Conventional corrosion science is interested in corrosion rates. Things are different in heritage science: the corrosion of heritage objects is quite slow (with few exceptions) and occurs over milennia or centuries in the atmosphere or in soil or water. Corrosion in the museum environment usually takes decades or at least years. Therefore, heritage science rarely studies rates (kinetics), but the products of corrosion (thermodynamics). As part of the ‘biography‘ of an object, they can give hints of what happened to them. Furthermore, even after 250 years of research on the corrosion of historic materials, there are still new products and mechanisms to be discovered. This rpapesearchr reviews glass-induced metal corrosion products on museum exhibits (GIMME) as studied at the Institute of Conservation Sciences of the Stuttgart State Academy of Art and Design. Typical for GIMME is metal corrosion in the contact zone with glass and often nowhere else on the metal. From 2006 to 2012, only 13 relevant objects could be discovered ^[1][2][1,2]. The number increased considerably during the intensive PhD research of Andrea Fischer [3]. Cases of individual compounds were published in a series of papers ‘When Glass and Metal Corrode Together I-VII’ (see references). Now some 500 samples from 350 objects in more than 40 collections were analysed providing a general overview of the GIMME phenomenon. Recently, other researchers also published GIMME corrosion product identifications ^[4][5][6][4,5,6].

Atmospheric metal corrosion needs oxygen and humidity, both provided by the air, but also electrolytes. The self-dissociation and, therefore, the conductivity of pure water is very low. In equilibrium with carbon dioxide from the air (currently 420 ppm), some carbonic acid yielding hydrogencarbonate and hydronium ions (pH ca. 5.6) is formed; the conductivity is increased but still quite low. Acid pollutant gases like SO₂ and NO_x forming sulfate and nitrate ions can add to ionic strength. It took until 2010 [1] for another important source for ions in the corrosion of metal exhibits to be realised: the contact of degrading historic glass with its alkaline surface films.

Quartz (SiO₂) as a glass network former consists of a three-dimensional network of strong Si-O bonds, and therefore has a melting point beyond the capability of historic coal or wood-fired furnaces. Network modifiers such as the fluxes soda (Na₂CO₃) or potash (K₂CO₃) or stabilisers like lime (CaCO₃) have to be added for glass production. They release CO₂ on heating and the O²⁻ anion from the remaining oxides attacks the Lewis acid Si(+IV) and scissors oxygen bridges in the network. In the case of soda this can be written as:

At normal humidity, glass is covered by ca. three monolayers of water [7]. The surface of glass tends to be alkaline as an ion exchange with surface water occurs:

The surface films on glass provide Na⁺ and/or K⁺ and OH⁻ ions for further reaction. In metal corrosion compounds, oxide (O²⁻⁾ ions can form when the oxide is thermodynamically more stable than the hydroxide, e.g., [8]:

The alkaline films on the glass absorb carbon dioxide from the atmosphere which forms carbonate ions (CO₃²⁻). As magnesium and calcium carbonate are quite insoluble at pH > 7, concentrations of Ca²⁺ and Mg²⁺ (from the stabiliser) in the alkaline films are very low and, therefore, do not find their way into metal corrosion products. This situation is different for church windows exposed to acid rain (pH 4.6), where gypsum and other sulfates appear in weathering crusts and alkali compounds are washed away. Despite the dominance of sulfates like brochantite in outdoor corrosion of copper alloys, no sulfates have been detected in GIMME indoor corrosion products. The same holds true for nitrate. Nitrates, such as the basic copper nitrates rouaite and gerhardtite, are in general very rarely found in atmospheric corrosion of not artificially patinated metals.

Carbonyl pollutants play an important role in corrosion reactions of museum objects [9]. They comprise the C₁ and C₂ alkanoic acids and aldehydes: formic (IUPAC: methanoic) and acetic (IUPAC: ethanoic) acids as well as formaldehyde (IUPAC: methanal, H₂CO) and acetaldehyde (IUPAC: ethanal, CH₃CHO). Wood and wood products are the major source of these pollutants [10]. They are incorporated into corrosion products as formate (methanoate, HCOO^-) and acetate (ethanoate, CH₃COO⁻). The alkaline glass films are able to absorb aldehydes and to transform them directly to the corresponding acid (without intermediate adsorption and oxidation steps) via the alkali-catalysed Cannizzaro reaction:

Volatile methanol (CH₃OH) is co-produced in this disproportionation reaction and has recently been detected [11]. Formate is collected in the films because formaldehyde is a ubiquitous air pollutant originating from many sources [12] and museum objects are exposed to it over decades or centuries. Therefore, it is not surprising that formate often dominates as anion on glass surfaces [13] and enters metal corrosion products.

Glass-induced metal corrosion could be successfully reproduced in desiccator experiments when metal coupons (soaked in alkali carbonate solution and then dried) were exposed to vapours from formaldehyde solutions [14].

