1Panel painting adhesives and the search for an ideal adhesive is a much debated issue. Almost every school of conservation has a different approach, using different adhesives depending on the nature of the problem, encountered cases, available adhesives and criteria of selection according to the properties and performance of a given adhesive. The choice is simplified by taking into account which adhesives are easily available: animal glues, PVAC dispersions and epoxy resins. The damage to wood panels caused by the water used to re-open cattle glue joints poses a problem for the structural conservation of panel paintings. A more suitable adhesive needs to be used in cases, such as those paintings with weak or damaged panels, where there is a risk of water damage.
2A potential alternative to animal glues in the treatment of such panel paintings, is the use of PVAC dispersions. However, there is a question mark over the suitability of their composition and stability. Recent research on PVAC dispersions has highlighted certain advantageous properties of this adhesive, and these theoretical conclusions appear to be confirmed in the behaviour of the PVAC dispersions in the many naturally aged joints being restored today.
3Our applied research on the behaviour of different adhesives - from the point of view of their stability, solubility and reversibility when they are used for gluing joints - will allow us to compare directly a cattle glue and various PVAC dispersions with different properties, and hopefully provide a solution to the issue of water damage that occurs in the conservation of panel paintings where water-soluble glues have been used.
4For the four adhesives considered in the study the test results presented here are divided into two parts: a study of the yellowing and the solubility of three PVAC dispersions before and after an artificial ageing; and a study of the reversibility and its limits, with a consideration of the eventual effects of the solvents on wood.
- 1 Young, C., Ackroyd, P., Hibberd, R., et al., “The Mechanical Behaviour of Adhesives and Gap Filler (...)
5The adhesives traditionally used for the conservation of panel paintings are animal glues, such as cattle glue, bone glue, fish glue and casein glue, which are still predominantly used, for instance, in Italy and in Germany1. The improvement and commercialisation of synthetic adhesives in the first half of the 20th century has led to the use of a greater variety of adhesives in conservation.
- 2 Cannon, A., “Major Development in Adhesives Manufacture”, in Adhesives and Consolidants for Conser (...)
- 3 Young, C., Ackroyd, P., Hibberd, R., et al., loc.cit.
- 4 Bria, C.F., “The History of the Use of Synthetic Consolidants and Lining Adhesives” in Western Ass (...)
6Urea-formaldehyde adhesives were commercialised as adhesives at the beginning of the 1930’s2 and used for the conservation of panel paintings – as adhesives – at the end of the 1940's, and as consolidants at the beginning of the 1950's3. PVAC resins were commercialised in the USA in 1929, followed by PVAC dispersions in the 1940's. As a resin or a dispersion, this polymer was used early on for the purposes of conservation. Dispersions were first used as an adhesive for lining canvas paintings, and then, from the 1950’s, for the conservation of panel paintings. Epoxy resins were commercialised and used in conservation during the same period as PVAC dispersions. The scientific literature on the materials used and studied in conservation already mention PVAC dispersions and epoxy resins at the beginning of the 1950's4. These types of adhesives are, for the most part, still in use now as adhesives for gluing panels, gap-fillers, or as reinforcements at the back of panels. The use of all these adhesives differs according to region, time period, and ideas concerning what is considered to be the ideal adhesive.
- 5 The word emulsion is incorrect. An emulsion is composed of a liquid dispersed in another liquid. A (...)
7The large quantity of manufacturers producing this kind of adhesive has led to the formulation of several different PVAC dispersions, with different physical and chemical properties, and different stabilities. The scientific literature and treatment reports tend not to mention the exact kind of PVAC dispersions used. They are often simply referred to as ‘PVA emulsions’ or ‘PVAC’5.
- 6 The data comes from articles, master theses and various studies on panel painting conservation.
Fig.1 The various PVAC dispersions used in the conservation of panel paintings6
Changes in the use of PVAC dispersions.
Credits: © Gwendoline Firmery
- 7 Glatigny, J.A., “Evolution des matériaux utilisés à l’IRPA, Bruxelles, à travers un exemple dans l (...)
- 8 Down, J.L., Macdonald, M.A., Tetreault, J., et al., “Adhesive testing at the Canadian Conservation (...)
