Mercury Emissions from Historical Tin Amalgam Mirrors Manfred Torge,* Sonja Krug, Michael Buecker, Ines Feldmann, Holger Scharf, and Heike Witthuhn BAM Federal Institute for Materials Research and Testing Berlin, Germany manfred.torge@bam.de Christoph Sander and Kerstin Fraenkler-Balhorn Spiegelmanufaktur Paderborn Paderborn, Germany Abstract With the implementation of glass casting in France in the second half of the 17th century, larger mirrors could be produced. So-called tin-mercury mirrors were made by coating glass with tin amalgam. Today, many historical mirrors are partly damaged, the image quality is compromised, and the material integrity of the mirror is threatened. The amalgam layer has often been destroyed, and restoration has become necessary. Conservators must follow strict safety precautions while handling amalgam mirrors because of possible mercury emissions and corrosion of the amalgam, during which elemental mercury accumulates near the mirror frame in droplets and is emitted into the air. In a research project, the quantity of the mercury emissions from historical mirrors was examined and a technique to decrease these emissions was developed. Keywords: mirrors, mercury, tin amalgam, deterioration Introduction From the 15th century to the end of the 19th century, mercury was used in the manufacture of mirrors in Europe. For a long time, coating glass with tinfoil and mercury was the only way to manufacture a flat mirror. Mercury or tin amalgam mirrors can be found in almost all historical buildings. In addition to private households, aris tocrats were among the principal customers for these objects because they decorated their castles and residences with halls of mirrors. Many mercury mirrors still exist today. The condition of these mirrors is quite varied, but most of the damage is to the tin amalgam. As the tin amalgam corrodes and separates from the glass, the original silvering fades, resulting in blind areas and discoloration. These phenomena compromise image quality and threaten the material integrity of the mirrors. The released mercury can be emitted from the back of the mirror, and it can enter the environment. During the conservation of such mirrors, beads of mercury are often found on the frames, and these may fall off, contaminating floors and polluting indoor air with mercury vapor. These beads may therefore be a health hazard to people who are exposed to such mercury emissions over long periods of time. The objective of the present work was to develop an emission protection concept for historical mirrors, reducing mercury emissions and protecting the back of the mirror while achieving long-term protection and conservation 156
of the mirror itself. The knowledge gained should become part of commercial conservation practice. Mercury Mirrors Mercury mirrors are glass with a coating of tin amalgam. Tinfoil was flooded with mercury and a sheet of glass was placed on top, applying slight pressure. The mirror was undisturbed for 10 20 hours and allowed to dry for up to three weeks, and it could then be transported. The history, manufacturing process, and material composition of such mirrors is described extensively in various publications (Cremer 1922, pp. 1 14 and 73 101; Krueger 1990, pp. 248 249; Hadsund 1993; Bronk and others 1999). According to Krueger, the use of mercury in making mirrors dates back to 1312. The backing of glass with tin amalgam began in the 15th century in the German/Flanders area (Cremer 1922, pp. 1 14). In 1507, the Dal Gallo brothers purchased a 20-year privilege to manufacture mercury mir rors on the island of Murano, near Venice, Italy. At that time, the size of mirrors depended on the technical capabilities of glass manufacturing. Glassblowing limited the surface area of mirrors to the size of the cylinders that could be created by lung power. With the implementation of glass casting in 17thcentury France (Cremer 1922, pp. 1 14; Hadsund 1993; Bronk and others 1999), larger surfaces could be produced. This development introduced new possibilities for using mirrors in interior design and decoration, including the splendid halls of mirrors of the Baroque period. Most of the larger mirrors at the courts of European aristocrats were wall installations, and mirrors inserted into wall panels were generally fixed in a wooden frame. The fronts of the frames featured richly carved ornament. In addition to wood, stucco