SWBSS 2017 Fourth International Conference on Salt Weathering of Buildings and Stone Sculptures

20 – 22 September 2017 – Potsdam

Verlag der Fachhochschule Potsdam

Proceedings of SWBSS 2017

Fourth International Conference on Salt Weathering of Buildings and Stone Sculptures

University of Applied Sciences Potsdam, Germany

20-22 September 2017

Edited by Steffen Laue

Verlag der

Verlag der Fachhochschule Potsdam Impressum Previous SWBSS CONFERENCES

Bibliografische Information der Deutschen Nationalbibliothek: SWBSS 2008, organized by L.M. Ottosen and colleagues, The National Museum Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Copenhagen, Denmark, 22-24 October 2008 Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über SWBSS 2011, organized by I. Ioannou & M. Theodoridou, University of Cyprus, Limassol, http://dnb.d-nb.de abrufbar. Cyprus, 19-22 October 2011

Dieses Buch ist auch als freie Onlineversion über die Homepage des Verlags sowie SWBSS 2014, organized by H. De Clercq and colleagues, Royal Institute for Cultural über den OPUS-Publikationsserver der Fachhochschule Potsdam verfügbar. Heritage, Brussels, Belgium, 14-16 October 2014 http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:kobv:525-15913

Steffen Laue (Hrsg.) Proceedings of SWBSS 2017 SWBSS 2017 Fourth International Conference on Salt Weathering of Scientific Committee Buildings and Stone Sculptures Potsdamer Beiträge zur Konservierung und Restaurierung Band 6 De Clercq, Hilde – Royal Institute for Cultural Heritage, Brussels, Belgium Diaz Gonçalves, Teresa – Laboratório National de Engenharia Civil, Lisboa, Portugal Verlag der Fachhochschule Potsdam Espinosa-Marzal, Rosa M. – University of Illinois at Urbana-Champaign, USA www.fh-potsdam.de/verlag Flatt, Robert – ETH Zürich, Switzerland Hamilton, Andrea – University of Strathclyde, Glasgow, UK © 2017 Fachhochschule Potsdam Ioannou, Ioannis – University of Cyprus, Nicosia, Cyprus Texte und Abbildungen in Verantwortung der Autorinnen und Autoren der Beiträge Laue, Steffen – University of Applied Science Potsdam, Germany Lubelli, Barbara – Delft University of Technology, The Netherlands ISBN 978-3-934329-93-5 (Druckausgabe) Ottosen, Lisbeth – Technical University of Denmark, Lyngby, Denmark URN urn:nbn:de:kobv:525-15913 (elektronische Ausgabe) Siedel, Heiner – Dresden University of Technology, Germany Steiger, Michael – University of Hamburg, Germany Layout: Semler Grafik oHG, Berlin Török, Ákos – Budapest University of Technology and Economics, Hungary Herstellung und Vertrieb: tredition GmbH, Hamburg Vergès-Belmin, Véronique – Laboratoires de Recherche des Monuments Historiques, Gesetzt in TheAntiquaSun Champs-sur-Marne, France

Organizing committee

Laue, Steffen – University of Applied Science Potsdam and Salze im Kulturgut e. V. Die elektronische Version dieses Werks ist lizensiert unter einer Creative Commons Steiger, Michael – University of Hamburg and Salze im Kulturgut e. V. Namensnennung 4.0 International Lizenz (CC BY 4.0): https://creativecommons.org/ De Clercq, Hilde – Royal Institute for Cultural Heritage, Brussels, Belgium licenses/by/4.0/

Sponsored by

Cover image: Photo: Hans-Jürgen Schwarz, Design: Benita Lohse The Deutsche Bundesstiftung Umwelt DBU (German Federal Environmental Foundation) Preface

Welcome at the University of Applied many reviewers of the Scientific Commit- Sciences in Potsdam! tee who so generously gave their time to ensure that the papers accepted are of a The first SWBSS event was held in high quality and represent a true advan- Copenhagen, Denmark in 2008, with Lis- ce in the state-of-the-art. beth Ottosen as driving force, the second Furthermore, I would like to thank Hil- was arranged by Ioannis Ioannou and de De Clercq, Royal Institute for Cultural Magdalini Theodoridou in Limassol, Cyp- Heritage, Brussels, and my colleagues of rus, in 2011, and the third SWBSS confe- Salze im Kulturgut e. V., especially Mike rence was organized by Hilde De Clercq Steiger and Hans-Jürgen Schwarz, for and colleagues at the Royal Institute for co-organizing this SWBSS conference. Cultural Heritage in Brussels, Belgium. Thanks also to the Prussian Palaces and This is the fourth conference in the Gardens Foundation Berlin-Brandenburg, series. It is a great pleasure to see that particularly to Kathrin Lange and Roland SWBSS 2017 has been met with interest Will, for organizing the excursion. And from so many different countries both in without the help of my colleagues and Europe and outside of Europe. the students of our University we could We received around 50 contributions, not have organized this international consisting of papers and abstracts and conference: thank you all! bringing together conservators, restorers, Last but not least, I would like to express engineers, scientists, young students and my deepest gratitude to our sponsor The experienced researchers. Deutsche Bundesstiftung Umwelt DBU The success confirms the significance (German Federal Environmental Founda- of the SWBSS agenda for the scientific tion) for supporting the conference. community. Being aware that salt crystallization is of major importance in the I sincerely hope that you will enjoy weathering of porous building materials, SWBSS 2017 and wish you a pleasant stay I do hope that this conference contribu- in Potsdam! te to an exchange of knowledge, create new solutions for the preservation of our cultural heritage and stimulate new research ideas and future collaborations Steffen Laue within this interesting field. Conference Chair On behalf of the Conference Organizing University of Applied Sciences Potsdam Committee and the participants I would Department for Conservation and like to take the opportunity to thank the Restoration Table of contents

Salt sources, transport and crystallization

• Deicing Salts: An Overview A. E. Charola, B. Rousset and C. Bläuer (Washington, DC, US)...... 16

• Traffic-induced salt deposition on facades M. Auras (Mainz, Germany)...... 24

• Wick action in cultural heritage L. Pel and R. Pishkari (Eindhoven, The Netherlands)...... 33

Measurement techniques and experimental studies

• A preliminary study on dynamic measurement of salt crystallization and deliquescence on a porous material surface using optical microscope M. Abuku, S. Hokoi and D. Ogura (Higashi-Osaka, Japan)...... 42

• Diagnostics and monitoring of moisture and salt in porous materials by evanescent field dielectrometry C. Riminesi and R. Olmi (Florence, Italy)...... 49

• Determination of the water uptake and drying behavior of masonry using a non-destructive method A. Stahlbuhk, M. Niermann and M. Steiger (Hamburg, Germany)...... 57

• Measurement of salt solution uptake in fired clay brick and identification of solution diffusivity E. Mizutani, D. Ogura, T. Ishizaki, M. Abuku and J. Sasaki (Kyoto, Japan)...... 65

• Local strain measurements during water imbibition in tuffeau polluted by gypsum M.A. Hassine, K. Beck, X. Brunetaud and M. Al-Mukhtar (Orleans, France)...... 74

• Assessment of the durability of lime renders with Phase Change Material (PCM) additives against salt crystallization L. Kyriakou, M. Theodoridou and I. Ioannou (Nicosia, Cyprus)...... 81

Salt crystallization tests

• Salt crystallization tests: Focus on their objective A. E. Charola, I. Rörig-Dalgaard, J. Chwast and J. Elsen (Washington, DC, US)...... 92

• Mitigating salt damage in lime-based mortars with mixed-in crystallization modifiers S. J. C. Granneman, B. Lubelli and R. P. J. van Hees (Delft, The Netherlands)...... 100 Studies for conservation issues • Desalination of Cotta type Elbe sandstone with adapted poultices: Optimization of poultice mixtures and application terms • Efficiency of laboratory produced water repellent treatments on limestone J. Maitschke and H. Siedel (Potsdam/Dresden, Germany)...... 208 C. Charalambous and I. Ioannou (Nicosia, Cyprus)...... 110 • Tracing back the origins of sodium sulfate formation on limestone as a consequence • Environmental control for mitigating salt deterioration by sodium sulfate on of a cleaning campaign: the case study on Charité and Espérance sculptures of Motomachi Stone Buddha in Oita prefecture, Japan Chartres cathedral K. Kiriyama, S. Wakiya, N. Takatori, D. Ogura, M. Abuku and Y. Kohdzuma S. Benkhalifa, V. Vergès-Belmin, O. Rolland and L. Leroux (France)...... 219 (Kyoto, Japan)...... 118 • Granite and schist masonry desalination by poultices at Jacobine Church in • Numerical analysis on salt damage suppression of the Buddha statue carved into Morlaix, France the cliff by controlling the room temperature and humidity in the shelter B. Brunet-Imbault, B. Reidiboym and C. Guinamard (Studiolo, Paris, France)...... 229 N. Takatori, D. Ogura, S. Wakiya, M. Abuku, K. Kiriyama and Y. Kohdzuma (Kyoto, Japan)...... 125 Case studies • Management of sodium sulfate damage to polychrome stone and buildings D. Thickett and B. Stanley (English Heritage, UK)...... 135 • Salt content of dust and its impact on the wall paintings of the church St. Georg at the UNESCO World Heritage site Monastic Island of Reichenau in Germany • Conservation of marble artifacts by phosphate treatments: influence of J. Frick, M. Reichert and H. Garrecht (Stuttgart, Germany)...... 242 gypsum contamination E. Sassoni, G. Graziani, E. Franzoni and G. W. Scherer (Bologna, Italy) ...... 143 • Investigation of salts sources at the Karadjordje‘s Gate on the Belgrade fortress M. Franković, N. Novaković, S. Erić, P. Vulić and V. Matović (Belgrade, Serbia)...... 252 • Electrode placement during electro-desalination of NaCl contaminated sandstone – simulating treatment of carved stones • Investigation and examination of a degraded Egyptian painted limestone relief L. M. Ottosen and L. Andersson (Lyngby, Denmark)...... 150 from Tell Hebua (Sinai) E. Mertah, M. Othman, M. Abdelrahman, M. Fatoh and S. Connor (Cairo, Egypt)...... 261 • How not to bother soluble salts while grouting C. Pasian, F. Piqué, C. Riminesi and A. Jornet (Lugano, Switzerland)...... 158 • Technique for transportation of stone sculptures damaged by salt crystallization Vinka Marinković (Zagreb, Croatia)...... 271 • Moisture transport during poultice application C. Franzen, M. Aulitzky and S. Pfefferkorn (Dresden, Germany)...... 168 • Investigation and conservation concept of salt damaged epitaphs in the church of Werben (Saxony-Anhalt) • The application of hydroxyapatite-based treatments to salt-bearing porous S. Laue, D. Poerschke and B. Hübner (Potsdam, Germany)...... 275 limestones: A study on sodium sulphate-contaminated Lecce Stone G. Graziani, E. Sassoni, G. W. Scherer and E. Franzoni (Bologna, Italy)...... 176 • Salt-induced flaking of wall paintings at the Mogao Grottoes, China L. Wong, S Bomin, W. Xiaowei, A. Rava and N. Agnew (The Getty Conservation Institute, L.A., US)...... 285 Salt reduction • Development of a network-based climate monitoring system for climate • Evaluation of desalination and restoration methods applied in Petra (Jordan) assessment and regulation W. Wedekind and H. Fischer (Göttingen/Berlin, Germany)...... 190 C. Leonhardt, S. Leonhardt and J. Heller (Kiel, Germany)...... 297

• Salt extraction by poulticing – results of a pre-investigation at the archeological site of Coudenberg S. Godts, H. De Clercq and M. Rich (Brussels, Belgium)...... 200 Salt sources, transport and crystallization SWBSS 2017 | 20-22 September 4th International Conference on Salt Weathering of Buildings and Stone Sculptures Deicing Salts: An Overview

Deicing Salts: An Overview ter. Antifreeze solutions of glycerol or crease the stability and permeability of various glycols are applied prior to the the soil, sodium ions will decrease them. event to prevent a strong bond between Furthermore, sodium, magnesium and A. Elena Charola1*, Bénédicte Rousset2 and Christine Bläuer2 1 Museum Conservation Institute, Smithsonian Institution, Washington, D.C., USA the pavement surface and frost by ap- calcium chlorides may contribute to the 2 CSC Sàrl, Conservation Science Consulting, Fribourg, Switzerland plying a freezing point depressant. These mobilization of trace metals from the * [email protected] are mainly used on aircraft, machinery soil to surface and groundwater. The so- and vehicles as they are non-corrosive, lid chloride deicers, i. e., NaCl, may cont- however, most of them are toxic.3 Other ribute to air pollution through particula- solutions have been developed based on tes released into the air.2,7 Abstract 1. Introduction special coatings.4

The world production of salt (NaCl) was The world production of salt (NaCl) was 2.2. Acetate based deicers and others over two hundred million tons in 2015. over two hundred million tons in 2015. 2. Deicing salt varieties The US is the second larger producer of The US, the second largest producer of Soil microorganisms will break down salt after China, produced over four mil- salt after China, generated over four mil- Deicing salts can be roughly divided acetate ions resulting in oxygen depleti- lion tons of which 43 % were consumed lion tons of which 43 % were consumed into inorganic salts, such as sodium on of the soil, which can impact vegeta- in highway deicing. While NaCl is the in highway deicing. While NaCl is the chloride (NaCl), organic salts, such as tion. A similar oxygen depletion is most most commonly used salt, other salts are most commonly used salt, other salts are magnesium calcium acetate (CaMg(CH3 likely to occur in slow flowing streams added to it to improve its performance, added to it to improve its performance, COO)4) and organic compounds such as and small ponds into which these ions such as CaCl , MgCl . To reduce the use 7 such as CaCl2, MgCl2. To reduce the use 2 2 urea (CO(NH2)2). Salts can be used in va- migrate. While the toxicity of calcium of the deteriorating NaCl, other salts are of the metal corrosive NaCl, other salts rious mixtures, and other substances magnesium acetate (CMA) to fish and in- also used, such as magnesium acetate, are also used, such as magnesium aceta- added, such as anticaking agents, e. g., vertebrates is low, when also containing calcium magnesium acetate or potassi- te, calcium magnesium acetate or potas- potassium ferrocyanide5 or anticorrosion potassium, CMAK (50 % CMA-50 % KA), um and magnesium formate. The additi- sium and magnesium formate, urea, and agents such as ammonium phosphate or they have higher toxicity. Acetate deicers on of sand and other inorganic insoluble even sugar containing solutions from eit- sodium hypochlorite.6 Also, they can be will result in the decrease of air pollution compounds to aid in making surfaces her sugar processing or equivalent proce- spread directly in granulated form, or as as sand use can be reduced; however, the less slippery is discussed, as well as the dures.1 a solution, i.e., brine. With the former, the solid deicers, CMA and sodium acetate, recent use of organic deicers and the pro- While the use of deicing salts is neces- mixture with sand (e.g., 75 % sand-25 % NAAC, may contribute fine particulates blems that these can induce. sary, they do have a negative impact on NaCl), or other equivalent materials such to the air increasing its pollution. These The paper aims to present an overview the environment, such as: soil conta- as fine gravel or expanded clay pellets deicers are mainly approved for use at of deicing salts, and the differences with mination, negative effect on plants and contributes to decrease the slippery sur- airfields and aircraft, as they are less cor- anti-icing or antifreeze solutions. It also trees near the highways/streets, conta- face of compacted snow2; however, they rosive, as is the case for potassium forma- discusses the problems they induce to ve- mination of water courses and eventual do increase small particulates in air by te, in either liquid or solid form. hicles, buildings and constructions, whi- drinking water, air contamination by about 45 %.7 Recently, potassium carbo- powdered salts, corrosion of reinforced le also considering the negative aspect nate (K2CO3) has been studied in compa- they have for the environment as well as concrete in bridges and structures, as rison to NaCl, and it was found that whi- 2.3. Urea their contribution to air pollution. Some well as of cars and trucks. For example, le it was more adsorbed to soil colloids, examples are presented to illustrate the in Austria it has been estimated that half the pH was elevated more than for NaCl, Urea [CO(NH2)2] is used as a deicing 9 problem and less aggressive alternatives the induced vehicle corrosion could be and the species composition of the area agent for airport runways though it has are discussed, especially with regard to attributed to de-icing salt.2 As deicing where it had been applied changed sig- been mostly discontinued in larger US 10 the conservation of valuable architectu- salts are distributed, the finer particles nificantly.8 airports. The main reason is that as a ral heritage. (<10 μm, usually referred to as PM10) can fertilizer (46 % by weight nitrogen con- remain suspended in the air, thus conta- tent) it contributes to environmental minating it. 2.1. Chloride based deicers pollution, e. g., acute toxicity to aquatic Keywords: deicing salts, building deterio- Differentiation between deicing and invertebrates and plants, as well as some 11 ration, environmental pollution anti-icing or antifreeze should be made. Chloride ions from deicing salts will fish. Several soil bacteria contain the Deicing salts are applied after snow mobilize and increase soil salinity near urease enzyme that catalyzes the de- + - events, their effectiveness being based the roadways where they are applied. composition of urea into NH4 and HCO3 . + on lowering the freezing point of wa- While magnesium and calcium ions in- Furthermore, NH4 (or NH3) is oxidized

16 17 A. E. Charola et al. Deicing Salts: An Overview

by nitrifying bacteria, Nitrosomonas and 2.5. Other organic deicers In certain streets, where regular and in- subsequently by Nitrobacter, in a two- tense winds are prevalent, the salt grains - step process to NO3 , an ion that is regu- In the USA, the Minnesota Department on the ground can be mobilized and larly found on building façades. of Transportation claims to have pionee- suspended in the air, as well as thrown Urea forms an eutectic mixture with red the use of sugar beet juice based on against the façades by whirl winds. If the water (at ~33 % by weight) with the eu- the huge sugar beet industry in the Red building surfaces are moist or rough, the tectic point at 11.5°C. Solubility is about River Valley of Minnesota/North Dakota, salts will be “attached” to them and dete- 1Kg/L at 20°C, the dissolution being endo- and the massive need for re-use of sugar rioration will eventually occur (Figure 3). thermic, and the equilibrium RH is 76.5 % beet waste helped create a market for it, Following the above mentioned proces- at 25°C. In dilute solutions (not specified and the fact that these states get a lot of ses, it is logical to ask what will be the but probably below 5 %), urea decompo- snow and ice contributed to the testing.15 eventual impact of the presence of these

ses to NH3 and CO2 (the formation of iso- The sugar beet syrup is mixed in with deicing salts in the air on the conservati- cyanic acid occurs upon heating, tempe- traditional salt, sand or chloride brines on of buildings and monuments. Recent rature not specified). The most common to improve performance and reduce the studies regarding the composition of fine impurity in synthetic urea results from impact on the environment.16 Not only particular matter suspended in urban air the condensation of two molecules to sugar beet syrup is used, but other re- have shown the recurrent presence of 7,18,19 form biuret (C2H5N3O2) or carbamylurea, sidues of distilled or fermented agricul- NaCl. In sea-side areas, it is obvious a compound that interferes with plant tural products7 such as corn, barley and that most of the salt present can be attri- growth. As a deicer, urea proves practi- even pickle brines. The addition of syrup buted to marine spray and fogs, however, cally useful, i. e., deicing within 15-20 from sugar processing to brines has been this cannot be applicable to inland areas. minutes, at temperatures below -9.4°C shown to improve their effectiveness and For example, in Putaud et al. 20 (see figure taking into account that its dissoluti- has been approved in Switzerland since 3, p. 2584), the measured annual average on is endothermic.12 Many studies have 2015.17 values of PM10 for NaCl (from January Figure 1: Lausanne (canton de Vaud, CH), Pierre Viret addressed the decomposition of urea in stairs (20.01.2010). Deicing salts are spread lavishly on 1998 to March 1999), is referred to as “sea aqueous solutions13, while others address the practically non-porous gneiss, and the salt and mel- salt”, in both rural areas, such as Chau- 3. Impact of deicing salts on buildings ted snow mixture accumulates at the base of the porous its use to decrease vehicular emissions of molasse-sandstone bridge wall that promptly powders mont, and urban areas, such as Basel and NOx which contribute to the formation When considering the effect of deicing and disaggregates. Zurich, as well as at street level (kerbside) of nitrates or nitrites in buildings along salts on buildings and constructions the the streets.14 immediate image that comes to mind is the damage at the foot of walls, resulting from the rising damp from the solution 2.4. Glycols and other alcohols of the melted snow and salts, as shown in Figure 1. Methanol was used as antifreeze in Experience has shown that to this de- windshield fluids, but because of health terioration mechanism two other direct concerns the amount added is restric- contamination processes have to be ad- ted. Ethylene glycol, commonly referred ded. The first one occurs in damp winter to as “glycol” is used as engine cooling conditions and affects buildings located antifreeze. The freezing point of ethy- along high traffic roads, where topogra- lene glycol is about −12°C, however, mi- phy contributes to accumulate the salt xed with water, this is depressed, e. g., a containing melted snow and that vehi- mixture of 60 % EG-40 % water freezes cular traffic and snow clearance vehicles at −45°C. Propylene glycol has replaced splash on to the building walls or disper- ethylene glycol in many uses because se into the air so that they enter directly of its lower toxicity. These products are at a certain height (Figure 2 left). The se- used for aircraft deicing fluids (heated cond process occurs during dryer winter Figure 2: Left: Belfaux (canton de Fribourg, CH), north façade of the Lanthen-Heid manor (1526).The damages that affect aqueous solution of ethylene glycol), and periods when the excess deicing salts ap- the render and the underlying molasse-sandstone are mainly the result of splashing traffic along this street (3.09.2008). as antifreeze, as undiluted, thickened Right: Lausanne (canton de Vaud, CH), rue Saint-Etienne (20.01.2010). During dry winter cycles the residues of recrystal- plied recrystallize and accumulate at the lized deicing salts can be seen at the foot of façades that can either be mobilized and suspended in the air or enter the propylene glycol. base of buildings (Figure 2 right). material when dissolved in water via hygroscopicity, rain, or more snow.

18 19 A. E. Charola et al. Deicing Salts: An Overview

Another topic that requires evaluation References is whether applying a polymer surface overlay system, such as SafeLane® having 1 Chemikalien-Risikoreduktions-Verord - an epoxy bonding agent, a special aggre- nung, Anhang 2.7, 2017. https://www. gate capable of storing a deicer that au- admin.ch/opc/de/classified-compilati- tomatically releases before frost and ice on/20021520/index.html#app27. can adhere to it4. The question is how fast is this overlay system worn down by the 2 Heisses Thema Eis und Schnee, Die Um- traffic and how much epoxy is released weltberatung, 2011. http://images.um- into the environment. weltberatung.at/htm/eis-und-schnee- ratgeber-wasser.pdf.