8PVAC dispersions have been widely used since the 1950's. Since the end of the 1980's, in Belgium, synthetic adhesives for the gluing of panel paintings have been partially abandoned, in favour of natural adhesives, such as animal or fish glues. However, PVAC dispersions are still employed for gap-filling or other non-structural tasks7. The list of adhesives presented here is not exhaustive. Jane Down's study on PVAC and acrylic dispersions include several adhesives used in conservation, some of which have been specifically manufactured for this purpose8.
9The scientific literature concerning PVAC dispersions and their use in conservation/restoration approaches this adhesive from a number of different points of view. It is time for a fresh reading of this literature. Many criteria can be involved in judging the appropriateness of an adhesive according to its use in the gluing of panel paintings. The adhesive’s solubility is often considered to be of critical importance. One difference between PVAC dispersions and animal glues is their solubility in water. Beyond the kind of solubility displayed by PVAC dispersions, which is characterised by a greater or lesser degree of swelling, PVAC dispersions are not sensitive to water, and for this reason they are described as irreversible. The technical data sheets of PVAC dispersion manufactured for conservation provide a good example of this kind of classification: modified PVAC dispersions swollen by water are described as reversible, whilst more classical types of dispersion, which are only sensitive to organic solvents, are classified as irreversible.
- 9 Howells, R., Burnstock, A., Hedley, G., et al., “Polymer dispersions artificially aged”, in Measur (...)
- 10 Young, C., Ackroyd, P., Hibberd, R., et al. loc.cit., p.86.
10The use of PVAC dispersions that have subsequently undergone too many changes during their ageing, such as a loss of solubility9 – as was the case for the Evo-Stik Resin at the National Gallery of London – can negatively influence the reputation of the whole class of PVAC dispersions; the potential for loss of solubility over time is thus generalised to all PVACs, which contributes to them being considered as irreversible10. These adhesives possess, however, many physical properties which plead in favour of their use, such as their vitreous transition temperature, superior to the ambient temperature, and by their tendency to form a sustainable joint, without any risk of creep, which will maintain the surface alignment of the panels when they are glued together.
- 11 Baer, n.s., Indictor, N., Schwartzman, T.I., Rosenberg, I.L., “Chemical and physical properties of (...)
11PVAC dispersions are used in every field of conservation. We must keep in mind that each field has different criteria for the use of adhesives, which depend on the substrates to which the adhesive is applied, and their sensibility to solvents. The conservation of paper, for example, requires the complete solubility of an adhesive in a solvent,11 which is usually water. PVAC dispersions that do not swell in water but rather partially swell in organic solvents are thus unsuitable for this field of conservation. In this case, they would be correctly described as irreversible.
- 12 Depuydt, L., Le collage des panneaux peints. Etude comparative entre différentes colles animales p (...)
12In parallel with their solubility, the adhesive’s behaviour when gluing, and the research conducted on this subject, have produced changes in the use of adhesives. The research carried out in 1988 at the ENSAV La Cambre by Lyvia Depuydt12 studied the performances of many animal glues compared to a PVAC dispersion – Keimfix – commonly used at this time in Belgium and at the KIKIRPA. The results showed that Keimfix created a bond that was too strong, thus risking damage to the substrate, whereas cattle glue had a more suitable adhesion strength for gluing panels, particularly those that didn’t need to support their own weight. As a result of this research, the KIKIRPA began to use cattle glue for gluing panels, whilst restricting PVAC dispersions for use in non-structural treatments.
- 13 Rothe, A., “Critical History of Panel Painting Restoration in Italy”, in The Structural Conservati (...)
13However, following the flood in Florence, Italy, observation of paintings damaged by water highlighted the benefits of the use of PVAC dispersions. Joints glued with this adhesive between 1950 and 1960 withstood an immersion of 18 hours in water. PVAC dispersions had been considered suitable to glue poplar panels because of the elasticity of this adhesive, whereas oak panels had been glued with epoxy resins. Despite this, due to the observed sensitivity to humidity observed in the adhesive Vinavil NPC, many Italian conservation experts still prefer to use Epoxy resins on all wood types13.
14Once again we see that the differences observed in the use of PVAC dispersions and its appreciation as an adhesive depend on various factors, such as cases encountered, conservation conditions, and primarily the performance and stability of these adhesives, which vary between different compositions.
