Solvent Abuse

Some observations on the safe use of solvents in the cleaning of painted and decorated surfaces

Alan Phenix


Successful cleaning depends on the principle that cleaning method X will remove material A (such as a harmful layer of dirt or other material) without affecting material B (the original finish). In this case scientific analysis of the decorative materials enabled the conservator to identify the organic solvent which was most likely to dissolve the varnish layer and least likely to dissolve the paint layer beneath. Cleaning proceeds cautiously, using cotton tips dipped in a carefully selected solvent mixture, to ensure that the risk of damaging the object is negligible

Painted and decorated surfaces can acquire a wide variety of deposits or coatings in their lifetime, any of which might be considered to impair their aesthetic, historical or physical integrity and therefore may warrant removal. These coatings or deposits vary enormously in their chemical and physical nature, ranging from simple superficial dirt to layers of paint and varnish. Successful chemical cleaning relies on identifying a cleaning agent which changes the properties of the coating without affecting the underlying material, so that the coating can be removed with minimum risk to the integrity of the original. The principle of removing coatings in this way - by 'chemical selectivity' as it is known - is currently finding increasing application in the cleaning of works of art and fine decoration.

All cleaning is an exercise in risk/benefit analysis. Even with the mildest cleaning agent, such as distilled water, there will always be some risk of damaging the object. The level of risk will depend very much on the specifics of each situation: clearly, the closer the properties of the original and the non-original material are, the greater the risk will be. In broad terms, the aim of the conservator is to be able to reveal the original materials or surfaces in their best possible condition, whilst reducing the associated risks (not only to the integrity of the object but also to the health of the conservator), ideally to the point where they are negligible. Good practice in cleaning therefore depends on an evaluation of risk and on a structured, progressive approach to testing and selecting cleaning agents and cleaning strategies, as well as on careful documentation of the work undertaken.


A solution is simply defined as a homogeneous mixture of the atoms, molecules or ions of one material in those of another. It follows then that solubility is the capacity for a solid, liquid or a gas (the solute) to be dispersed at the molecular level through the medium of another substance (the solvent). This phenomenon can occur for very different types of substances: water, for example, can dissolve both a crystalline ionic solid such as common salt (sodium chloride) and a complex organic polymer, gelatin.

Three main classes of cleaning agent can be identified: neutral organic solvents, reactive organic solvents and cleaning formulations based on water. Only the first of these are true solvents of relevance within the context of this article, as the other two categories rely on other properties for their cleaning action.

Neutral organic solvents: A 'true' solvent does not, in the process of dispersing the fundamental particles of the solute, cause any alteration to the essential chemistry of the substance: rather it acts at the secondary level, breaking up bonds between atoms/ions or molecules rather than within them. Therefore if the solvent then evaporates, the substance will be left essentially unchanged. This is an important factor in favour of the use of volatile solvents (solvents which evaporate). Non-volatile cleaning agents must be actively cleared from original surfaces, usually by dilution.

Reactive organic solvents: Unlike a 'true' solvent, these cleaning agents combine solvent action with a chemical reaction, altering the nature of the material. They include acids such as acetic acid and, more commonly, bases such as triethanolamine. Reactive organic solvents are among the most powerful cleaning agents, often used for breaking down and solubilising old oil and oleo-resinous paints.

Cleaning formulations based on water: To deal with a variety of surface coatings or deposits the cleaning properties of water can be enhanced - and, in some instances, made remarkably selective - by additions of any of the following types of substance: acids or alkalis, pH buffers, soaps or detergents, soluble salts, chelating or sequestering agents, enzymes, and certain others. Many of these substances are not volatile and need to be removed thoroughly in order to prevent possible long-term effects on original materials.


Surface dirt: The layers of dirt acquired by an artefact can be a complex mixture of organic and inorganic materials, including particulates and amorphous substances which vary according to the history of the object and its environment. The deposits can adhere strongly to decorated surfaces and in some instances may even be absorbed into the body of the coating. Most effective methods of removal usually involve the use of water for its varied solubility and dispersal properties to gently wash the dirt from the surface. However, organic solvents are occasionally used, usually in situations where water cannot be tolerated. They are most effective where the dirt has a strong greasy or oily character.

Decorative coatings fall into two categories: those which are applied as a solution and dry simply by loss of solvent; and those which dry or harden by chemical change.

Decorative coatings which dry by solvent evaporation
Those which dry simply through the loss of their solvent include natural resin varnishes like mastic (dissolved in turpentine), beeswax polishes (in white spirits) and egg white (in water). Provided that, once dried, the solid material does not undergo significant chemical change, these types of coatings can also be expected to be removable with the same solvent. However, oxidation of the coating may mean that a different solvent will be required.