elements, primers, and gold plating were often present. The back of the mirror was closed off by the frame structure, or was further braced with a cover. Sometimes, it was protected from damage with wool fabric, but more often, the tin amalgam layer was left exposed. In all cases, however, any mercury that was released could be freely emitted into the environment. Poisoning from exposure to mercury during the manufacturing of mercury mirrors was not recognized until the 19th century. Factory workers suffered from mercurialism (Spoerer, Busl, and Krewinkel 1997, p. 76), but today s conservators must also take strict safety precautions (using gloves, face masks, and exhausters) when handling tin amalgam mirrors. Around 1835, Justus von Liebig developed a method of depositing silver onto glass. The wet-chemical silvering of glass using silver nitrate was the basis for the industrial production of the silver nitrate mirrors we still use today. But only when the manufacture of mercury mirrors was forbidden in 1886 because of their toxicity did silver mirror production become widespread. During the process of outfitting the Grünes Gewölbe (Green Vault) in the restored Dresden Castle in 2006, the original mercury mirror coatings were reconstructed and new mirrors were manufactured using the historical method (Glaser 2006, pp. 144 149). The revival of this technology took place under the strictest occupational safety and environmental protection constraints. After several years of development, historically accurate modern tin amalgam mirrors were produced to decorate the rooms of the Grünes Gewölbe. Damage to Mercury Mirrors Compounds of mercury and tin are called tin amalgam and form a two-phase system (Hadsund 1993). Crystals of a tin-mercury compound form the solid phase 1, which is surrounded by a mercury-rich liquid phase 2. The solid and liquid phases are in equilibrium. The crystals of the solid phase contain approximately 81% tin and 19% mercury. 1 In addition to pure mercury, the liquid phase contains only about 0.5% tin. Because of its low vapor pressure, mercury evaporates at room temperature and is released into the environment. The slow evaporation of mercury, initially from the liquid phase but later also from the solid phase, results in damage to the reflective areas. Tin oxides are formed as corrosion products. In severely damaged mirror coatings, the tin amalgam was converted back into tin, tin oxide, and metallic mercury. The reaction ends with the complete dissolution of the two-phase system into a structureless tin oxide layer, which contains sporadic beads of mercury. This results in material loss in the reflective glass area. 1. All percentages in this paper are mass percent. 157
Figure 1 ESEM image of a historical mirror, depicting damage from SnO 2 precipitation. Figure 2 ESEM image of a historically accurate modern mirror with intact surface. Damage to the mirrors is directly related to the emission of mercury. To what extent this may represent an environmental health hazard indoors has not yet been examined. One goal of this study was to measure the emissions from mercury mirrors and to develop concepts for emission reduction. Analysis Methods The development of a workable emission protection concept for historical mirrors required various approaches. First, the material properties of tin amalgam were examined more closely. Documenting the corrosion process helped us to develop an experimental setup for measuring airborne mercury emission and to select materials for reducing mercury emissions from mirrors. Cold vapor atomic fluorescence spectrometry (CV-AFS, mercur Analytik Jena AG) and atomic absorption spectrometry (CV-AAS) are well suited to the determination of mercury in air. To select suitable absorber substances, their ability to absorb mercury was also tested. For this purpose, different materials, along with mercury mirrors, were weathered in climate test chambers. Mercury absorption testing was conducted using an environmental scanning electron microscope (ESEM XL30, FEI) and an advanced mercury analyzer, AMA 254 (LECO Instrumente GmbH, Mönchengladbach, Germany). Examination of Mercury Mirrors Several original mercury mirrors made by different producers and historically accurate modern tin amalgam mirrors (Glaser 2006, pp. 147 149) were available as sample material. ESEM examination of the surfaces showed that the original and modern mirrors are characterized