Conclusions 3 Brunning A., Periodic Graphics: Deicers

Figure 3: Lausanne (canton de Vaud, CH), rue de la Barre, Château Saint-Maire (10.04.2012). The sanding and disaggrega- and Antifreeze. C&E News, (93:2), (2015), tion that affects the base of the building and up to 2 m in some areas is mainly due to the presence of NaCl. It is clear that to maintain the econo- 30. http://cen.acs.org/articles/93/i2/Peri- mic system that has been developed since odic-Graphics-Deicers-Antifreeze.html. the industrial revolution, many changes in Bern. The PM10 immission of this city cannot be used freely elsewhere, no such have been made in the environment and 4 Cargill, Safelane Surface Overlay, 2017. is strongly influenced by traffic and sho- proviso has been considered for our cul- its ecology. But these changes take time http://safecote.com/project/anti-icing- wed a value of about 1.5 μg/m3 while the tural heritage. to implement, so it is important that, for surface-overlay-safelane/, https://www. other sites range between 0.5 μg/m3 for Another problem is the introduction of those of us concerned with the conser- cargill.com/industrial/winter-road- the urban sites to 0.2 μg/m3 for the rural additional hygroscopic salts into salt-con- vation of our architectural heritage, we maintenance/safelane-surface-overlay. site of Chaumont taining structures, as the hygroscopic should point out the problems that thre- The chemical composition and the salts will mobilize and activate those aten it to raise the awareness of the gene- 5 Mansfeldt T., Rennert T., Götzfried F., quantity of the coarser fractions in air present within the structure so that they ral public so eventually some actions can Eisencyankomplex-Gehalte in nord- pollution reflect the contribution of mi- will migrate and eventually crystallize be taken to protect them. As every buil- rhein-westfälischen Strassenrandböden neral dusts put into suspension by ve- in other areas. This has been observed ding and its situation are different, solu- nach dem schnee-reichen Winter 2009- hicular traffic. These include remaining on monuments, where salt efflorescence tions adapted to the individual case have 10 Strasse und Autobahn, (6), (2011), deicing salts and explain the relatively changes places after events introducing to be found. Often it is possible to use 389-393. high concentration of NaCl, as clearly sta- moisture and new salts, as described by gravel and no deicing salt at the proximi- ted by Gianini et al.21, p. 104 Rolland et al.24 who called it the “trans- ty of the walls of historic buildings, or it 6 Wrochna M., Malecka-Przybysz M., Ga- porting brine hypothesis”. But the pro- is possible to cover the basis of the walls, wronska H., Effect of road de-icing salts blem is now increased with the use of e.g., with boards to keep salt containing 25 with anti-corrosion agents on selected products from the food industry that snow away from the walls, similar to the 4. Discussion plant species, Acta Sci. Pol. Hortorum are added to deicing salts. No one has covers used to protect statues during Cultus, (9:4), (2010), 171-182. as yet raised questions regarding their winter. In some instances, a French-drain There is no question as to the contribu- long term effect when retained together could be installed by the wall where dei- 7 tion of deicing salts to the deterioration with the deicing salts in structures. For cing salts are applied, so that the melted Fischel, M., Evaluation of selected deicers of our architectural heritage as well as example, sugar beet syrup is hygroscopic, snow and salt will get trapped in it, or as based on a review of the literature, Re- to the environment. But it is also clear with a DRH ~60 % 26, so supposing that in Switzerland, many staircases are half port No. CDOT-DTD-R-2001-15, Colorado that deicing salts, as well as antifreezing this, together with NaCl is taken up into closed and only a small part kept free Department of Transportation Research formulations for airports and aircraft are the structure, in theory, less salt should of ice or snow. Ingenuity has been the Branch, 2001. https://www.codot.gov/ necessary and have to be used to avoid crystallize out. Of course, other problems mark of humankind, and it is time for it programs/research/pdfs/2001/deicers. traffic accidents, disruptions in the eco- could set in with the introduction of or- to come to the rescue should we want to pdf. nomy, and taking into account that the ganic materials, such as biocolonization, preserve our architectural heritage. negative impact of closing roads far and deterioration of the stone matrix, 8 Erhart E., Hart W., Effects of potassium exceeds the cost of snow and ice remo- should it have some solubility in water. carbonate as an alternative de-icer on val.22,23 While airport and aircraft are al- But this is an area that still needs to be ground vegetation and soil. Annals of lowed the use of some formulations that studied formally. Applied Biology, (136:3), (2000), 281-289.

20 21 A. E. Charola et al. Deicing Salts: An Overview

9 Meesen J. H., Urea, Ullmann’s Encyclope- bafu/de/home/themen/chemikalien/ 24 Rolland O., Vergès-Belmin V., Etienne dia of Industrial Chemistry, Vol. 37, Wi- fachinformationen/chemikalien--be- M., et al., Desalinating the Asyut dog ley-VCH Verlag GmbH & Co. KGaA, Wein- stimmungen-und-verfahren/auftau- in the Musée du Louvre, Science and heim, 2012, 657-695. mittel.html. art: a future for stone, Proc. 13th Int. Congress on the Deterioration and 10 Ground Support Magazine, Snow and 18 Van Dingenen, R., Raes F., Putaud J.-P., Conservation of Stone, Hughes J. and Ice Control Chemicals for Airports et al., A European aerosol phenome- Howind T. (eds), University of the West Operations, 2005. http://www.cryo- nology - 1: Physical characteristics of of Scotland, Paisley, 2016, vol. II, 1247- tech.com/snow-and-ice-control-chemi- particulate matter at kerbside, urban, 1253. http://research- portal.uws.ac.uk/ cals-for-airports-operations. rural and background sites in Europe, portal/files/397446/13th_ICDCS_Pais- Atmospheric Environment, (38), (2004), ley_2016_VOL_II.pdf. 11 MeltSnow.com, Urea, Material Safe- 2561–2577. ty Data Sheet, 2010. http://meltsnow. 25 Boller M., Bryner, A., Questions fré- com/development2010/wp-content/ 19 Putaud, J.-P., Van Dingenen R., Alastu- quentes sur le salage des routes, Eawag, uploads/2010/08/UREA_MSDS_MSWS. ey A., et al., A European aerosol phen- Swiss Federal Institute of Aquatic Scien- pdf. omenology - 3: Physical and chemical ce and Technology, 28.03.2017.http:// characteristics of particulate matter www.eawag.ch/fileadmin/Domain1/ 12 Peters Chemical Company, Breaking from 60 rural, urban, and kerbside sites Forschung/Oekosysteme/Oekosyste- the Ice, 2006. http://www.petersche- across Europe, Atmospheric Environ- me/FAQs_Salage_Eawag-update2016. mical.com/break-the-ice-compari- ment, (44), (2010), 1308-1320. pdf. son-of-ice-melting-chemicals/. 20 Putaud, J.-P., Raes F., Van Dingenen R., 26 Maudru J.E., Paxson, T.E., The Relati- 13 Alexandrova A.N., Jorgensen W.L., Why et al., A European aerosol phenome- onship of Sugar Moisture to Relative urea eliminates ammonia rather than nology - 2: chemical characteristics of Humidity. American Society of Sugar hydrolyzes in aqueous solution, J. Phys. particulate matter at kerbside, urban, Beet Technologists, 6th Biennial Mee- Chem. B, (111:4), (2007), 720–730. rural and background sites in Europe, ting, Detroit, Michigan, 1950, 538-540. Atmospheric Environment, (38), (2004), http://assbt-proceedings.org/1950Pro- 14 Durickovic I., Thiébault L., Bourson P., 2579-2595. ceedings.htm. Kauffmann T., Marchetti M., Spectros- copic characterization of urea aqueous 21 Gianini, M.F.D., Fischer, A., Gehrig, R., et solutions: Experimental phase diagram al., Comparative source apportionment of urea-water binary system. Applied of PM10 in Switzerland for 2008/2009 Spectroscopy, (67:10), (2013), 1205-1209. and 1998/1999 by Positive Matrix Fac- torisation, Atmospheric Environment, 15 MacDonagh L.P., Deicing with beet (54), (2012), 149-158 doi:10.1016/j.atmo- juice, Green Infrastructure, Deeproot, senv.2012.02.036. January 29, 2014. http://www.deeproot. com/blog/blog-entries/deicing-with- 11 Kuemmel D.A., Hanbali R.M., Accident beet-juice. analysis of ice control operations, Report, Marquette University, 1992. 16 Kinney T., Cities, states testing beet http://www.trc.marquette.edu/publica- juice mixture on roadways, USA to- tions/IceControl/ice-control-1992.pdf. day, February 21, 2008. http://usato- day30.usatoday.com/weather/resear- 23 HIS Global Insight, American Highways ch/2008-02-21-beeting-ice_n.htm. Users Alliance, The economic costs of disruption from a snowstorm, 2014. 17 Schweizerische Eidgenossenschaft, http://www.highways.org/wp-cont- Bundesamt für Umwelt, Auftaumit- ent/uploads/2014/02/economic-costs- teln, 2015. https://www.bafu.admin.ch/ of-snowstorms.pdf.

22 23 SWBSS 2017 | 20-22 September 4th International Conference on Salt Weathering of Buildings and Stone Sculptures Traffic-induced salt deposition on facades

Traffic-induced salt deposition on facades City Area (km2) Inhabitants Motor vehicles

Michael Auras* Bamberg 54.62 70,863 44,168 Institut für Steinkonservierung e. V., Mainz, Germany * [email protected] Würzburg 87.63 124,577 71,342

Mainz 97.74 202,756 107,004

Essen 210.30 566,862 307,943 Abstract from road traffic. Besides their negative health effects, their impact on materials Munich 310.70 1,388,308 771,625 Air pollution has been reduced signifi- has to be considered. This is particularly cantly in Germany during the last deca- true for historic buildings, because they Table 1: Size, inhabitants and number of registered motor vehicles of the selected cities. Data from 2011, from 3 des. Nevertheless high pollution levels have been exposed to air pollutants for are recorded in urban environment along decades or even centuries and they also heavy-trafficked roads. Relevant portions will be affected by future emissions. of the emissions of nitrogen oxides and The paper will report some results from Rack: Horizontal Average traffic Rack: Height fine particulate matter are caused by car recent research on the impact of the ac- Rack code City Building volume distance from above street traffic. The change of environmental con- tual pollution regime on historic stone (vehicles/d) street ditions has led to a change of salt depo- buildings.1-3 Especially results regarding BA Bamberg Nonnenbrücke 1 15,000 1 m 5 m sition on facades. Actual research results the input of salts and salt-forming subs- show high concentrations of nitrogen tances are presented. WÜ, up Würzburg Residenz 15,000 4 m up: 23 m oxides at historical facades but contradic- tory data for the deposition of nitrates. WÜ, low - - - - low: 4 m Additionally, the redispersion of de-icing 2. Methods salt by car traffic leads to a deposition of Mainz - - up: 35 m up: 31 m chlorides on facades not only in the base In the research project presented here, MZ, up MZ, low zone. the impact of traffic-related immissions - Christuskirche 27,000 low: 24 m low: 13 m on the building materials of historic mo- Keywords: air pollution, traffic emissions, numents in several German cities was E Essen Wasserturm 30,000 2 m 6 m salt input studied. Data on traffic volume and air Bayer. National- quality, numerical modelling of the dis- M Munich 52,000 5 m 3 m persion of air pollutants in the surroun- museum 1. Introduction ding of selected monuments, exposure experiments, as well as laboratory and Table 2: Exposure sites (2 racks at Mainz and Würzburg - up: upper rack, low: lower rack; from 3 In the last decades air pollution in onsite measurements were combined to Germany has been reduced significant- allow for an evaluation of the traffic-rely. Most notably the emission of sulphur lated pollution and its effects on stone number of buildings positioned along sive samplers (Tab. 2). Special exposure dioxide and its secondary products are buildings. More details are given in a main road and thus exposed to increased racks were constructed and fixed at the reduced now to a level of less than 10 %, specific paper.2 traffic emissions was determined. facades of the buildings in traffic-near compared to the amount emitted in the Data from five cities, Bamberg, Würz- Using data from air quality survey and positions. In Würzburg and Mainz, racks late 1980ies. Consequently the pH value burg, Mainz, Essen, and Munich were meteorological stations, the dose-respon- were mounted at two different heights of of rain increased to a level being inhe- used to obtain an overview of the situ- se functions from the MULTI ASSESS pro- the buildings. Then slabs of three stone rent by the saturation with carbon dioxi- ation in these cities being characterized gram4 were calculated for the recession varieties, Portland limestone, Baumber- de. Also the emissions of nitrogen oxides by different climate and traffic volume of the reference material Portland Limes- ger sandstone, and Carrara marble were and particulate matter could be reduced, (Tab. 1). tone. exposed in sheltered and unsheltered po- but they still are at problematic levels, re- Maps of traffic volume were drawn In each of the five cities one historic sitions. Only in Mainz a Mank’s carousel garding the actual threshold values. High and superimposed to maps of the stock building in a heavy-trafficked road was was additionally installed to allow for a proportions of these pollutants originate of cultural heritage buildings. Then the chosen for the exposure of several pas- comparison with literature data.

24 25 M. Auras Traffic-induced salt deposition on facades

On all racks horizontally oriented 3. Results rain-sheltered passive samplers made of a boron substrate were used to collect Air pollution by sulphur dioxide has particulate matter for ESEM studies and been diminished drastically. The emis- EDX-analyses. The samplers were replaced sions of nitrogen oxides and fine parti- bi-monthly during one year. The collec- culate matter were also reduced, but to a ted particles were characterized morpho- much lower extent (Fig. 1). logically and chemically by semi-auto- Calculation of recession rates of the mated individual-particle analysis with a reference material Portland limestone Scanning Electron Microscope (FEI Quan- via the MULTI ASSESS formulae5 show ta 200 Feg) combined with energy-disper- a decline of recession caused by the lo- sive microanalysis (EDX). Due to the large wering of pollutant concentrations with spectrum of particle sizes, 2,000 particles time. These calculations need an input

from each sample were analysed within of data of air pollution (SO2, HNO3, pH of two size classes (300 nm to 3 µm and > precipitation) and climate (relative humi-

3 µm). Image analysis showed that the dity). Missing HNO3 data are calculated

coverage of the substrates was mainly from the concentrations of NOx (calcula-

due to the deposition of particles of the ted as NO2) and ozone, and relative hu- > 3 µm fraction. Therefore so far only the midity5. By setting one of these factors larger fraction is evaluated. Altogether after another to zero, the influence of Figure 1: Development of pollutant emission from 1990 to 2010. Data: Umweltbundesamt 8 TSP: Total suspended approximately 42,000 particles were ana- the single pollutants on the calculated particles. lysed. Each particle was checked visually recession rate can be evaluated. Fig. 2 and after elimination of artefacts and shows the result of such calculations particles with poor images about 31,000 with environmental and meteorological particles were evaluated. Based on che- data from Munich. Although the recessi- mical composition and morphology, the on rate decreases continually with time,

particles were classified into 14 groups, an increasing impact of HNO3 (more pre-

and the relative particle number, as well cisely NO2 and O3) on the weathering of as the relative and absolute area coverage limestone is observed. of each group was calculated. The question of the role of the actual Surface Active Monitors (SAM) accor- pollutant regime on historic buildings ding to Rumpel5 are filter papers impreg- is not an academic one. In several Ger- nated with a solution of alkali carbona- man cities the stock of historic buildings te. They were used to sample sulphate, being positioned along main roads and nitrate and chloride and were replaced thus being exposed to increased traffic bi-monthly. emissions caused by more than 5,000 ve- The contents of chloride, sulphate and hicles daily is considerable (Tab. 3). nitrate of the SAM filters and of the stone Single particle analysis of deposited slabs that were exposed under sheltered particulate matter shows a distinct in- conditions were analysed by ion chro- crease of chloride particles in the winter matography (Dionex ICS 1000). months at several buildings (Fig. 3). This 6,7 Diffusive samplers NO2 and HNO3 proves, that the deposition of chloride were provided and analysed by courtesy particles is mainly due to the redispersi- of the IVL Swedish Environmental Rese- on of de-icing salt by car traffic. Calcula- arch Institute. ting annual surface coverage rates from the bi-monthly samples, it can be shown, Figure 2: Calculated recession rates of Portland limestone under environmental conditions of the Karlsplatz at Munich that at Munich and at Bamberg about 9 during the last 35 years. Modified from 10 % of a horizontal, rain-sheltered surfa-

26 27 M. Auras Traffic-induced salt deposition on facades

Exposed to traffic Listed Therefrom at Number of historic buil- volume > 5,000 monuments main roads dings along main roads vehic./d

Traffic volume (x103 < 5 5-20 20-40 >40 Total % vehicles/d

Bamberg 1348 507 163 344 - - 344 26 (city)

Würzburg 802 318 222 87 9 - 96 12 (city)

Mainz 866 328 167 94 57 10 161 19 (part of city)

Munich 4882 1212 90 702 344 76 1122 23 (part of city)

Table 3: Stock of listed historical buildings and monuments exposed to elevated traffic volume. Data from10

Figure 4: Calculated annual area coverage by particle deposition. From 11

ce might be covered by chloride particles after one year (Fig. 4). The deposition of redispersed chloride particles predominantly affects the lower parts of buid- lings, but in two cases it was found also in remarkable height (MZ, low and WU, up). Using surface activated monitors (SAM), the deposition rates are higher for nitrate compared to sulphate. Using however natural stone samples, more sulphate than nitrate is deposited (Fig. 5). It is supposed, that this is caused by the ability of the SAM-samplers to bind nitrogen oxide gases chemically, while at stone surfaces gaseous substances are only bound physically by weak adhesion forces. So the input on SAM-filters seems to depend on the concentration of NOx-gases, while the input on stone samples is assumed to

be determined by the formation of HNO3.

The measurements of HNO3-concentra- Figure 5: Annual deposition of salts on passive samplers tions in air directly at the facades show from historic buildings situated at traffic-rich urban Figure 3: Variation of the composition of deposited particles > 3 µm at Munich (relative surface coverage). Letters on main roads. Left: Results for SAM-samplers. Right: Re- x-axis indicate bi-monthly sampling. From 11 two trends: First there is a distinct seaso- sults for stone samples.

28 29 M. Auras Traffic-induced salt deposition on facades

nal dependency with maximum values 6. Acknowledgements in summer (Fig. 6) and secondly a positive correlation with the height of the ex- Many thanks to Silvia Beer, Stefan posure racks at the building is indicated Brüggerhoff, Petra Bundschuh, Thomas (Fig. 7). Dirsch, Joachim Eichhorn, Martin Ferm, Anette Hornschuch, Dirk Kirchner, Inga Kraftczyk, Karin Kraus, Martin Mach, 4. Discussion Wolfgang Mühlschwein, Dirk Scheuvens, Klemens Seelos, Björn Seewald, Rolf While the observation from figure 6 is Snethlage, Roger Thamm, Stefan Wein- due to the intensity of solar insolation, bruch and Tim Yates for their coopera- the dependency on building height could tion and assistance. Thanks also to all

indicate that the formation of HNO3 from the involved authorities for the supply of

NO2 takes some time. HNO3 seems not to data. Financial support by the Deutsche be formed immediately at the source of Bundesstiftung Umwelt is gratefully ack- the exhaust emissions, but in some dis- nowledged. tance. This could indicate that the reaction of nitrogen oxides with water and oxygen might to be rather slow. A consequence of this time-dependency might

be a formation of HNO3 not only at the Figure 6: Results from passive sampling of HNO at various facades. Data by courtesy of the Swedish Corrosion 3 higher parts of the buildings but also in References Institute, from 12 the side roads, where it is superimposed by the dilution effect due to the minor 1 Auras, M., Beer, S., Bundschuh, P., Eich-

NOx emission rates. horn, J., Mach, M., Scheuvens, D., Schor- ling, D., v. Schuhmann, J., Snethlage, R. & Weinbruch, S., Traffic-related immis- 5. Conclusions sions and their impact on historic buildings: implications from a pilot study Although the environmental condi- at two German cities, Environmental tions have been significantly improved Earth Sciences 69 (2013), 1135 - 1147. over the last decades, there is still an impact of pollutants on many historic buil- 2 Institut für Steinkonservierung, Bau- dings. Besides accelerated soiling by tire denkmäler unter dem Einfluss ver- wear and others the deposition of salts kehrsbedingter Immissionen, Institut takes place on facades. Chloride and sul- für Steinkonservierung e.V., Mainz, phate is deposited in form of fine parti- IFS-Bericht Nr. 49 (2015), 163 pp. culate matter. Chlorides originate mainly from the redispersion of de-icing salt. 3 Auras, M., Bundschuh, P., Eichhorn, J., Kirchner, D., Mach, M., Seewald, B., The role of nitrogen oxides and HNO3 is not very clear up to now. The contradicti- Scheuvens, D. and Snethlage, R., Traf- onary results obtained by using different fic-induced emissions on stone buil- sampling materials need further rese- dings, In: Hughes J.J. and Howind T. arch on the formation and deposition of (eds.), “Science and Art: A Future for Figure 7: Mean values of the data from figure 6, ordered by the distance from the emission source. Data: see fig. 6 nitrates. Stone, proceedings of the 13th Interna- tional Congress on the Deterioration and Conservation of Stone”, University of the West of Scotland, Paisley (2016), 3-11.

30 31 SWBSS 2017 | 20-22 September M.Auras 4th International Conference on Salt Weathering of Buildings and Stone Sculptures

4 Kucera, V., Model for multi-pollutant sionen und Belastung des städtischen Wick action in cultural heritage impact and assessment of threshold le- Denkmalbestandes, In: Baudenkmäler vels for cultural heritage, project period unter dem Einfluss verkehrsbedingter Leo Pel* and Raheleh Pishkari 1 January 2002 to 30 April 2005, Publis- Immissionen. Institut für Steinkonser- Eindhoven University of Technology, The Netherlands hable Final report, www.corr-institute. vierung e.V., Mainz, IFS-Bericht Nr. 49 * [email protected] se (2005), 52 pp. (2015), 19-33.

5 Rumpel, K., Ein Verfahren zur Feststel- 11 Scheuvens, D., Dirsch, T., Moissl, A., lung potentieller Immissionsraten: Küpper, M., Weinbruch, S., Partikulä- Oberflächenaktive Monitore SAM (Sur- re Schadstoffe an Baudenkmälern. In: Abstract 1. Introduction face active monitoring), Monatsberichte Baudenkmäler unter dem Einfluss ver- aus dem Messstellennetz des Umwelt- kehrsbedingter Immissionen, Institut Salts crystallization is one of the main Salts crystallization is one of the main bundesamtes 7 (1984), 2-15. für Steinkonservierung e.V., Mainz, degradation mechanisms of historical degradation mechanisms of historical IFS-Bericht Nr. 49 (2015), 79-99. objects, e. g., masonry. In this study we heritage objects, such as masonry. Mois- 6 Ferm, M., Svanberg, P.-A., Cost-efficient looked at a special case often encoun- ture penetration can advect dissolved techniques for urban- and background 12 Bundschuh, P., Auras, M., Kirchner, tered in marine environment, i.e., wick ions along with it, into a porous material. measurements of SO2 and NO2, Atmo- D., Scheuvens, D., Seelos, K., Expo- action. This is a steady state situation The salt ions can accumulate and crystal- spheric Environment 32 (1998), 1377- sitionsprogramm zur Wirkung ver- in which one end of an object is conti- lize in the pores and can as a result pro- 1381. kehrsbedingter Immissionen auf Na- nuously absorbing a salt solution, e. g., duce cracks due to crystallization pres- tursteinoberflächen, In: Baudenkmäler sea water, whereas at the same time at sure. To get a better insight into the salt 7 Ferm, M., De Santis, F. Varotsos, C., Nitric unter dem Einfluss verkehrsbedingter other side there is continuous drying. As transport mechanism we picked up on acid measurements in connection with Immissionen, Institut für Steinkonser- a result there will be a continuous flux a special case when a porous material is corrosion studies, Atmospheric En- vierung e.V., Mainz, IFS-Bericht Nr. 49 of ions towards the drying surface and in contact with salt solution on one side vironment 39 (2005), 6664-6672. (2015), 53-77. the concentration at the drying surface and, at the same time, exposed to drying will slowly increase, resulting eventually conditions on the opposite side. This si- 8 Umweltbundesamt, Nationale Trend- in crystallization. In this study we look- tuation is often encountered in cultural tabellen für die deutsche Berichter- ed at wick action for a 1 m NaCl solution heritage in marine environment. stattung atmosphärischer Emissionen using a biomicritic limestone from Sar- A well-known example is the housing 1990–2014 klassische Luftschadstoffe dinia, which is found in many cultural along the canals in the historic city of Ve- (Endstand 09.02.2012). Umweltbun- heritage objects. To measure both mois- nice. Here there is a continuous supply of desamt, Dessau, http://www.umwelt- ture and salt content simultaneously, we fresh sea water as a source of salt which bundesamt.de/sites/default/files/ have used a specially designed Nuclear is absorbed, whereas the top of the ma- medien/376/dokumente/emissionsent- Magnetic Resonance (NMR) set-up. The sonry is drying, giving rise to continuous wicklung_1990_-_2014_fuer_klassi- wick action experiment was performed salt damage. This continuous transport sche_luftschadstoffe.xlsx (2016). for over 40 days. The results show that of a salt solution, combined with drying, the concentration over 40 days slowly is often referred to as wick action (see 9 Bundschuh, P. & Auras, M., Anwendung increases at the top until the saturation also Fig. 1). von Dosis-Wirkungsbeziehungen auf concentration is reached. It is shown that During wick action salt ions will move das Münchener und Mainzer Unter- the concentration profiles can be model- by two mechanisms, advection and dif- suchungsgebiet, In: Wirkungen ver- led by a simple analytic solution of the fusion.1,2 Advection is the process of ions kehrsbedingter Immissionen auf Bau- advection-diffusion equation describing moving along with the moisture flow, denkmäler – Eine Pilotstudie zu den the ion transport. whereas diffusion is dependent on the Innenstädten von Mainz und München. concentration gradient, i. e., diffusion Institut für Steinkonservierung e.V., Keywords: Wick action, drying, salt crys- tries to level off any concentration gra- Mainz, Bericht Nr. 37 (2011) 47 – 53. tallization, Nuclear Magnetic Resonance dient. As a result the net ion flux will be a competition of these two processes. 10 Bundschuh, P., Auras, M., Seewald, B., Due to the continuous flow of salt ions Snethlage, R., Verkehrsbedingte Immis- towards the drying face there will be a continuous accumulation and as soon as

32 33 L. Pel and R. Pishkari Wick action in cultural heritage

2. Theory In the case of drying as studied here, which were vacuum saturated with dis- the boundary condition for ions at the tilled water. In order to perform 1D ex- In order to get a better understanding drying surface is given by a no flux periment the specimens were isolated of the salt concentration profiles which boundary, i. e., q=0, and hence: on the sides with an epoxy coating. The will develop during wick action we can experimental setup is given in Figure 2. have a look at combined ion and mois- The sample holder is a Teflon cylinder ture transport. As long as the sample (3) with a reservoir at the bottom and air stays saturated during the wick action, flow inlet and outlet at the top to provi- the ion transport can be described by an de drying at the top of the sample. The advection-diffusion equation as given by: As in our experiment we have a cons- bottom of the sample is in contact with tant boundary condition we can assume a 1m NaCl solution. In order to maintain that in first order the liquid flow is cons- a constant level the pump is controlled (1) tant, i. e., u= constant. As also the concen- by using an electrical sensor (max level tration of the absorbed salt solution is fluctuations are in the order of 3 mm). To constant, i.e., c= co , we can in this case induce drying, the top of the sample is where c [m] is ion concentration, which solve the differential equation giving the will be a function of both the time t (s) concentration c(x,t) as a function of time and position x (m), D [m2/s] is the effec- and position, i. e.; tive diffusion coefficient of the ions in the porous material and u [m/s] is the macroscopic liquid velocity of the liquid (4) in the porous material, i. e, the Darcy speed. This equation will be valid as long as there is no crystallization taking place Where a(t) is a constant which is a fun- else a sink term has to be added to the ction time. Hence this solution tells us right-hand side of this equation. that the concentration in the sample can Therefore the right hand side is describe described by an exponential decay. In bing the total ion flux, consisting of dif- addition, it indicates that characteristic Figure 1: A schematic representation of wick action: one fusion, whereas the second part descri- width of the salt concentration peak gi- side of an object is absorbing a salt solution, whereas at the same time other side is drying. As a result there will bes the advection of the ions along with ven by 4D/u, is determined by the ratio be a continuous flow of ion towards the drying surface. the liquid flow. The competition of these of the liquid velocity and the diffusivity. two can be characterized with a dimensi- onless number, i. e., the Peclet number. In the maximum solubility limit has been this case, based on Eq 1. it can be defined as: reached, there will be crystallization, eit- 3. Material and methods her inside the materials (subflorescence) or outside (efflorescence). In this study (2) For this study we have looked at a bio- we have focused on wick action in the micritic limestone from Sardinia which stable situation where the drying front has been used in many objects there. Is stays at the top of the sample, i. e., the where L is a so-called characteristic has a porosity of 0.34, a pore size with the surface where the evaporation takes pla- length scale which in this case can be maximum distribution around 1.6 µm ce. We have used Nuclear Magnetic Reso- chosen as the length of the sample. In and has only very minor magnetic impu- nance (NMR) to measure both moisture the case Pe>1 advection will be dominant rities. For the experiments we used cylin- and ion profiles non-destructively and and there will be concentration gradient. drical samples with a diameter of 20 mm quasi-simultaneously during wick action Whereas in the case of Pe<1, diffusion is and length of 100 mm, which were dril- experiments. These experimental results dominant and we expect a homogenous led out of a larger block. After rinsing the have been compared to a simple analytic distribution of salt. This Pe-number was sample in clean water, the sample was Figure 2: A schematic representation of the setup for solution of the advection-diffusion equa- also found to be very useful to give an dried at 40o°C until constant weight. The measuring the wick action using NMR. With an electro- nic level control and a pump the reservoir level is kept 3 tion describing the ion transport. indication of the effect of poulticing. experiments were started with samples constant.

