1 INTRODUCTION: MEASURING THE SWELLING OF PAINT AND POLYMER FILMS DUE TO SOLVENT SORPTION

In the removal of varnishes and overpaint from pictures using organic solvents, one of the key elements of risk to the original paint comes from the possibility that it will be swollen due to sorption of the cleaning solvent. In the swollen, gelled condition, the pigment binding power of the paint is diminished and the paint is vulnerable to loss of pigment from the mechanical action of the swab delivering the solvent and removing the varnish. Understanding the relative swelling powers of solvents and solvent mixtures is an essential foundation of safe cleaning practice. For several decades the work of Stolow has been the primary reference source for data on the swelling of artist's paints (Stolow 1955, 1963; Feller et al. 1985).

There have, over the years, been many approaches to the measurement of swelling of paint films and polymers (Thiessen 1925; Faraday Society 1946). These approaches generally fall into two categories: determination of dimension change either in the plane of the film or in the perpendicular direction; and measurement of weight gain due to solvent sorption. Weight gain is perhaps the most widely used of all methods to determine swelling of paint (Rinse and Wiebols 1937; Browne 1956; Brunt 1964; Long et al. 1967). This method—or its inverse, measurement of weight loss from material saturated with solvent—has been applied to measure swelling of many other materials in addition to paint, mostly polymer networks of various types (Barr-Howell and Peppas 1985; Errede 1992; Byun et al. 1996), including poly (vinyl chloride) (Parker and Ranney 1995, 1996) and rubber (Gee 1942; Salomon and van Amerongen 1947; Scott and Magat 1949; Bristow and Watson 1958), as well as photographic gelatin (Johnsen 1996). Degree of swelling, for example, is accepted as one indicator of the solvent barrier properties of rubbers used in protective gloves (Zellers et al. 1996a). It is common in studies of polymer swelling for the results to be interpreted using various solubility parameter treatments, of which the Hildebrand solubility parameter, ∂ (Bristow and Watson 1958; Mangaraj 1963; Stolow 1963), the Hansen solubility parameters, ∂d, ∂p, ∂h (Zellers et al. 1996b), and the Flory-Huggins interaction parameter, χ (Huglin and Pass 1968; Peppas and Merrill 1976; Parker and Ranney 1995), are the most common.

The main problem associated with the gravimetric approach to measuring swelling is that samples must be removed from the swelling liquid and handled at intervals, and reliability is strongly affected by the presence in the subject material of a significant soluble phase, as in the case of oil paints. Consequently, there has been considerable interest in noncontact methods for measuring swelling in terms of change in sample dimensions. Stolow's method (Stolow 1954) was itself a development of that devised by Lewis and Soper (1950) for measurement of the swelling of gelatin films. The principle common to these two methods is that the swelling liquid impinges on the sample surface from a jet. As the sample swells or contracts, the jet is retracted or advanced to maintain a constant fluid pressure. Calibrating the movement of the jet tip provides a measure of the change in sample thickness over time without actual contact with the surface. The “swellmeter” of Green and Levenson (1972) was also developed for measuring perpendicular swelling of gelatin films, but this is a contact method: a solid feeler probe connected to a solenoid transducer measures the displacement of the probe electrically as the sample expands or contracts. Compression of the sample due to the force of the feeler probe is minimized by using a broad-faced probe and low contact time (0.25 seconds). This type of apparatus has been used within the field of conservation to measure the swelling of gelatin photographic film emulsions exposed to atmospheric pollutants (Nguyen et al. 1999). Recently, various other high-sensitivity noncontact methods have been proposed for measuring swelling of polymer films due to solvent sorption from liquid or vapor. These include ellipsometry (Chen et al. 1999; Parbhoo et al. 2000), quartz crystal microbalance (QCM) (Chen et al. 1999), and x-ray or neutron reflectivity (Tan et al. 1998). The ellipsometry method, which relies on reflection of polarized light from the boundary interfaces of the films, is capable of subnanometer resolution. It is not, however, suited to the measurement of opaque, pigmented films due to light scattering.

The usual alternative approach to noncontact measurement is some form of microscopical measurement, often with a traveling microscope. This method has been applied to the cross sectional swelling of various textile fibers by atmospheric humidity (Morehead 1952). A microscopical method was used by Browne to measure paint film swelling as a complement to weight gain measurements (Browne 1956). During the 1960s, further studies appeared that employed the microscopic measurement technique: particularly relevant here are studies by Brunt (1964) and by Eissler and Princen (1968, 1970), both of which follow the same remarkably simple method. Small triangular-shaped pieces of paint (edges ca. 0.5 mm) cut from unsupported paint films are placed on a microscope slide under a cover glass, and the length of the sides of the triangular pieces is measured on a traveling microscope or with a micrometer eyepiece. The sample is then immersed in liquid, and at regular intervals the length of the edges is measured until equilibrium is reached. The magnitude of linear swelling is expressed as a percentage increase in the perimeter of the samples compared to the initial value. It has been rightly observed that measuring swelling by changes in sample dimension can be time consuming and potentially unreliable (Down 1999).

Michalski's compilation of paint swelling data from various sources (Michalski 1990) demonstrated good correlation between swelling results on oil paints obtained by different experimental methods. Interestingly, the results obtained by the relatively simple methods of Browne (1956) and of Eissler and Princen (1968, 1970) compare favorably with those of the more sophisticated, higher-sensitivity method of Stolow (1954, 1955).