by similar morphology and composition. The damage on the surface of a historical mirror can be documented by ESEM (Figs. 1 and 2). Figure 2 shows the intact mirror surface with the solid phase of tin and mercury (gray area = phase 1), with an average particle size of 20 100 μm, and the mercury-rich liquid phase (white area = phase 2), which fills the gaps. The historical mirror in Figure 1 shows corrosion damage. The width of the liquid phase is reduced, and tin oxide crystals have formed. A distinct change in the morphology accompanies this phase conversion. The transformation of phase 1 into tin oxide starts on the surface and continuously progresses to the glass surface as the tin amalgam phase is destroyed. The examination shows that phase 2 initially breaks down, followed by the escaping of mercury from the solid phase 1 and then by the converting of tin amalgam particles into tin oxide. The tin amalgam layer is transformed into a tin oxide layer with numerous cracks and holes, and the surface becomes visibly damaged. X-ray diffraction of the corroded surface layers permits an accurate specification of the phases formed. In addition to an amorphous portion from the residue of liquid mercury, 158 torge AND others
there is a phase very similar to β tin. Cassiterite (SnO 2 ) is clearly present, and romarchite (SnO) is much less so (DBU 2002). The analytical investigations of Herrera (Her rera and others 2008) on historical mercury mirrors from Spain show similar damage mechanisms and the same corrosion products. The glass surface under the damaged tin amalgam lay er does not show any sign of degradation (DBU 2002). There is no difference in composition between the contact surface and the external surface of the glass. Analyses of three samples from the Merseburger hall of mirrors (Merseburger Spiegelkabinett, MS) and the Green Vault (Grünes Gewölbe, GG) in Dresden indicate the composition, as given in Table 1. Table 1 Range of Chemical Composition of Historical Mirror Glass Samples (EDX Analysis in Mass %) Sample Na 2 O K 2 O CaO MgO Al 2 O 3 SiO 2 MnO Cl MS 3.6 4.7 18.3 22.8 7.9 11.2 0.9 1.4 1.9 2.1 62.3 62.6 0.2 0.3 GG 1.1 4.6 18.1 23.6 4.9 10.1 0.4 1.3 0.9 1.5 62.6 66.0 0 0.9 0.2 0.4 Because of their composition, these glasses could be dated to the 18th century. They exhibit a relatively high K 2 O content and, compared with modern glass, a still relatively low SiO 2 content. Measuring Emissions Mercury emissions from tin amalgam mirrors were measured in accordance with the German Industrial Stan - dard DIN 51865-3: 1998-2, Determination of Mercury Content of Gaseous Fuels and Absorption in Solutions, Part 3: Absorption in Solutions, published by Beuth-Verlag in Berlin in 1998. Historical and modern tin amalgam mirrors 10 x 10 cm 2 were used as sample material. The tests took place under accelerated conditions in a climate test chamber at 40 C and a relative humidity of 50%. Mercury emitted by the mirror was enriched in a 33% nitric acid solution. The mercury enrichment setup is shown in Figure 3. The air was enriched with mercury emissions from the mirror sample placed in the closed desiccator over 24 hours. Subsequently, a valve on one side of the desiccator was opened. On the other side, air was pumped from the desiccators (D) and passed through to a gas wash-bottle with frits and filled with 100 ml of 33% nitric acid (C). One gas wash-bottle, empty, acted as overflow protection (B), while a second bottle was filled with sodium hydroxide and activated charcoal filter (A). The air was fed from gas wash-bottle (A) through the flow controller and the gas meter to the pump and finally out through the exhaust. Figure 3 Sketch of mercury enrichment setup: (A) gas wash-bottle with NaOH and activated charcoal, (B) empty gas washbottle, (C) gas wash-bottle with HNO 3 (33%), and (D) desiccator with mirror sample. mercury emissions from historical tin amalgam mirrors 159