34 35 L. Pel and R. Pishkari Wick action in cultural heritage

exposed to a constant air flow of 10 L/m moisture and Na-profile takes in the or- at approximately 0% relative humidity. der of 3 hours. The total experiment las- An plastic evaporation shield has been ted up to 40 days. added as to separate the drying and ab-

sorption part, i. e., to limit the influence of the drying on the bath concentration. 4. Experimental results Both the moisture and Na-content are measured using Nuclear Magnetic Reso- The NMR signal was collected from nance (NMR).4,5 Using NMR we are able to cross sections of the sample in 1.27 mm non-destructive and quantitatively mea- steps along the axis of the cylindrical sure the moisture and Na-content at a fi- sample. Both the moisture and Na-con- xed position, with a 1D-resolution in the centration profiles were measured every order of 2 mm for hydrogen and 8 mm 3 hours for 40 days. The resulting Na con- for Na. Hence every point represents an centration profiles for every 4 days are average over 2 to 8 mm. A stepper motor given in figure 3. was used to move the sample holder ver- At position x=100 the sample is in con- tically inside the NMR to be able to mea- tact with the reservoir of 1 molal NaCl sure the moisture and Na profiles over solution, whereas at x=0 the sample is the complete sample. Measuring both a drying. As can be seen the concentration

Figure 4: The measured Na-concentration profiles during wick action of a sample of 100 mm in length and a reservoir of 1m. The profiles are given for every 4 days for a total time of 40 days. The smooth curves represent fits of the model to each individual profile.

at the top is slowly increasing with time on reached within the sample, i. e., the indicating Pe>1, whereas at the bottom top, will be higher. of the sample the concentration remains In order to correct for this problem, we almost constant reflecting the constant have fitted the model as derived see( Eq. concentration of the reservoir. 4) to each individual profile. The results However at the top of the sample the are given in figure 4. As can be seen each concentration does not raise to 6.1 molal profile can be approximated well by an as to be expected. Here one has to take exponential function. We can also see into account the 1D resolution of 8 mm, that the interpolation with x=0 indicates i. e., each point represents the average that the maximum salt concentration of over 8 mm. Indeed looking in more detail 6 molal is reached after 40 days, which one can see the concentration, which is could also been seen from crystals for- measured every 1.27 mm, is slowly rising ming at the top of the sample. from almost zero 0 molal at -4 mm representing a point outside of the sample, to 5 molal at 4 mm inside of the sample. 5. Conclusions Hence the 1D resolution is smoothing off the real concentration profile within the Wick action in combination with Figure 3: The measured Na-concentration profiles using NMR during wick action of a sample of 100 mm in length and a reservoir of 1m at x=100 mm. The profiles are given for every 4 days for a total time of 40 days. sample, and the maximum concentrati- evaporation at the top of a sample can

36 37 L. Pel and R. Pishkari

be described by a simple analytic model, which only takes into account the liquid speed and the diffusivity. The measured concentration profiles match well with a simplified model. Also NMR has been shown to be an effective method to measure the concentration profiles during wick action over a long time and non-destructively.

References

1 Y.T Puyate and.C.J Lawrence, Effect of solute parameters on wick action in concrete. Chemical Engineering Science (54) (1999) 4257-4265.

2 Y.T Puyate and.C.J Lawrence, Steady state solutions for chloride distribution due to wick action in concrete, Chemical En- gineering Science (55) (2000) 3329-3334.

3 L. Pel, A. Sawdy and V. Voronina, Physi- cal principles and efficiency of salt extraction by poulticing, Journal of Cultu- ral Heritage (11) (2010) 59–67.

4 L. Pel, P.A.J. Donkers, K. Kopinga and J.J. Noijen, 1H, 23Na and 35Cl imaging in cementitious materials with NMR, Appl Magn Reson (47) (2016) 265–276.

5 L. Pel and H.P. Huinink, Building Materi- als Studied by MRI, Encyclopedia of Ma- gnetic Resonance, by John Wiley & Sons, Ltd 2012 (DOI: 10.1002/9780470034590. emrstm1294).

38 Measurement techniques and experimental studies SWBSS 2017 | 20-22 September A preliminary study on dynamic measurement of salt crystallization and deliquescence on a porous material surface 4th International Conference on Salt Weathering of Buildings and Stone Sculptures using optical microscope

A preliminary study on dynamic measurement of salt volume is basically reduced due to the 2. Methodology presence of salt crystals, the specific pore crystallization and deliquescence on a porous material volume for some pore sizes can increase.4 Dynamic laboratory measurements surface using optical microscope This means that salt crystals can provide of salt crystallization and subsequent a new pore structure that exhibits diffe- deliquescence on a surface of a porous Masaru Abuku1*, D. Ogura2 and S. Hokoi2 1 rent water and salt transport properties material specimen (Autoclaved Aerated Faculty of Architecture, Kindai University, Japan 5-6 2 Graduate School of Engineering, Kyoto University, Japan in the material. Concrete (AAC) as a model material) are * [email protected] Compared to crystallization in pores, carried out with use of a high-resolution less attention has been paid to crystalli- digital optical microscope that allows Abstract 1. Introduction zation at surfaces of porous materials in generating three-dimensional images. the past7-15, because in-pore crystallizati- The spatial resolution at the top surface This paper reports on our preliminary Cycles of crystallization and delique- on is more harmful for damage of poro- of the specimen is horizontally 10 μm x attempt to perform dynamic laboratory scence of salt can be often observed on a us materials. However, recently, the work 10 μm; vertically the stage moves with measurements of salt crystallization and surface of building and heritage materi- of Veran-Tissoires13-15 using 1 mm glass a resolution of 0.1 μm, to calculate the subsequent deliquescence on a surface of als such as stones, bricks, mortar, etc. In beads gave an important motivation to height of the object (the specimen sur- a porous material specimen (autoclaved general, salt crystals grow in cold seasons understand the phenomena at the ma- face or precipitated salt) based on focus. aerated concrete as a model material) and disappear in warm seasons, often as terial surface, indicating that salt struc- The dry bulk density of the specimen is by using a high-resolution digital opti- a result of hygrothermal responses of the tures forming efflorescence are porous 507 kg/m3. The moisture transport pro- cal microscope that allows generating materials to changes of ambient air tem- and transfer salt solution by capillarity, perties of the same type of AAC as this three-dimensional images. A specimen perature and humidity, rain, solar radia- influencing the water evaporation rate specimen is given in e. g.16 The mass of that is initially filled with a NaCl soluti- tion and so on. To physically understand at porous material surfaces, which is cal- the specimen as well as the temperature on is dried in a room at ~26°C and ~40% such phenomena and propose a suffi- led the screening effect. Rad and Shokri10 and humidity of the ambient air is also RH for 14 days and is then wetted in a cient countermeasure, it could be useful also studied salt precipitation on sand measured simultaneously. A temporal desiccator at ~95% RH during the next to quantitatively determine the rates of grains with an average particle size of change of the height h of salt crystals on 14 days. A temporal change of the spa- water evaporation and the subsequent 0.48 mm and the relation of evaporation the specimen surface is determined ba- tial distribution of the volume occupied salt crystallization and those of water with salt precipitation. Rad et al11 further sed on contrast between two microscopic by salt crystals on the specimen surfa- absorption and the subsequent salt deli- conducted three-dimensional imaging of images successively taken with a certain ce is determined based on the contrast quescence on the surface of the material. using x-ray microtomography, to under- interval of time. between two microscopic images succes- Water and salt transport in porous stand more in detail the relation of the The following steps are taken: sively taken with a certain interval of building materials has been steadily and pore sizes and the initial salt concentra- time. The mass of the specimen is also rather well studied over the last years. tion with the dynamics and patterns of (1) A cubic specimen of 27 cm3 is dried measured simultaneously to determine This phenomenon can be characterized salt precipitation during drying. Our cur- completely to determine the porosi- the rate of water evaporation/absorption. by two main material properties: the rent study focuses on salt on a material ty and dry density of the specimen. The results show that salt crystallization sorption isotherm and the liquid water surface that crystallizes during drying Then, the surfaces of the specimen during drying is not spatially uniform conductivity, both of which are depen- and furthermore deliquesces during sub- except the top and bottom surfaces throughout the specimen surface, which dent on the salt concentration, water sequent re-wetting. It reports on our pre- are made vapor tight. likely depends on the spatial distributi- content and temperature. Apart from liminary attempt to perform dynamic on of the pore structure in the specimen, water and salt transport, crystallization laboratory measurements of salt crystal- (2) The specimen is immersed in pure wa- and salt crystal deliquesces faster near of salt in pores of a porous material or lization and subsequent deliquescence ter and then a drying test is conduc- the edges of the specimen than in the at/near the surface of the wall still needs on a surface of a porous material speci- ted to know how fast the specimen is centre, which indicates a stronger depen- fundamental studies for quantification. men (autoclaved aerated concrete as a dried in the laboratory. The temporal dency of salt deliquescence on the geo- Regardless of difficulty of measurement, model material) by using a high-resoluti- change of the mass of the specimen is metry of the specimen than on the inner recent non-destructive measurement on digital optical microscope that allows measured. pore structure of the specimen. techniques enabled to visualize the volu- generating surface three-dimensional me of salt in pores of the material.1-3 Ano- images. A temporal change of the height (3) The specimen is again completely Keywords: light microscopy, efflorescen- ther approach to quantify the amount of of salt crystals on the specimen surface dried and then immersed in satura- ce, sodium chloride, absorption/desorp- salt in a porous material is to measure is determined based on contrast between ted NaCl aqueous solution with a salt tion the pore size distribution of the material, two microscopic images successively ta- concentration of 0.1 kg/kg. Note that i. e. to show that although the total pore ken with a certain interval of time. in this paper, the salt concentration is

42 43 A preliminary study on dynamic measurement of salt crystallization and deliquescence on a porous material surface M. Abuku et al. using optical microscope

defined by the mass of salt divided by at the drying stage is calculated. ∆h and (a) 0.71 days (b) 1.79 days (c) 2.79 days (d) 6.79 days the mass of the solution. Afterwards, h at t = 1.79, 2.79 and 6.79 days are illus- the bottom of the specimen is made trated in Figure 2. Note that h is equal to vapor tight; only the top of the speci- 0 at t = 1.79 days and for example, h at t = men is open to vapor transfer. 2.79 days is the sum of h at t = 1.79 days and ∆h at t = 2.79 days. (4) The specimen is stored at a constant The results show that during drying, temperature (~26°C) and humidity there is a large spatial distribution in (~40% RH). The height h of salt crystals the crystallization rate in a shorter time on the specimen surface is recorded scale ( one day) as also measured by Rad ~ Figure 1: Photographs of the center (2.4 cm x 2.4 cm) of the top surface of the specimen taken at the drying stage with a certain interval of time using et al11; but at the end of drying, salt is the microscope that can allow to ge- rather uniformly distributed on the sur- nerate three-dimensional images. The face. Although salt crystals first grow at (a) ∆h (1.79 days) (b) ∆h (2.79 days) (c) ∆h (6.79 days) weight of the specimen is also mea- some specific areas, it seems that crystal (x10μm) μm (x10μm) μm (x10μm) μm sured to determine the evaporation growth stops after some time. However, rate. The drying process is continued next salt crystal growth occurs at other for approximately 14 days. areas, where successive water evaporation occurs due to a lower liquid water con- (5) The specimen is stored at almost the ductivity. This could be explained by the same temperature but at a humidity intrinsic spatial distribution of the pore 11 of ~95% RH. Salt crystals on the sur- structure of the specimen , which deter- face and the specimen’s weight are mines the in-pore liquid water transport recorded, by the latter of which the and the evaporation at the surface. (x10μm) (x10μm) (x10μm) rate of water vapor absorption is de- Looking at e. g. Figure 2b, a very larger (d) h (2.79 days) (e) h (6.79 days) termined. The wetting process is cont- increase of h is observed at some large (x10μm) μm (x10μm) μm inued for approximately 14 days. open pores on the specimen surface seen in e.g. Figure 1a. This can be attributed to In the following sections, results are the difficulty of determining the height discussed based on the properties of the of the surface of the material in the po- specimen and salt at the given tempera- res. ture and humidity conditions. Photographs of the center of the top surface of the specimen taken at t = 0.47, 0.86, 1.24 and 1.86 days during the wet- 3. Results and discussion ting stage are given in Figure 3. The (x10μm) (x10μm) decrement ∆h of the height h of salt on Figure 2: The spatial distribution of (a, b, c) the height change ∆h of salt crystallizing on the top surface (2.4 cm x 2.4 Figure 1 shows photographs of the cen- the specimen surface at the wetting sta- cm) of the specimen for three consecutive drying periods and (d, e) the height h at the end of the last two periods (2.79 ter (2.4 cm x 2.4 cm) of the top surface of ge and h at t = 0.86, 1.24 and 1.86 days and 6.79 days). (a) ∆h for 26 hours between 0.71 and 1.79 days; (b) ∆h for 24 hours between 1.79 and 2.79 days; and (c) ∆h for 96 hours between 2.79 and 6.79 days. the specimen taken at t = 0.71, 1.79, 2.79 are illustrated in Figure 4. Note that h is and 6.79 days during the drying stage. equal to 0 at t = 1.86 days. To omit the effect of the edges of the Although the spatial distribution of top surface on three-dimensional image salt crystals that are deliquescing is more when the specimen absorbs moisture, se of the low temperature dependency of analysis, only a square of 2.4 cm x 2.4 cm uniform compared to that of crystalliz- the temperature of the surface gets hig- the solubility of NaCl. is taken though the top surface area of ation and independent of the in-pore her due to heat of condensation. Becau- A clear difference in time is observed the specimen is 3 cm x 3 cm. Based on structure of the specimen, it seems clear se the change rate of the temperature is between crystallization and delique- the difference of two three-dimensional that deliquescence occurs faster near the larger near the edges of the specimen, scence processes. This can be explained images successively taken with a certain edges of the specimen than the center of the solubility near the edges may have as follows. The specimen was comple- interval of time, the increment ∆h of the the surface due to a geometry effect on become more increased due to the tem- tely wetted with the salt solution at the height h of salt on the specimen surface the heat and moisture transfer. Note that perature increase. However this effect is beginning of drying and salt solution considered to be extremely small becau- transfer in the specimen took some time.

44 45 A preliminary study on dynamic measurement of salt crystallization and deliquescence on a porous material surface M. Abuku et al. using optical microscope

(a) 0.47 days (b) 0.86 days (c) 1.24 days (d) 1.86 days ent temperature and relative humidity, solution digital optical microscope that and so on. allows generating three-dimensional images. The obtained results are analysed to quantify the spatial distributions 4. Conclusions of the rates of salt crystallization and subsequent deliquescence on the surface In this paper, we reported our preli- and their spatially averaged values. The minary dynamic measurements of salt results also showed that salt crystallizati- crystallization and subsequent delique- on during drying is not spatially uniform scence on a surface of a porous material on the specimen surface and depends on Figure 3: Photographs of the center (2.4 cm x 2.4 cm) of the top surface of the specimen taken at the wetting stage specimen (autoclaved aerated concrete as the spatial distribution of the pore struc- a model material) with use of a high-re- ture in the specimen as demonstrated by Rad et al11, and salt crystals delique- (a) h (0.47 days) (b) h (0.86 days) (c) h (1.24 days) (x10μm) μm (x10μm) μm (x10μm) μm sce faster near the edges of the specimen than in the centre, which indicates a stronger dependency of salt deliquescence on the geometry of the specimen than the inner pore structure of the specimen. The obtained data and future measurements with different conditions are to be used for model development and valida- tion. (x10μm) (x10μm) (x10μm)

Figure 4: The spatial distribution of the height h of salt deliquescing on the top surface (2.4 cm x 2.4 cm) of the specimen at t = 0.47, 0.86 and 1.24 days of wetting periods. h is defined to be 0 at 1.86 days. Acknowledgements

This work was supported by JSPS KA- Figure 5: The average moisture content w, the estimated KENHI Grant Number 26709043. The au- This means that the drying rate was de- condition (>10% RH) which was not well precipitated salt content wc and the average height have of salt on the specimen surface for the drying process of thors would like to thank the reviewer pendent on the salt solution permeabili- controlled during the experiment. The 13 days. The solid line shows the measurement result of for the valuable comments. ty of the specimen. As a result, crystal- evaporation rate from the specimen with w for pure water evaporation from the same specimen for 5 days. have is defined to be 0 at 0 day. lization continued for at least 7 days of the salt solution is much lower than that the drying process. Figure 5 shows the from the specimen with pure water. temporal change of the average moisture If wc > 0, the water content is not suffi- content w in the specimen that cont- cient to dissolve all the salt crystals both ained salt water, calculated from the dry in the specimen and at the surface of the References mass of the specimen and the mass of specimen. In Figure 6, wc > 0 after 2 days water in the specimen, and the content indicates that salt crystals still surely re- 1 H. Derluyn, M. Griffa, D. Mannes, I. Jer- of precipitated salt estimated based on w main during absorption, which can lower jen, J. Dewanckele, P. Vontobel, A. Shep- and the initial salt content; Figure 6 does the vapor absorption rate. pard, D. Derome, V. Cnudde, E. Lehmann, the same during wetting. In Figure 5, the In our current work, we only conducted J. Carmeliet, Characterizing saline upta- data for the specimen that contained measurements under limited conditions, ke and salt distributions in porous li- pure water are also plotted. Although a but future work should address more dif- mestone with neutron radiography and rapid change of the evaporation rate for ferent conditions to enable to develop a X-ray microtpmgraphy, Journal of Buil- pure water is observed at approximately model that can predict cycles of crystal- ding Physics 36 (2013) 353-374. 2 days, this would be mainly due to the lization and deliquescence of salt on and typical transition from a capillary -dri- near the specimen surface as a function Figure 6: The average moisture content w, the estimated 2 I. Daher, Salt Transport Experiments in ven to a vapor-driven drying and partly of the type of a porous material and salt, precipitated salt content wc and the average height have of salt on the specimen surface for the wetting process of Fractured Media, PhD thesis, Imperial due to a large change of the humidity the initial salt concentration, the ambi- 8 days. have is defined to be 0 at 1.86 days. College London, 2016.

