS spectrum, showing the luminescence nature of the lipid medium. Such luminescence is highly material-specific and depends on the chemical composition of the surface⁶⁸.

In Fig. 3d, e show that oil separation also occurred within some of the cracks. This is confirmed by the luminescence response within the crack in UVL observed in SET3-ZnMM-T.

Finally, UVR analysis allowed the determination of specific characteristics related to the surface and chemical composition of the materials. UVR is the equivalent to IR and VIS reflected photographs, but in the UV band. An opaque effect was observed on the surface of the mock-ups containing a Golden titanium white oil paint film in SET3 and SET2, and subjected to both natural and %RH artificial ageing. This could indicate either the presence of materials with high UV reflectance or scattering properties or a surface alteration. Due to the interaction of UV radiation with materials, this technique highlights alterations such as photodegradation, usually not visible in VIS light or other spectral bands, which provides valuable information about the condition of the mock-ups.

Chemical and physical interactions at the interface in-between layers

This section focuses on the characterization of those samples that showed the most relevant failure issues at the interface between the acrylic preparatory layer and the oil paint film by means of non-invasive techniques.

SET1-ZnMM-RH% and SET3-ZnGD-RH% were selected to represent the mock-ups having protrusions which, in some cases, also showed medium separation (Fig. 9). In order to study these phenomena in detail, HR-FESEM-EDX was carried out, allowing a micro-morphological examination of the alteration and the detection of elements³⁰. Cross-sectional analysis of the samples revealed two different types of protrusions: – protrusions with separation of the (lipid) medium (Fig. 11); – protrusions where the separation consists of metal soaps (Fig. 12).

Figure 11 shows the cross-section of SET1-ZnMM-RH% in the area corresponding to the protrusion and the separation of the medium previously observed in VIS on the surface. The analysis provided detailed insights into the structure of the mock-up and the distribution of elements in the organic components (carbon and oxygen) and the inorganic materials (zinc, calcium, magnesium, titanium, aluminium, sodium, and silicon) within the paint film.

The imaging revealed:

• a homogeneous distribution of zinc, sodium and calcium ions in the paint film;

• a distinct region, corresponding to the protrusion, where this homogeneity was disrupted, and only organic components were detected.

EDX confirmed the exclusive presence of carbon and oxygen, indicating that the region was basically composed of an organic compound, i.e., lipid medium, without any metal ions.

Figure 12 shows the analysis of SET3-ZnGD-RH%. As in the previous example, the cross-section was taken from an area where lipid medium separation was observed on the surface in the visible (VIS) band. The VIS image revealed a distinct region corresponding to the protrusion, enriched not only with carbon and oxygen—characteristic of the oil medium—but also with a significant concentration of zinc metal ions. The EDX spectrum confirmed the higher presence of an organic fraction along with metallic ions originating from the acrylic primer layer.

The samples in Figs. 11, 12 exhibit several common characteristics: • Both displayed a degradation mechanism involving the separation of the lipid fraction.

• Both contained a Zn-based paint layer, a pigment known to promote this type of deterioration.

• Both underwent artificial ageing with % RH fluctuations, a process that accelerates hydrolysis and the migration of metal ions. The key difference lies in the forms of degradation observed:

• in SET1-ZnMM-RH%, the separation consisted exclusively of the oil medium.

• in SET3-ZnGD-RH%, the separation involved an accumulation of organic material and zinc ions, which are probably related to the formation of metal soaps, a structure composed of metal ions present in the paint layer and free fatty acids from the lipid binder.

GC–MS and ATR-FTIR investigated the chemical interactions that led to the degradation phenomena observed^{50,63,69,70,71},

The TIC (total ion current) chromatograms show all the characteristic fatty acids (as their corresponding methyl esters) detectable in oil paint films. Table S1, supplementary material, summarises the results of qualitative and quantitative analyses performed with GC–MS on the following mock-ups:

• SET3-ZnMM-T, which showed cracks with lipid medium separation;

• SET3-TiGD-T, which showed surface cracks;

• SET3-ZnGD-RH%, which showed protrusions probably consisting of metal soaps.