The maximum flow of the pump was about 6 L/min. The gas flow was adjusted to 0.45 L/min. (27 L/h), pumping for five hours. The tube connectors were made of polyethylene. The mercury concentration in the absorption solutions was determined using CV-AFS. Ionogenic mercury from the solution was reduced to elemental mercury (Hg 0 ) using tin(ii) chloride and then stripped from the solution with argon. The gas flow was continuously dried and if necessary, after enriching the Hg 0 at a gold collector and subsequent thermal desorption transported to the measuring cell (fluorescence cuvette) of the spectrometer. This is where the excitation of the element-specific fluorescence radiation took place by means of a high-energy, low-pressure mercury lamp, and where the intensity of the radiation emitted at a resonance wavelength of 253.7 nm was measured which was directly proportional to the mercury concentration in the sample. Certain gases, water vapor, or aerosols can lead to quenching (partial absorption and thus attenuation of the emitted fluorescence radiation) of the atomic fluorescence signal. It removes water vapor and aerosols from the carrier gas with a tubular diaphragm dryer and prevents interference from gaseous species by separating the Hg 0 on a gold collector. The analyses were conducted in accordance with Environmental Protection Agency (EPA) Method 1631, Rev. E of August 2002: Mercury in Water by Oxidation, Purge and Trap, and Cold Vapor Atomic Fluorescence Spectrometry. A certified mono-element mercury solution of CertiPur (Merck, Darmstadt, Germany) quality was used for producing the calibration samples. Before each series of measurements, a multi-point calibration was executed and the calibration range (up to a maximum of 100 ng/l) was adapted to the mercury concentration in the samples. In order to minimize the risk of interference from high nitric acid concentrations, all measurements were made using diluted absorption solutions in addition to a gold collector. For quality assurance, each series of measurements employed the analysis of reference samples that were produced using certified mono-element mercury sol utions different from those used for the calibration sam ples. By comparing the analyses, the margin of error for the measurement results is assumed to be lower than five percent. Emission Test Results Several test series were conducted to optimize the experimental setup, test time, and analytical evaluation. The reproducibility of the measurement results was improved by adding KBr/KBrO 3 solutions to stabilize the mercury dissolved in nitric acid, and the dilution of the analytical measuring solutions was optimized. At 20 C and 1 bar, saturated mercury vapor contains 13.19 * 10 3 ng/l (Paz 2003). According to criteria of the World Health Organization recommendation published by Link (1999, p. 173), the maximum exposure limit for mercury in indoor air should be less than 0.35 ng/l. The blank reading for mercury throughout the entire experimental setup was approximately 0.17 ng/l. Mercury emissions from modern and historical tin amalgam mirrors were determined under reproducible conditions, both by repeated measurements of the same mirror and by taking measurements from different mirrors of the same dimensions. The results of the analysis were normalized to the actual mercury-enriched airflow pumped through the nitric acid solution. Table 2 shows some results of the mercury emission from modern and historical mirrors. Under the selected test conditions, mercury emissions from modern tin amalgam mirrors were lower than those from historical mirrors by a factor of approximately 10. It is assumed that damage from aging processes and the associated corrosion of the tin amalgam layer lead to an increased release of mercury from the mirror. The emissions measured represent relatively high mercury concentrations, but we should remember that the measurements were made at 40 C in the desiccator in the climate test chamber and therefore represent accelerated conditions. Possible Approaches to Reducing Emissions Various procedures were proposed for decreasing mercury emissions from historical and modern tin amalgam mirrors. Enclosing the back of the mirror with a layer impermeable to mercury vapor is one possibility. Another is to bind mercury vapor with absorption materials. In both cases, the objective is to significantly reduce or altogether stop emissions of mercury into the environment. Glass is one material that has been used to reduce mercury emissions in this manner. The suitability and effectiveness of different absorption materials, such as iodized activated carbon, Mercurisorb, and gold sputter-coated paper, were 160 torge AND others
