Atmospheric Corrosion Monitoring on Silver in Museums

Michel Dubus, Marc Aucouturier & Brice Moignard

Centre de Recherche et de Restauration des Musées de France

Abstract
The Preventive Conservation Department of the C2RMF has developed a methodology to identify the nature and the causes of the corrosion of metallic collections in museums, in order to assess the risk of corrosion in cultural heritage environments and to improve the preventive conservation policy. This methodology, based on the kinetic study of corrosion on metallic objects, coupons and sensors exposed to not much corrosive museum atmospheres, has been tested in several museum environments. The nature of the pollutants responsible for corrosion can be inferred from the nature of the corrosion products and the origin of pollution can be identified. Quantitative information on the corrosion rate is given by atmospheric sensors based on resistivity measurement of thin silver films. These real time corrosion monitoring methods are useful to classify "low corrosivity" atmospheres in museums, to detect environmental changes and predict the future evolution of the corrosion rate.

Introduction
Atmospheric corrosion is the reaction of air with a solid at ambient temperature. Atmospheric corrosion is complex because chemical, electrochemical and physical processes occur in the liquid phase, in the solid phase, in the gaseous phase and in the interfaces between them. Opinions about silver corrosion rate dependence on relative humidity are contradictory [Leygraf and Graedel, p. 123; p. 325; Rogers, p. 9] : "Silver is the only metal that has higher corrosion rate values indoors than outdoors. It is likely that the principal reason is the independence of the silver corrosion rate on relative humidity. Another potential reason is the marked dependence of silver corrosion rate to the concentration of reduced sulphur corrodants... Reduced sulphur concentrations indoors are about the same as outdoors, although their outdoor deposition velocities are higher". This case study has been launched in November 2000 in five conservation workshops at the Pavillon de Flore of the Musée du Louvre (figure 1). Three years before the Pavillon de Flore was converted into workshops for the conservation of sculptures, modern metals, painting on canvas, and tapestries.

Figure 1: the Pavillon de Flore in the Louvre

The workshops can be described with several parameters like the orientation of the room, the level of the room in the building, the volume of the room, the type of air-conditioning device, the location of the fresh air intake, the climate (temperature and relative humidity - temperature within the range +10°C to +25°C should have little direct influence on corrosion rates indoor [Leygraf and Graedel, p. 195], lighting because many important reactions are photolytic [Leygraf and Graedel, p. 111], the nature of surfaces (floor, walls, ceilings may play important roles as absorbing surfaces but also because they allow heterogeneous reactions to take place), the professional activity, the time of exposure, the localisation of the probes (table 1).

Table 1: description of the workshops

	sculptures	archaeological metals	modern metals	paintings	tapestries
air conditioning	rh t	rh t d	t	rh t	t
hrmin	21	22	23	44	15
hrmed	49	39	47	52	41
hrmax	60	57	79	65	77
fresh air intake (store)	0	0	0	7	7
orientation	North	North	South	North	North
lighting	artificial	day light	sun light	day light	dark
activity (days)	619	0	844	?	619
store	-1	0	0	3	4
floor	stone	stone	rubber	ceramic	plastic
walls	stone	stone, plaster, concrete	stone, plaster, concrete	stone, plaster	plaster
vol (m3)	350	600	720	700	270
probes localisation	wall	window	window	window	HVAC
height (m)	1.5	2.5	1.6	1.2	1.5
exposure (days)	619	660	844	660	619

Symbols: rh = relative humidity, t = temperature, d = dehumidification, hrmin = minimum rh, hrmed = mean rh, hrmax = maximum hr

Experimental
Mass change
Real time monitoring provides a quantitative measurement of the overall corrosion potentiality of the atmosphere. Metal loss from corrosion is obtained from the increase in electrical resistance upon atmospheric corrosion of a 250 nm thick metallic strip exposed to the environment (figure 2). Measurements are performed in situ without electric supply, the detection limit is between 2 and 4 nm (a few milligrams per square meter) under favourable circumstances that is, when the ohmic resistance of the corrosion products is high and the corrosion attack over the metal surface is uniform. Comparison with other techniques has shown that resistance sensor may overestimate the corrosion effect in case of non-uniform corrosion attack (edge effects, pitting).

Figure 2: resistance sensor

Phase identification
Metallic coupons have been exposed to collect information on the cumulated corrosion products (passive sampling). The corrosion products are identified by X-ray diffraction. We use a Bruker-Siemens D5000 diffractometer with a cobalt radiation. A multilayer crystal ("Göbel mirror") generates a high intensity K alpha beam. The result is a monochromatic parallel beam with high brilliance and without Bresmstrahlung. We designed a sample holder dedicated to diffraction at grazing incidence (GID) and X-ray reflectometry (figure 3). The solid-state PIN diode detector, thermoelectricaly cooled, is stable over high count rates and has a typical resolution of 300 eV.

Figure 3: grazing incidence sample holder

Results
Climate
The average RH value lies between 40 % (workshops for archaeological metals and tapestries) and 52 % (paintings). The lowest level goes beyond 20%; the highest level almost reaches 80% (figure 4).