46 47 SWBSS 2017 | 20-22 September M. Abuku et al. 4th International Conference on Salt Weathering of Buildings and Stone Sculptures

3 C. J. Graham, A petrographic investi- Student report, Delft University of Tech- Diagnostics and monitoring of moisture and salt in gation into the durability of common nology, 2012. replacement sandstones to the crys- porous materials by evanescent field dielectrometry tallisation of de-icing salts, PhD thesis, 10 M. N. Rad, N. Shokri, Nonlinear effects University of Glasgow, 2016. of salt concentrations on evaporation Cristiano Riminesi*1 and R. Olmi2 1 from porous media, Geophysical Rese- Institute for the Conservation and Valorization of Cultural Heritage (ICVBC), National Research 4 Council, Firenze, Italy N. Aly, M. Gomez-Heras, A. Hamed, M. arch Letters 39 (2012) L04403. 2 Institute of Applied Physics (IFAC), National Research Council, Firenze, Italy Alvarez de Buergo, F. Soliman, Porosi- *[email protected] ty changes after different temperature 11 M. N. Rad, N. Shokri, M. Sahimi, Po- regimes for a salt weathering simu- re-scale dynamics of salt precipitation Abstract 1. Introduction and research aims lation test on Mokkattam limestone in drying porous media, Physical Re- (Egypt), In: Proceedings of the 3rd In- view E 88 (2013) 032404. Moisture and salts are the main causes In this contribution the authors descri- ternational Conference on Salt Weathe- of decay of porous materials, like wall be the Evanescent-Field Dielectrometry ring of Buildings and Stone Sculptures 12 M. N. Rad, Pore-scale investigation of paintings, stones, plasters and cement-ba- (EFD) system called SUSI©* for determi- (SWBSS2014), Brussels, Belgium, Octo- salt precipitation during evaporation sed artefacts. Water is the ‚driving force‘ ning the water content and detecting the ber 14-16, 2014, 247-257. from porous media, PhD thesis, Univer- of decay, such as the detachment of the presence of soluble salts inside porous sity of Manchester, 2014. painted layer, the whitening of surfaces materials, and present its application. 5 S. Gupta, H. P. Huinink, M. Prat, L. Pel, due to the crystallization of salts (efflore- The water and salt content are key issues K. Kopinga, Paradoxical drying of a fi- 13 S. Veran-Tissoires, M. Marcoux, M. Prat, scence), and the weakening of the cemen- for the maintenance of works of art, such red-clay brick due to salt crystallizati- Discrete salt crystallization at the sur- ting binder. Early diagnostics of water as wall paintings, stone artefacts, mason- on, Chemical Engineering Science 109 face of a porous medium, Physical Re- content and detection of the presence of ry and cement-based artefacts. (2014) 204-211. view Letters 108 (2012) 054502. soluble salts inside the material is a key The SUSI© system is an electromagnetic issue for understanding the degradation (EM) diagnostics method based on the 6 L. Pel, S. Gupta, Paradoxical drying due 14 S. Veran-Tissoires, M. Marcoux, M. Prat, processes in such kind of materials and existence of the relation between phy- to salt crystallization; the effect of fer- Salt crystallization at the surface of a for improving their schedule maintenan- sical characteristics, in particular water rocyanide, In: Proceedings of the 3rd heterogeneous porous medium, Euro- ce. In this contribution a non-invasive content, and the complex permittivity of International Conference on Salt We- physics Letters 98 (2012) 34005. microwave system based on evanescent hygroscopic porous materials.1-6 athering of Buildings and Stone Sculp- field dielectrometry is described. The EM diagnostics methods can be use- tures (SWBSS2014), Brussels, Belgium, 15 S. Veran-Tissoires, M. Prat, Evaporation method was tested in the laboratory on fully employed for such tasks, when ap- October 14-16, 2014, 77-87. of a sodium chloride solution from a moistened plaster samples, some of them plied with non-destructive modalities. saturated porous medium with efflore- containing salts at different concentra- Among the EM tools, dielectric spectros- 7 S. Dai, H. Shin, J. C. Santamarina, For- scence formation, Journal of Fluid Me- tions. Measurements on water-saturated copy appears as an eligible technique, i. e. mation and development of salt crusts chanics 749 (2014) 701-749. and oven-dry samples provide the basis the study of the spectral response of the on soil surfaces, Acta Geotechnica 11 for calibrating the instrument for on-si- dielectric permittivity of porous mate- (2016) 1103-1109. 16 D. Ogura, S. Hokoi, T. Shimizu, H. No- te measurement of masonry structures, rials can help in determining their sta- guchi. Influence of hysteresis in sorp- wall paintings and concrete historical te of conservation. In particular, thanks 8 M. Dueñas Velasco, P. Duru, M. Marcoux, tion isotherm and moisture conducti- buildings too. The obtained results pro- to the dielectric contrast between water M. Prat, Efflorescence fairy ring and salt vity on condensation and evaporation ve the usefulness of the method as a tool and dry material, the presence of mois- centripetal colonization at the surface processes, Journal of Environmental for diagnostics and for monitoring the ture is easily detectable. Moreover, ionic of a drying porous medium containing Engineering (Transactions of AIJ) 643 effectiveness and durability of restoring conductivity – related to the presence of a salt solution. Impact on drying cur- (2009) 1065-1074. interventions. salts in solution – is measured as well. ve., In: Proceedings of the 4th European A system for dielectric spectroscopy is Drying Conference (EuroDrying‘2013), Keywords: sub-surface investigation, generally too complex and too expensive Paris, France, October 2-4, 2013, 1-9. moisture and salt content, dielectrome- to be used for real-time monitoring. Nar- try, resonant technique, SUSI© system, row-bandwidth, resonant dielectrometry 9 F. R. Janmahomed, Salt crystallization at plaster, stone, concrete. is a viable solution, because it allows to the surface of consolidated porous me- realize portable, low-weight instruments dia as determined by microCT imaging, independently measuring the moisture and salt content. Humidity content mea-

* Italian acronym for Sensore per la misura di Umidità e Salinità Integrato, integrated sensor for measuring humidity 48 and salinity 49 C. Riminesi and R. Olmi Diagnostics and monitoring of moisture and salt in porous materials by evanescent field dielectrometry

surements based on EFD techniques have ty of a desalination treatment on a wall Therefore, the „calibration“ parameters been developed for the diagnostics of painting, we have to monitor the residu-  and β relative to a material can simply wall paintings, masonry and cement-ba- al quantity of salts in the substrate. The be obtained by two measurements, on a sed materials.7-11 availability of a non-invasive and porta- dry sample and on a water-saturated one. In particular, water inside building ma- ble tool for sub-surface investigation of The salinity index SI is related to the terials is known to be the main source moisture content and for detecting the quality factor Q of the resonant probe. Q of damage in masonry for the following presence of salts, like the SUSI system, can be simplified as the sum of the re- reasons: can be of great help in the preservation ciprocals of an „unloaded“ term, depen- and quality control of the treatment. ding on the geometry and on the elec- - water diffusion inside masonry and its The SUSI system has been successfully tric/dielectric properties of the materials evaporation from the surface can de- used in the past for the following appli- constituting the microstrip cavity, and teriorate the finish of the surface, and cations: of a „loaded“ term depending only on in the presence of a wall painting may the dielectric properties of the material degrade the painted layer or the prepa- (1) – preliminary screening of state of facing the sensor. Operatively, SI is com- Figure 2: Principle of the SUSI© system ration layer – detachment and loss of conservation of the support - maps puted by using the following expression: cohesion; of moisture content and salts content can be achieved for focusing the MC and SI related to the material in con- - the water diffusion can induce the gro- sampling9; (2) wing of microorganisms on the exposed tact with the probe are calculated in real time by means of the numerical code. surfaces that induce biodeterioration; (2) – monitoring the effectiveness and The measurement principle is shown in where fr is the resonance frequen- durability of restoration or main- Figure 2. The normalized S21 measured - or, if the support contains salts, a high 10 cy measured on the material, f0 is that tenance interventions ; -1 risk of whitening and salts crystallization the material (dashed line) allows to in air, and ∆(Q ) is the variation of the on could appear as water diffusion af- calculate the shift of the resonance fre- reciprocal of Q among air and material (3) – giving insight about the absorption fects the surface. quency (∆f) with respect to the response conditions. dynamics of water-based products in air (solid line), and the width Lw at 3 dB SI has a relatively simple link with in areas to be treated during the A sub-surface diagnostic can reduce or the quality factor Q of the bell-shaped the loss tangent (tan δ) of the material. restoration phase.8 and prevent such kinds of risks. The same curve. The resonance frequency ranges Following a definition for SI used in soil 12 result could be obtained by sampling the between 0.9-1.5 GHz for stone materials. science , in terms of the dependence of The probed volume directly depends support in depth, but this practice can’t material electrical conductivity (σc) on its 2. Materials and methods on the size and geometry of the sensor; be applied on surface or material with dielectric constant (ε‘c) artistic value (wall paintings, monumen- in this case the investigated volume is tal stones, etc.). For example, if we want The SUSI is based on EFD technique approximately a semi-spherical volume to assess the effectiveness and durabili- that operates in the microwave range. of 2 cm radius within the material. (3) The diagnostics parameters achieved by The diagnostic parameters (MC and SI) this instrumentation are the moisture are directly computed from the measu- content (MC) and the Salinity Index (SI) red quantities ∆f and Q, by resorting to and using the relation between con- related to the presence and amount of a calibration procedure.8 The dielectric ductivity and dielectric losses, the follo- salts. A photo of the system is shown in permittivity, if needed, can be obtained wing expression for SI, in terms of the Figure 1. It consists of a two-port reso- offline by an inversion procedure invol- saturation degree S of the porous media, nant probe developed for measurements ving the development in orthogonal mo- for a given porosity φ (the water content on solid materials, connected to a scalar des at the coaxial opening.7 is φS), is obtained: network analyser (SNA) for measuring The MC parameter has been demons- the sensor response (the S21 scattering trated to be linearly related to ∆f 8: parameter).7 The system is completed by (4) a numerical code, running on a PC. (1) The probe (on the bottom left side) is an open resonant cavity able to check the with K an arbitrary constant. Eq. (4) Figure 1: The SUSI© system (US Patent Specification 7,560,937 B2) material without damaging it.7 Both, the states that SI includes a linear term in

50 51 Diagnostics and monitoring of moisture and salt in porous materials by evanescent field dielectrometry C. Riminesi and R. Olmi

the tan δ and a function of tan δ, ε‘ and S. the water content, but SI values relative c Name Description Photo The relation between (4) and (2) can be to different material conductivities are obtained by specifying a dielectric model very well separated. for the material under investigation. In The measured SI, computed by eq. (2), M1 Size: 4x4x16 cm terms of a simple three-phase model13, can be easily demonstrated to be related using the so called Complex Refractive to the dielectric properties as in (4), with Index Method (CRIM), the composite ma- a suitable choice of the K constant. terial is represented as a mineral/water/ Composition: Portland air mixture with a complex permittivity M2 CEM I cement, binder:sand given by: 3. Results (weight ratio) : 1:3

3.1. Calibration procedure Size: 4x4x16 cm (5) Composition: Hydraulic The calibration procedure is adopted lime, binder:sand (weight to relate the measured MC with a water P1 ratio) : 1:3 where ε is the permittivity of the so- content calculated by gravimetric appro- Size: 4x4x16 cm m Composition: Natural lid matrix, ε and ε are those of air and ach. In this regard, on each type of poro- a w prompt cement, no sand water, respectively. Although developed us material we can apply the following for one-dimensional layered structures, procedure: Table 1: Cement mortar samples made in laboratory the CRIM model can be demonstrated to satisfy the Hashin & Shtrikman bounds14 1. Samples in ambient conditions are and, therefore, to be a coherent model weighted and measured by SUSI sys- also for more complex structures. tem; The calibration procedure consists in 3.2. Application for diagnostics and Figure 3 shows the dependence of SI determining the parameters  and β of monitoring eq. (1), relating the SUSI-measured un-ca- computed by (4), with ε‘c given by (5), on 2. Samples are dried in oven at 65°C for the saturation degree and on the conduc- three days, until to steady-state condi- librated MC (uMC) with the gravimetric Two applications of the SUSI system for tivity of a saline solution filling the solid tion, then weighted and measured by MC (MCg) obtained on a dry basis. diagnostics and monitoring are briefly matrix, for a frequency of 1 GHz and for a SUSI system; Figure 4 shows the calibration curve described in this sub-section: porosity φ = 20%. The SI is arbitrarily nor- for sample M1 (Table 1). We observe an malized, choosing a value for the K cons- 3. Samples are water saturated in a vacu- uneven spreading of uMC values among - Mapping of the MC and SI on the wall tant in (4), such as to assume a full-sa- um cell with inner pressure of about samples at low water content. This de- painting of St. Clement at mass and the turation value of 10 for a conductivity 2-3 kPa, in order to obtain maximum pends on the characteristics of the sur- legend of Sisinnius in the St. Clement 9 of 10 S/m. We observe that the SI slightly filling of pores accessible to water, face of the material, less density on the Basilica, Rome (Italy) ; depends on the saturation degree, i. e. on then weighted and measured by SUSI surface than in the bulk, due to a no-ho- system. mogeneity associated with the sample - The assessment of the effectiveness of preparation. This affects the water con- an extractive poultice of salts from the The SUSI measurements are taken tent in different way in the surface res- wall paintings in the Allori’s loggia in when the sample weight reaches stea- pect to the bulk. the Pitti Palace in Firenze (Italy).10 dy-state conditions. At least two cycles of For any given type of porous materials, drying/wetting should be performed on the parameters  and β can be implemen- About the first application, the St. Cle- each sample. Steps 2 and 3 are functional ted in the calibration section of the soft- ment Basilica is built on three different to obtaining the calibration parameters, ware managing the instrument. levels. The middle level is located below  and β of eq. (1), for each class of mate- The SI measurement does not requi- the road level at about 12 m. At this level rials. re any calibration. Actually, the salinity are located several precious ancient wall The calibration procedure for cement index is assumed only as a semi-quanti- paintings. These wall paintings are sub- mortar samples is described here as an tative measure of the salt content, as we ject to hard environmental conditions: example. Table 1 summarizes the cement are not able to distinguish among salt high relative humidity of the ambient air mortar samples made in laboratory and species only on the basis of the dielectric (about 90% – constant during the year - Figure 3: SI versus the saturation degree and the water conductivity used in this study. measurement. with a temperature ranging from 12°C to

52 53 C. Riminesi and R. Olmi Diagnostics and monitoring of moisture and salt in porous materials by evanescent field dielectrometry

Figure 4: Gravimetric MC (MCg) versus uMC for sample M1

Figure 7: The wall painting of the Allori’s small loggia in Figure 8: Plots of Salinity Index for two kinds of extractive the Pitti Palace. poultices

Firenze are summarized. Figure 7 shows 4. Conclusions the portion of the wall painting where the poultices were applied. On the area The SUSI system based on EFD labelled with A, a cellulose pulp (bc1000 technique can be employed for in situ di- and bc 200) was applied; on area B the agnostics of buildings materials, in par- mixing of cellulose pulp, sepiolite clay ticular for diagnosis and monitoring of and sand in different ratios (6 parts of the presence of water salts up to 2 cm in Figure 6: Maps of the distribution of (a) moisture content and (b) of the salinity index arbocel bw 40/6 parts of sepiolite clay/6 depth (for the probe used in this work). parts of sand was applied); and, on area C The resonant approach guarantees an the Cocoon® by Westox was applied. The excellent sensitivity on the measurement The measurements were performed on effectiveness of each poultice was evalua- of the moisture content and also allows the points indicated on Figure 5, keeping ted by SUSI measurements applying the to simultaneously and independently de- the sensor in contact with the surface for following protocols: tect the presence of soluble salts. A rough less than 30 s (the time necessary for the quantification of the salt concentration data acquisition). The maps, in terms of a. preliminary measurement by SUSI sys- is made based on a Salinity Index, empi- MC and SI, are obtained by interpolating tem before the application of extracti- rically introduced and theoretically justi-

Figure 5: St. Clement at mass and the legend of Sisinnius the collected data. ve poultices on the selected area on a fied based on a simple 3-phase model of wall painting. The solid red framework indicates the area SUSI measurements showed that the regular grid; material. The measurement system, cali- investigated by SUSI system and circles represent the points where measurements were performed wall painting exhibited a decreasing gra- brated on samples with proper characte- dient of MC from the bottom to the top b. new measurements after from the re- ristics (plaster, stone, concrete, etc.), is a (until 1.50 m in height) of the investigated moval of extractive poultices. promising technique to be applied in dia- 22°C), and capillary rise of water from the area (Figure 6a), ranging from 3.5% (red) gnosis and monitoring of materials used ground below.15 On the wall paintings, to 8% (blue). The SI was higher in the first Figure 8 shows the results regarding in historical buildings and artworks. the typical degradation processes in- meter from the floor (Figure 6b), ranging two kinds of poultice: cellulose pulp and duced by rising damp were present: whi- between a minimum of 1.5 (negligible Cocoon. The Cocoon poultice demons- tening of the painted layer, salt crystalli- salts) in the upper part of the investiga- trates a better effectiveness after 7 days zation, biological microorganism attacks ted area to a maximum of 5.5 (quite high with respect to the cellulose pulp, while (algae, bacteria, etc.). In particular, on the salt concentration) in the lower part. the poultice made by mixing cellulose St. Clement at mass and the legend of Si- As regards the latest application, the pulp and sand didn’t provide good re- sinnius, wide areas of salt efflorescence results of tests on the wall paintings at sults. and thick encrustations were present. the Allori’s loggia in the Pitti Palace in

54 55 SWBSS 2017 | 20-22 September C. Riminesi and R. Olmi 4th International Conference on Salt Weathering of Buildings and Stone Sculptures

5. References 10 R. Manganelli Del Fà, et al., “Non-des- Determination of the water uptake and drying behavior tructive testing to perform service of 1 A. Kraszewsk, Microwave Aquametry the evaluation of conservation works”, of masonry using a non-destructive method (New York: IEEE), 3-34, 1996. Proc. of the 6th Inter. Conf. on Emer- ging Technologies in Non-destructive Amelie Stahlbuhk*, Michael Niermann and Michael Steiger 1 Department of Inorganic and Applied Chemistry, University of Hamburg, Germany 2 R. Rouveure, et al., “A microwave sensor Testing (ETNDT 2016), 457-463, 2016. *[email protected] for agricultural implements”, IGARSS ’02, IEEE T Geosci Remote, 5, 3020–2, 11 C. Riminesi, et al., Moisture and salt 2002. monitoring in concrete by evanescent field dielectrometry, Meas Sci Technol, Abstract Keywords: non-destructive method, wa- 3 R.A. Yogi, et al., “Microwave microstrip 28(1), 2017. ter uptake and drying, encrustations ring resonator as a paper moisture sen- Historic masonry of different types sor: study with different grammage”, 12 M. A. Malicki, et al., „Evaluating soil sa- often suffers from various enrichments Meas. Sci. Technol., 13, 1558–62, 2002. linity status from bulk electrical con- in surface and near-surface areas. Salts, 1. Introduction ductivity and permittivity“, Eur. J. Soil atmospheric pollutants and other mate- 4 B.L. Shrestha, et al., “Prediction of mois- Sci., 50, 505-514, 1999. rials from external sources are possible Masonry of different kind is often af- ture content of alfalfa using density-in- causes for these enrichments which can fected by various types of enrichments at dependent functions of microwave die- 13 K. Roth, et al., “Calibration of time do- directly influence the water uptake and surfaces and near-surface regions. Such lectric properties”, Meas. Sci. Technol., main reflectometry for water content drying of the masonry. Thus, a compacti- masonry includes e. g. bare brickwork, 16, 1179–85, S 2005. measurement using a composite die- on of the superficial porous network can plastered walls, wall paintings and other lectric approach”, Water Resour. Res., be expected in the case of crust formati- monuments. Porous building materials 5 W. Meyer and W.M. Schilz, “Feasibility 26, 2267– 2273, 1990. on, an enhanced water uptake in the case are used for the construction of those ob- study of density-independent moisture of hygroscopic salts. Both of these lead jects since antiquity. Due to their surface measurement with microwaves”, IEEE 14 Z. Hashin and S. Shtrikman, „A varia- to a change of the intrinsic and unaf- and their porous network they are direc- Trans. Microw. Theory Tech., 29, 732–9, tional approach to the elastic behavior fected water balance. Besides damaging tly exposed to the atmosphere. Having 1981. of multiphase minerals“, J. Mech. Phys. effects, possibly provoked by crusts, and an additional connection to the ground Solids, 11(2), 127-140, 1963. elevated salt contents, it is of interest to water, atmospheric and anthropogenic 6 S. Trabelsi and S.O. Nelson, “Density-in- investigate how affected walls behave at influences can act on them. Enrichments dependent functions for on-line micro- 15 M. Coladonato, et al., “Environmental a given relative humidity, e. g. threshold can be present in form of salts, atmo- wave moisture meters: a general discus- study for the safety of frescos in hypog- values for enhanced water uptake or re- spheric pollutants and other materials sion”, Meas. Sci. Technol., 9, 570–8, 1998. eous site: the relationship between the lease influenced by salts. The present stu- from external sources. Salts are mostly water quantity and the soluble salts dy reports on a non-destructive chamber introduced via capillary rise of ground 7 R. Olmi, et al., “Diagnostics and moni- into the masonry”, Proc. 6th Internat. method used to investigate the impact of water or anthropogenic effects like usa- toring of frescoes using EFD”, Meas. Sci. Conf. Non-destructive Testing. Micro- encrustations and salts on the masonry’s ge of de-icing salt, agriculture, animal Technol., 17, 2281-88, 2006. anal. Methods and Environ. Eval. Study behavior concerning drying by determi- excrements and cleaning or conservative and Conserv. Works of Art, Rome, 1996. ning moisture flow. Advantages of the actions.1 Also anthropogenic emissions, 8 R. Olmi and C. Riminesi, “Study of wa- chamber are its easy and non-destructive e. g. caused by the combustion of oil and ter mass transfer dynamics in frescoes application on the masonry using a sea- coal, lead to the deposition of pollutants by dielectric spectroscopy”, Il Nuovo Ci- ling material that is solely pressed on the on the porous objects which are a sour- mento, SIF, 2008. wall without leaving residues and the use ce for salts as well, especially for the of harmless water vapor. An investigati- formation of gypsum.2 In addition, cle- 9 V. Di Tullio, et al., “Non-destructive map- on of wall paintings in the cloister of the anings and conservative procedures can ping of dampness and salts in degraded St. Peter Cathedral in Schleswig which in introduce other materials like resins or wall paintings in hypogeous buildings: parts suffer from massive yellowish en- fixatives whose ageing products can also the case of St. Clement at mass fresco crustations is presented to demonstrate enrich over the course of years.3,4 Sulfur in St. Clement Basilica, Rome”, Anal Bio- the promising results obtained with the dioxide pollution was a problem in in- anal Chem, 396, 1885–96, 2010. chamber method. dustrial areas until the 1990s. With the distribution of exhaust gas purification for power plants and a change in lifestyle

56 57 A. Stahlbuhk et al. Determination of the water uptake and drying behavior of masonry using a non-destructive method

towards a declining use of coal in dome- ke and drying behavior of masonry, thus, stic heating the problem was reduced, at they may change the natural water ba- least in Europe.5,6 Regarding the formati- lance of the considered object. While hy-

on of crusts, gypsum as a result of SO2 groscopic salts enhance the water upta- impact, might be the most popular one.7,8 ke, the formation of superficial crusts of The processes of water uptake and re- slightly soluble salts or non-saline depo- lease in a porous medium in contact with sits result in a compaction of superfici- its surroundings are described by the al pores leading to a hindrance of water sorption equilibrium. As a function of sorption. Another effect that leads to an the materials’ pore size distribution, its impairment of the natural water balance water content and the ambient tempe- of porous materials is hydrophobing. In rature and relative humidity (RH) water this case the pores are not blocked but vapor may be adsorbed from or released hydrophobic agents coat the pore walls to the atmosphere. Salts in the pores of and the material surface, thereby affec- Figure 1: Schematic diagrams of the chamber used for the measurements of moisture flow, showing a front view (left) and a 3D representation (right). Numbers represent (1) cylindrical chamber, (2), (3) inlet and (5), (6) outlet ports for the building materials not only have a high ting water uptake and drying. air flow, (4) external power supply for the ventilating fan and (7) positions for the RH/T sensors at the entrance and the damage potential due to the generation In this work we report on the appli- exit of the chamber. of crystallization pressure.9 Salts also cation of a chamber method10 that was influence the water uptake and drying originally developed to investigate the behavior of such porous matter. Each deposition of air pollutants such as SO investigation using sealing strips (Te- such that typical volume fluxes were 3.9 2 –1 salt has a characteristic, temperature onto material surface.11,12 In contrast to samoll P-profile) which can be removed L·min . Incoming air was dried in a si- dependent value of the relative humi- the deposition of atmospheric pollutants, without residue after the measurements. lica gel drying tube positioned between dity – the deliquescence humidity – at the water sorption is a reversible process. An adjustable air flow with controlled membrane pump and chamber entrance. which it starts to pick up water from the Nonetheless, the chamber method is also relative humidity passes the chamber For other applications, the drying tower atmosphere to form a saturated soluti- suitable for the observation of moisture through the inlet (2, 3) and outlet ports may be replaced with a humidification on. Regarding salts in building materi- flows across the material surface in both (5, 6), respectively. The chamber is venti- system using saturated salt solutions als, the hygroscopic behavior of the salt directions, i. e. hygroscopic water uptake lated using a fan that is installed on the to control the relative humidity at the and that of the material both influence and evaporation during drying. Ques- lid and connected to the external power chamber entrance. The flow rate was water uptake and release. Very hygrosco- tions concerning the behavior during supply via inlet (4). Ventilation is neces- measured using a gas meter connected sary to achieve ideal mixing in the cham- to the chamber exit (6). pic salts, like nitrocalcite, Ca(NO3)2·4H2O, drying of masonry containing salts or with a high solubility take up high other foreign substances and threshold ber, thus, to avoid influences of Brownian During the measurements the only amounts of water also at moderate rela- values for augmented water uptake and diffusion on the overall moisture exch- data captured are relative humidity and tive humidities2, theoretically, leading to release due to salts may be examined ange rates. In a well-mixed chamber, the temperature at the entrance and the exit a saturation of the pore space with salt with this non-destructive method. relative humidity at the exit equals the of the chamber. Known quantities are 3 −1 solution and consequently to a much hig- RH inside the chamber and transport of the volumetric flow rate Q (in m ·h ) of 2 her water content within the masonry water vapor to the surface or evaporati- the air flow and the surface area A (in m ) than in the salt free material at the same 2. Experimental on from the surface are the rate limiting of the wall that is exposed to the cylin- relative humidity. On the other hand processes. The base plate also carries two drical chamber during the experiment. salts with an extremely low solubility 2.1. Chamber method RH/T sensors (Hygroclip, Rotronic Mess- The relative humidity φ is defined as geräte GmbH) that are used to monitor like gypsum, CaSO4∙2H2O, pick up water φ = p /p (1) vapor only at extremely high humidities For the measurement of moisture flow continuously the relative humidity and w w,0 (>99.9% RH). Furthermore the behavior a chamber as shown in Figure 1 was temperature both at the entrance and where p and p are the water vapor concerning water uptake and release is used. The cylindrical chamber (1) made the exit of the chamber (7). The two sen- w w,0 more complicated in the case of salt mix- of POM (polyoxymethylene) and PMMA sors (±0.8 % RH and ±0.2 K) are conected pressure and the saturation water vapor tures like they are commonly present in (polymethylmethacrylate) with a diame- to a Hygrolab C1 station (Rotronic Mess- pressure, respectively. The latter was cal- 13 real objects. ter of 100 mm is mounted on a POM-base geräte GmbH). culated using the Wagner Pruss equation . Next to the enrichment of salts also plate with dimensions 300x150 mm2. In the experiments that were carried The absolute humidity f is given by: other materials of various origins accu- For the measurements, the plate with a out on site, the drying behavior of the f = p M /(RT) (2) mulated on the material surface may cylindrical aperture on the back side is masonry was investigated. The airflow w w have a direct influence on the water upta- gently pressed against the wall under was generated by a membrane pump

