PREVENTIVE CONSERVATION RESEARCH AND PRACTICE AT THE BRITISH MUSEUM
3 INDOOR POLLUTANT GASES
When Werner (1972) published a short paper on the corrosive effects of materials used in showcases, he could not have imagined that 36 years later, scientists in the Museum would still be conducting research into that issue and that there would be an annual conference on indoor air quality. The development of a simple test to screen materials used in the storage and display of objects (Oddy 1973) was followed by research into the gases given off by materials (Blackshaw and Daniels 1978), the conditions under which gases are given off, adsorption of gases by objects, and the formation of uncharacterized mixed salts on objects. The role of the external pollutant gases hydrogen sulfide and carbonyl sulfide in the tarnishing of silver has also been an important area of research in the Museum.
3.1 ACCELERATED CORROSION TESTING OF MATERIALS
3.1.1 The Oddy Test
Simple corrosion test methodology in which the test material and a metal coupon were heated to 100�C for 3 days was adapted to carry out the first testing of showcase materials. Preliminary studies were conducted to determine the optimum test temperatures using both pure metal and metal alloys. From these tests it was decided that 60�C for 28 days was optimal for testing. At 100�C the test materials degraded beyond recognition, and a lower temperature of 50�C did not cause enough evolution of gases to corrode the test coupons in a reasonable time. Pure silver, copper, and lead coupons were seen to adequately represent the metals of antiquity based on the analytical data available in the Research Laboratory and in the literature at that time.
Initially only the fabrics used inside showcases were tested, but it soon became apparent that a much wider range of materials such as paints, woods, adhesives, sealants, and fittings needed to be tested. Two materials emerged as highly problematic for use in showcase construction, wood and wool. The problem of acetic acid being given off by wood and corroding lead had already been published by Scott during the early years of science in the Museum (Scott 1922), and had been known in antiquity (Rackham 1968). In the tests it emerged that wool fabrics always caused silver to tarnish. This was because the sulfide linkages in the protein chains degraded, giving off reduced sulfur gas. Wool was banned from use, but it was not possible to ban the use of wood as at that time all of the showcases in the Museum, including new showcases, were constructed of wood.
Since its publication in 1973, the Oddy test has undergone modifications (Oddy 1975, Blackshaw and Daniels 1979). In 1992 a major review of the methodology was carried out by two of the Museum scientists. They conducted an interlaboratory comparison of the testing and found that there was a great variation in test results obtained (Green and Thickett 1993). As a result of this study, a revised methodology was developed with a step-by-step guide to carrying out the tests (Green and Thickett 1995). A number of spots tests or quick tests were introduced to cope with the many occasions when there was not enough time to carry out the corrosion test (Daniels and Ward 1982; Zang et al. 1994). All of the test methods were brought together in a booklet with an introductory chapter on why testing is necessary (Lee and Thickett 1996; Thickett and Lee 2004).
3.1.2 The 3-in-1 Test
Carrying out large numbers of accelerated corrosion tests is time-consuming, and following the publication of a test methodology where one test contained all three metal coupons (Bamberger et al. 1999), a 3-in-1 test was developed for use in the British Museum (Robinet and Thickett 2003). The test methodology devised by Bamberger et al. was not used after finding that the method of deploying the coupons by bending them over the edge of a beaker resulted in contact with water condensation, inducing corrosion that would not normally be observed. The method set out by Robinet utilized a reliable supply of disposable silicone stoppers which fitted the quickfit tubes used for the accelerated corrosion test. The coupons were inserted into slits in the stoppers and the rest of the test set up was as before. For a six-month evaluation period the 3-in1 test and the normal test were run on every material which came in for testing. Currently the 3-in-1 test method is used for the routine testing of materials for use in storage or display of the collection. More complex methods for evaluating materials have been suggested and one has been published (Reedy et al. 1998). This method requires equipment, expertise, and time that are not available in most museum science laboratories.
The Oddy test and 3-in-1 test are pass/fail tests which can be carried out by a scientist or conservator who is trained in the test procedure and has a good understanding of laboratory practice. These tests provide a simple way of ensuring that the risk to objects from indoor pollutant gases given off by materials used in storage and display is minimized.
3.2 CARBOXYLIC ACIDS AND ALDEHYDES
Indoor pollutant gases are those which are given off by the materials used inside a building. The gases of concern in conservation are those which react on object surfaces to form corrosion. The lower carboxylic acids, up to C4, were identified in emissions from a range of hardwoods, softwoods, and wood products by gas chromatography (GC) headspace analysis and shown to cause corrosion of lead (Blackshaw and Daniels 1978). In practice the main gases of concern are acetic and formic acid, possibly formaldehyde and acetaldehyde, and the reduced sulfur gases, hydrogen sulfide and carbonyl sulfide.
3.2.1 Acetic and Formic Acids
For manufactured wood products, the most important indoor pollutant is formaldehyde. When composite wood products such as particle boards and Medium Density Fiber board (MDF) were developed, they were quickly taken up by the furniture industry because of their low cost and easy working properties. However, these products give off large quantities of formaldehyde derived from the adhesives used in their manufacture. Formaldehyde was found to affect the health of people and was eventually identified as a carcinogen. Because of this, a passive test for measurement of formaldehyde was developed and was used in several museums including the British Museum; at the time there was no easy quantitative test for acetic and formic acid.