Table 2 Mercury Emissions from Modern and Historical Tin Amalgam Mirrors (CV-AFS) Test Sample Number Flow Rate [L/h] Hg Concentration in 100 ml of HNO 3 [ng] Hg Concentration per Liter of Air Pumped [ng/l] Ma 20 27.40 193 1.4 Mb 20 28.90 346.3 2.4 Mc 20 29.40 344.9 2.3 Ha H1 31.10 4290.0 27.6 Hd H1 30.60 3510.0 22.9 He H1 32.10 3070.0 19.1 Ma, Mb, and Mc = modern mirrors; Ha, Hd, and He = historical mirrors tested. There are several groups of materials that bind to mercury, and the selection of suitable materials was made according to their mercury absorption capacity, as well as the aging behavior of the absorber, its capacity to absorb over long periods of time, its reversibility, and its potential for practical implementation. To protect the mirror, no direct physical contact of the absorbing agent with the mirror coating was permitted. Glass Enclosures Thin (2 mm) glass panels were glued to the backs of modern and historical mirror samples, using different adhesives and techniques. From a conservation perspective, applying adhesive over the whole surface was not an option, so the bond between the glass panes was created by gluing only the edges. On older mirrors, the edges are worn and cracked, and small pieces of the material are missing. Since smoothly cut edges cannot always be expected, model mirrors with uneven glass edges were also tested. The edges of the mirror and the glass were bonded using an adhesive made by saturating Wagami ( Japanese paper) with Araldite 2020 or Paraloid B-72. The emission behavior of the enclosed mirrors was examined using the experimental setup shown in Figure 3. Absorbing Agents The absorbing agents iodized activated carbon and Mercurisorb are commercially available and have a fine-grained form. To create an even absorption area, the particles were scattered on an adhesive paper that was the same size as the mirror surface (10 x 10 cm 2 ). The gold absorber was created from gold sputter-coated (Scancoat 6, Edwards, London, U.K.) sheets of 5 x 5 cm 2 acid-free, aging-resistant paper. The thickness of the gold layer was ap proximately 180 200 nm. By using four sheets of paper, absorption areas 10 x 10 cm 2 were assembled. The paper absorber sections were mounted on a glass frame, about 1 cm away from the tin amalgam layer of the test mirror. The emission behavior of the mirror/absorber unit was examined using the experimental setup seen in Figure 3. Test Results The effectiveness of emission protection was tested on modern and historical mercury mirrors under conditions comparable to those of the tests on mirrors without rear enclosures and those without absorption materials. The results of the emission measurements are shown in the logarithmic representation in Figure 4. This clearly indicates that historical tin amalgam mirrors emit more mercury than modern mercury mirrors (compare M and H). Emission rates can be reduced by sealing the back of the mirrors with a glass pane (compare M and MG, and H and HG). However, mercury emissions from the sealed historical mirror are still clearly higher than those of the unsealed modern mirrors. Although the back of the historical mirror is made airtight by the glass, the adhesive on the edges is still permeable to mercury, so only limit - ed emission reduction was achieved with this method. A mercury emissions from historical tin amalgam mirrors 161
Figure 4 Logarithmic representation of analysis results for emission of mercury from modern (M) and historical (H) tin amalgam mirrors, with and without a glass enclosure (G), and effects of gold sputter-coated paper (Au), iodized activated carbon (JC), and Mercurisorb (Mer) as absorbing agents. distinct decrease in mercury emissions can be realized using gold sputter-coated paper, as well as iodized activated carbon and Mercurisorb as absorbing agents. The effectiveness of this method is evident on both modern and historical mirrors (compare MAu and HAu). The emissions using these absorbing agents are close to the blank reading (0.17ng/L) of the experimental setup and should therefore be considered extremely low. Gold sputter-coated paper is more easily applied than the fine-grained iodized activated carbon and Mercurisorb, particularly in avoiding direct physical contact between the tin amalgam layer and the absorber. The technique of sputtering gold onto paper also makes it easy to produce mirror-sized absorption areas. Transfer to Conservation Practice A mirror model was produced as a sample for practical application. It consists of a tin amalgam mirror in a historical frame over which acid-free Tyvek fabric is stretched to protect the back of the mirror. Gold sputter-coated