58 59 A. Stahlbuhk et al. Determination of the water uptake and drying behavior of masonry using a non-destructive method

where Mw is the molecular mass of water dissolved most of the time which is also mals (area 2) a patchy yellowish encrusta- that, the results from areas 2 and 3 with (0.0180153 kg∙mol−1), R is the gas constant the reason for the high moisture cont- tion was clearly visible. The same situati- patchy encrustations show a significant- (R = 8.3144 J∙mol−1∙K−1) and T is the absolu- ent of the masonry. Even though success on was noted in area 3, while in area 4 on ly reduced water release. For test area 4 te temperature. was noted in the last years concerning the upper right side of the wall painting a the moisture flow is even smaller and The difference of the absolute humidi- salt reduction and gypsum conversion, massive and continuous crust was obser- the curves reach a plateau after less than

ties at the exit (fout) and at the entrance the removal of yellowish encrustations ved. To all these areas the chamber was one minute of drying which means that

(fin), i.e. Δf = fout−fin, is the result of either on parts of the bays was not effective so applied to measure the moisture flow du- there is no further release of water from evaporation (Δf>0) or water uptake from far. These inhomogeneous encrustations ring drying, implemented by passing a the wall. the air (Δf<0). The moisture flux F in units most likely comprise organic materials dry airflow through the chamber. Two or The curves representing replicate mea- of g·m−2·h−1 is then given by: which are probably alteration products three replicate measurements were con- surements in the same areas show some of organic substances introduced in for- ducted on all areas except area 3. variability which might reflect minor F = Δf Q/A (3) mer restoration campaigns. With their shifts of the chamber position on the massive appearance it was expected that wall between two measurements. This For visual presentations the mass of they would not only affect the object in 3. Results and discussion includes e.g. a higher or lower extent of water, evaporated during drying, (or the an aesthetic manner but also the natural encrustations. Different initial moisture mass of water, adsorbed from the atmo- water balance of the walls by clogging su- The results of the multiple measure- contents of the exposed parts of the wall sphere in case of water uptake, respecti- perficial pores. ments at the different positions on the might also be an important cause of the vely) per unit area (in g∙m−2) will be used The chamber method was used on wall are represented in Figure 3. In all observed differences. Every measure- which is obtained by multiplying the different parts of the most affected bay measurements an increase of the mass ment causes a reduction of the moisture moisture flow rate with the correspon- no. 5 which is depicted in Figure 2. The flow with time is observed which de- content at a test position due to partial ding time (in h). Hence, the transported area under the frieze is not pigmented monstrates the ongoing release of water drying. If there is not sufficient time for quantity of water vapor in g·m−2 is depic- and comprises a fresh and even lime vapor from the wall to the atmosphere. re-humidification between two consecu- ted at any point of the measurement. plaster which does not show any visible The curves representing the measure- tive measurements, the initial moisture encrustations (area 1). Within the frieze ments in the freshly limed area 1 show content is lower in the next run. Nevert- which is decorated with paintings of ani- the highest mass flow. Compared to heless, the curves belonging to the same 2.2. Cloister of St. Peter Cathedral

In this study, wall paintings in the cloister of the St. Peter Cathedral in Schles- wig, Northern Germany, were investigated. The so-called “Schwahl” is furnished with wall paintings from the 14th century showing monochrome scenes of Christ’s life in 22 bays and polychrome paintings of apostles, mythical creatures and leaf tendrils in the vaults and on the courtyard-sided walls. The restauration history of this object started quite early and involves several interventions including revisions, whitewashes and reparations which are not completely recorded in the years before the late 19th century 14. Regarding the state of conservation the walls of the object showed high contents of nitrates and chlorides in most parts, gypsum enrichments on the surface and damage in the form of scaling and

sometimes loss of pigmented areas. Due Figure 2: Bay no. 5, cloister of St. Peter Cathedral in to a very humid climate the salts remain Schleswig. Measurement areas 1 to 4 are depicted. Figure 3: Evaporated water versus time curves for the four areas in bay no. 5 (see Figure 2).

60 61 A. Stahlbuhk et al. Determination of the water uptake and drying behavior of masonry using a non-destructive method

measurement position or to areas with of measurement leading to variations of areas adjacent to the crusts causing a servations potentially critical situations a similar state of the surface can be cle- the duration of the first drying stage. The risk. Also flaking of the crust is a pos- for the object considered as a consequen- arly grouped together and show similar decreasing slope of the curves of areas 2 sible scenario e.g. in the case of heavy ce of hindered water vapor transport and drying behavior. The relative drying ra- and 4 represents the decreased drying accumulations and small cracks within accumulation of salts can be derived. tes observed for these groups seem to be rate at the beginning of and during the the dense layer which allows drying of These promising results induce other plausible as it was found that an area wi- second drying stage. In case of area 1, the underlying wall and crystallization interesting applications of the chamber thout any visible encrustation shows sig- a change of the slope is not clearly ap- of salts. The chamber method is a useful method on real objects like, for example nificantly higher drying rates than areas parent, consequently assuming that the tool to detect and validate such critical an investigation of the hygroscopic water covered with patchy or even extensive first drying stage was not terminated circumstances on the object. The main uptake of walls contaminated with salts, crusts. during the duration of the experiment advantage for the use on sensitive ob- or, the monitoring of the dynamics of the Before the results will be discussed, seems reasonable. Without hindrance in jects is the non-destructive character en- drying and water uptake behavior in re- the two distinct drying stages of porous form of a crust like in area 1, water vapor abled by a sealing strip which is normal- sponse to daily or seasonal cycles of the materials may be discussed. During the is transported uninhibitedly to the sur- ly applied for the sealing of windows and climatic conditions. first stage climatic conditions, air flow face and the drying front can recede into doors. Limitations result for very uneven rate and surface texture are the limiting the material in the second stage. If crusts or bent walls where it is not possible to factors. Capillary moisture transport to represent an obstacle (see area 4), capil- seal the room between chamber aperture Acknowledgments the drying front at the surface provides lary transport to the surface is hindered and wall for level differences greater more moisture than the amount that can as it was already mentioned in the prece- than the height of the strip. However, Financial support of this research by evaporate, as climatic and transition con- ding section. From microscopic images of bare brickwork and other non-decorated the Deutsche Bundesstiftung Umwelt is ditions limit evaporation in this stage. samples from bay no. 5 it is known that walls might be investigated by using ano- gratefully acknowledged. Consequently, these conditions determi- the yellowish crusts cover parts of the ther sealant (Terostat IX, Loctite, Henkel) ne the slope of the curve (see Figure 3) render and block the pores to a large ex- which adheres to the wall but can be re- while the material inherent properties, tend as they comprise a dense layer. Thus, moved without residues as well. e. g. the moisture transport properties, it can be assumed that pore clogging is influence the duration of the first stage. not a consequence of salt crystallization With decreasing moisture content within close to the surface induced by drying 4. Conclusions References the material the liquid flow to the surfa- but an effect caused by (most likely orga- ce declines. As less water is transported nic) deposits on the wall paintings which A non-destructive chamber method 1 A. Arnold, K. Zehnder, Monitoring wall to the surface, the drying rate decreases were already present before the experi- was successfully used to study the influ- paintings affected by soluble salts. In: and the drying front shifts into the ma- ments. In case of area 4, the course of the ence of encrustations on wall paintings S. Cather (Ed.), The conservation of wall terial.15, 16, 17 Regarding the specific case of mass loss curve reaches a plateau after on the drying behavior of the wall. Ori- paintings. Proceedings of a symposium salt contaminated materials, as the pore only 20 seconds, suggesting that only the ginally, the chamber method was used to organized by the Courtauld Institute of solution is transported to the surface condensed water on the surface of the determine the deposition of atmospheric Art and the Getty Conservation Institu- due to capillary forces, salts crystallize crust evaporated during drying. pollutants to material surfaces. In the te, London, July 13-16, 1987. Getty Con- and accumulate on the surface or in near These observations demonstrate that present study, a simplified version of the servation Institute, Los Angeles, 1996, surface regions. Hence, they provoke the simple setup with the flow cham- chamber was applied in an investigation 103–135. pore clogging, hindering capillary trans- ber yields useful and promising results of water release from wall paintings. It port to the surface and slowing down the concerning the water transport across was shown that complete coverage with 2 M. Steiger, A. E. Charola, K. Streflinger, drying rate.18 the surfaces of encrusted wall paintings. encrustations hinders the drying process Weathering and deterioration. In: S. Sie- The curves shown in Figure 3 are qui- Besides, this hindered water vapor trans- of the render and the underlying wall by gesmund, R. Snethlage (Eds.), Stone in te similar during the first seconds with port caused by the encrustations has clogging superficial pores and blocking architecture. Springer, Berlin, Heidel- a nearly linear initial increase, i. e. a another consequence for the affected ob- the passage for the intrinsic water vapor berg, 2014, 225–315. constant drying rate, characteristic of jects which could pose a risk. Dissolved exchange between masonry and atmo- the first drying stage. The limiting con- salts within the porous network tend to sphere. In addition, it was found that 3 P. Mora, L. Mora, P. Philippot (Eds.), ditions at this stage were quite similar in crystallize on the surface when the rela- the extent of encrustation controls the Butterworths series in conservation all experiments thus, it is not surprising tive humidity decreases to form subflore- drying behavior of the surfaces. Patchy and museology, Butterworths, London, that equal initial drying rates are obser- scences or efflorescences. If parts of the and irregular crusts show mass flow cur- 1984. ved. In contrast, the material characteri- surface are covered with impermeable ves between unaffected areas and fully stics might differ between the positions crusts, salts accumulate in uncovered covered areas. With these accessible ob-

62 63 SWBSS 2017 | 20-22 September A. Stahlbuhk et al. 4th International Conference on Salt Weathering of Buildings and Stone Sculptures

4 L. Borgioli, P. Cremonesi, Le resine sin- nes: A chamber study at very low SO 2 Measurement of salt solution uptake in fired clay tetiche usate nel trattamento di opere concentration, Environ. Earth Sci. (69) policrome, Il prato, Saonara (Padova), (2013), 1125–1134. brick and identification of solution diffusivity 2005. 13 W. Wagner, A. Pruss, International Etsuko Mizutani1*, D. Ogura1, T. Ishizaki2, M. Abuku3 and J. Sasaki2 5 P. Brimblecombe, Air pollution and ar- equations for the saturation properties 1 Graduate School of Engineering, Kyoto University, Kyoto, Japan 2 Tohoku University of Art and Design, Yamagata, Japan chitecture: Past, present and future, J. of ordinary water substance. Revised 3 Faculty of Architecture, Kindai University, Higashi-Osaka, Japan Architect. Conserv. (6) (2000), 30–46. according to the international tempe- * [email protected] rature scale of 1990, J. Phys. Chem. Ref. 6 S. J. Smith, J. van Aardenne, Z. Klimont, Data (22) (1993), 783–797. Abstract 1. Introduction R. J. Andres, A. Volke, S. Delgado Arias, Anthropogenic sulfur dioxide emissi- 14 B. Löffler-Dreyer, Von Truthähnen Salt solution uptake rates in fired clay Salt weathering is one of the main cau- ons: 1850-2005, Atmos. Chem. Phys. (11) und Füchsen – zur Restaurier-ungsge- brick were measured with γ-ray to in- ses of the deterioration of brick masonry (2011), 1101–1116. schichte der Wand- und Gewölbema- vestigate the influence of salt on soluti- walls, which can be found at the Hagia lereien im Schwahl des Schleswiger on diffusivity for appropriate numerical Sophia (Aya Sophia Museum) in Istan- 7 M. Steiger, Salts and crusts. In: P. Brim- Doms, DenkMal! Zeitschrift für Denk- analysis of salt solution transfer and bul (see photo 1). There are many sour- blecombe (Ed.), The effects of air pollu- malpflege in Schleswig-Holstein (21) crystallization in porous materials. NaCl ces of salt in the environment, including

tion on the built environment. Imperial (2014), 98–105. and Na2SO4 are the main salts confirmed ground water, atmospheric deposition, College Press, London, 2003, 133–182. in Hagia Sophia (Aya Sophia Museum), and originally contaminated salts in the 15 C. Hall, W.D. Hoff, M.R. Nixon, Water Istanbul, Turkey, and are used in this materials. In Hagia Sophia, atmospheric 8 M. Steiger, Air pollution damage to movement in porous building mate- experiment. We identified solution dif- deposition and originally contaminated stone. In: P. Brimblecombe (Ed.), Urban rials – VI. Evaporation and drying in fusivity as a function of solution cont- salts in the materials are assumed to be pollution and changes to materials brick and block materials, Build. En- ent by a numerical analysis of solution the main sources of salt that cause we- and building surfaces. Imperial College viron. (19) (1984), 13–20. transfer. The main results are as follows. athering of the structures.1 In the Hagia Press, London, 2016, 65–101. Solution uptake tests demonstrated that Sophia case, there exist complex humid– 16 G.A. Scheffler, R. Plagge, Ein Trock- saturated solutions take approximately dry and wet–dry cycles that complicate 9 M. Steiger, Crystal growth in porous ma- nungskoeffizient für Baustoffe, Bau- six times longer to attain a steady state the deterioration process that comprises terials – I: The crystallization pressure physik (31) (2009), 125-138. than the time needed for pure water to adhesion, dissolution, and infiltration of large crystals, J. Cryst. Growth (282) obtain steady state. This was thought to processes that transfer salts from the (2005), 455–469. 17 T. Diaz Gonçalves, V. Brito, L. Pel, Water be caused by the viscosity of the dissol- sources in the environment. Therefore, vapor emission from rigid mesoporous ved salts and the presence of crystallised the process is much more complex than 10 A. C. Hill, A special purpose plant en- materials during the constant drying salt in the material. We identified solu- when ground water is the main salt sour- vironmental chamber for air pollution rate period, Drying Technol. (30) (2012), tion diffusivity by multiplying moisture ce. The objective of this study was to de- studies, J. Air Pollut. Control Assoc. (17) 462–474. diffusivity for pure water by a constant velop an analytical model of the simulta- (1967), 743–748. to reproduce the measured solution con- neous transfer of heat, moisture, and salt 18 R.M. Espinosa-Marzal, G.W. Scherer, Im- tent in the cases where large amounts and the subsequent crystallisation and 11 A. Behlen, C. Wittenburg, M. Steiger, pact of in-pore crystallization on trans- of efflorescence was observed at the ma- dissolution of the salt for reproducing W. Dannecker, Dry deposition of NO, port properties, Environ. Earth Sci. (69) terial surface. In the case of a saturated the deterioration mechanism caused by

NO2, HNO3, and PAN on historic buil- (2013), 2657-2669. solution of Na2SO4, which is considered these complex salt deterioration proces- ding stones. In: J. Riederer (Ed.), Inter- the salt that most frequently precipitates ses. national congress on deterioration and into material, it was necessary to correct Salt, which exists in the dissolved pha- conservation of stone – Proceedings. the solution diffusivity equation by con- se or solid phase (crystal) in porous ma- Möller Druck und Verlag GmbH, Berlin, sidering the effect of in-pore salt preci- terial, causes significant changes to the 1996, 377–386. pitation. material properties of the surface it con- tacts. This is due to a rise in viscosity of 12 U. Wiese, A. Behlen, M. Steiger, The in- Keywords: Solution diffusivity, γ-ray at- the solution and the descent of vapour

fluence of relative humidiy on the SO2 tenuation, fired clay brick, sodium salts pressure of the salt solution and chan- deposition velocity to building sto- ging pore structure in the solid phase.2, 3 Therefore, material properties conside-

64 65 E. Mizutani et al. Measurement of salt solution uptake in fired clay brick and identification of solution diffusivity

quantitatively measured by a γ-ray atte- Number of specimen Test 13 Test 14 Test 17 nuation apparatus. In addition, we characterised solution diffusivity as a fun- Salt NaCl Na2SO4 ction of volumetric solution content by a numerical analysis of the salt solution Saturation ratio at 20°C 100% 100% 20% transfer process. This was done to inves- Temperature (°C) Relative humidity (%) 20.5°C, 40–48% 20.5°C, 40–48% 20.5°C, 43–44% tigate the effects of dissolved salt and salt crystallisation. Table 1: Experimental conditions

2. Methodology during the uptake process was experimentally measured using γ-ray attenua- 2.1. Outline of solution uptake test tion. Volumetric solution content ψsol [m3/m3] was evaluated using γ-ray trans- A salt solution uptake test of NaCl and mission of the material during the expe-

Na2SO4 was conducted to investigate the riment in equation 1: influence of dissolved salts and crystalli-

sed salts on solution transfer. Fired clay (1) Figure 1: Schematic representation of the measurement bricks with dimensions of 3x4x10 cm set-up were used. An epoxy resin and alumi- where ρsol is the density of solution nium film were applied to the surfaces [kg/m3], μ is the mass attenuation coef- of the bricks, with the exception of the ficient [cm2/g], l is the thickness of spe- top and bottom face (3x4 cm), to provide cimen [cm], and Io and I are the intensity water and vapour proof seals and accom- of γ-ray radiation of material at dry state plish one-dimensional solution transfer. and at the time of experiment [cps], res- Specimen bricks were hung above a pure pectively. μ is the specific value of a ra- water/salt solution water bath. The bot- dioisotope, and the value of the water of tom surface was exposed to the bath. The americium is 0.2059. Prior to the experi- top surface was in contact with room air ments, the mass attenuation coefficient, only. Initially, pure water uptake tests μ of the salt solution was identified from were conducted using specimen bricks the measured values of the γ-ray trans- in a dry condition. Subsequently, salt so- mission of an acrylic water container Figure 2: Mass attenuation coefficients of the salt solutions Photo 1: Deterioration of the inner walls due to salt lution uptake tests were conducted with crystallisation at Hagia Sophia, Istanbul, Turkey filled with a salt solution. Figures 1 and the same specimen that had been dried Figure 2 show the schematic representa- again. This process allowed for a compa- tion of the measurement set-up and the rison of the uptake of water vs salt solu- mass attenuation coefficient of the salt used to identify the solution diffusivity ring these effects of salt are needed to tion. Experimental conditions during the solutions, respectively. The identified are shown. The Boltzmann transforma- predict the real phenomenon appropri- experiments, such as the saturation ratio mass attenuation coefficients of the salt tion method is also an effective method ately. of the solution, the ambient tempera- solutions, which depend on salt concen- for the determination of solution diffusi- 4, 5 In this research, we investigated the so- ture, and the room humidity are shown tration and the type of salt used, were vity when gravity can be neglected. In lution diffusivity, which is an important in Table 1. utilised to calculate the volumetric salt this research, we conducted a numerical aspect that influences the transfer rate solution content. analysis of the salt solution and moisture of dissolved salts. We conducted pure transfer that included the effect of gra- water and salt solution uptake tests on 2.2. Measurement set-up of volumetric vity.6, 7 We identified the solution diffu-

fired clay brick material. Na2SO4 and NaCl salt solution content 2.3. Fundamental equations for dissol- sivity Dsol in Eqn. (2) using time change are the main salts found in the walls of ved salt and moisture transfer of measured volumetric solution content Hagia Sophia, and they were used in this The time evolution and spatial distri- distribution. For this research, we assu- experiment. The spatial distribution of bution of the volumetric content of the The fundamental equations for dissol- med that the salt crystallisation and salt the solution during the experiment was pure water/salt solution in the material ved salt and moisture transfer which are diffusion due to the salt concentration

66 67 E. Mizutani et al. Measurement of salt solution uptake in fired clay brick and identification of solution diffusivity

gradient did not occur during the absorp- μo is the water chemical potential [J/kg], 3. Results and discussion solution uptake process and continued tion of the solution. In addition, we assu- and g is the gravitational acceleration approximately 5 days after the solution med that dissolved salt moved by advec- [m/s2]. In this paper, the literature valu- 3.1. The time evolution of water / salt uptake experiment. On the other hand, solution content distribution tion only. Under these assumptions, the es of solution density for NaCl solution in the case of the Na2SO4 saturated so- 8 mass balance of liquid phase which is and Na2SO4 solution at 20 °C are use. We lution, efflorescence was not confirmed

consists of liquid phase water and dissol- identified the solution diffusivity D , Figures 3 and 4 show the change in the during nor after the uptake test. Na SO ∆ sol 2 4 spatial distribution of volumetric soluti- ved salt ions, is expressed by equation (2): used in equation (2) as ψsol. The sorption easily crystallised within the specimen on content during TEST 13 (Left: pure wa- isotherm for the material was identified because Na2SO4 can change phase at a (2) from the literature.9 ter, Right: NaCl saturated solution) and higher in relative humidity than NaCl. TEST 14 (Left: pure water, Right: Na SO 2 4 In the case of the Na2SO4 solution with saturated solution), respectively. Since where Dψsol is the solution diffusivity in function of solution content [kg/ms], we assumed that the change in the concentration of salt solution would be very little during the solution uptake tests, constant values of the mass attenuation coefficient of the solution for each experimental condition were used to calculate the volumetric solution content. The moisture uptake rate of the specimens was variable. The specimens reached a near steady state in approximately two to three hours after the beginning the absorption of the pure water. The upta- TETS 13 NaCl sat. Figure 3: Change in the spatial distribution of the volumetric solution content used in TEST 13（Left: pure water, Right: NaCl ke rate of the saturated solutions of both saturation) NaCl and Na2SO4 were significantly smaller than that of pure water, as it took approximately 12 hours for saturated solutions to reach a steady state. In addition, the uptake rate of the saturated solutions at upper locations on the specimens (7.7 cm and 8.7 cm from the bottom) became gradually slower. This tendency was assumed to be due to gravity or salt crystallisation within the material during the uptake process. This was signifi- TEST 14 Na2SO4 sat. cant in case of Na SO . Figure 5 illustrates Figure 4: Change in the spatial distribution of the volumetric solution content used in TEST 14 (Left: pure water, Right: 2 4

Na2SO4 saturation) the change in the spatial distribution of the solution in the specimen that absor-

bed the Na2SO4 solution with a saturation ratio of 20%. The salt solution uptake rate was twice the rate of pure water, with a decreasing uptake rate at the upper locations of the specimen not seen. Photo 2 shows a comparison of salt crystallisation at the upper surface of the specimen after enough time had passed to allow TEST 17 Na SO 20% crystal growth dissipation. In the case of 2 4 the saturated solution of NaCl, salt preci- Figure 5: Change in the spatial distribution of the volumetric solution content used in TEST 17 (Left: pure water, Right: Photo 2: Salt crystallisation at the upper surface of the

Na2SO4 20%) pitation on the surface began during the specimen

68 69 E. Mizutani et al. Measurement of salt solution uptake in fired clay brick and identification of solution diffusivity

the saturation ratio of 20%, salt crystals es used to identify the solution diffusi- sation behaviour within the specimen. is required. The relationship of solution emerged at the top surface. When we con- vity in the cases of pure water and a sa- Therefore, the solution diffusion model content and λ is not able to be formalised ducted the same experiment at a lower turated solution of NaCl. The calculated including a reduction of effective porosi- as one function because the data scatte- relative humidity (about 25%), the pro- values agree with experimental values in ty due to salt crystallisation is necessary red, especially at high moisture content.