Use of the formaldehyde test put undue emphasis on formaldehyde as a main source of corrosion of metals, although laboratory tests in the Museum showed that at concentrations of 0.5 and 5 ppm it did not corrode lead or copper test pieces at 50% RH and temperatures of 15, 25, and 35�C. Slight corrosion of lead test coupons occurred using the same experimental set-up at 100% RH. This suggested that at higher relative humidity, conversion of formaldehyde to formic acid occurs (Thickett et al. 1998). Other researchers suggested that formaldehyde was a more serious problem (Hatchfield and Carpenter 1987). Researchers at the University of East Anglia showed that formaldehyde could be oxidized to formic acid at ambient temperature. However the experimental levels of oxidant were high compared to what normally would be expected in the air (Raychaudhuri and Brimblecombe 2000). Another mechanism suggested is that of the high-temperature Cannizaro reaction (Schmidt 1992), but this seems unlikely to occur at ambient temperature.
The wood industry changed formulations and lowor zero-formaldehyde MDF emerged. These products still gave off copious acetic and formic acid and failed the accelerated corrosion test. For more than thirty years in which the accelerated corrosion test has been in use, most of the woods and wood products tested have corroded the lead test coupon. Even ancient wood can corrode lead. An eighthcentury BC lead figurine (1880-12-16-46) formed from tiny lead and ivory squares had a wood core which was found to be the source of the regular corrosion of the lead (Duncan 1986a).
In the British Museum the corrosion product identified most frequently on lead objects by X-ray diffraction (XRD) is hydrocerrusite, PbCO3. Pb(OH)2. The corrosion of lead to basic lead carbonate has been described as a two-stage process via an intermediate, lead acetate, which reacts with water and carbon dioxide in the air to form basic lead carbonate. In the last few years, more findings have been made of acetateand formate-containing corrosion products on lead and other metals, and of mixed salts on porous stone, ceramics, and glass. These occur where composite wood products or untested paints are in use. Recent improvements in analysis by ion chromatography (IC) in the Museum have identified the presence of carbonates in salt mixtures which would previously have been identified only as formate and/or acetate. Measurement of the acetic and formic acid levels using the diffusion tube method described by Gibson et al. (1997a) have shown acetic acid to always be present at a higher concentration than formic acid. The prevalence of formatecontaining corrosion and salts suggests that either formic acid is more reactive than acetic acid at some object surfaces, or oxidation of formaldehyde to formic acid is occurring.
3.2.2 Conditions for the Formation of Mixed Salts
From a number of incidents of rapid formation of mixed corrosion products and mixed salts on objects, it was suspected that although acetic and formic acid may have adsorbed onto the surface of some objects, the formation of corrosion and efflorescence was dependent on high temperature and high relative humidity, either promoting out-gassing from wood, or promoting the reaction of the adsorbed gases on object surfaces. The formation of acetateand formate-containing corrosion and salts has been investigated on a range of materials including Egyptian copper alloys, limestone, marble, glass, and enamels. On Egyptian bronzes a previously uncharacterized compound, sodium copper carbonate ethanoate (acetate) was identified and characterized (Bradley and Thickett 1999; Thickett and Odhlya 2000).
On an Egyptian limestone stela (EA1332) a salt efflorescence containing methanoate (formate), nitrate, and chloride in the ratio 3:2:1 was identified. This is likely to be the corrosion product characterized by Gibson et al. (1997b). During conservation the stela had been poulticed to remove salts and was returned to its oak storage box before fully drying. It is likely that the moisture promoted a reaction between acetic acid from the wood and the soluble salts in the stone forming the mixed salt. The soluble salt profiles through the thickness of this stela and one which had not been treated with water and did not have salt efflorescence present were compared. In addition to the expected chloride and nitrate, acetate and formate ions were present in amounts of 0.01–0.06% w/w throughout both stele (Bradley and Thickett 1999). This shows that gases given off by the oak storage boxes had been adsorbed not only on the surface of the porous limestone but throughout its structure. A white efflorescence had formed on the surface of marble relief (MLA OA 10562) following leakage from a water pipe. This was identified as a calcium acetate formate hydrate (Thickett 1995), a mixed salt which had previously been identified on shells (Tennent and Baird 1985). The relief was mounted in a glass-fronted wood box that was a source of acetic and formic acid.
Acetates and formates have been found on the surface of glass and enamels. Many gases readily adsorb onto the surface of glass, and it is highly likely that acetic and formic acids were taken up during long periods of storage in wood cupboards or display in wood showcases.
Like many objects around the world, those in the British Museum have traditionally been stored in wood cupboards, resulting in long-term exposure to emissions of acids and aldehydes, which have been adsorbed onto the surface or even throughout the structure of many of the objects. This appears to be a substantial problem, but there are not that many instances of corrosion or salts on objects containing acetate or formate. On investigation, incidents of formation of corrosion or salts containing acetates and formates have involved the presence of water, applied during conservation, leaking onto the object, or in high relative humidity.
From the analysis of acetic and formic acid levels in showcases and store cupboards, an empirical relationship between high humidity and temperature and the rate of out-gassing from materials has emerged (Bradley 2003b). In the non-air-conditioned galleries of the Museum there is a seasonal variation in levels of acetic and formic acid and aldehydes, with winter levels considerably lower than summer levels. Since wood is still in use in showcases, albeit wrapped in a barrier film to reduce out-gassing (Thickett 1998), acids and aldehydes are present. However an examination of at-risk objects on display showed that they were not being affected by the gases and lead coupons in showcases did not corrode (Bradley and Thickett 1999). Even very high levels of acetic acid in cupboards used for the storage of Egyptian copper alloy objects did not corrode lead coupons. The relative humidity was at or below 45%. In general, the presence of moisture is needed for mixed corrosion products or mixed salts to form on objects; when conditions are favorable, formation is rapid. Further work is needed to establish if these types of reactions can be eliminated by keeping objects at a low relative humidity.