3.3. Drying
The analysis of the drying time provided the kinetics of water evaporation from either an adhesive thick layer or from a relevant support.
Figure 4a shows the drying curve of a 1.2 cm thick layer of foam deposited onto a Petri dish, and it is noteworthy that the complete removal of the solvent (about 47 wt.%) that occurred within 24 h was not significantly affected by the environmental relative humidity.
Furthermore, 0.02 g/cm
2 of foam was spread onto the surface of 5 cm diameter Origam and Ispra canvases, used as the lining canvases of a cotton one, in order to simulate the real application. As indicated in
Figure 4b, only 3 h is required for the complete water evaporation, independently of the lining canvas used. Moreover, the weight loss at complete evaporation is 47%, as for the Petri dish drying test, in line with the product specification. Therefore, no water can be trapped in the solidified external regions of the foamed adhesive, thus avoiding possible water vertical diffusion phenomena at longer times.
The drying behavior of the other commercial glues (Beva Gel, Plextol B500, glue paste, and Liquid EVA water-based emulsion) was tested for the sake of comparison, to inspect the drying time of a 0.8 cm thick layer spread on a Petri dish (
Figure 5). Although the first three systems were the most common adhesives for lining paintings, it was found that the time required to the complete evaporation was four times longer with respect to the foamed EVA. Moreover, the long time needed for the evaporation of the solvent led to the formation of a solidified skin that slowed down the evaporation process, and part of the solvent remained trapped. Such an effect when occurring between the painting and the canvas may lead to permanent damages to the painting.
Table 4 (and
Figure 5) highlights that the use of a foamed adhesive can shorten drying time, and it prevents any possible formation of solid and less permeable skin, which may hinder solvent evaporation. These aspects are crucial not only from the technical/technological point of view, but since the solvent (i.e., toluene for Beva Gel) is often toxic in most of the commercial adhesives in the field, safety concerns may be associated with the adhesive application.
Furthermore, an additional advantage of the foamed solution is related to the amount of adhesive needed to obtain the total coverage of the surface and thus to ensure the complete adhesion of the lining canvas. The amount of the foamed EVA, indeed, is always significantly lower than that required by the liquid EVA resin. Such differences are reported in
Table 5 and the results are also influenced by the material considered for the preparation of mock-ups.
3.4. Peel and Lap Shear
The mechanical tests carried out on the mock-ups allowed us to compare the adhesion strength of the foamed EVA adhesive with the other commercial solutions. It is useful to remind readers that all samples tested were fabricated considering the same canvas, using the prescribed method for re-lining, and spreading the adhesives by means of a spatula. Therefore, the amount of adhesive deposited may vary (see, e.g.,
Table 5), as well as the thickness of the bonding layer. However, such differences are not expected to significantly affect the adhesion tests.
Table 6 highlights the results obtained using either Origam or Ispra as the lining canvas. Relevantly, when Origam was used in the mock-up preparation, the failure of the canvas occurred before the end of the adhesion tests for all of the samples, except for foamed EVA and paste glue. Therefore, the comparison with the optimal performances in terms of adhesive capacity and adhesive strength indicated in the technical literature (280–400 N/m for peeling and to 0.24 MPa for shear [
32]) was performed on Ispra canvas samples only. Plextol B500, foamed Evart, and Beva Gel provide values close to the optimal ones, but it is noteworthy that during the tests, the use of both Plextol and Beva Gel led to the detachment of the layer that simulated the painting with the Ispra canvas, although the adhesion strength results were quite large. Conversely, foamed EVA shows satisfactory results, without damaging the canvas to any extent; the adhesion failure occurs right within the adhesive layer, with no possible consequences for the painting.