cess of crystallisation of the 20% Na2SO4 both cases. Therefore, we concluded that for reproducing the measured values in Figure 9 shows the identified solution

solution was similar to that observed in the solution diffusivity can be found by the case of the Na2SO4 saturated solution. diffusivity, Dsol as a function of solution

the Na2SO4 saturated solution, TEST 14 multiplying the moisture diffusivity of The Boltzmann data transformed with content. The influence of salt on soluti- (see Photo 2). Therefore, a small differen- pure water by a constant. Figure 7 illus- the Boltzmann variable λ=x▪t-1⁄2 obtained on diffusivity is so large that the value ce in the concentration of the solution, trates the comparison between the mea- from the measured values of the satura- for the saturated solution of NaCl is 0.17

ambient temperature, or ambient humi- sured values and the calculated values of ted Na2SO4 solution are shown in Figure times smaller than that of pure water. In

dity may affect the crystallisation beha- solution content in the case of the Na2SO4 8 as a reference to demonstrate that the addition, a higher solution concentration

viour of Na2SO4, which may increase the solution. In the case of a 20% saturated consideration of in-pore crystallisation causes smaller solution diffusivity. This solution uptake rate. solution, the calculated values of solution diffusivity can be found by multiplying the moisture diffusivity of pure wa- 3.2. Identified solution diffusivity ter by a constant. Generally, the results correspond with the measured values In this research, solution diffusivity (see Figure 7, Right). Alternatively, in the

was approximated by the exponential case of the saturated Na2SO4 solution, it C2ψsol function Dψsol=C1e , and the fixed was difficult to replicate the measured

values C1 and C2 were determined by values by simply changing the solution matching the solution uptake tests. Figu- diffusivity to that of pure water (see Fi- re 6 shows the comparison between the gure 7, Left). This discrepancy is thought measured values and the calculated valu- to be due to the difference of crystalli-

Figure 8: Boltzmann transformation of the measured solution content of the saturated Na2SO4 solution

Figure 6: Comparison between the measured and calculated values of the volumetric solution content used in TEST 13 (Left: pure water, Right: NaCl sat)

Figure 7: Comparison between the measured and calculated values of the volumetric solution content used in TEST 14 and

17 (Left: Na2SO4 sat, Right: Na2SO4 20%) Figure 9: Identified solution diffusivity Dsol as a function of solution content

70 71 E. Mizutani et al. Measurement of salt solution uptake in fired clay brick and identification of solution diffusivity

result can be explained by the rise in References 9 C. Iba, ‘A Study on Freezing-Thawing viscosity. From these results, we conclu- Processes in Porous Building Walls’, ded that the moisture diffusivity model 1 E. Mizutani, D. Ogura, T. Ishizaki, M. Ab- Master thesis of Graduate School of En- requires the consideration of both visco- uku, J. Sasaki, ‘Degradation of the wall gineering, Kyoto University, 2002. sity and in-pore crystallisation. paintings of Hagia Sophia in Istanbul’, Central European Symposium of Buil- ding Physics, Dresden, Germany, 2016. 4. Conclusions 2 R. M. Espinosa-Marzal, G. W. Scherer, We conducted a solution uptake test of ‘Impact of in-pore salt crystallization two kinds of sodium salts (NaCl, Na SO ) 2 4 on transport properties’, Environ. Earth which are believed to affect the deterio- Sci 69, 2013, 2657-2669. ration of the inner walls of Hagia Sophia. Fired clay brick was used to investigate 3 D. Ogura, M. Abuku, S. Hokoi, C. Iba, S. the influence of salt on solution diffusi- Wakiya, T. Uno, ‘Measurement of salt vity. The influence of salt on solution dif- solution uptake by ceramic brick using fusivity is significant due to its viscosity γ-ray projection’, 3rd international Con- and salt deposition characteristics. The ference on Salt Weathering of Buldings uptake rate of a salt solution is, at most, and Stone Sculptures,2013, 529-532. approximately six times slower than that of pure water. It was observed from signi- 4 J. Carmeliet, et al. ‘Determination of the ficant change of moisture behaviour that Liquid Water Diffusivity from Transient Na SO easily crystallises in the material 2 4 Moisture Transfer Experiments, Journal compared to NaCl. A numerical analysis of Thermal ENV. & BLDG. SCI., Vol. 27, of dissolved salt moisture transport was No. 4, 2004, 277-305. also conducted to identify solution diffusivity as a function of solution content. 5 L. Pel, K. Kopinga, H. Blocken ‘Moisture The solution diffusivities for the satura- transport in porous building materials’, ted solution of NaCl and the 20% satura- HERON, Vol. 41, No. 2,1996,95-105. ted solution of Na2SO4 can be found by multiplying the moisture diffusivity of 6 M. Matsumoto ‘Energy conservation in pure water by a constant including sa- heating cooling ventilating building: turation degree of solution. The solution heat and mass transfer techniques and diffusivity model takes into account the alternatives (ed. Hoeogendoorn C.J. and change due to salt crystallisation, which Afgan N.H.)’, Washington: Hemisphere is required to predict salt solution trans- Pub. Corp.: 1978, 1-45. fer with the influence of pore clogging, especially in the case of the high concen- 7 M. Abuku, S. Hokoi, S. Takada ‘Heat and tration Na SO solution. 2 4 Moisture Transfer in Cloth Considering Salt Influences Part 2-Numerical model Acknowledgements of simultaneous heat, moisture and salt transfer and analysis of moisture upta- This work was supported by JSPS KA- ke and evaporation’, Journal of heating, KENHI Grant Number 16H06363 (Grant- air-conditioning and Sanitary Enginee- in-Aid for Scientific Research), Grant rs of Japan, 131 ,2008. Number 26709043 (Grant-in-Aid for Young Scientists (A)) and Grant Number 8 O. Söhnel, P. Novotny, ‘Density of 26870897 (Grant-in-Aid for Young Scien- Aqueous Solutions of Inorganic Subs- tists (B)) tances’, Elsevier, Amsterdam, 1985.

72 73 SWBSS 2017 | 20-22 September 4th International Conference on Salt Weathering of Buildings and Stone Sculptures Local strain measurements during water imbibition in tuffeau polluted by gypsum

Local strain measurements during water a nerated in the laboratory with the precipitation of gypsum in depth within the imbibition in tuffeau polluted by gypsum fresh stone samples. The work presented aimed to assess the mechanism of spal- Mohamed Ahmed Hassine*, Kévin Beck, Xavier Brunetaud and Muzahim Al-Mukhtar ling generated by mechanical stresses University of Orleans, INSA-CVL, PRISME - EA4229, Orleans, France and strains developed due to imbibition *[email protected] and the link with atmospheric pollution.

Abstract the surface and the core of the stone due 2. Studied stone and methodology to water infiltration demonstrate unu- The research presented in this commu- sual behaviour not restricted to expan- Tests were carried out on tuffeau, the nication aimed to evaluate the mecha- sion alone: a local contraction zone and stone used as building material in the nism of spalling generated by mechani- expansion zone in each direction separa- construction of most cultural heritage cal stresses and strains developed due to tely for the stone. Pollution by gypsum buildings in the Loire Valley in France. It imbibition and the link with atmosphe- has an effect on the strain measure- is of Turonian age (Upper Cretaceous pe- ric pollution generating gypsum in the ments mainly on the first few centimet- riod between 88-92 million years ago). It stone. Tests were carried out on a French res of the stone. is mainly affected by spalling. Tuffeau is limestone, called tuffeau. Local strains composed of a major calcite phase (50%), were measured using strain gage rosettes Keywords: gypsum, spalling, local strain, a high siliceous fraction (40%: opal Cris- during water imbibition in polluted and imbibition, tuffeau tobalite-Tridymite and quartz) and a si- unpolluted samples. Three rosettes were gnificant clay content (10%: glauconites, placed on a sample at different distances smectites, illite). It is very porous (45% from the surface (1, 4, 7 cm) to measure 1. Research aim of porosity), with a bi-modal (first peak locally the strain during water infiltrati- at 8 µm; second peak at 0.01 µm) porous on. These rosettes determine the strain Limestones constitute the main con- network.4 The mechanical characteristics in three directions 0°/45°/90°. struction materials in historical monu- depend on the degree of water saturati- Results concerning the behaviour of ments and are affected by several de- on and on the orientation with respect polluted stone during imbibition are terioration mechanisms, among which to the bedding plane.5 The stone tested compared to unpolluted stone. Results spalling.1, 2 This decay is defined as a de- in this experimental campaign was ex- concerning differential strains between tachment of thick plates (1 to 3 cm thick) tracted from the Usseau quarry, located gradually formed on the surface of the b in the Vienne department, in the Cen- stone (Fig. 1). Once the plate falls off, the ter-West of France. Two cylindrical stone resulting stone surface turns to powder. samples, 40 mm in diameter and 80 mm Mineralogical analysis of the degraded in height, were cored in the direction pa- stone, throughout its depth, shows the rallel to the bedding plane (i. e. the cylin-

presence of gypsum (CaSO4.2H2O) located der axis is parallel to the bedding plane) mainly within a crack network parallel to (Fig. 2). This direction was chosen to si- 3 the surface at 1 or 2 cm depth. Therefo- Figure 2: Sample pollution process mulate a real in-situ imbibition process re, the study of the effects of pollution (a), sample of tuffeau with strain gage rosettes (b) due to rain on the stonework. on these stones is one of main aims to For each sample, three strain gage ro- understand the physico-chemical me- settes were glued on the lateral surface chanism leading to spalling. There is also this paper presents the results of an ex- of the cylindrical sample, at different a consensus on the role of water, which perimental campaign applied to a French heights: 10 mm (J1); 40 mm (J2); 70 mm is the source of physical changes and the limestone: tuffeau. The experiment con- (J3). Each strain gage rosette was compo- dissolution/ crystallization of phases. In sisted in subjecting stone samples (both sed of three strain gages, oriented at 0°, order to further our knowledge of the polluted and unpolluted) to imbibition, 45° and 90° depending on the eigenvec- reaction of stones with respect to water Figure 1: Example of spalling in tuffeau at the “Chateau and in monitoring strain at different tors of the loading, which correspond to de Chambord” (France) 3 migration and the effect of pollution, heights in the samples. Pollution was ge- the same directions as the axes of sym-

74 75 M. A. Hassine et al. Local strain measurements during water imbibition in tuffeau polluted by gypsum

metry of the sample. The rosettes are a b „KFG 120“ type from Kyowa and were glued using a cyanoacrylate glue (CC- 35A). The diameter of each rosette is 1 cm. The error of measurement is about 1E-06 m/m and the quantification limit is about 1E-08 m/m. Before testing, the samples were dried in an oven at 60°C during 92 hours. Then, they were cooled

down in a desiccator with a drying salt Figure 3: Imbibition curves of polluted and unpolluted that maintains an almost zero relative tuffeau humidity. One of the samples was subjected to pollution by gypsum inside the Figure 4: Vertical strains of unpolluted tuffeau (a) and polluted tuffeau (b) stone. Pollution was generated in the la- zed water to be imbibed into the sample boratory by artificial ageing through an by a capillary process. During the test, sion, or a stabilization. When the water by gypsum decreased the total strain by injection of dry gaseous SO2 in the po- the stone sample was subjected to imbi- rous network of the stone followed by a bition in controlled conditions (20°C and is below the rosette, the stone is locally 33%, which is significant. partial imbibition of samples with water 50% of relative humidity). contracted. The amplitude is estimated Figures 5.a and 5.b show respectively (Fig. 2). For the creation of gypsum in the at 1.6E-04 m/m which represents 28% of the results of strain measurements of porous network, the following chemical total strain. Once the water reaches the horizontal gages for unpolluted and equations were used: 3. Results rosette height, there is an abrupt exten- polluted tuffeau. H10, H40, H70 refer re- sion up to a value of about 6E-04 m/m for spectively to the vertical strains of gages First, we present the well-known im- unpolluted tuffeau and around 3.8E-04 at 10mm, 40mm and 70mm of height. SO2 + H2O → H2SO3 THEN H2SO3 + ½O2 → H2SO4 (1) bibition curves based on visual measu- m/m for polluted tuffeau. Then, a small The horizontal strains show monotonous rements of water front height and the contraction (relative to the amplitude of evolution in extension for polluted and time of capillary rise. The curve of wa- the extension) is measured, followed by unpolluted tuffeau. The same reasoning SO2 + ½O2 → SO3 THEN SO3 + H2O → H2SO4 (2) ter front height (cm) over square root of a stabilization step. These stages are pre- was adopted for the water front. When time (min½) is linear (Fig. 3). The slope sent for polluted and unpolluted tuffeau. the water front is below the rosette, stone represents an intrinsic property of the Two differences were reported between extension begins. Then, a sudden exten- CaCO3 + H2SO4 + H2O → CaSO4.2H2O + CO2 (3) material. the two states: The magnitude of total sion occurs when the water front reaches The imbibition curves show that the vertical strain was higher for unpolluted the rosette. A second stabilization plateau The first two equations (1) and (2) are hydric properties of tuffeau changed due stone (around 6E-04 m/m against 4E-04 is obtained when the water front exceeds two possible reactions to obtain sulfuric to the presence of gypsum. The slope of m/m), and the imbibition rate was hig- the rosette. Pollution has no real effect acid and then the reaction of sulfuric acid, the curve decreased from 0.97 to 0.81. her for unpolluted stone. The pollution on the total horizontal strain value. water and calcite (dissolved from the sto- The imbibition rate decreased by 16%. ne) makes it possible to obtain gypsum This is the first evidence for the effect of in the porous network inside the stone pollution. and not only at the surface (3). The injec- Figures 4.a and 4.b show respectively a b

tion of sulfur dioxide (SO2) was carried the results of strain measurements of out in the dry state in samples previously vertical gages for unpolluted and pollu- dried under vacuum for about 12 hours. ted tuffeau. V10, V40, V70 refer respec- The sample was therefore saturated with tively to the vertical strains of gages at

SO2 gas (≈ 12 hours) and after opening of 10 mm, 40 mm and 70 mm of height. the desiccator, it was subjected to partial While the water penetrates into the capillary imbibition (≈ 2 cm) of distilled stone, local vertical strains change. Du- water. Afterwards, the samples were storing the capillary imbibition and de- red in an oven at 50°C for 15 days. The pending on the relative location of the stone sample, in the dry state, was placed water with respect to the rosette, the on a plastic grid that allows deminerali- strain may be a contraction, an exten- Figure 5: Horizontal strains of unpolluted tuffeau (a) and polluted tuffeau (b)

76 77 M. A. Hassine et al. Local strain measurements during water imbibition in tuffeau polluted by gypsum

4. Discussion Then, the third stage begins when the zontal contraction perpendicular to the 1 and 2 cm, the depth of the crack would water front is above the rosette. There surface of the stone. This contraction is be around those values. Hence, the pre- As pointed out in the Results section, is a slight contraction in vertical strain restrained by the polluted area, resulting sence of gypsum and the presence of ver- water has an effect on local strain chan- and a continuity of horizontal extension. in tensile stress, situated just after the tical contraction observed on a sample ges. In vertical strains, four distinguisha- This effect is due to two phenomena: the water front, i. e. around 1 to 2 cm depth. submitted to imbibition could be a key ble stages are observed for tuffeau (pollu- continuity of capillary imbibition in ca- The amplitude of this tensile stress is not element in the clarification of the origin ted and unpolluted). pillary pores and the slower saturation negligible compared to the low tensile of spalling. Indeed, this study is the first The first stage starts when the water of small pores which decrease the satura- strength of tuffeau. This could be the ori- step for understanding the mechanisms front has not yet reached the rosette; the- tion of the capillary pores. This stage is gin of a crack parallel to the surface at 1 of spalling of tuffeau and this hypothesis re is vertical contraction and the onset restricted because of the bi-modal poro- to 2 cm depth, which corresponds to the must be supported by further tests. of horizontal extension. This behaviour sity network of tuffeau as explained by description of spalling. This work will be pursued on a second could be the consequence of the horizon- Hassine.5 The stone has the same behavi- stone, Richemont stone, used as a repla- tal extension of the imbibed zone just our even in the presence of gypsum. The cement stone during restoration work at under the rosette. Vertical contraction is final stage concerns the stabilization of 5. Conclusion the end of the 20th century. The behavior due to the Poisson effect.5 The horizontal strain in all directions, so all porosity is of the two stones will be compared in or- tensile strain is accompanied by vertical fully saturated by water. Table 1 sums up This paper evaluated the effect of der to understand the effect of gypsum compressive strain. the main phenomena affecting tuffeau pollution during water migration on the pollution during imbibition and to ex- When the water front crosses the zone stone during imbibition. strain of a French limestone, tuffeau, and plain why no degradation by spalling is of the rosette, the extension is in all di- The main interesting feature of the proposed an interpretation for the de- observed on Richemont stone. Finally, rections due to pore saturation. This phe- results concerns the effect of pollution gradation process causing spalling. The research on both tuffeau and Richemont nomenon is the expected reaction of sto- during imbibition on the local vertical local strains were measured by using stone will be extended to investigate cy- ne due to imbibition.6–8 The total vertical strain. As shown by Hassine et al.5, the strain gage rosettes. On tuffeau, which is cles of water imbibition – drying with strain of tuffeau is higher than the total effect of imbibition is not limited to mo- a homogeneous stone, the strain measu- and without gypsum pollution to better vertical strain due to the high quantity notonous extension, as was previously rements during an imbibition on several understand their long-term behavior. of clays in tuffeau (10%).7 The presence thought.7, 8 Firstly, the existence of ver- samples of unpolluted tuffeau showed of gypsum salt has an effect on the ex- tical contraction with significant ampli- reproducible results.5 The evaluation of tension of tuffeau and reduces the ma- tude may contribute to spalling. Second- strain measurements in a sample pollu- ximum strain value by 33%. In polluted ly, the presence of salt (gypsum) in the ted and unpolluted is reproducible. The sample, the gypsum which replaced the stone stiffens the skeleton and creates a tests have been repeated 3 times in each calcite acted as a local reinforcement of difference in stiffness between polluted case and the error is about 1E-06 m/m. References the stone structure. Gypsum is a soluble and unpolluted areas. Subsequently, the Gypsum does not affect the contracti- salt with a low solubility (2 g/l) but moves polluted zone (the most rigid one) crea- on phase in the vertical direction, and 1 V. Vergès-Belmin (Ed.), Illustrated glos- slightly during an imbibition. The main tes restraint stresses and a shear plane decreases the maximum vertical strain sary on stone deterioration patterns, problem during partial imbibition is the between the two zones. By comparison value by 33%. However, no change is de- (2008) http://www.icomos.org/publica- accumulation zones of gypsum due to with in situ conditions of stones in a wall tected horizontally. The imbibition ki- tions/monuments_and_sites/15/pdf/ tidal effect with the movement of water. submitted to a rain event where polluti- netics of the tuffeau are slowed down. Monuments_and_Sites_15_ISCS_Glos- In these zones, gypsum content could be on is located on the first few centimeters, The presence of vertical contraction sary_Stone.pdf. very high, and this effect of reduction of the vertical contraction observed during just above the water front can contribu- strain could be higher. the experiments corresponds to hori- te to the discussion about the cause of 2 C. Walbert, J. Eslami, A.-L. Beaucour, A. spalling. This vertical contraction may Bourges, A. Noumowe, Evolution of the correspond to horizontal contraction mechanical behaviour of limestone - No effect on contraction phase for in-situ stones in a wall exposed to subjected to freeze–thaw cycles, En- Vertical strain rain. If the displacement of the stone in viron. Earth Sci. 74 (2015) 6339–6351. - Decreases the maximum strain value by 33% the direction of depth is restrained by doi:10.1007/s12665-015-4658-2. Horizontal strain - No effect on strain changes harder stone (polluted area), it could result in significant tensile stress, able to 3 S. Janvier-Badosa, K. Beck, X. Brunet- Imbibition kinetics - Slows the imbibition rate generate a crack parallel to the surface. aud, M. Al-Mukhtar, The occurrence As the typical depth of water imbibition of gypsum in the scaling of stones at Table 1: Effect of gypsum pollution on tuffeau behaviour during imbibition during a rain event in tuffeau is between the Castle of Chambord (France), En-

78 79 SWBSS 2017 | 20-22 September M. A. Hassine et al. 4th International Conference on Salt Weathering of Buildings and Stone Sculptures

viron. Earth Sci. 71 (2014) 4751–4759. Assessment of the durability of lime renders with doi:10.1007/s12665-013-2865-2. Phase Change Material (PCM) additives against 4 A. Al-Omari, X. Brunetaud, K. Beck, M. salt crystallization Al-Mukhtar, Effect of thermal stress, condensation and freezing-thawing ac- Loucas Kyriakou, Magdalini Theodoridou* and Ioannis Ioannou tion on the degradation of stones on the Department of Civil and Environmental Engineering, University of Cyprus, Nicosia, Cyprus Castle of Chambord, France, Environ. * [email protected] Earth Sci. (2013) 1–13. doi:10.1007/ s12665-013-2782-4. Abstract when assessed both quantitatively and qualitatively following 15 full immersi- 5 Energy consumption in buildings is M.A. Hassine, K. Beck, X. Brunetaud, M. on wetting and drying cycles in Na2SO4 Al-Mukhtar, Strain changes during the mostly associated with the use of hea- solution. progress of water infiltration in tuffeau ting and cooling systems. Renders with stone, in: Int. RILEM Conf. Mater. Syst. the addition of Phase Change Materials Keywords: lime, composites, PCMs, salt Struct. Civ. Eng. Conf. Segments Hist. (PCMs) have the ability to absorb and crystallization Mason., Lyngby, 2016. release thermal energy, when the temperature changes accordingly, thus en- 6 E. Colas, J.D. Mertz, C. Thomachot-Schnei- hancing the thermal comfort and energy 1. Introduction der, V. Barbin, F. Rassineux, Influence of efficiency of buildings. Nevertheless, the the clay coating properties on the dila- performance of such renders with tradi- Buildings nowadays account for about tion behavior of sandstones, Appl. Clay tional binders lacks international expe- 40% of the global energy consumption.1 Sci. 52 (2011) 245–252. doi:10.1016/j. rimental data, especially regarding their The continuously augmenting energy clay.2011.02.026. durability against salt weathering. consumption in buildings is directly lin- This paper focuses on the effect of ked to thermal comfort, as well as to the 7 J. Berthonneau, P. Bromblet, F. Cher- different percentages of commercial growth rate of population.2 Therefore, it blanc, E. Ferrage, J.-M. Vallet, O. Grauby, microencapsulated powder PCMs on the is imperative that efficient construction The spalling decay of building bioclastic properties of hydrated and hydraulic technologies and materials are used or limestones of Provence (South East of lime renders, investigating at the same developed for the reduction of energy France): From clay minerals swelling to time the durability of the end-products consumption in the built environment hydric dilation, J. Cult. Herit. 17 (2016) against salt crystallization. The aim is and for the preservation of energy re- 53–60. doi:10.1016/j.culher.2015.05.004. to produce energy efficient and durable sources. lime-based renders for the upgrading of Conventional thermal insulating buil- 8 Dessandier D., Role of the clays in the contemporary buildings, as well as for ding materials are usually inappropriate modifications of hydric properties: con- conservation purposes. for traditional building envelopes, since sequence on the mechanism of plates The modified composites have signifi- normally they are used in thick layers exfoliation of Tuffeau., in: Proc. 9th Int. cantly lower thermal conductivity and and most of the times they cannot offer Congr. Deterior. Conserv. Stone, Venice, increased specific heat capacity at 90 the desired results without collateral hit- 2000. days after laboratory production, thus ches, such as load bearing problems.3 The confirming the great potential of PCMs application of latent heat storage is pro- in enhancing the thermal performance bably one of the most efficient methods of the aforementioned renders. Compara- for enhancing the thermal behaviour of tive tests show that the addition of PCMs buildings, since it is based on the phase has an adverse effect on the mechanical change enthalpy of a material, which can properties of the renders and a note- store heat within a temperature range.4 worthy reduction of their bulk density. Phase Change Materials (PCMs) are wi- Nevertheless, the salt crystallization re- dely known due to their ability to absorb sistance of the modified renders impro- and store energy and release it back to ves with the percentage of PCM addition, the environment, under certain condi-

80 81 L. Kyriakou et al. Assessment of the durability of lime renders with Phase Change Material (PCM) additives against salt crystallization

Additive Workability rent percentages (5%, 10% and 20% w/w tests were carried out at different curing Mixture Aggregates Binder W/B (% w/w of solids) (mm) of solids). The binder consisted of either times (28, 56 and 90 days); this paper, ho- hydrated (aerial) lime (A: CL80 supplied wever, presents only the results recorded A H by Hellenic Mining Public Co.) or natural 90 days after casting, due to page limit- REFA 3 1 - - 1.06 174 hydraulic lime (H: NHL 3.5 supplied by ation. Fresh hydrated mortars were sto- Lafarge). The aggregate fraction consis- red in a room with constant temperature PCMA5 3 1 - 5 1.00 176 ted of Latouros sand, a local calcarenite (23±5) °C and humidity (50±5%), while 11 PCMA10 3 1 - 10 1.13 174 fine aggregate with 0-2 mm particle size. hydraulic ones were cured in closed pla- A commercial PCM (Micronal DS 5038 X stic containers at constant temperature PCMA20 3 1 - 20 1.50 175 supplied by BASF) was also used; this is (23±5) °C and high humidity conditions. REFH 3 - 1 - 0.83 174 microencapsulated in powder (dry) form Mercury Intrusion Porosimetry (MIP) and has its main melting and crystalli- was carried out on bulk samples to PCMH5 3 - 1 5 0.87 176 zation peaks in the range of 24-25 °C. evaluate porosity (po), apparent density The selected binder/aggregate ratio was (p ), average pore diameter and pore size PCMH10 3 - 1 10 1.05 180 a 1:3 by weight, based on the prevalence distribution. Moreover, the capillary ab- PCMH20 3 - 1 20 1.34 176 of this ratio in ancient and traditional sorption coefficient (s) was measured on composites in Cyprus and other areas of prismatic specimens (40x40x160 mm), 12, 13 Table 1: Mix design of laboratory composites. All quantities are measured by mass. A: hydrated (aerial) lime; H: hydrau- the world. The water demand of each using water as the wetting liquid. Pris- lic lime. mixture was subsequently estimated af- matic specimens were also used for ter achieving a constant workability in testing the materials under flexural the range of 175±5 mm according to EN load (three-point bending). Compressi- tions, during the melting and solidifying This study focuses on the effects of 1015-3.14 The water to binder ratio (W/B) on strength tests were executed on the process respectively. PCMs on the performance of lime-based along with all the other mix design de- fragments of each specimen resulting While the introduction of Phase Chan- renders, and includes research on the tails are given in Table 1. from the flexural tests. The analysis of ge Materials in structural elements often durability of PCM-enhanced lime renders the temperature response of the test involves the dispersion of PCMs into the against salt weathering. Due to the lack specimens (175x50x30 mm) to heat flow matrix, encapsulation of the raw materi- of a suitable standardized methodology 2.2. Experimental tests impulses, following a standardized bulk al can prevent leakage without altering to study the salt weathering phenome- measurement in accordance with ASTM 5 the thermal energy transfer efficiency. non on renders with traditional binders 2.2.1. Physico-mechanical and thermal E1530-11 15, was used for the determi- The incorporation of microencapsulated under laboratory conditions, the concept properties nation of the thermal properties of the PCMs in composite construction materials, of full immersion in sodium sulphate renders investigated. The test was carried 6, 7, such as renders, plasters and concrete (Na2SO4) solution and subsequent drying A series of laboratory tests and analyti- out in the curing room, at a stable tempe- 8, 9 , shows the inordinate interest of the is adopted. It should be noted that this cal techniques were used for the deter- rature of 23±5 °C; this temperature is lo- research community in investigating the test is considered to be highly aggressive mination of the physical and mechani- wer than the phase change temperature potential of these additives towards the and inconsistent with natural environ- cal performance of the samples. All the of the PCM particles (25 °C). enhancement of the energy performance mental conditions, let alone the fact that of buildings. in this study it is performed on relatively Nevertheless, although PCMs have weak construction materials, such as the been used as additives in renders, the lime renders under investigation. Level Description durability of PCM-enhanced renders has not yet been investigated. Furthermore, Perfect Specimen intact additional research is needed on the ef- 2. Materials and methods Good Very minor damage or minor cracks fects of PCM addition on the physico-mechanical properties of renders, since 2.1. Materials and sample preparation Moderate Rounding of corners and several cracks or detachment of small fragments existing related data is rather inconclu- Bad Specimen with several major cracks or broken sive and contradicting (e. g. some studies Eight lime-based mixtures were desi- report a negative impact on the mecha- gned and produced in the laboratory; two Fail Specimen in pieces or disintegrated nical properties, whereas others suggest reference mixtures and six optimized 6, 10 a positive effect). ones with the addition of PCMs at diffe- Table 2: Five-level rating classification for visual inspection of the materials resistance to salt crystallization.

