1The diamond magnate Julius Werner had a passion for medieval and Renaissance art and amassed the finest collection in the twentieth century. He displayed his outstanding collection of paintings, bronzes, majolica, and an important corpus of Limoges enamels in his London residence, Bath House, in specially constructed showcases. In 2000 the Werner foundation signed a 100-year loan with English Heritage to display this material to the public at Rangers House. In a £2,000,000 project, the ground floor was transformed into a historic interior, while the upper floor displayed the objects in showcases reproducing Werner’s early twentieth-century ornate cases. Two original showcases were reused and refitted to modern standards and fifteen new cases installed.
2The relative humidity (RH) in the cases was controlled with conditioned silica gel (Artsorb) hidden within the display plinths. The RH requirement for glass was the dominating parameter, and the range was set at 35–40% following the work of Ryan et al. (1996). The copper supports for the enamels also required a low RH to retard corrosion, but 40% is more than adequate in the relatively unpolluted atmosphere of Rangers House. Salts on the surface of some of the enamels were analysed by scanning electron microscopy with energy dispersive analysis (SEM-EDX), which indicated the presence of sodium sulphate. The concentration of sulphur dioxide in the room and the showcases was measured with passive diffusion tubes. The results showed that sulphur dioxide was in very low proportion to cause a problem and to be responsible for the sulphates formation in the showcases. The salts on the surface were inherited from an older environment.
3A thermal camera was used to measure the surface temperature of two enamels on display in the window showcase. A vertical thermal gradient on the objects induced by the tungsten-halide top lighting in the display case was found. This translated to a 2.5ºC temperature range across the surface of the enamels displayed higher in the cases. It was unclear whether such temperature fluctuations posed a risk to the enamels and research was initiated. To gain this type of dynamic information, non-destructive and non-invasive instant measurement techniques (i.e. that do not destroy or require samples to be taken from the object) are required to monitor in real time the behaviour of enamels under different environmental regimes. Until now suitable methods for doing this have not been fully investigated. A promising possible non-destructive (not affecting the enamel) method for real time monitoring of micro-crack formation in enamels (and also other materials used in cultural heritage objects) is acoustic emission testing (AE).
4AE testing is a very sensitive testing method, able to detect high frequency acoustic signals beyond human hearing capabilities. This technique can provide important information by correlating the recorded signals with progressive damage and micro cracking inside and on the surface of enamels. When a load or stress is applied to a material, small cracks may appear within the material. To become stress free the surface of the cracks move and release elastic energy in the form of elastic waves. They propagate to the surface, the motion of which can then be detected by a sensor attached to the surface. If the frequency components of the wavefront are within the bandwidth of the sensor then a voltage signal is created (McIntire, 1987). The data acquired can then be analysed through graphs via a layout file. This can then give an instant record of the damage, which has occurred within the test object. This research project aimed to test the feasibility of acoustic emission technique for monitoring instant damage in Limoges enamels.
5The painted enamels, produced during the Renaissance period in Limoges, are unique and extraordinary because of their figurative composition and creative achievements, and also because of the technological knowledge about the melting and expansion properties of the glass flux that the workshops had (Speel and Bronk, 2001).
- 1 The counter-enamel was an innovation of Limoges. This had the function of equalizing stress and pr (...)
6Painted Limoges enamels are made from several layers of translucent or opaque glass powder of different compositions, which were applied and fused in up to ten firing stages onto a thin copper plate. The fabrication of the painted enamels involved the shaping of the copper base, applying a counter-enamel1 and a grounding and several layers of coloured enamel to create a picture. The straight-sided copper plaques were shaped with a slight curvature in the 16th century, whereas the earliest plaques appear to have been used as flat sections. The slight curvature helped them to withstand warping or sagging during numerous stages of refiring (Speel, 1998).
7The enamels can show signs of deterioration, induced through the chemical composition of the enamels, humidity, temperature and atmospheric pollution. The implications can be irreversible loss of material, changes in colour and transparency of the glass (Bonne et al., 1998).
8There are two different ways in which Limoges enamels can be degraded:
through chemical processes forming visible corrosion products at the surface from the glass or the copper sheet,
through mechanical stress due to changes in temperature forming cracks in the enamel.
9These degradations are all complex and may have a great impact on the object’s preservation.
10The exposed coloured glass layer can be subject to moisture- and pollution-induced deterioration, the extent of which is dependent on the chemical composition of the glass. Caused by hydration, micro fractures in the glass surface form a gel layer. Additionally there can be a precipitation of different corrosion products, as a white powder, on the surface (Bonne et al., 1998).
- 2 Network modifiers (fluxes) are added to the silica (SiO2), breaking certain Si-O bonds and thus ch (...)