Figure 4: relative humidity recorded in the workshops

Above the critical value of 50% RH at room temperature, the metal is covered by physical adsorption with more than 3 molecular water monolayers (table 2). This aqueous layer works as an electrolyte either through the incorporation of gas or through the deposition of airborne particles. When the aqueous layer evaporates, a film of corrosion products may precipitate (hydroxides and oxy-hydroxides). If condensation-evaporation cycles repeat, like here, this film of corrosion products can hinder the transport of ions. Hence the anodic reaction rate is lowered, and therefore the atmospheric corrosion rate.

Table 2: approximate number of water monolayers on various metals versus relative humidity
after [Marcus, p. 534]

RH%	water monolayers
90	8 on Ag
80	5 to 10
60	2 to 5
40	2
20	1

Corrosion products
Acanthite (Ag₂S) and chlorargyrite (AgCl) have been detected (figure 5). The relative contents change from one workshop to another: in the tapestries workshop, only silver chloride has been identified. In the metals workshop, silver sulphide has been detected, plus talc and quartz due to professional activity. These particles possibly could attract water and create differential aeration cells responsible for dramatic localised corrosion. In the sculptures, archaeology and paintings workshops, both silver chloride and sulphide are identified. Acanthite is more difficult to form in solution than chlorargyrite because it contains two silver atoms, but silver is known to be less sensitive to chlorine than to sulphur [Leygraf and Graedel, p.326].

Figure 5: X-ray diffraction patterns of the corrosion products

Acanthite:
The formation of silver sulphide is mostly due to the reduced sulphur in the indoor atmosphere. At neutral pH, the HS ^- ion is the main reduced sulphur constituent from either H₂S or carbonyl sulphide (COS). This ion can either react directly with silver ions or sorb onto the surface [Leygraf and Graedel, p. 321]. HS ^- could be produced outdoors by the catalytic exhausts of the cars.

Chlorargyrite:
Two mechanisms may possibly generate silver chloride: chloride ions enters the aqueous layer at the surface of silver either through the incorporation of gaseous HCl or through the deposition of chlorine-containing airborne particles [Bouquet, Bodin, Fiaud, 1993] and a solid product may form by precipitation of the aqueous ionic salt AgCl (this process is the most likely to happen); the chloride ion may also be sorbed onto the surface and form silver chloride when the aqueous layer evaporates [Leygraf and Graedel, p. 326].Indoor, the origin of chloride ions may be surface cleaning agents.

Kinetics
According to Graedel and Leygraf [Graedel and Leygraf, p. 199] studies of the corrosion kinetics in museum environments show that some metals obey a parabolic corrosion law, some a linear corrosion law, when other metals have an intermediate behaviour. We observe here two different corrosion laws on the same metal but with different corrosion products (figure 6).

Figure 6: corrosion kinetics in two workshops

The formation of silver sulphide in the workshop for the conservation of modern metals has a linear behaviour, indicating that the corrosion rate is limited by the transport of corrodants from the atmosphere to the corroding surface, or by the reaction kinetics.
In the workshop for the conservation of tapestries the formation of AgCl follows a parabolic corrosion law indicating the formation of a fully protective corrosion layer in which the corrosion rate is limited by diffusion of the species through the film, like here in the workshop for tapestries with AgCl (figure 7). In this case we observe, according to Bénard [Bénard 1964], a generalised description of corrosion rate versus time, which includes an induction period, a period for establishing stationary conditions and a stationary period. During the induction period (phase 1), the metal is covered with a spontaneously formed oxide and by the aqueous layer, which afford a certain protection depending on the aggressiveness of the atmosphere. During the transition period (phase 2), the oxide layer locally transforms and coalesces into a fully developed layer of corrosion products. Finally, the stationary period (phase 3) is characterised by the full coverage of the surface by corrosion products.

Figure 7: parabolic corrosion law in the tapestry workshop

Instantaneous corrosion rate:
The silver chloride layer formed in the workshop for the conservation of tapestries is protective, as shown in figure 8. When compared with the cumulated metal loss, one sees that the instantaneous corrosion rate is slowing down with time. The corrosion rate derivate over 30 days makes possible the comparison between corrosion rates after different exposure times: it is much higher in the metals workshop (broken line) than in tapestries, especially for long exposure time.

Figure 8: instantaneous corrosion rate () and metal loss (X) in the tapestries workshop compared to the corrosion rate derivate over 30 days in the modern metals workshop (broken line)

Mean corrosion rate
30 days exposure
The results obtained in the different workshops can be compared, as long as the exposure time is the same (figure 9). Looking at the mean corrosion rate on a 30 days exposure we can compare it to the classification proposed by Purafil [Muller, 1999] for the semiconductor industry, in which the class S3 corresponds to a "moderate contamination slightly above the acceptable level for efficient fabrication process"; above class S4 contamination is severe and the fabrication process must stop. Due to its high reflectivity, silver starts to show visible coloration at levels of corrosion corresponding to weight increases as low as 20 mg/m² and is clearly degraded with a mass increase of 40 mg/m² like here in the sculptures workshop. A mass increase of 20 mg/m² corresponds to 30 nm/30 days.