15The extent of the use of PVAC dispersions has been changing since their introduction in the conservation of paintings. Beyond the habits and trends which determine the use of different adhesives for the gluing of panel paintings, it is important to consider the stability of certain compositions, and the fact that, as we shall see, PVAC dispersions can provide a valuable alternative to aqueous adhesives in the case of damaged or weak panels.
16Every PVAC dispersion has a different composition. The polymer, the potential additives and their quantity influence the properties of dispersions and films:
17Following natural or artificial ageing, the properties of the films can change, due to a degradation of the polymer or the additives. However, it is difficult to establish with certitude the exact influence of the polymer or the individual additives on the film’s properties after ageing. These different degradations such as changes in the polymer's chain (cross-linking or chain-breaking) may or may not occur simultaneously, but the following effects can be observed:
18PVAC films are mainly swollen by organic solvents, such as alcohols, ketones, esters and aromatic hydrocarbons, unlike animal glues which are only sensitive to water. Some of the adhesives available specifically for conservation or restoration are sensitive to water after drying. PVAC films, when they are dry, are generally considered to be insoluble: additives inhibit the formation of intermolecular bonds between the solvent and the polymer molecules, therefore limiting the formation of a solution. However, they can be swollen by solvents.
19The choice of a solvent to swell a dried film from a PVAC dispersion cannot only be established upon the basis of the solvent efficiency on the film alone. We also need to consider the substrate. The required solvent must only produce a small amount of swelling of the wood, compared to water. To avoid damage to the other layers of a painting by the diffusion of solvents, we also need to consider the volatility, retention and penetration of each solvent.
20The swelling of PVAC films with various solvents is particularly advantageous when the substrate is wood, since the films remain very viscous, and do not spread into the substrate, thus allowing for the mechanical removal of the adhesive. Protein glues, on the other hand, return to a liquid form with water, which allows them to penetrate the wood.
21Wood is a complex substrate. The wood structure and its components influence the behaviour of solvents in contact with the wood, in the case of both water and organic solvents. In the field of panel paintings conservation, the wood swelling will also depend on its conservation state. Thus, if the substrate contains insect galleries, for example, the penetration of water will be more significant, leading to loss or irreversible damage to the substrate.
- 17 Inter alia G.I., Mantanis, R.A., Young, R.M., Rowell, W., Baker, et D., Grattan.
22Studies on wood swelling17 in solvents permit us to identify tendencies in the capacities of various solvents to swell wood, which indicates the importance of the size of the molecules, the polarity of the solvents used, and the density of the wood. This information will allow us to choose organic solvents that produce much less wood swelling than water. Thus, in the use of PVAC dispersions sensitive to organic solvents, we can choose the appropriate solvents to limit the wood swelling during the opening of a joint. Although we do not possess any data about oak swelling in organic solvents, we know that swelling increases with the relative density of the type of wood under consideration. Oak has a higher density than maple, and so the swelling will be more significant.
- 18 Mantanis, G.I., Swelling of lignocellulosic Materials in water and organic liquids, Madison, Unive (...)
Fig. 2 Maximum swelling relative to that in water at 23° C18
The tangential swelling of wood in acetone compared to the swelling in water remains acceptable.
Credit: © G.I. Mantanis
23Amongst the solvents available, acetone is often used for opening joints glued with a PVAC dispersion. Acetone produces much less swelling in wood than water. This set of data for the swelling of wood was obtained after the immersion of wood samples in various solvents. During conservation treatment, a compress filled with solvent may be locally applied at the back of a panel along the adhesive joint. Among other things, the high volatility of acetone helps limit the extent of wood swelling.
24In this experiment, we evaluated various PVAC dispersions and an animal glue comparatively, focusing on the stability and reversibility of these adhesives. Tests were both carried out on PVAC films, and on glued joints, before and after an accelerated ageing.
- 19 This adhesive is regularly used for gluing panel paintings. It is employed warm, diluted in water, (...)
25Various PVAC dispersions have been tested in order to compare them to a cattle glue19. A selection of adhesives, amongst those on the market, was made according to various criteria: their uses, their composition and the availability of these adhesives in the field of conservation. Three PVAC dispersions were selected:
-
- 20 Supplier: Wacker Chemie AG.