11Glass network modifiers2, through hydrolysis, migrate to the surface as potassium hydroxide or sodium hydroxide, reacting with carbon dioxide and sulphur dioxide from the atmosphere. These compounds may further attack the silica network and alkali salt deposits can form (Goffer, 2007). These compounds can further react with organic pollutants. Work has been undertaken to replicate the deterioration of glass through exposure to formaldehyde (Schmidt, 1992), acetic and formic acid (Robinet et al., 2004; Robinet et al., 2005). It also has been reported that copper can form corrosion at higher levels of RH when acetic acid (Thickett and Odlyha, 2000; Paterakis, 2003) and formic acid are present (Tétreault et al., 2003). Finally any increase in temperature will intensify the chemical degradation of the enamel reacting with humidity or pollutants (Goffer, 2007).
12When heated, materials expand and conduct the heat throughout their mass. Those changes in temperature can induce physical changes in the material, which will affect the rates of chemical reactions. The thermal conductivity of the solid determines the rate at which the heat is transferred. Metals, such as copper, have a high thermal conductivity, whereas non-metals such as glass, have a low thermal conductivity. When heated up or cooled, temperature will be distributed differentially causing different temperature gradients throughout the volume of the glass. Expansion and contraction will create internal mechanical stresses that can only be accommodated through crack formation.
13Materials change in length or in volume when heated or cooled by an amount proportional to the original length and the change in temperature. This fractional change in length for each unit change in temperature is the coefficient of expansion of a material (Goffer, 2007, p. 425). Since the expansion coefficient of copper is higher than glass it will expand more with elevated temperatures and also contract more than glass when cooled.
14Enamels made up of two different materials (glass and copper) when undergoing temperature changes, will undergo large stresses, where one component will be compressed by the change, whereas the other component is forced apart, resulting in physical degradation, such as crazing or detachment of the glass.
- 3 The initial specifications for temperature increase inside the cases was set at less than 5ºC per (...)
15A number of methods were employed to evaluate the cases where Limoges enamels are displayed at Rangers House. Diffusion tubes were placed in the showcases to measure the volatile organic pollutants and two showcases were monitored for the RH/T distributions within. This helped to illustrate how vulnerable the enamels are to deterioration. These monitoring tests were done parallel to the acoustic emission (AE) tests and shed light on the complex display environment of the enamels’ showcase. Identifying the increase in temperature induced by the display case tungsten halogen lamps3, knowing the air exchange rates of the showcases and knowing the materials used for the case design all help in the analysis of the diffusion tube and AE results.
16Particular glass formulations are known to deteriorate more rapidly in the presence of certain concentrations of the carbonyl pollutants formaldehyde and formic acid (Robin et al., 2005). The water-based coating Dacrylate 103 applied to the medium-density fiberboard (MDF) used in the interior of the showcases is known to retard emission of formaldehyde but not formic acid to any significant degree (Thicket, 1998). Acetic and formic acid are known to accelerate the corrosion of copper, but only at much higher concentrations than found in the showcases (concentrations found in the three showcases containing Limoges enamels at Rangers House were 118-388 µg m-3 ). High concentrations would cause corrosion of the enamel copper plates and could generate additional confusing acoustic emission signals.
17The formaldehyde concentrations were negligible. The concentrations of acetic and formic acid were well below the reported damage levels for copper but not for glass. It is possible that the levels of formic and acetic acid are accelerating the deterioration of the Limoges enamels glass surface. Concentrations were lower in the window showcase because of the high air exchange rate (4.5 ± 0.1 per day) and higher in shallower, larger cases because of the higher surface to volume ratio of the coated MDF. Silica gel used in the case to control RH is probably also adsorbing some of the acetic and formic acid, and the formaldehyde (Grosjean and Parmar, 1991); when the gel was reconditioned, acetic and formic acid were emitted (formaldehyde was not measured).
18The capacity for RH control of a showcase depends on its air exchange rate, the volume and type of adsorbent that may be used, and how often the adsorbent can be changed. One of the showcases with Limoges enamels had shown a dramatic increase in air exchange rate over the five years since installation, from 0.5 per day (well under the original specification of 1.0 per day) to over 4.5 per day. The space for silica gel was limited by the case design and the case could no longer hold its specified RH conditions. Silica gel in the form of Artsorb was used originally but replaced subsequently with Rhapid gel. These gels have increasing buffering capacity in the desired RH range. With Rhapid gel the RH conditions were retained (Fig. 1).
Fig. 1 Artsorb is exchanged with the Rhapid Gel in the window showcase
19The temperature and RH distribution inside the showcases was measured at eight points, using Meaco radio telemetry transmitters with Rotronic Hygrostat probes in each corner of the baseboard and Rotronic dataloggers (HygroLog) for the top corners (front and rear) for two months to identify differences in temperature and RH. RH was never lower than 36% and the highest daily fluctuations were between 5,5-7% RH, depending on location. Tungsten lighting used in the top of the showcases induced a daily temperature increase of 4–7ºC (Fig. 2), again depending on location.