Decorative coatings which harden by chemical change
The second type of decorative coating might be described as 'convertible', meaning that it changes chemically during the process of drying or hardening. Usually this involves either 'polymerisation' (in which small carbon-based molecules or 'monomers' link up to form the larger, chain-like molecules which make up 'polymers') or 'cross-linking' (in which soluble polymer molecules form larger and more complex polymer molecules). Oil paints and oil/resin varnishes are typical examples of this kind of coating. These may include solvents as 'thinners' to assist their application, but the solvents quickly evaporate and the hardening process (sometimes called 'curing') proceeds by oxidative polymerisation. More modern examples include alkyd paints and polyurethane varnishes. Often these coatings are not soluble in organic solvents when dry simply because of the size of the molecules which are created. However they may be so affected by a solvent that they cease to adhere to the substrate and can be removed.

It is important to make a distinction between the removability of a coating and its solubility - indeed, many coatings and deposits may not be truly soluble. Several other processes may additionally combine in the use of a liquid agent to change the properties of one or more undesired layers to the extent that they can be separated from lower layers.

Most of the organic materials that make up paint and surface decoration layers are not pure, homogeneous substances as they usually contain molecules of different types and sizes. Paints made from drying oils or from egg tempera are typical examples: for any given solvent some of their components (including original constituents and any degradation products) will be soluble and others, usually cross-linked polymeric networks, will be insoluble.

Although such materials may not be soluble in the true sense of the word, solvents still may affect them. Solvents can bring about removal of the soluble components, a process that has come to be called 'leaching'. This process can lead to the physical deterioration of paint films, but its occurrence during the cleaning of old paint films is still largely undocumented.

Cross-linked polymeric materials like dried oil paints may be prevented from dissolving in organic solvents by the size and immobility of the molecules. They may, however, still absorb solvent molecules, swelling and softening in the process to form a 'gel'. Softening by gelation may be such that pigment binding is greatly reduced to the extent where it is easily removed by mechanical action. This phenomenon explains why even hard oil overpaint can often be removed by neutral solvents, and importantly it explains why original paints are vulnerable to solvent damage.

Solvents are not, however, highly specific in their action. They will have an effect, albeit possibly very slight, on virtually all organic materials they contact; but the magnitude of that effect will depend essentially on the chemical similarity of the solvent and solute and on the duration of their contact. Achieving selectivity in solvent cleaning generally relies on a combination of refined empirical testing and observation and on informed application of solvent theory.


The saying 'like dissolves like' helps to explain why water and ethanol mix, and why water and oil do not. If we are to define solvent behaviour more precisely, however, we need to consider the types of chemical bond which operate within organic molecules, inasmuch as these influence the forces between one molecule and the next.

We therefore make a distinction between primary bonding forces in organic compounds (strong covalent bonds which join the atoms making up the molecules) and the weaker and secondary forces that attract molecules to each other. These latter forces are largely responsible for the cohesion of the substance, its state and physical properties, and most importantly, its solubility.

For a solid substance to dissolve in a liquid, the solute/solute intermolecular forces must be broken down, to be replaced by solute/solvent interactions. This is most likely to happen when the balance of intermolecular forces in the solvent is similar to that in the solute. It is convenient to divide the intermolecular forces into three main types: Dispersion forces, Polar forces, and Hydrogen bonding. These classifications are used in many of the model systems for specifying solubility behaviour.

It should be stressed that many systems for describing solubility properties (including the Teas solubility parameter system described below) make critical simplifications in the treatment of hydrogen bonding, which is the strongest of the secondary intermolecular forces. This will limit their reliability.

A liquid will be a strong solvent for a solute which has a similar balance of intermolecular forces and likewise, a poor solvent for materials with a rather different balance of forces. The terms 'strong' and 'weak' must be used with care when referring to solvents, as they are relative, not absolute, properties: they only have meaning if the solute is specified. Ethanol is a strong solvent for shellac, but a weak solvent for beeswax.

Systems for visualising the solubility behaviour of materials have proved to be useful aids to the conservator in selecting solvents for cleaning and for establishing a hierarchy of solvent power for different materials. The solubility characteristics of solvents and solutes can be defined numerically by solubility parameters.


Although it is not without its practical and theoretical limitations, the system most widely used by conservators is the fractional solubility parameter system of J P Teas. Individual solvents are assigned three numbers, Fd, Fp and Fh according to the relative strength of their dispersion, polar and hydrogen forces respectively. These numbers can be plotted on a triangular diagram as in Figure 1 which is annotated here to show families of solvents having similar properties. Solvents lying close to each other on the chart will be expected to have similar solvent properties and indeed to mix. Aliphatic and aromatic hydrocarbon solvents occupy the lower right corner of the chart. They have very low contributions of polar and hydrogen-bonding forces and are commonly referred to as non-polar. They are also described as hydrophobic (water-hating) and lipophilic (oil-loving). By contrast, oxygen-containing solvents such as acetone and ethanol have a high contribution from the two types of polar forces and are therefore described as polar solvents. Water is the most polar of the solvents. The more polar solvents are also hydrophilic (water-loving).