3.5. Surface and Color Alteration
The results of colorimetric analysis on the mock-ups front side (
Table 7,
Figure 6 and
Figure 7) showed a similar chromatic alteration due to a post degradation process on the specimens with and without adhesive application on the two lining canvases, with Δ
E∗(ab) values always remaining well below the alert threshold (Δ
E∗(ab) > 6). Concerning the “M” series (Ispra support), the Δ
E∗(ab) values were always kept below 3 (i.e., the limit of perceptibility), with a lower value for the foamed adhesive (Δ
E∗(ab) foamed EVA = 1.88; Δ
E∗(ab) liquid EVA = 2.82). Regarding the samples supported by Origam, Δ
E∗(ab) rose above the value of 3 (alteration defined as perceptible but not alarming), with no significant differences between the two forms of resin application (Δ
E∗(ab) foamed EVA = 3.11; Δ
E∗(ab) liquid EVA = 3.12). Observing the spectral curves, a slight decrease in the reflectance is apparent in the range between 600 and 740 nm for all of the specimens of the “M” series after the degradation process, compared to the original color. In the “A” series, the Δ
E∗(ab) values with Ispra support rose above the value of 3 (Δ
E∗(ab) foamed EVA = 3.93; Δ
E∗(ab) liquid EVA = 4.19), while with the Origam as the lining canvas, Δ
E∗(ab) was lower than 3 in the case of EVA resin applied in foamed form and higher than the limit of perceptibility in liquid one (Δ
E∗(ab) foamed EVA = 2.58; Δ
E∗(ab) liquid EVA = 5.54). The spectral reflectance curves show, for both forms of resin applications, a significant decrease in reflectance, especially in the 570–620 nm range, and more highlighted in the 620–740 nm range, as a consequence of a strong darkening of the specimen surface after the degradation process. However, the same surface darkening also occurs in specimens without the EVA resin application, as also demonstrated by Δ
E∗(ab) values (
Table 7 and
Table 8); therefore, the alteration can mainly be ascribed to the natural ageing of oil paint layers.
3.6. Penetration of the Adhesive
The relined canvas treated with Pounceau S dye was cut and the resulting cross-sections were examined using an optical microscope.
Figure 8 shows the diffusion and the extent of penetration of the adhesive through the painting stratigraphy. The specimens fabricated with the foamed EVA resin show the marker visible across the original and the lining support layers, while no penetration was observed inside the ground layer. On the contrary, the preparation layer resulted moderately pink and partially spotted due to the application of liquid EVA resin, which confirms its unsuitability for such an application in its original form (liquid water emulsion).
Such observations are also validated by the FTIR analysis on the preparation layer, as illustrated in
Figure 9, which reports the spectra collected on the preparatory ground layer (the exact spots are indicated in the optical images also present in the figure). The IR spectra are characterized by a predominance of absorption bands related to calcium sulphate dihydrate (strong S-O stretching in the region 1200–1050 cm
−1), two sharp peaks at 1620 and 1680 cm
−1 due to O-H bending and O-H stretching in region 3600 and 3200 cm
−1 [
33], related to proteins with amide I, amide II, and amide III, and less intense absorption bands near 1650, 1550, and 1450 cm
−1, respectively [
33], due to the animal glue as a binder for gypsum. No vibration bands related to the water-based EVA resin [
34,
35] were detected on the samples with the foamed product applied, confirming that no significant penetration occurs in the ground layers above the canvas. Conversely, linear maps on samples prepared with liquid EVA water-based adhesive show, along the thickness of the cross-section, the continuous presence of the carbonyl band at 1735 cm
−1, the ester C-O stretching at 1240 cm
−1 and 1022 cm
−1 as the small shoulder, and the methyl C-H rocking at 795 cm
−1 [
34,
35], which are attributable to the ethylene-vinyl acetate resin. A gradual decrease in the intensity of such absorption bands is observed until they disappear altogether with the beginning of the paint layers.
Such findings have to be ascribed to the larger drying velocity of the EVA resin when foamed (thanks to its porosity), which prevents the adhesive penetration into the painting: the faster the drying process is, the less the penetration in the substrate. Similarly, water can be readily removed from the lining layer, thus not affecting to any extent the painting. This is also supported by the rheological tests, indicating that the foamed product has a higher viscosity at small shear rates than the liquid one, and it drops only at high stress values, as during the drawing up, thus allowing one to obtain a good spread ability of the adhesive. Such a larger viscosity, coupled with the increase in volume that allows the use of smaller amounts of resin to be obtained, but a uniform coverage of the same surface, enables the prevention of the penetration of the adhesive and of the water along the stratigraphy of the mock-ups.
The obtained result is a key factor, as it influences not only the interactions between the adhesive and the materials upon which it is applied, thus preserving the integrity of the painting, but also the invasiveness and successive reversibility of the lining operations. In fact, the amount of product that could remain on the support of a painting after a de-lining procedure and the degree of penetration, as well as the possibility of removing it in a non-invasive way, are other key factors that support the use of this solution for such an application.
The micro-FTIR analysis revealed that no appreciable aging can be observed on the samples subjected to artificial degradation, as no footprint of EVA resin degradation products was observed in appreciable quantities according to the technique used. The IR spectra, in fact, show neither absorption bands in the 1700–1720 cm
−1 region, relating to the formation of ketones, nor at 1785 cm
−1 for lactone formation, resulting from the typical hydrolysis and acetic acid formation by the de-acetylation of the vinyl fractions present in the EVA polymer. No decrease in intensity of the carbonyl band at 1735 cm
−1 was registered [
36,
37]. The absence of such components supports the stability of the EVA water-based adhesive and its suitability to the targeted application.