Fig. 2 Temperature distribution in the window showcase
20The risk from such medium scale heating events is not easy to assess, even with microscopy. Observing the harmful changes that can occur between the enamel and metal interface is difficult; therefore acoustic emission has been developed and validated for use on enamels.
21The AE technique uses piezoelectric sensors applied to the surface of the object to detect the high frequency noise emitted when brittle materials undergo micro-cracking. It has been used in conservation for monitoring damage on wood, copper alloys and from salt activity in limestone for measuring acoustic emission from stress damage that is not visible to the human eye (Grossi et al., 1997, Caneva et al., 2004, Jakiela et al., 2007, Jakiela et al., 2008).
- 4 The sensors can have a high sensitivity if they have a good contact to the surface of the material (...)
22Prior to applying the acoustic emission technique on original enamels at Ranger’s House, which show signs of deterioration, tests were performed on pre-damaged replica enamels (kindly supplied by Veerle Van der Linden and Eva Annys, Royal Academy of Fine Arts, University of Antwerp, Belgium) and on 1.2mm thickness copper plates to discern signals generated from the copper substrate and from the enamel layer. The coupling4 of the acoustic emission sensors (Physical Acoustics R15A running onto a Pocket AE-2) applied to these samples was optimized to increase the signal. Samples were partially immersed in a hot water bath and deformed with three point bend flexure tests and the amount of acoustic emission recorded to assess their response to stress induced deterioration. The intention was to simulate the stress they could be exposed to in the showcase. This helped analyse the acoustic emission signals received when applied on the originals.
- 5 When the lights were on in the showcases, Laser interferometry (Schmitt Acuity AR200) indicated an (...)
23To induce a stress difference on the enamel test samples a three-point bend flexure test was applied to simulate the tension or compression the Limoges enamels are under as they are slightly curved in shape. In reality it would be the copper plaque expanding mainly under elevated temperatures that causes the stress. 5. Stress was generated and a force applied towards breaking the enamel test sample. To detect real-time deformations during those tests, the acoustic emission sensor was applied to the surface of the test sample on the side of the enamel with and without a couplant (Sil-Glyde) (Fig. 3).
Fig. 3 Set up of the flexure test
24The acoustic emission signal’s amplitude and counts/hits were recorded at the same time with the flexure extention and load signals.
25Beforehand flexure tests were carried out on 1.2 thickness copper plates to be able to discern signals generated from the copper substrate and from the enamel layer.
26Further tests were carried out on the enamel samples with flexure test and looking at the enamel surface under a Nikon SMZ80 as soon as a signal is generated to see if any cracks had formed. Since the enamel samples had undergone accelerated ageing test beforehand and the surface is covered with cracks, they were first coloured with red ink (Dormy® stamp pad ink).
27Tests were carried out using a Instron 4466 machine. With this type of equipment material testing is carried out and it is a universal testing machine, where tensile, compression, flexure, cyclic and bend tests can be performed. With the 3-point flexure test the stress area is under the loading point along the midline.
28A 1kN load cell was used. Test speed ranged between 2-0.1mm/min and the desired extension was fixed in mm for the end of the test. The flexure extension, load and time (sec) was saved by a program from Instron Bluehill2, Version 2.5 which was also used to set up the test.
29The flexure experiments showed that the acoustic emission signals is originated from the glass or glass metal interface and not from the copper. A copper plaque requires at least 100 µm flexure extension to produce acoustic emission. Only very few signals were recorded while the copper plaques have undergone flexure tests (Fig. 4).
Fig. 4 Flexure extension, load and acoustic emission hits of copper sample test without couplant
30Those signals may be attributed to dislocations and line-defects or imperfections, which exist in the crystalline lattice of the copper plaques, as the samples show deformation without rupture. Far more hits were recorded when doing a flexure test on the enamel samples than the copper plaques. It has been shown that using a couplant (Fig. 5) between the sensor and the surface helps recording much more signals than without couplant (Fig. 6).
Fig. 5 Flexure extension, load and acoustic emission hits of enamel sample test with couplant
Fig. 6 Flexure extension, load and acoustic emission hits of enamel sample test without couplant
31One hit can be related to new cracks forming on the enamel surface on the test samples demonstrating that the sensor can be relatively sensitive even without couplant (Fig. 7).