82 83 L. Kyriakou et al. Assessment of the durability of lime renders with Phase Change Material (PCM) additives against salt crystallization

2.2.2. Salt crystallization enhancing the thermal performance of lime-based renders.16, 17 PCM addition cau- For the assessment of the renders’ re- ses a significant reduction in the thermal sistance to salt crystallization, six speci- conductivity of the modified end-pro- mens (50x50x50 mm) from each mixture ducts, compared to the reference ones. were subjected to a maximum of 15 wet- Higher reductions correspond to the ren- ting and drying cycles using a 10% soluti- ders with the higher percentage of PCM th on of Na2SO4. The 16 cycle corresponded addition. to a full immersion of the specimens in Increase in the specific heat capacity deionised water for 24 hours and then is further recorded for the modified ren- drying until constant mass was achie- ders (Table 3); this is also proportional to ved. The test was carried out when the the PCM content in the mix design. Other samples had completed at least 90 days researchers16 also noted a positively cor- Figure 1: MIP pore size distributions for REFA & PCMA5 (left) and REFH & PCMH5 (right). of curing. The durability of each speci- relation between the specific heat capaci- men after each crystallization cycle was ty and the PCM addition. assessed based on two parameters: (i) vi- The thermal diffusivity of the PCM- relatively higher porosity values can be Regarding the capillary absorption sual inspection (qualitative evaluation) enhanced composites is notably lower attributed to the higher water demand results (Table 3), the addition of PCMs and (ii) percentage loss in mass (quan- (Table 3). As in the case of thermal con- of the PCM-enhanced mixtures in order seems to lead to significantly reduced titative evaluation). With regard to the ductivity, this decrease is related to the to attain the desired workability16, 18; this values in both the hydrated and hydrau- qualitative evaluation, a five-level rating amount of PCM in the mixture. may, in turn, be associated with the in- lic renders (over 65% and 90% respecti- classification was devised (Table 2). Higher open porosity and lower appa- creased volume fraction of solids, the fi- vely). Similar results were also recorded rent density values are observed (Table 3) neness of the microcapsules and the hy- by other researchers.19 for the mixtures with the PCM addition, drophylic nature of their polymeric wall. 3. Results and discussion compared to the reference composites (REFA and REFH); the fluctuation is gene- The results of the thermal tests (Table 3) rally proportional to the percentage of demonstrate the potential of PCMs in the PCM additives in the mix design. The

Av. Pore Mixture p p s FS UCS λ Cp a o a Diam. (mm/ (102J/ (10-7 (%) (g/cm3) (nm) (MPa) (W/mK) min1/2) kgK) m2/s)

REFA 30.8 1.65 300.3 2.01 1.32 2.21 0.575 8.147 4.040

PCMA5 32.1 1.53 231.9 1.19 1.03 3.31 0.452 9.200 3.110

PCMA10 34.8 1.51 133.7 0.88 0.91 3.08 0.387 9.663 2.690

PCMA20 33.1 1.44 76.9 0.64 0.72 3.22 0.316 10.361 2.250

REFH 28.8 1.66 69.8 0.82 2.85 10.76 0.687 8.336 4.740

PCMH5 35.9 1.57 90.5 0.06 2.28 8.70 0.570 9.351 3.970

PCMH10 38.1 1.34 108.3 0.23 1.36 5.75 0.426 10.395 2.970

PCMH20 39.0 1.32 96.3 0.39 1.32 3.92 0.309 10.760 2.230

Table 3: Physico-mechanical and thermal properties of laboratory composites measured at 90 days after casting. po: open porosity, pa: apparent density, s: water capillary absorption coefficient, FS: flexural strength, UCS: uniaxial compressive strength, λ: thermal conductivity, Cp: specific heat capacity, a: thermal diffusivity. Figure 2: Weight loss of specimens after salt crystallization cycles.

84 85 L. Kyriakou et al. Assessment of the durability of lime renders with Phase Change Material (PCM) additives against salt crystallization

The pore size distributions recorded by flexural strength values are also recor- compared to hydrated lime). At the same capacity of the modified composites. MIP (Figure 1) show changes in the pore ded in the case of hydrated lime renders, time, PCM addition in the case of hydra- The modification in thermal properties structure of the mixtures with and wi- compared to the reference materials (Ta- ted lime renders seems to lead to better is proportionally linked to the amount thout PCMs. More specifically, the PCM ble 3). These observations are in line with durability results; it is worth noting that, of PCMs added to the mixture. This de- addition causes a shift of the main peak other researches reporting on similar in the case of hydraulic composites, the monstrates the potential of PCMs in en- towards larger pore sizes in comparison lime-based renders17. It is worth noting PCM-enhanced specimens completed the hancing the thermal performance of li- with the reference mixtures; in the hy- that, despite the fact that PCM addition durability test without any weight loss. It me-based renders. drated mixtures, the main peak shifts leads to a decrease in flexural strength, is therefore evident that better durability The relatively higher open porosity from ca. 500 to 900 nm (REFA and PCMA5 it is found to enhance the compressive results correspond to the PCM-enhanced and lower apparent density values of the respectively) and in the hydraulic ones strength in the case of hydrated lime renders, when compared to the referen- PCM-enhanced mixtures are associated from ca. 100 to 430 nm (REFH and PCMH5 mortars. The latter may be attributed to ce mixtures. This is found to be in good with the fact that mixtures with higher respectively). Furthermore, in the case of the decrease of the bigger pore volumes agreement with the materials’ capillary PCM content require higher amounts of hydrated lime-based renders, the PCM recorded by MIP (Figure 1). absorption results, as well as with their water to reach the desired workability. addition causes a decrease of the larger The determination of the renders’ re- high values of porosity and the shift of Nevertheless, PCM addition significantly

pore volumes (i. e. peaks at ca. 9000 and sistance to Na2SO4 attack through quan- the main peak of pore volume towards reduced the capillary absorption coef- 60000 nm). Other researchers10 reported titative evaluation is given in Figure 2, bigger pore sizes (see Table 3 and Figure 1); ficient in both hydrated and hydraulic similar changes in the pore structure of whereas the qualitative assessment of this may have led to less damage linked renders. Noticeable changes in the pore PCM-enhanced hydrated lime composites. their condition following salt crystalliz- to high crystallization pressures due to size distribution were also reported sin- Generally lower flexural and compres- ation is presented in Figure 3. crystal growth within the pores.20 It is ce PCM-modified renders were characte- sive strengths are reported for the hy- Only six mixtures survived all 15 cycles worth noting that the increase obser- rized by a shift of the main peak of the draulic mixtures with the addition of of the artificial weathering test (PCMA5, ved in salt crystallization resistance is in pore size distribution towards bigger size PCMs, compared to the reference compo- PCMA20, REFH, PCMH5, PCMH10, constrast to the decrease reported in fle- pores. sites (Table 3). This agrees well with the PCMH20). Their noteworthy resistance to xural strength; the latter is possibly attri- PCM enhancement had in general an higher porosity values observed and may salt crystallization is primarily attribu- buted to localised flaws in the specimens inverse effect on the mechanical probe attributed to the higher water demand ted to the binder nature (hydraulic lime containing PCMs, however further inves- perties of the end products. However, of the PCM-enhanced mixtures. Lower is associated with enhanced durability tigation using enhanced microscopy (e. g. PCM addition benefited the compressi- SEM) is needed to confirm this. ve strength in the case of hydrated lime The positive effects of PCM addition on mortars; this can be attributed to the the durability of the lime renders inves- decrease of pore volumes in the relati- tigated in this study is confirmed by the vely larger pore sizes. outcomes of the visual inspection (Figure The alterations in several physical 3). All PCM-enhanced hydrated renders values (e. g. lower capillary absorption) show a better performance against salt due to the addition of PCMs led to noti- weathering, compared to the reference. ceable changes in the salt crystallization In the case of hydraulic renders, and des- resistance of the mixtures. Better dura- pite the fact that the mixture without bility results corresponded to the modi- PCMs (REFH) survived until the end of fied composites when compared with the the crystallization test, the addition of reference mixtures. This is probably due PCMs is associated with efficient damage to the reduced amount of salt absorbed prevention, since all relevant specimens owing to the low capillary absorption of (PCMH5, PCMH10, PCMH20) are found in- the PCM-enhanced composites; a note- tact after the end of the test. worthy increase of the salt crystallization resistance may also be attributed to the hydraulic binder. Visual inspection 4. Conclusions of the samples studied confirmed the enhanced performance of PCM modified PCMs significantly reduced the ther- renders.

Figure 3: Qualitative evaluation of salt resistance by visual inspection of render mixtures. For explanation on the scale mal conductivity and thermal diffusivi- The results of this research are overall used see Table 2. ty, while they increased the specific heat positive in terms of producing innovati-

86 87 L. Kyriakou et al. Assessment of the durability of lime renders with Phase Change Material (PCM) additives against salt crystallization

ve thermally upgraded, yet durable, lime chitecture, Energy Procedia, (97), (2016), 13 A. MOROPOULOU, A. BAKOLAS, A. renders. Therefore their incorporation in 539-545. ANAGNOSTOPOULOU, Composite ma- the construction industry of southern terials in ancient structures, Cement European countries for upgrading the 7 C. CARBONARO, Y. CASCONE, S. FAN- and Concrete Composites, (27), (2005), energy efficiency of contemporary buil- TUCCI, V. SERRA, M. PERINO, M. DUTTO, 295-300. dings and for conservation purposes is Energy assessment of a PCM-embedded deemed promising. plaster: embodied energy versus ope- 14 European Committee for Standardizati- rational energy, Energy Procedia, (78), on. Methods of test fot mortar for ma- (2015), 3210-3215. sonry. Determination of consistence of fresh mortar (by flow table), EN 1015-3, 8 M. KHERADMAND, M. AZENHA, J. AGUI- 2007. AR, J. CASTRO-GOMEZ, Experimental References and numerical studies of hybrid PCM 15 ASTM International. Standard Test embedded in plastering mortar for en- Method for Evaluationg the Resistan- 1 L. YANG, H. YAN, J. LAM, Thermal com- hanced thermal behaviour of buildings, ce to Thermal Transmission of Mate- fort and building energy consumption Energy, (94), (2016), 250-261. rials by the Guarded Heat Flow Meter implications – A review, Applied Energy, Technique. ASTM E1530-11, 2016. (115), (2014), 164-173. 9 A. JAYALATH, R. SAN NICOLAS, M. SOFI, R. SHANKS, T. NGO, L. AYE, P. MENDIS, 16 L. VENTOLÁ, M. VENDRELL, P. GIRAL- 2 M. LACHHEB, Z. YOUNSI, H. NAJI, M. Properties of cementitious mortar and DEZ, Newly-designed traditional lime KARKRI, S. NASRALLAH, Thermal beha- concrete containing micro-encapsula- mortar with a phase change material as viour of a hybrid PCM/plaster: A nume- ted phase change materials, Constructi- an additive, Construction and Building rical and experimental investigation, on and Building Materials, (120), (2016), Materials, (47), (2013), 1210-1216. Applied Thermal Engineering, (111), 408-417. (2017), 49-59. 17 J. VIEIRA, L. SENFF, H. GONÇALVES, L. 10 S. S. LUCAS, V. M. FERREIRA, J. L. BAR- SILVA, V. M. FERREIRA, J. A. LABRINCHA, 3 R. BAETENS, B. JELLE, A. GUSTAVSEN, ROSO DE AGUIAR, Latent heat storage Functionalization of mortars for cont- Phase change materials for building in PCM containing mortars – Study of rolling the indoor ambient of buildings, applications: A state-of-the-art review, microstructural modifications, Energy Energy and Buildings, (70), (2014), 224- Energy and Buildings, (42), (2010), 1361- and Buildings, (66), (2013), 724-731. 236. 1368. 11 R. FOURNARI, I. IOANNOU, D. VATYLIO- 18 S. CUNHA, J. AGUIAR, PACHECO-TOR- 4 A. SHARMA, V.V. TYAGI, C.R. CHEN, D. TIS, A study of fine properties and their GAL F., Effect of temperature on mort- BUDDHI, Review on the thermal ener- effect on the quality of cementitious ars with incorporation of phase change gy storage with phase change materials composite materials, Proceedings of materials, Construction and Building and applications, Renewable and Sus- the IAEG XII Congress Engineering Materials, (98), (2015), 89-101. tainable Energy Reviews, (13), (2009), Geology for Society and Territory – 318-345. Volume 5: Urban Geology, Sustainable 19 Z. PAVLÍK, M. PAVLÍKOVÁ, V. KAULICH, Planning and Landscape Exploitation, A. TRNÍK, J. ONDRUŠKA, O. ZMĚŠKAL, 5 L. COPPOLA, D. COFFETTI, S. LORENZI, G. LOLLINO, A. MANCONI. F. GUZZETTI, R. ČERNÝ, Properties of a New Type of Cement-Based Renders Manufactured M. CULSHAW, P. BOBROWSKY, F. LUI- Plaster Containing Phase-Change Mate- with Phase-Change Materials: Applica- NO, Springer International Publishing, rial, IPCSIT, (28), (2012), 122-126. tions and Feasibility, Advances in Ma- Switzerland 2015, 33-36. terials Science and Engineering, (2016), 20 G. SCHERER, Stress form crystallization 1-6. 12 M. THEODORIDOU, I. IOANNOU, M. of salt, Cement and Concrete Research, PHILOKYPROU, New evidence of ear- (34), (2004), 1613-1624. 6 M. THEODORIDOU, L. KYRIAKOU, I. IO- ly use of artificial pozzolanic material ANNOU, PCM-enhanced lime plasters in mortars, Journal of Archaeological for vernacular and contemporary ar- Science, (40), (2013), 3263-3269.

88 89 Salt crystallization tests SWBSS 2017 | 20-22 September 4th International Conference on Salt Weathering of Buildings and Stone Sculptures Salt crystallization tests: Focus on their objective

Salt crystallization tests: Focus on their objective 2. Crystallization tests microfissures formation. Benavente et al.10 characterized material parameters

2.1. Sodium sulfate (Na SO4) of several different rocks and applied a A. Elena Charola1*, Inge Rörig-Dalgaard2, Jacek Chwast3 and Jan Elsen3 2 1 Museum Conservation Institute, Smithsonian Institution, Washington, DC, USA non-standard Na2SO4 crystallization test 2 DTU, Department of Civil Engineering, Group of Construction Materials, Lyngby, Denmark The oldest salt crystallization test to determine their resistance to it. Using 3 KU Leuven, Department of Earth and Environmental Sciences, Leuven, Belgium uses a Na2SO4 solution to impregnate a principal component analysis they cor- * [email protected] the samples and then subject them to related the resulting salt weathering to wet-drying cycles2, 3 since it can result the pore structure of the stone, its water in intense damage to the substrate. It transport properties, as well as its me- Abstract 1. Introduction was used to test durability and mecha- chanical strength finding that the latter nical resistance. The deterioration of the was the most important parameter in Many factors influence the durability It has long been known that laborato- substrate can be attributed to the crys- resisting this particular salt weathering of a building material, such as its me- ry tests have limitations and rarely can tal growth that may occur both during in agreement with previous tests carried chanical resistance, exposure conditions reproduce the deterioration induced on the drying as well as the wetting step4 out on bricks.11 It would appear that this and the presence of soluble salts in it. building materials over time. Lubelli et and that both thenardite, the anhydrous non-standard crystallization test proved Since the latter interact with each other, al. 1 have recently summarized the short- form, and mirabilite, the decahydrate, to be more useful than the accepted stan- it is difficult to relate any of them to the coming of these tests and of the mathe- can crystallize simultaneously, one at dard test. specific damage observed. Lubelli et al. 1 matical models based on the accepted the expense of the other, thus inducing have recently summarized the shortco- salt crystallization theories. repeated crystallization cycles.5 ming of some salt crystallization tests The deterioration induced by salt crys- From the study of the actual behavior 2. Sodium chloride (NaCl) and of the mathematical models based tallization has been studied for over a of salts as they crystallize out, both Püh- on the accepted salt crystallization the- century and the present paper aims to ringer 6 and Zehnder and Arnold7 show Sodium chloride, halite, is probably one ories. The net result is that there is no point out relevant issues with regards that crystal size and habit will depend of the most ubiquitous salts originating single salt crystallization test that can to the most common salts present in on the relative humidity (or moisture both from marine environments and the provide all answers since crystallization monument and structures, such as the availability) during the crystallization use of deicing salts in northern coun- kinetics, depending on specific circum- high damage inducing sodium sulfate, process and strongly related to crystalli- tries.12 This salt induces deterioration stances, play a critical role in the induced the ubiquitous sodium chloride, and the zation kinetics. such as powdering, which is more likely deterioration. Nonetheless, specific tests less damaging gypsum. Because sodium Rodriguez Navarro et al. 8 showed that to develop when isometric crystals grow, have been developed which have proved sulfate induces such fast and intense de- under normal conditions, i. e., non-equili- as observed on Venetian brick shown in to be practically viable in assessing par- terioration in materials, it has been the brium, 40% RH and 20ºC and relative fast Figure 1 left; or flaking, which in our ex- ticular material compatibility or potenti- salt most used for crystallization tests to evaporation, the anhydrous thenardite perience can be attributed to columnar al damaging sources. Two such tests are evaluate mechanical resistance (durabili- will crystallize directly from solution crystals growth as shown on another Ve- described, one using sodium chloride to ty) of materials. However, one of the pro- (although it is below the 32.4ºC thenar- netian brick sample, Figure 1 right. This determine compatibility of restoration blems with this test is that the conditions dite-mirabilite transition point) inducing particular sample was collected in Venice mortars, and another where the efflore- under which it is undertaken strongly in- significant deterioration to the substrate but only analyzed some years later. As scence of gypsum for brick masonry is fluence the results obtained.1 Therefore, and thereby also reflecting the import- shown in the photomicrograph, initially evaluated. These methods have proven results from different laboratories are ance of crystallization kinetics. Further- NaCl formed the cubic crystals, but with their reliability and lead to the conclusi- practically impossible to compare. more, no direct hydration of thenardite the change in environmental conditions on that salt tests should be designed for Meanwhile, other crystallization tests occurred, with mirabilite being produced (closed sample holder) over years, the specific objectives. have been developed for specific materi- only from the dissolution of thenardi- slow drying out resulted in the change of als, such as the sodium chloride for repla- te and its reprecipitation, as previously habit of the crystal growth following the Keywords: sodium chloride test, sodium cement mortars, and the gypsum efflore- shown 5. For materials with a relatively pattern described by Zehnder and Arnold7 sulfate test, masonry materials, crystalli- scence for evaluation of both brick as homogeneous pore system, those with and clearly showing that crystals grow zation kinetics well as mortar formulations. Tests with finer pores tend to suffer more deteriora- from the bottom up. a more focused objective have proven tion from salt crystallization than those Sodium chloride is far less damaging more useful than the generalized sodi- with larger pores. Further studies by Cul- than sodium sulfate. When tested in im- um sulfate crystallization test. trone and Sebastián9 confirmed the in- mersion/drying cycles far more cycles crease of smaller pores (<1 μm) after salt are needed to produce damage13 than

crystallization tests is possibly caused by with Na2SO4, even though studies have

92 93 A. E. Charola et al. Salt crystallization tests: Focus on their objective

An interesting example of the deteri- A more prosaic of gypsum efflorescen- oration that gypsum can induce was re- ce was found on a brick retaining wall in ported when a Romanesque mural pain- the garden at the Cowper Hewitt Muse- ting was uncovered in Austria.19 It had um as shown in Figure 2. The gypsum, been covered by a thick plaster and had probably found in the soil of the garden suffered quite some damage in the lower on the other side of the wall, migrates so- part by rising damp problems. Once un- lubilized in water (from watering plants covered, the mural paintings were des- and rain) through it and crystallizes out alinated by poulticing. The desalination as a white efflorescence at about the top extracted the more soluble NaCl whose soil level on the garden side. presence significantly increases the so- It is interesting to compare the gypsum lubility of the gypsum.20, 21 However, its crystals formed from direct crystallizati- Figure 1: Left: A brick sample collected from the S.Stefano church in Venice and analyzed some years later show the ch- removal reduces the solubility of the on with a fairly constant supply of mois- ange in crystal growth habit resulting from the changed environmental conditions. Right: Cubic halite crystals growing gypsum that started “bloom” lifting up ture, to those formed by reaction of air on brick from the Abbazia da Misericordia in Venice (ca. 1980), where brick powder can be seen being pushed away by the growing crystals. the paint layers. Therefore, further con- pollution with a calcareous material, servation measures were required to sta- such as limestone. Figure 3 shows the bilize the mural painting.19 well-formed crystal shapes that resulted shown that NaCl also tends to crystal- cycle, or number of cycles, the samples lize in smaller pores (>1μm).14 Lubelli15 are weighed and visually evaluated, presents an overview of salt weathering using the following scale: U unaltered; tests and concludes that sodium chlori- SD surface disaggregation; HD half disag- de crystallization damage is enhanced gregate; and, D disaggregated. by repeated dissolution/crystallizati- The test has been used extensively on cycles induced by fast drying of the and has proved useful, particularly for moistened sample or by changing RH evaluating restoration mortars for situa- conditions. Based on these premises, a tions where sodium chloride is the main test16 that uses this latter approach was deterioration factor.18 developed by Prof. Henriques and his team17, 18 to evaluate replacement mort-

ars for historic buildings, so as to ensure 2.3. Gypsum (CaSO4∙2H2O) that the replacement mortar will be com- Figure 2: Detail of the brick wall showing the gypsum efflorescence formed on the parallel wall below grade to that patible with both the remaining original Most of the papers dealing with seen in the garden at right. mortar and the weathered masonry ma- gypsum address the black crusts formed terial. The test is briefly described below. on calcareous stones from air pollution The test16 uses regular shaped speci- and the deterioration they caused. Fe- mens for easy visualization of any chan- wer papers deal with the efflorescence of ges. The specimens are dried at 105 ± 5°C gypsum. These are currently considered for 24 hours and weighed. Once cooled as being mainly an aesthetic problem, to room temperature they are immer- rather than the potential deteriorating sed in a supersaturated solution of so- agent it will prove to be with time. When dium chloride at 20-25ºC for 24 hours. in the 1960s the main monuments and They are left to dry for 1 hour and then buildings in Paris were cleaned of their dried at 105 ± 5°C until constant weight, black crusts using an aggressive sanding i. e., when the weight between two subse- procedure, the reaction of the conser- quent 24-hour is equal or less than 0.1% vation community was that some black of the sample weight. Once this has been crusts could be considered “protective”. It achieved, they are cycled in a climate took some years until the phenomenon chamber between dry (<40% RH) and hu- of black crust formation and its deterio- Figure 3: SEM photomicrographs of the gypsum efflorescence on brick, left (see Fig.2 left), and the framboidal growths mid (>90%RH), every 12 hours. After each rating effect was clearly understood. resulting from air pollution on limestone of the Cowper Hewitt Museum, right.