As the strength of a solvent on a particular solute depends on the similarity of its properties, it is possible to gauge the affect of any solvent on the solute by testing a material to see if it dissolves in a broad range of solvents from different positions on the Teas chart. This can be expressed as a solubility region for the material. Solvents and mixtures of solvents whose parameters lie within that region should be effective solvents for that material. The solubility regions of fresh beeswax and the resin shellac are illustrated in Figure 2.

As the solubility regions of some materials change with age, primarily through oxidisation, so they generally become more polar. This change may also be accompanied by a decrease in overall solubility through the formation of insoluble polymeric matter. The change in solubility which occurs with age is illustrated in Figure 3 for the natural resin mastic. Mastic is initially soluble in the hydrocarbon solvent xylene, but within a relatively short time (less than 30 years) it will stop being soluble in xylene and will require more polar solvents for its removal.

The implication for such changes on cleaning with solvents is also indicated in It should be stressed that, at the present time, our knowledge about the swelling characteristics of old oil paint and of paints made with other binders is virtually nonexistent and this is an area in much need of further scientific study. Also, the real situation is very much more complicated than the above picture presents. Practical experience suggests that oil paint films can have a much broader swelling region than indicated in Figure 3, and we would also expect a shift towards solvents of greater polarity as paints age, for the same reason as occurs with mastic.

Figure 3 also clarifies a common misconception about solvents: the idea that the solvent power of a liquid can be reduced by dilution, in much the same way as the strength of an acid can be reduced by addition of water. This is often manifest in cleaning practices in which an apparently inactive solvent like white spirit is used as a 'stopper' or 'restrainer', to quench or dilute the action of an active solvent such as acetone or ethanol. It must be stated emphatically that the idea of dilution of solvent power is a potentially dangerous misconception. Because of the relative nature of solvent power, mixing solvents in this way does not necessarily diminish the strength of the solvent. On the contrary, it may actually increase its effect on certain materials. For example, oil paint is more strongly affected by a 50:50 mixture of ethanol and white spirit than by either of the pure solvents. This may be a useful property in improving the removability of old oil overpaint, but it can also expose original oil paint to unnecessary risk. Figure 3 shows the solubility parameter position of a 50:50 mixture of ethanol and white spirit. The enhanced effect on oil paint becomes immediately obvious.

The Teas Chart is perhaps most useful to conservators in this way, for predicting the power of solvent mixtures. A common practice in removing natural resin varnishes (the situation depicted in Figure 2) involves progressive increase in solvent polarity by addition of, say, ethanol to a non-polar solvent like white spirit. Essentially this method explores the boundary of the resin's solubility region on the non-polar side and can help the conservator limit the effect of the cleaning solvent on paint layers. Alternatively it may guide the conservator in selecting alternative solvents, for example where necessary on the grounds of safety. A common example is that of aromatic solvents (toluene, xylene etc.) which can often be substituted with mixtures of safer solvents (typically acetone and white spirit) in proportions determined by calculation.

Although the Teas Chart has its limitations, it fulfils a useful function as a map of solubility behaviour through which the conservator can navigate a course to safer cleaning. Critically, however, it will not provide insight into the speed at which things will happen, and for this the conservator must rely on information from elsewhere and on close observation. Also, it applies only to neutral solvents and it will not directly provide information on the response of substances, in solubility terms, to acidic or alkaline conditions.


Recommended Reading

  • A Moncrieff, and G Weaver, Science for Conservators Book 2: Cleaning, Conservation Unit/Routledge, 1987
  • G Torraca, Solubility and solvents for conservation problems, ICCROM, Rome, 1978
  • G Hedley, 'Solubility parameters and varnish removal: a survey', The Conservator No 4, 1980, pp12-18
  • V Horie, Materials for Conservation, Butterworth-Heinemann, 1987
  • H Ruhemann, The Cleaning of Paintings, Faber and Faber, London, 1968
  • RL Feller et al, On Picture Varnishes and their Solvents, National Gallery of Art, Washington, 1985
  • S Michalski, 'A physical model of varnish removal from oil paint', Preprints to International Institute for Conservation Congress, Brussels, 1990, pp85-92

Information sheets on Shell hydrocarbon and chemical solvents can be obtained from Shell Chemicals (UK) Ltd, Heronbridge House, Chester Business Park, Chester CH4 9QA



This article is reproduced from The Building Conservation Directory, 1997


ALAN PHENIX is a lecturer in the Conservation of Easel Paintings at the Courtauld Institute of Art, University of London. He trained and worked as a paintings conservator after a first degree in Chemistry and Colour Chemistry. His teaching and research are directed to the application of science and technology to the conservation, restoration and examination of works of art.

Further information





Fine art conservation
Site Map