Vinnapas DPX 27120 : this PVAC dispersion is used in the manufacture of commercial adhesives. It only contains the additives needed for its production. It can't be used for conservation because its properties are not suitable.
-
Wood glue Pattex Classic D221 : a standard wood glue available in hardware stores. Its pH is relatively high compared to other PVAC dispersions.
-
Wood glue Pattex Waterproof D322 : also widely available, and more resistant to humidity than the Pattex Classic D2.
- 23 The samples were aged for 3 weeks under tube lights UVA-351 until an adequate whitenng of a Blue wo (...)
26Accelerated ageing tests are frequently used to establish the stability and the reversibility of materials used in conservation-conservation. In museum conditions, light is one of the main sources for accelerating the degradation of the materials in works of art. We have thus decided to use this type of accelerated ageing for our tests. Every sample is composed of 3 sets: 1 normal set, 1 set aged using light, and 1 set aged in the same environment, but excluding any contact with the light23.
- 24 Adhesives are applied in successive layers on a silicone polyester film up to a thickness of 1mm. (...)
- 25 Two set of oak planks, quarter-sawed, of 2cm and 5mm thick were glued with the 4 adhesives, and th (...)
27Three kind of analytical tests were carried out on PVAC films24 and glued joints,25 before and after the accelerated ageing:
Fig. 3 Film samples for the colorimetry tests
The samples for each adhesive were classified accordingly: from C1 to C3 for non-aged samples, from C4 to C6 for samples artificially aged in the light, and from C7 N to C9 N for samples artificially aged in the dark.
Credit: © Gwendoline Firmery
- 26 The weights were measured with a precision scale.
28Three film sets (1,5 cm side) of each adhesive were immersed in either acetone or toluene for 30 minutes. The weight and dimensions of each samples were measured before and after the immersion. The observed differences in these two measures provides us information regarding the amount of material removed from the films26.
Fig. 4 Film samples for the solubility tests
The sets of samples for each adhesive are composed of 3 films.
Credit: © Gwendoline Firmery
- 27 The application time varied during the experimentation: water has a low volatility, so it was not (...)
29A compress filled with solvent (water or acetone) was applied to each sample during 30 minutes27. This process was repeated until the joint opened. The time required to open these joints, the amount of swelling in the wood, and the extent of diffusion of the solvent allow us to evaluate the reversibility of the tested adhesives.
Fig. 5 Joint samples for opening tests
For each adhesive, 4 mm thickness joints are designated from 1A to 5A, and 2 cm thickness joints from 6A to 10A.
Credit: © Gwendoline Firmery
- 28 In this graph, L is the Lightness from black to white (from 0 to 100), and a and b, the two colour (...)
30The graph28 obtained from our experiments permits us to quantify what was visually observed on the films. Aged VDPX 271 films underwent the most significant yellowing. Classic D2 from Pattex films underwent the least yellowing. We made the following observations for the films tested:
-
VDPX 271 films heavily yellowed during the accelerated ageing. The two other sets of samples (non-aged and aged in the dark) had a similar yellowing. These non-aged films were more yellow than non-aged films from the other adhesives. This confirmed our first impressions during the preparation of the samples.
-
Non-aged and aged in the dark, D2 films had the same values for b. We noticed a slightly increased yellowing in the aged samples. This was barely noticeable to the naked eye.
-
D3 films had initial non-aged b values between the non-aged films of D2 and VDPX271. It confirmed what we observed during the preparation of the samples. Films that were aged, whether in the light or in the dark, had a similar yellowing.
Fig. 6 Films tested in the CIELab color space
The 3 samples of each set had closed values for b, except for artificially aged samples of VDPX 271.
Credit: © Gwendoline Firmery
31The adhesives D2 and VDPX271 only yellowed under the light, and this confirmed that the yellowing is produced by light and not by heat. The possible degradation of the PVAC is probably not responsible for the yellowing of the films and likely results from changes in the additives. It is worth noting that VDPX271 only contains the additives needed for its fabrication. However, it is also possible that commercial dispersions contain additives that prevent yellowing.