paper is applied to the fabric as an absorber, with the gold-coated side facing away from the tin amalgam layer in order to avoid direct contact. The frame is closed, using the original back cover. This unit was then tested under the same conditions in the climate test chamber, and it shows distinctly reduced mercury emissions in comparison with the mirror without an absorption layer. Conclusion Modern and historical mercury mirrors are distinguished by different levels of mercury emissions. Our measurement results indicate a deterioration of the amalgam layer of the mirrors under environmental conditions. The best way to reduce the emission rate of mercury into the environment is to use gold sputtered paper as an absorption material that is affixed to the back side of the mirror. Additional data on the stability of the absorber in situ and the aging effects of the amalgam layer will be collected in long-term studies under conditions in line with 162 torge AND others
standard usage. Measurements in mercury mirror cabinets are also being prepared. Acknowledgment We thank the German Federation of Industrial Research Associations Otto von Guericke e.v. (AiF) Berlin for its financial support (codes KF0133713 PK 6 and KF 0396301 PK6) of this project. References Bronk and others 1999 Heike Bronk and others, Das Merseburger Spiegel kabinett: Wie sind Spiegelflächen zu restaurieren, Restauro, v. 105, no. 7, 1999, pp. 506 509. Cremer 1922 Ferdinand Cremer, Die Fabrikation der Silber- und Quecksilber-Spiegel oder das Belegen der Spiegel auf chem ischem und mechanischem Wege, Vienna and Leipzig: Hartlebens Verlag, 1922. Krueger 1990 Ingeborg Krueger, Die Glasspiegel im Mittelalter: Fakten, Funde und Fragen, Bonner Jahrbücher, v. 190, 1990, pp. 233 319. Link 1999 B. Link, Richtwerte für die Innenraumluft Quecksilbe, Bundesgesundheitsblatt-Gesundheitsforschung-Gesundheitsschutz, no. 2, 1999, pp. 168 174. Paz 2003 Boaz Paz, Die simultane Bestimmung von Quecksilber-, Arsen- und Selenspezies mit der ICP-AES nach deren Abtrennung aus der Luft mit der Chromatomembran- Methode, diss., Freie Universität-Berlin, 2003. Spoerer, Busl, and Krewinkel 1997 Mark Spoerer, Adalbert Busl, and Heinz W. Krewinkel, 500 Jahre Flachglas, 1487 1987: Von Waldhütte zum Konzern, Schorndorf: Hofmann, 1997. DBU 2002 Deutsche Bundesstiftung Umwelt, Abschlussbericht, Modellhafte Untersuchungen zu Umweltschäden an national wertvollen historischen Quecksilberspiegeln am Beispiel des Grü nen Gewölbes/Dresden und des Merseburger Spiegel kabi nettes/berlin, AZ 15499/45, 2002. Glaser 2006 Gerhard Glaser, Das Grüne Gewölbe im Schloss zu Dresden, Dresden: E. A. Seemann Verlag, 2006. Hadsund 1993 Per Hadsund, The Tin-Mercury Mirror: Its Manufacturing Technique and Deterioration Processes, Studies in Conservation, v. 38, no. 1, 1993, pp. 3 16. Herrera and others 2008 L. K. Herrera and others, Studies of Deterioration of the Tin-Mercury Alloy within Ancient Spanish Mirrors, Journal of Cultural Heritage, no. 9, 2008, pp. e41 e46. mercury emissions from historical tin amalgam mirrors 163
Glass and Ceramics Conservation 2010 Interim Meeting of the ICOM-CC Working Group October 3 6, 2010 Corning, New York, U.S.A. Hannelore Roemich, Editorial Coordinator ICOM Committee for Conservation in association with The Corning Museum of Glass
2010 International Council of Museums ICOM-CC Glass and Ceramics Working Group Committee: Gerhard Eggert Coordinator Hannelore Roemich Editorial Coordinator Review Panel for Papers Renske Dooijes, Leiden, The Netherlands Gerhard Eggert, Stuttgart, Germany Agnès Gall-Ortlik, Barcelona, Spain Isabelle Garachon, Amsterdam, The Netherlands Stephen P. Koob, Corning, New York, U.S.A. Laurianne Robinet, Gif-sur-Yvette, France Kate van Lookeren Campagne, Amsterdam, The Netherlands For The Corning Museum of Glass: Editor: Richard W. Price Design and Typography: Jacolyn S. Saunders Editorial Adviser: David Whitehouse Proofreader: Monica S. Rumsey Reference Librarian: Gail P. Bardhan ISBN: 978-0-87290-182-7 Library of Congress Control Number: 2010931220 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, whether electronic or mechanical, including photocopying, recording, or otherwise, without the prior permission in writing of the publisher. Cover Image: Cire perdue figure made by Frederick Carder in the 1930s or 1940s, with a repair in which the epoxy is badly yellowed. The Corning Museum of Glass (59.4.426).