• SET1-ZnMM-RH%, which showed surface protrusions without lipid medium separation.

Figure 13 presents a focused wavelength region of the ATR spectrum and GC–MS chromatogram of sample SET3-ZnGD-RH%.

The spectra analyzed presented the characteristic bands of an oil binder:

• the broadband centered at ca. 3440 cm⁻¹ is assigned to the stretching of alcohol and hydroperoxide bonds;

• the band at ca.1740 cm⁻¹ assigned to the ester stretching;

• the bands corresponding to the CH stretching are ca. 2928 and ca. 2860 cm⁻¹.

ATR-FTIR analysis highlighted:

• the band at 1710 cm⁻¹ corresponding to the C=O stretching vibration related to the formation of carboxylic acids, which may form as a result of both the hydrolysis and oxidative degradation of triglycerides.

• Moreover, a broad band at ca.1580 cm⁻¹ related to amorphous zinc carboxylates was detected, since the band is shifted by 40 cm⁻¹ toward higher wavenumbers with respect to the absorptions of their corresponding crystalline Zinc soaps at ca. 1540 cm^−1 72.

• Finally, the peaks at 1465–1464 m (CH₂ δ bending) and 1399–1397 m (symmetric carboxylate stretching ν COO), due to the presence of carboxylate groups.

Overall, these features support the presence of non-crystalline zinc carboxylates and carboxylic acid groups within the protrusions, but do not allow for a definitive identification of specific zinc soap species. Zinc carboxylates were detected in all mockups shown in Fig. 9, confirming that this degradation phenomenon occurs in both zinc- and titanium-based oil paints. Literature suggests that zinc has a strong tendency to bind with carboxylate groups, forming metal soaps^33,35,36,73. This explains their presence in the titanium oil paint mock-ups, which contain a 10% mixture of zinc oxide.

GC–MS results highlighted the presence of a variety of oxidized acids (e.g. oxo, hydroxy-, and epoxy-octadecanoic acids) related to an advanced degree of oxidation of the unsaturated fatty acids present in the fresh oil: these compounds are formed through the oxidative pathway of unsaturated fatty acids during the propagation reactions of the oxidation process^{25,26,27,74,75}. The evidence of degradation processes is also supported by the detection of dicarboxylic fatty acids, such as sebacic acid, suberic acid, and azelaic acid, which are known to be the tertiary oxidation products of unsaturated fatty acids^{37,60,76,77,78,79}.

The palmitic to stearic acid ratios (P/S) were calculated to help in defining the kind of oil used¹⁹ and compared with the information about oils reported in the technical data of color tubes. It was, therefore, possible to confirm that Golden Artist Colors Inc. (GD) employs alkali-refined linseed oil (P/S 0.9 ± 0.3). Linseed oil has typically a P/S ratio of approximately 1.6 ± 0.3, but in the SET3-TiGD-T and SET3-ZnGD-RH% samples, the P/S ratios were notably lower (0.95 and 0.8, respectively) (Table S1). This might be linked to the common addition of metal stearates as dispersion agents in oil manufacture. The alkali-refining process used to produce alkali-refined linseed oil (ARLO) is highly effective in purifying the oil, removing mucilages, phospholipids, and other impurities, and improving its drying properties^74,80. However, this refining process involves the addition of free fatty acids that may have favored the formation of metal soaps, as observed in this sample.

Moreover, mock-up SET3-ZnGD-RH% exhibited a higher glycerol concentration compared to other samples with the same paint layer (ZnO) from the same brand (Golden) and subjected to the same accelerated ageing conditions (RH%), but with different ground structures. Specifically, the glycerol percentage was 16.26% in sample SET3-ZnGD-RH%, 5.25% in sample SET2-ZnGD-RH%, and 3.32% in sample SET1-ZnGD-RH%.

This suggests that mock-up SET3-ZnGD-RH% underwent increased hydrolysis of ester-glycerol bonds, leading to the formation of di-, mono-, and triglycerides, glycerol, and free fatty acids^74,80. The enhanced formation of metal soaps observed in this sample can therefore be attributed to its structural composition rather than the alkaline refining process used to produce ARLO.