32The results obtained for D3 films may invalidate the previous remarks. Indeed, the degree of yellowing was similar for all the samples artificially aged. It may be the case that additives are responsible for the yellowing, or perhaps the polymer itself deteriorated. However, the solubility tests allowed us to clarify this matter.
- 29 A Poly(Vinyl acetate/Ethylene) dispersion.
33The increase of all the films dimensions occurs quickly after their immersion in acetone. The adhesives D2 and D3 showed a similar increase in size, and this increase was slightly inferior for the VDPX271 films. Despite the interesting results obtained by Jane Down in her research on the swelling of Dur-O-Set 15029 in toluene, our results here showed minimal swelling in all adhesives treated with toluene. The adhesive VDPX271 swells slightly more in toluene than the other adhesives. This can be explained by the presence of fewer additives.
34Other studies which have carried out the same kind of tests allow the solvents to evaporate during 24 hours. However, we felt it necessary to go beyond this short time-frame. In our presentation of the results obtained, large differences between the 3 dispersions could be seen:
-
The three sets of VDPX271 samples retained toluene to a similar degree, and lost weight after their immersion in acetone. The weight losses before, and after, ageing in the dark were similar. The weight loss exhibited by the light-aged films after immersion in acetone was superior to the weight loss observed in the other films.
-
Every D2 film showed a similar level of toluene retention (around 9%) and a small amount of weight loss after the immersion in acetone. Aged films showed a higher retention than other films. The samples that were aged in the dark didn't lose weight, but showed instead an increase in their weight, because of the retention of acetone in the films.
-
The D3 films produced irregular results: we saw a small decrease of weight with toluene, and a slight increase with acetone, which stays in the film. After the accelerated ageing, the weight loss was similar whether the film was immersed in acetone or in toluene. The results for the films aged in the dark and not aged were similar. It was therefore the accelerated ageing that modified the films’ behaviour in the solvents.
Fig. 7 Results of the solubility tests on the films
L'adhésif D3 a un comportement différent dans les deux solvants par rapport aux deux autres adhésifs testés.
Credit: © Gwendoline Firmery
35It was difficult to interpret the results concerning the solubility, because of the continued presence of solvents in the films for at least one week after the experiment. The swelling, however, provided a good indication of the adhesives’ sensitivity to solvents.
36The weight loss of the VDPX271 films in acetone was greater than that of the other adhesives. This may be explained by the low quantity of additives it contains. D2 and D3 adhesives exhibited different behaviour with solvents, but their swelling was similar. D3 films lost weight in toluene, whereas D2 films had a tendency to retain toluene.
37Given the results obtained for the swelling and solubility of films in toluene, and after our first attempts at opening joints, we decided not to carry out any tests with toluene, and to focus our study on the use of acetone.
38Solvents spread in wood from one growth ring to another. These swelling fronts form an obstacle to a deeper spreading in the wood. For both solvents, we noticed that solvent diffusion became restricted at a thickness between 0,5 cm and 1 cm. Indeed, the solvents tended to spread across the wood instead of penetrating into it. The solvents penetrated more successfully into the wood at the extremity of the joints, and less so in the middle. Joints glued with cattle glue confirmed this observation. We noticed a distribution of cattle glue in wood at the extremities of some joints, inducing a partial opening of joints at these extremities.
Fig. 8 Irregular solvent distribution in the joint
La diffusion est plus important vers les extrémités qu'au centre du joint.
Credit: © Gwendoline Firmery
39Compress application time was relatively long on thick joints. This produced an irregular pattern of wood swelling, with more swelling at the extremities of the joint that in the middle. This phenomenon was more noticeable when water compresses were used, and can be explained by the larger access to solvents provided by the joint’s extremities. Water applied to cattle glue joints produced swelling of up to 1,5mm, or 7%. Water takes more time to evaporate from wood than acetone, and one week after the end of these tests, the cattle glued joints still showed a thickness increase of 0,5 mm (2% swelling)
40Non-aged joints opened more quickly and easily than aged joints. The swelling of PVAC dispersions was instantaneous, once the acetone reached the adhesive film. Far fewer compresses were needed to open cattle glued joints, compared to joints glued with PVAC dispersions. Artificially aged cattle glue joints took the longest time to open. It took aged joints glued with VDPX271 and D2 330 minutes to open, where each acetone compress was applied for 30 minutes. The joints which opened most quickly were those glued with cattle glue, and not aged.