Figure 9: mean corrosion rate over 30 days

Mean corrosion rate
1 year exposure
The mean corrosion rate calculated on the basis of a one year exposure (figure 10) gives complementary information: all the curves have the same behaviour but they start at different levels, showing the influence of initial exposure conditions on the corrosion rate [Leygraph, in Marcus, p. 548]. The tapestries and paintings workshops have a mean corrosion rate of 7 - 8 nm/month whereas the mean corrosion rate is twice in the other workshops. In the first case the fresh air comes from the 7th floor, in the second case, it comes from street level. This result shows the direct relationship between the quality of fresh air and the indoor corrosion rate on silver.

Figure 10: mean corrosion rate on a 1 year exposure

Summary
The corrosion kinetics could not only depend on metals but could also be related to the nature of the electrolyte, thereby to the nature and origin of pollutants and of the corrosion products.
Ag₂S seems to obey a linear corrosion law; AgCl seems to have a parabolic (protective) behaviour.
The tarnishing of silver seems to be more or less protective, depending on the nature of the corrosion products.
In this study, the corrosion rate is directly related to the quality of the fresh air intake.
After Leygraf and Graedel [Leygraf and Graedel p. 199], standards should be more stringent for the cultural heritage than for than for electronics, because art objects are irreplaceable. But industrial silver is used for its physical properties, corrosion resulting in failures; in museums, when corrosion products change the visual appearance of the art objects, they also form a protective coating, slowing down the corrosion process.
To give useful information on the long-term conservation of silver artefacts, a museum classification should take into account the appearance of the objects and should be based on mean corrosion rates on long term exposure.
A comparative study is presently running on copper in the same conditions and will give complementary results.
A similar investigation has been launched in November 2002 with the collaboration of the art objects and sculptures departments of the Louvre museum to reduce the tarnishing of silver artefacts.
We are also separately studying the relative influence of SO₂, NO₂, temperature and relative humidity on the corrosion of Cu and Ag in collaboration with the Centre de Recherche sur la Conservation des Documents Graphiques.
Finally, the influence of wetness duration on the corrosion rate on Cu and Ag is investigated by using quartz crystal monitoring (QCM).

Suppliers
BRUKER S.A.: 34 rue de l'industrie, 67166, Wissembourg Cédex, Tél. 03 88 73 68 00, Fax. 03 88 73 68 79, E-mail. bruker@bruker.fr

CIMEL Electronique, 172 rue de Charonne, 75011, Paris, Tel: (+33) 1 43 48 79 33 - Fax: (+33) 1 43 48 62 61

GOODFELLOW, 229, rue de Solferino, 59000 Lille

Instruments & Controls, 4 rue Panhard, 91830 Le Coudray Montceaux, Tel:(33) 01.64.93.92.30, Fax:(33) 01.64.93.91.57, E-mail : insco@insco.fr

Testo Sarl, 19, rue des maraîchers, Immeuble Testo, 57602 Forbach, Tel: 03 87 29 29 00, Fax:03 87 87 40 79, E-mail: info@testo.fr

Acknowledgements
J.-C. Dran, L. Pichon, J. Salomon, Centre de Recherche et de Restauration des Musées de France, Research Department

L. Vilmont, Centre de Recherche sur la Conservation des Documents Graphiques, Muséum national d'histoire naturelle, 36, rue Geoffroy-Saint-Hilaire, 75005 Paris, France, Téléphone : +33 (0)1 44 08 69 90, Fax : +33 (0)1 47 07 62 95

A. Marcellan, S. Chauveau, Université Paris 12 Val-de-Marne, Faculté des Sciences et Technologies; S. Chouadra, C. Solazzo, Université de Paris 11

References
J. Bénard, Corrosion des métaux, Dunod, Paris, 1964

S. Bouquet, C. Bodin and C. Fiaud: Influence relative des composés soufrés et chlorés sur le ternissement de l'argent en milieu atmosphérique, C. R. Acad. Sci., Paris, t. 316, série II, p. 459-464, 1993

C. Leygraf and T. Graedel: Atmospheric corrosion, Electrochemical Society Series, Wiley Interscience, New York, 2000, 354 p.

P. Marcus: Corrosion mechanisms in theory and practice, Marcel Dekker, New York, 2002, 742 p.

C. Muller : reactivity monitoring : an alternative to gas monitoring in semiconductor cleanrooms, Journal of the Institute of environmental sciences and technology, vol 42, no.5, sept-oct 1999, 25

G. D. Rogers : Particular aspects of silver tarnishing, CCI, exhibition conservation research, International Institute for Conservation - Canadian Group Annual Conference, Kingston, Canada, 1975 05 17, 1st IIC-CG bulletin, Ottawa, vol 1, no. 1, 1976, 5-8

Part 2: Slides from presentation

Author to whom correspondence may be addressed:

Michel Dubus
Centre de Recherche et de Restauration des Musées de France - C2RMF
UMR 171 du CNRS
6, rue des Pyramides
75041 PARIS Cedex 01
France

E-mail: michel.dubus@culture.fr