The formation of metal soap reflects extensive degradation processes in the mock-ups of SET3, which not only affect the oil paint film, but also appear to originate in the preparatory layers, as supposed by comparing RL and IRT images (Fig. 8).

As is known from the literature, coalescence is the process where acrylic-based polymers consolidate into a dense, compact structure ^82,83. During this process, as water and other volatile compounds evaporate, the micrometer-sized polymer particles merge, forming a strong hexagonal, honeycomb-like polymer network. This creates a dense microstructure, crucial for the mechanical stability and integrity of the resulting acrylic paint film. This also occurs in the case of the acrylic ground used in this experiment. Ideally, coalescence would form a continuous polymer film once dried. However, in this case study, the coalescence of the polymer spheres leads to a polymer film with interstitial spaces, resulting in somewhat porous acrylic films with channels running along the walls of the hexagonally deformed particles. These pores are pathways for water to move in and out of the film. In addition, because acrylic films require hydrophilic, water-attracting additives to ensure compatibility with water, they will inevitably retain some water, even after appearing fully dry. Thus, residual moisture can accumulate in some areas of the interface between the oil paint film and the acrylic ground, making the polymer system more susceptible to hydrolysis, as the ester bonds remain exposed to water molecules^13,72,81. Moreover, environmental factors significantly influence drying: the fluctuation of %RH can promote water absorption into the acrylic layer, further increasing the risk of hydrolysis^72,82.

These processes highlight the key role of both chemical composition and environment in influencing the degradation kinetics of oil paint systems when applied to acrylic grounds^81,83.

Cross-sectional HR-FESEM-EDX images confirmed the observations made through optical microscopy (Fig. 3) and multiband imaging analysis (Figs.4,5):

• crack patterns were observed exclusively in samples from SET3;

• the different behavior of zinc and titanium white paint films under the same environmental conditions (artificial ageing with temperature variation) and the same structural composition (SET3).

In Fig. 14, the cross-sectional analysis of the mock-ups with titanium white (SET3-TiGD-T) shows that the observed crack is exclusively confined to the paint film, thereby confirming the superficial nature of the cracks as previously described.

Figure 15 provides critical micro-scale insights into the structural damage and degradation mechanisms of the mockups made of a zinc-oil paint film (SET3-ZnW&N-T). Figure c highlights that the cracks start within the acrylic primer, traverse the entire acrylic gesso ground layer and terminate at the paint film. This observation is consistent with previous studies indicating that zinc oxide can reduce the flexibility of paint films, increasing the risk of cracking, delamination, and metal soap formation^{33,34,35,36,37,73}. In addition, the crack propagation through the layers from the bottom (acrylic primer) to the top (oil paint film) was demonstrated by comparing the VIS, IR, and IRT images of this sample. (Fig.5).

The damage observed in mock-ups made with a zinc white oil paint film is attributed to its chemical and mechanical properties. It is well known that zinc oxide forms a brittle and inelastic paint film. Furthermore, these properties are enhanced by the tendency of zinc ions to bind to the carboxylate groups of the polymer network, acting as anchor points for the carboxylic acid groups during the paint-drying process. This increases the density of cross-links and leads to a decrease in elastic properties over time, resulting in increased rigidity and brittleness. The brittleness of the paint film compromises its structural integrity, making it more susceptible to deeper friable cracking under stress. In contrast, the titanium white samples exhibited superficial cracks confined to the oil paint film, suggesting a more stable structural integrity^{18,78,84,85,86,87}. It is known from the literature that titanium white does not form stiff films, but has limited strength and behaves in a somewhat flexible manner. This would explain the superficial cracking observed, which only affected the paint film, leaving the underlying ground layer intact, as acrylic films are more elastic.^{17,81,84,85,86,87}.

This distinction highlights the key role of pigments in the mid-to-long-term durability and stability of oil paintings. This study shows that the interaction between the pigmented oil medium at the interface with acrylic grounds may significantly govern some physical properties of the oil paint film, such as its flexibility, adhesion to the underlying substrate, and sensitivity to moisture (and environment fluctuations).