Fig. 7 New crack on surface after recording the first signal with a load of 1.8 N at an extension of 0.01 mm (original magnification 40x)
32To induce a real temperature grade difference on the enamel test samples, as measured on the original enamels with the thermal camera, heat induced testing was applied. This is done to reproduce the thermal stress those enamels could be undergoing. For that purpose one part of the enamel was placed in a waterbath while the other part was at room temperature. The waterbath was gradually heated to 42.5ºC and then cooled down to 3ºC. These temperatures were chosen to be not too far off the circumstances in the showcase but implying a bigger difference and stress grade on the sample. This was monitored with acoustic emission sensors (Fig. 8). The presence of water makes micro-crack growth easier (Rees Rawling, pers. comm.).
Fig. 8 Test set up in the waterbath
33The test were carried out using a Neslab Endocal RTE-100 refrigerated/heating and circulating waterbath. The equipment has a temperature range of -15 Celcius to + 100 Celcius. The water temperature was measured with a thermometer.
34Two enamel test samples were subjected to variations of temperature in the waterbath, having a thermal gradient over the surface as being only partly subjected in the water. One enamel test sample had been used previously for a flexure test where the enamel was bent to an extension of 5.5 mm to induce delamination of the glass surface (Channel 2). Far more signals, 127 hits in total, were recorded from Channel 2, whereas Channel 1 recorded 48 hits (Fig. 9).
Fig. 9 Temperature variations and AE signals during the waterbath test
35Already damaged enamel test samples seem to be more sensitive to variations in temperature with a temperature grade difference on the sample, producing more hits than undamaged test samples.
36It has been shown that using a couplant between the sensor and the surface helps to record many more signals, but the sensor can be relatively sensitive even without a couplant. Since the monitoring is carried out on the original Limoges enamels it was not possible to apply a liquid couplant and some loss of the acoustic signal has to be accepted. Tests had been carried out previously to compare different inert couplants on a ceramic plate with a pencil break test to receive a signal. Comparisons were made between signals using no couplant, with a Melinex sheet, with a piece of Nitrile rubber glove and with grease (Sil-Glyde ®) as couplants. As grease could not be applied it was decided to use a sheet of Melinex between the sensor and the original surface as it gave the next highest signal energy and had a low acoustic impedance (Fig. 10).
Fig. 10 Pencil test 1. no couplant, 2. Melinex, 3. Nitrile plastic, 4. grease (Sil-Glide®)
37These experiments demonstrated that the AE method was sufficiently sensitive to be applied to the Limoges enamels. The sensors were applied to the front side of the Pseudo-Monvaerni enamel ‘the Betrayal of Christ’ and on the Pénicaud enamel ‘Thetis emptying a pitcher of water’. To hold the sensors against the surface a stand with attached extension clamps was used and placed inside the showcase, so no pressure would be applied on the Limoges enamel surface. Holding the sensor onto the glass surface with a Melinex sheet inbetween gave enough coupling to collect acoustic emission signals.
38Four series of AE measurements were made on the enamel plaques thought to be most at risk from the highest temperature increases observed. One serial was carried out with one sensor applied on the surface of the Pseudo-Monvaerni while the other sensor (Channel 2) was applied to the backboard (covered with cotton velvet) to record the background noise (Fig. 11). The threshold level was lowered to 30 dB.
Fig. 11 Sensors applied on Pseudo-Monvaerni enamel and backbord
39Signals were recorded from original enamels in the showcase and could be distinguished from the background noise (Fig. 12).
Fig. 12 AE signals, time and temperature rise of the Pseudo-Monvaerni monitoring. The first signal recorded was when the light was switched on in the showcase
40The amplitude never exceeded 33 dB. With rising temperatures there seemed to be an accumulation of signals. No signals were received after the showcase lighting was switched off. The signals recorded are most likely from the glass layer, as could be shown through the flexure tests and the laser interferometry measurements.
41Acoustic emission events were recorded from original enamels only when the lights were turned on and heated up. This indicates that the enamels are under a certain temperature induced stress. Fluctuating and rising temperatures within the showcase does appear to be causing damage to the enamels in the showcase. As a consequence, the tungsten lamps were replaced with light-emitting diodes (LED) that had less than 25% of the near infra-red emission of the original tungsten halogen lamps. The LED lighting produced temperature increases of less than 1.4ºC and no acoustic emission events were recorded.
42Acoustic emission analysis has a great potential for assessing whether deterioration is occurring within the material and demonstrated that fluctuating and rising temperatures within showcases can cause damage to enamels on display.
43The reported actions have improved the preventive conservation of an important collection of Limoges enamels, thus ensuring their continued benefit to the public and protecting the generous loan from the Werner Collection for the future. The acoustic emission work has provided a basis for thermal specifications for fused glass on metal substrates, and the study supports the ongoing evaluation of microenvironments for sealed display cases and the specifications for their environmental parameters.