The corrosive effects of acetic acid emitted from oak have long been known in museums. Calcium carbonate (calcite or aragonite) based materials as found in Natural History collections (shells, eggs, pearls, etc.) are prone to acetate containing efflorescence (‘Byne’s disease’) and a number of compounds have been characterised [15]. As formic acid, acetic acid is absorbed and neutralised in the glass films, providing acetate ions for further reactions.

Antiquity and Mediaeval Europe knew seven metals which were aligned to the days of the week and the movable celestial bodies: Gold (Au), silver (Ag), iron (Fe), mercury (Hg), tin (Sn), copper (Cu), and lead (Pb). Mercury (present in fire-gilded layers), gold, and silver are too noble for GIMME corrosion. However, historic silver always contains at least some percent of copper from production or intentionally alloyed to increase hardness. Copper is preferentially corroded in such alloys; hence, only copper compounds have been discovered on silver. Tin and iron form very stable oxide/hydroxides during corrosion; no specific compounds containing other ions were discovered during GIMME research. Lead is known to favour basic carbonates in corrosion, but in the presence of alkali carbonates, compounds containing Na⁺ or K⁺ can form (see below). Copper and its alloys (brass, bronze) in contact with glass frequently show contact corrosion; most often, formates are formed. This also holds true for zinc occurring in brass. Pure zinc (recognised in Europe since the 18th cent.) and other zinc alloys have not been surveyed yet for GIMME. The same holds true for other modern metals like nickel or cadmium.

Many types of heritage objects consist of glass in contact with metal.

The closest contact is achieved when glass (e.g., enamel) is directly fused onto metal. Early unstable Limousin painted enamels are particularly at risk of glass-induced metal corrosion, but examples of corrosion were also identified on enamel champlevé and enamel cloissoné. Glass figures were often fused onto a wire support and show corrosion in cracks were metal is exposed.

Other objects achieve close contact by mechanical pressure: metal-mounted glass vessels, glass gems in bezels (often found in folk jewellery), and cover glasses on metal (glass-framed miniatures or daguerreotypes with metal passe-partouts, watches), lenses held by metal (spectacles, optical instruments), glass levels mounted in brass, miners‘ lamps, and electric bulbs.

The contact can also be loose as in glass elements (e.g., beads) on metal wires. Often the loss of glass elements was caused by the total corrosion of thin wires on traditional bridal crowns, reliquaries, and Christmas tree baubles. Only in these objects, the contact between metal and glass can be prohibited by the use of coatings (Paraloid B-72 works for both materials) or plastic separation layers.

The induction of metal corrosion by glass is mediated by liquid electrolytes caused by glass degradation. Such electrolytes are somewhat mobile. Therefore, corrosion has also been observed outside the direct contact zone when drops run down a vessel over a metal mounting or drip from cover glasses on metal below.

In Stuttgart, the identification of compounds was mainly performed by Raman microscopy, but energy dispersive X-ray spectroscopy in the scanning electron microscope (EDX-SEM) and X-ray powder diffraction (XRPD) were also employed (for experimental details see [16]). Attenuated total reflection Fourier transform infrared spectroscopy (ATR-FTIR) could also be used for the identification of compounds [5]. Marchetti et al. [6] demonstrated the usefulness of the novel optical photothermal infrared (O-PTIR) spectroscopy that allows to noninvasively obtain chemical information at a submicrometric scale with high spatial resolution without need for sample preparation. The spectral results are comparable to traditional FTIR.

A breakthrough in characterising unknown compounds was achieved by modern high precision XRPD. Sophisticated methods of data evaluation at the Max Planck Institute of Solid State Research even allowed the determination of complicated crystal stuctures with low symmetry from powders [17]; single crystals are no longer mandatory. Chemical formulae of compounds automatically follow from such structure determinations.

Depending on the metal and the availability of anions, a number of corrosion compounds can form by glass-induced metal corrosion. These are systematised here along with the anions present (carbonate or formate, Table 1).

In a first approximation, glass-induced metal corrosion can be understood as metal corrosion in the presence of alkali carbonate solutions.

3.2. Formates ⑤–⑧

Formates are rarely found as corrosion products on metals without glass contact ^[23][32]. However, the majority of the analysed glass-induced metal corrosion samples contained formate due to formaldehyde pollution and the Cannizzaro reaction in the alkaline surface films on glass (Equation (4)). The corrosion can be simulated in desiccator experiments by exposing metal coupons soaked in alkali carbonate solutions and then dried to formaldehyde vapours [14].

3.3. Uncharacterised Compounds

Because of a lack of suitable samples of pure and ideally crystalline material there are still a number of GIMME compounds waiting for their full characterisation [3]. They contain Na, K, Cu, and/or Zn (EDX-SEM) and are often carboxylates (Raman). One example is the corrosion product of the enamel of a procession cross in the British Museum (K, Cu, and formate). As it was also found in simulation experiments [14], it might be possible to obtain sample material for further studies in the future. The crystalline compound labelled Zinc A contains zinc and sodium and possibly a little copper and is most likely an acetate [21]. ‘Zinc B’ contains more potassium than zinc and often occurs together with the copper formates ⑤ and ⑥ [21].