Fig. 9 Amount of compresses needed to open the joints (“+” means the experimentation lasted 2 days)
Le temps d'application des compresses d'eau a du être noté du fait de la faible volatilité de ce solvant.
Credit: © Gwendoline Firmery
- 30 A more significant whitening means a higher movement of extractives.
41Wood whitening was the most noticeable alteration on the surface of the samples. The whitening was more significant with water30 than with acetone, particularly in the case of aged samples. The whitening increases as the number of compresses applied to the wood increases. It remained difficult to establish the extent of the whitening beneath the surface of the wood. While opening cattle glued joints, the areas that were filled with water were noticeably softer that the other areas. This phenomenon did not appear to the same extent on joints opened with acetone.
Fig. 10 Whitening of the thin joints after its opening, before and after accelerated ageing (D3, D2 and cattle glue)
The whitening was more significant for artificially aged samples.
Credit: © Gwendoline Firmery
42In order to open joints, a perpendicular force was applied to the samples between each compress. Because samples were only 2 cm wide, it was not possible to proceed in the same way as when opening the joint on a panel painting, where the applied force is low. The joints usually opened once the solvent had penetrated half-way through the joint. As a consequence, there was some loss of small fibres in the middle of the joints, especially in the case of the aged samples.
43Joints glued with PVAC dispersions opened more quickly, and with more elasticity, before accelerated ageing, in comparison with the aged samples. The diffusion of acetone in the wood was deeper and more rapid than the diffusion of water.
44Only a few of the thick joints tested could be opened: two aged samples, one of which was glued with VDPX271, and the other with cattle glue (non-aged). Due to the fact that these samples opened abnormally quickly, we might suppose that these adhesive films were irregular or insufficiently thick.
45The fact that this sample had been quarter-sawn didn’t seem to have influenced the rapidity with which the joint opened. The two other joints which we succeeded in opening took two and a half days, with a total of six compresses each (3ml each). These were both cattle glue joints.
46The surface whitening of samples did not differ substantially from that of the thin joints. The greater amount of solvents used for thick joints did not seem to produce more whitening, but did seem to repel extractives further.
Fig. 11 Whitening of the thin joints before and after the accelerated ageing (cattle glue, D2, D3 and VDPX 271)
The whitened area was larger because of the greater amount of compresses applied on the joints.
Credit: © Gwendoline Firmery
47The use of adhesives in Panel painting conservation a complex issue, where we must take into account the adhesives’ nature and properties, as well as the structure and behaviour of the wood substrate. PVAC dispersions have complex compositions, with various additives, which determine its properties and characteristics, before and after ageing. The small amount of yellowing caused, and the constant swelling of the adhesives, indicate that the polymer is stable. We have seen that an adhesive film in a glued joint will only manifest minimal yellowing with ageing. The films’ swelling facilitates the opening of joints glued with this adhesive.
48The wood swelling was found proportional to the amount of time the compresses were applied to the wood, and the quantity of solvent used. The swelling and shrinking of the samples was limited with acetone partly because of its high volatility. Water is less volatile than acetone, and is thus retained for a longer time in wood, modifying its mechanical properties. This is deeply problematic for a panel with weak areas, since the application of water to the painting will weaken it further, and possibly cause breakage. Our failure to open the thick joints puts a question mark over the reversibility of these adhesives when used for panel paintings which are thicker than 1 cm. To open such a joint, in the case of an animal glue, the time spent applying compresses would be very long, which would provoke a significant diffusion of the solvent (in this case water) into the surrounding wood. Whereas, if the adhesive were a PVAC dispersion, the solvent (in this case acetone), would not be able to penetrate sufficiently into the joint to open it. Whilst our results concern a very specific group of adhesives, woods and methods, they nevertheless allowed us to observe and understand the nature and limits of an adhesive’s reversibility from the perspective of the swelling of the wood, the joints’ thickness, and the adhesives’ own nature. The aim of this research was not to generalise the use of PVAC dispersions for gluing panel paintings, nor to champion the use of this adhesive as a substitute for animal glues, but to reconsider PVAC dispersions as a valuable alternative to animal glues for panel paintings which have a damaged substrate, or which are sensitive to water.