94 95 A. E. Charola et al. Salt crystallization tests: Focus on their objective

from growth in the presence of sufficient sue was first undertaken by Bowler and fect of admixtures can be evaluated alo- specific material and cannot be used to moisture7 and those formed as a result collaborators.23, 26 However, these tests ne (mixed with a gypsum solution), as a evaluate different ones. Finally, the fact from air pollution reacting with the li- had several drawbacks as discussed by component of carbonated cement paste, that salt distribution in porous materials mestone of the Cowper Hewitt Museum Chwast27 who then tried to improve the or added to the brick powder. The amount is not homogeneous29 makes this an even (see Figure 2) with limited moisture, testing procedure, as discussed below. of the tested sample is calculated to more complex situation. highlighting the importance of crystalli- A wick-test setup (see Figure 4) was ad- quantitatively reflect the mortar to brick The above considerations lead to the zation kinetics in producing significantly apted to evaluate the risk of gypsum eff- ratio in real masonry. The efflorescence conclusion that salt tests should be desi- different crystallization patterns and de- lorescence formation on brick masonry.27 appearance and thus human appreciati- gned for specific objectives, such as the grees of deterioration, as discussed else- The test uses a thoroughly desalinated on of its extent substantially depend on two test examples mentioned for sodi- where.8 The poorly crystallized gypsum brick core sample (Ø = 30mm, height = the light conditions and surface dryness. um chloride and gypsum. By focusing in the black crusts will migrate into the 62mm) which serves as the transport For this reason, high quality efflorescen- on the dominating deteriorating effect, substrate when enough moisture is avai- medium by being kept in constant cont- ce photographs are collected under well as done by Henriques16, relatively simple lable for its dissolution. Once in the sub- act with the test solution (that represents controlled conditions from dry samples tests can be elaborated that will provide surface it will re-crystallize into larger the source of the salt). Evaporation takes allowing for a reliable evaluation of the the information required. However, the and better formed crystals, thus slowly place only over the brick core sample discoloration extent by means of visual complexity of the salt induced deterio- inducing more damage to the material promoting continuous efflorescence for- and digital image analysis. ration as outlined above, suggests that it with time as described by Domaslowski.22 mation. Gypsum can be found both in Furthermore, efflorescence compositi- cannot be expected that one single test Over the past 35 years, gypsum efflore- brick23 but also in carbonating cement on is determined by XRD and HCl tests, provide all the information that would scence was found to form on new const- paste27, so to include this latter source while its effect on the wick performance be desirable. ruction in the UK, The Netherlands and carbonated cement paste, and if desi- is evaluated by a drying rate analysis.27 Belgium.23-25 The origin of this efflore- red, mortar admixtures, can be added. scence was attributed mainly to the used For this purpose, both powdered brick masonry mortar formulations, mostly and hardened cement paste are added 3. Discussion and conclusion ready-to-use based on Ordinary Portland to the solution. The sulfate and Ca++ ions Cement, and containing additives such as migrate then into the brick core and will Salts induce deterioration in materials References air-entrainers, surfactants among others. precipitate efflorescing or subflorescing when they migrate into it and crystallize The development of tests to study this is- as gypsum on the drying surface. The ef- out. Some salts induce far more damage 1 Lubelli B., van Hees R.P.J., Nijland T.G., than others, sodium sulfate being the Salt crystallization damage: how reali- prime example. Therefore it has been stic are existing ageing tests? 3rd Inter- used to evaluate material durability. But national Conference on Salt Weathering the durability of a material depends on of Buildings and Stone Sculptures, ed. the material itself as well as on the con- De Clercq H., KIK-IRPA, Royal Institute ditions under which the salt crystallizes. for Cultural Heritage, Brussels, (2014), Since the former may be heterogeneous, 259-273. as is the case for bricks28; and the latter cannot be predicted in practice, the re- 2 Schmölzer A., Zur Entstehung der Ver- sults of standard crystallization tests are witterungsskulpturen an Bausteinen. not completely reliable. Chem. Erde (10), (1936), 479–520. Sodium chloride will also induce damage, albeit at a slower rate, while gypsum 3 de Quervain F., Jenny V., Verhalten der is probably the salt that is slowest in Bausteine gegen Witterungseinflüsse in inducing deterioration. Crystallization der Schweiz. In Beiträge zur Geologie der kinetics will vary depending on environ- Schweiz, Geotechnische Serie, 30 Liefe- mental conditions; therefore, the same rung, ed. Schweizerische Geotechnische amount of salt will produce different de- Kommission, Kümmerly & Frey, Geogra- terioration patterns, which will also de- phischer Verlag, Bern, (1951), 1–66. pend on the nature of the material. This suggests that the use of a single salt test 4 Steiger M., Asmussen S., Crystallizati- Figure 4: A sketch of the optimized wick-test set up. needs to be specifically developed for a on of sodium sulfate phases in porous

96 97 A. E. Charola et al. Salt crystallization tests: Focus on their objective

materials: The phase diagram Na2SO4– Scrivener J.C., eds. University of Mel- parative evaluation of lime mortars for rious brick/mortar combinations for H2O and the generation of stress. Geo- bourne, Melbourne, (1985), 253-260. architectural conservation. Journal of mortar durability, efflorescence poten- chimica and Cosmochimica Acta (72), Cultural Heritage, (9), (2008), 338-346. tial and resistance to rain penetration. (2008), 4291–4306. 12 Lewin S.Z. and Charola A.E., The physi- Proc. British Masonry Society, (1998), cal chemistry of deteriorated brick and 19 Leitner H., The treatment of wall pain- 31–36. 5 Charola A.E., Weber J., The hydrati- its impregnation technique. In Il Matto- tings affected by salts: An interdiscipli- on-dehydration mechanism of sodium ne di Venezia. Laboratorio per lo Studio nary task as seen from a conservator’s 27 Chwast J., Gypsum efflorescence on clay sulphate. In: 7th International Congress della Dinamica delle Grande Masse del perspective. Restoration of Buildings brick masonry. Ph.D. Thesis. KU Leuven, on Stone Deterioration and Conservati- CNR and Universitá di Venezia, Venice, and Monuments, 11[6] (2005) 365-380, (2017). on, eds. Delgado Rodrigues J., Henriques (1979), 189-214. and, European Research on Cultural F., F.Telmo Jeremias F., Laborátorio Naci- Heritage 5, ARCCHIP Workshops, eds. 28 Rörig-Dalgaard I., Ottosen L.M., Hansen onal de Engenharia Civil, Lisbon, (1992), 13 Binda L., Baronio G., Charola A.E., De- Simon S., Drdácký M. Institute of The- K.K., Diffusion and electromigration in 581-590. terioration of porous materials due oretical and Applied Mechanics: Prague clay bricks influenced by differences in to salt crystallization under different (2006), 201-314. the pore system resulting from firing, 6 Pühringer J., Salt disintegration. Salt thermohygrometric conditions. I. Brick. Construction and Building Materials, migration and degradation by salt – a 5th International Congress on Stone 20 Charola A.E., J.Pühringer J., Steiger M., (27), (2012), 390-397. hypothesis. D15:1983. Swedish Coun- Deterioration and Conservation, Félix Gypsum: a review of its role in the de- cil for Building Research: Stockholm, G., ed. Presses Polytechniques Roman- terioration of building materials. En- 29 Arnold A., Zehnder K., Monitoring Wall (1983). des, Lausanne, (1985), 279-288. vironmental Geology, (52), (2007), 339- Paintings Affected by Soluble Salts. The 352. Conservation of Wall Paintings, Carter 7 Zehnder K., Arnold A., Crystal growth 14 Amadori M.L., Lazzarini L., Massa S., Il S. ed. The Getty Conservation Institute, in salt efflorescence. Journal of Crystal deterioratmento da sodio cloruro di 21 Larsen P.K., The salt decay of medie- Los Angeles, (1991), 103-135. Growth (97), (1989), 513-521. rocce compatte e porose a Venezia. La val bricks at a vault in Brarup Church, conservazione dei monumenti nel ba- Denmark. Environmental Geology, (52), 8 Rodriguez-Navarro C., Doehne E., Sebas- cino Mediterraneo. 1st International (2007), 375-383. tian E., How does sodium sulfate crys- Symposium 1989, ed. Zezza F., Grafo, tallize? Implications for the decay and Brescia, (1990), 83-89. 22 Domaslowski W. La conservation pré- testing of building materials. Cement ventive de la pierre. Musées et Monu- and Concrete Research, (30), (2000), 15 Lubelli B.A., Sodium chloride damage to ments, XVIII, UNESCO, Paris, (1982), 51- 1527-1534. porous building materials. Ph.D. Thesis. 57. Technical University of Delft, (2006). 9 Cultrone G., Sebastián E., Laboratory si- 23 Bowler G.K., Winter N.B., Investigation mulation showing the influence of salt 16 Henriques F.M.A., Pedras naturais e into causes of persistent efflorescence efflorescence on the weathering of com- artificiais. Análise de argamassas. De- on masonry. Masonry International, posite building materials. Environmen- terminação da resistência a acção dos (11), (1997), 15–18. tal Geology, (56), (2008), 729-740. cloretos. Universidade Nova de Lisboa, 24 Caparica. (1996), Fe 12. Brocken H., Nijland T.G., White efflore- 10 Benavente D, Cueto N., Martínez-Martí- scence on brick masonry and concrete masonry blocks, with special emphasis nez J., García del Cura M.A, J.C. Cañaver- 17 Henriques F.M.A., The effects of soluble on sulfate efflorescence on concrete as J.C., The influence of petrophysical salts on a vaulted ceiling. Proc. 7th Int. blocks. Construction and Building Ma- properties on the salt weathering of Congress on Deterioration and Conser- terials, (18), (2004), 315–323. porous building rocks. Environmental vation of Stone, eds. Delgado Rodrigues Geology, (52), (2007), 215–224. J., Henriques F, Telmo Jeremias F. Labo- 25 Chwast J., Todorovic J., Janssen H., Elsen ratório Nacional de Engenharia Civil, 11 Somsiri S., Zsembery S., Ferguson J.A., A J., Gypsum efflorescence on clay brick Lisbon, (1992), 1355-1362. study of pore size distributions in fired masonry: Field survey and literature clay bricks in relation to salt attack re- study. Construction and Building Mate- 18 Faria P., Henriques F.M.A., Rato V., Com- sistance. In Proc.7th International Brick rials, (85), (2015), 57-64. Masonry Conference, eds. McNeilly T., 26 Bowler G.K., Sharp R.H., Testing of va-

98 99 SWBSS 2017 | 20-22 September 4th International Conference on Salt Weathering of Buildings and Stone Sculptures Mitigating salt damage in lime-based mortars with mixed-in crystallization modifiers

Mitigating salt damage in lime-based mortars salts: sodium chloride and sodium sul- having a different effect on sodium sulfate, fate crystallization. One phase favours with mixed-in crystallization modifiers the crystallization of hydrated sodium ii) Study the modifier-salt interaction in sulfate (mirabilite) at or near saturation, Sanne J. C. Granneman1, Barbara Lubelli1* and Rob P. J. van Hees1, 2 bulk solution to elucidate the working meaning that no or lower crystallization 1 Delft University of Technology, Delft, The Netherlands mechanism and find a suitable con- pressures can develop. The other phase 2 TNO Technical Sciences, Delft, The Netherlands centration, modifies the habit of anhydrous sodium *[email protected] sulfate (thenardite) to elongated needles. iii) Test the effect of modifier addition This habit modification can be supposed Abstract 1. Introduction on mortar properties (which might to lead, thanks to a larger evaporation limit their application), and iv) Test surface, to enhanced salt transport to This paper presents the most import- Salt crystallization damage in porous the durability of additivated mortars the drying surface. Similar to ferrocyan- ant results of a research project which building materials is a ubiquitous threat in an accelerated salt weathering test. ide, this would favour the formation of focused on the use of crystallization mo- to our built cultural heritage. Low mecha- harmless efflorescence.8 difiers mixed in lime mortar to mitigate nical strength makes lime-based mortars From literature research, sodium fer- In this paper the major findings rela- salt crystallization damage. The research especially susceptible to salt damage. In rocyanide and borax (sodium tetraborate ting to the effect of modifier addition on focused on two of the most damaging restoration or renovation works, replace- decahydrate) were identified as potential mortar properties and to the durability salts, sodium chloride and sodium sulfa- ment of renders and plasters often con- modifiers for sodium chloride and sodi- of additivated mortars are reported. First, te, and suitable crystallization modifiers stitutes a large part of the total costs of um sulfate, respectively.7 Ferrocyanide is an experiment to test whether borax’ ef- (sodium ferrocyanide and borax). We re- the project. Current solutions, such as a well-known modifier for sodium chlo- fectiveness is affected by the carbonation port the major findings related to the ef- using a stronger binder or changing the ride. Therefore, this research focused on process is described. Then, the effect of fectiveness of the modifiers when mixed moisture transport properties of the the study of its effect on fresh and har- modifiers on mortar properties such as in the mortar, and the results of charac- mortar, usually have a limited resistance dened mortar properties and on its capa- workability, water absorption and drying terization of the additivated mortars in to salt decay and low compatibility with bility of mitigating salt decay in mortar. and flexural and compressive strength comparison with reference mortars. Mo- the existing materials.1, 2 Ferrocyanide could reduce crystallizati- are reported. Finally, the durability of reover, the durability of the developed Alternatively, the use of crystallization on damage in two ways: the additivated mortars with respect to mortars to salt decay is discussed, based modifiers has been proposed.3 Crystalli- salt crystallization damage is discussed on the results of an accelerated salt we- zation modifiers do not aim to alter the i) it keeps the salt longer in solution fa- based on the results of an accelerated salt athering test carried out in laboratory. material properties, but the damaging vouring transport to the drying surfa- weathering test. No major effects of the modifiers on the mechanism itself. Using crystallization ce, and fresh and hardened mortar properties modifiers in porous building materials were observed which might restrain the has gained wide research interest in the ii) it alters the crystallization habit of so- 2. Materials and Methods application of crystallization modifiers last years (see e. g.3-5). However, the use of dium chloride from cubic to dendritic, in restoration mortars. Additionally, the modifiers in a fresh mortar, thereby gi- fact which also favours transport to 2.1. Mortar characterization mortars with mixed-in modifiers showed ving the mortar “self-healing properties”, the surface and enhanced drying. a considerable improvement of the salt is relatively new. By mixing modifiers Two types of specimens were prepared: resistance when compared to reference in during mortar production, they can Both factors lead to enhanced trans- lime only specimens and mortar (lime + mortars. Considering these results an become active as soon as the damaging port to the surface, where subsequently sand) specimens. The first, used to study outlook for future research pathways is salts enter the porous material. Promi- the salt can crystallize as harmless eff- the effect of borax on sodium sulfate crys- given. sing results have already been obtained lorescence.8 Borax on the other hand is tallization, were prepared by mixing cal- in a pilot study.6 less well-known and was therefore first cium hydroxide powder (Sigma-Aldrich, Keywords: crystallization modifiers, A research project was started to studied in bulk solution experiments. ≥ 96% purity) with distilled water only or self-healing, lime mortar further study the feasibility of the use of In this research, the effect of borax ad- with water additivated with the modifier. crystallization modifiers to mitigate salt dition on solution properties and its After carbonation, a blank specimen was crystallization damage. This project had effect on crystallization of sodium sul- treated with borax solution. Then, the the following aims: fate were studied. As has been reported blank and the two borax specimens con- in Ref.9, depending on the starting contaminated with sodium sulfate solution. i) Identify suitable modifiers for two of centration of the solution, two different After drying, the specimens were broken the most abundant and damaging phases of borax can precipitate, each and the cross section was studied using

100 101 S. J. C. Granneman et al. Mitigating salt damage in lime-based mortars with mixed-in crystallization modifiers

SEM. By comparing specimens additiva- ment. Recrystallization of sodium sulfate used for rewetting, in order to replenish ted with borax prior to carbonation and was obtained only by rewetting with li- the brushed off salt. After rewetting, any those to which borax was added later on, quid water at the end of a cycle. To have a loose material was brushed off and the the effect of carbonation on the effecti- comparable test protocol, the specimens specimens started a new cycle. The brus- veness of borax as modifier of sodium with sodium chloride were also rewetted hed off material was washed and dried in sulfate crystallization was assessed. Full with liquid water after each cycle. order to separate the salt efflorescences details of this experiment can be found Before the test, the specimens were from the debris. The debris was weighed in Ref.10 contaminated with salt solution via ca- and the amount of salt calculated by the The second type of specimens, used to pillary absorption from the bottom. A difference. In total the specimens were test fresh and hardened mortar proper- precise amount of solution was used to tested for 5 cycles. Full details of the ex- ties and assess the durability towards salt ensure contamination with 1 wt% so- periment can be found in Ref.11 decay, was prepared according to EN1015- dium sulfate or 2 wt% sodium chloride 2. Commercial air lime (Supercalco90) with respect to the mortar. In total 3.46 and standard sand (EN 196-1, sieved to (reference NaCl), 3.91 (ferrocyanide), 1.77 3. Results and discussion

select the grain size between 0.25 and 1.0 (reference Na2SO4) or 1.91 (borax) gram mm, were used. The modifier was added salt was added to the specimens during 3.1. Mortar properties to the water used to prepare the speci- the complete test. These values include mens: 0.94 wt% sodium ferrocyanide and the brushed off efflorescences during The effect of borax on sodium sulfate 3.2 wt% borax were added with respect the test. For each mortar type, 3 replicas crystallization can be observed in the to the lime. Several fresh and hardened were tested. SEM pictures in Figure 2. When Fig. 2a mortar properties were tested according After each complete cycle, all speci- is compared to 2b/c, it is clear that the to standard procedures or techniques: mens were rewetted with demineralized crystal habit of sodium sulfate without workability (EN1015-3), water absorpti- water equal to the initial amount used borax is different from the crystal habit on and drying (EN1015-18), porosity and to contaminate the specimens. After in the presence of borax. If subsequently pore size distribution (Mercury Intrusion 3 cycles (63 days), salt solution (exact Figures 2b and 2c are compared, a simi- Porosimetry) and flexural and compres- amount to obtain again 1 wt% sodium lar crystal habit can be seen in both fi- sive strength (EN1015-11). Full details on sulfate or 2 wt% sodium chloride) was gures. This means that the carbonation the preparation of the mortars and the process of the mortar has no effect on testing methods can be found in Ref.10 the effectiveness of borax as modifier for sodium sulfate. With this experimental technique it is not possible to investigate 2.2. Accelerated salt weathering test the nature of the precipitated phases, but this would be interesting for future work. The salt crystallization resistance of A selection of the measured fresh and the reference and additivated mortar hardened mortar properties is summari- specimens was tested with a custom de- zed in Table 1 (additional characterizati- signed salt weathering test, shown in Fi- on results can be found in Ref.10). When gure 1, simulating circumstances found the values for additivated and reference in practice. 80 RH% is above the equili- specimens are compared, no notable dif- brium relative humidity of sodium chlo- ferences can be observed. It can therefore ride (RHeq = 75%), but below that of the be concluded that there are no negative sodium sulfate phases. Consequently, the consequences to mixing these quantities Figure 2: SEM images of lime-only specimens contamina- sodium chloride crystals will deliquesce ted with sodium sulfate. A: reference specimen; of modifiers in the mortar during pro- B: specimen with 3.2wt% borax mixed in during prepara- when the humidity goes up and recrystal- tion (thus before carbonation); C: Specimen additivated duction, as the addition of these modi- lize when the humidity goes down again. Figure 1: Temperature and RH cycles used in the accele- with borax after full carbonation of the specimen. A fiers does not negatively affect the mort- rated salt weathering test. This entire cycle was repeated clear difference in crystal habit can be observed between This ensures multiple crystallization cy- 5 times (in total 105 days). The diamonds correspond to the reference specimen and both specimens with borax. ar properties. cles for sodium chloride, a requirement wetting by capillarity at the start (day 0) or after each Contrarily, no distinction can be observed between B and 3 week cycle (day 21) at 22.9°C ± 0.3/29.1RH% ± 2.4. At C, meaning that the carbonation process has no effect for this specific salt for damage develop- day 0 and day 63 salt solution was used, the other times on the borax effectiveness as modifier of sodium sulfate demineralized water. crystallization.

102 103 S. J. C. Granneman et al. Mitigating salt damage in lime-based mortars with mixed-in crystallization modifiers

4. Conclusions Property Method Reference Ferrocyanide Borax

Workability Flow table test 170 mm 161 mm 161 mm The additivation of mortars with crystallization modifiers during production Water content - 15.95 wt% 15.14 wt% 14.59 wt% has been proposed here to mitigate salt crystallization damage in porous buil- WAC Capillary rise 8.05 7.62 7.84 [kg/m2h1/2] ding materials. Suitable crystallization modifiers for sodium chloride (sodium Density [kg/m3] Saturation at 1943 1964 1933 ferrocyanide) and sodium sulfate (borax) Open porosity atm pressure 26.7 ± 0.19 25.9 ± 0.24 27.1 ± 0.06 (%V/V) were identified to be mixed in a mortar during production. In this paper, at first Bulk density Mercury the effectiveness of borax as a modifier [g/ml] 1.977 1.971 1.961 Intrusion Open porosity 25.1 ± 0.11 25.4 ± 0.36 25.1 ± 0.55 for sodium sulfate crystallization when Porosimetry (%V/V) Figure 3: Cumulative material loss, comparison between mixed in lime was assessed and confir- reference and additivated specimens. med. In a next step, additivated mortars Tensile strength - 0.79 ± 0.11 0.85 ± 0.03 0.92 ± 0.11 were characterized and compared to refe- (N/mm2) rence mortars in order to identify poten- Compressive tial (negative) effects on fresh and harde- strength - 2.01 ± 0.33 2.08 ± 0.18 2.61 ± 0.22 ned mortar properties. None of the tested (N/mm2) properties was affected by the addition of the modifiers, meaning that there are Table 1: Fresh and hardened mortar properties of the different 1:3 lime:sand mortar mixtures. Results previously reported in 10 no contra-indications to mixing them in the mortar in the used concentrations during production. Finally, the salt crystallization re- 11 3.2. Salt durability loss, but extensive efflorescence, which sistance of the additivated mortars was developed very fast already just after assessed using a custom designed acce- During the accelerated salt weathering rewetting via capillarity and brushing lerated salt weathering test. The mortars test, the specimens were monitored both of the specimens. In Figure 6 the speci- with mixed-in modifiers showed a consi- visually and gravimetrically. The weight mens contaminated with sodium sulfate derable improvement of the salt resistan- loss of material (with respect to the mort- are compared. At the end of the test, both ce when compared to reference mortars. ar) is plotted in Figure 3, and the weight specimens show damage at the surface, Combining all these results it can be con- loss of salt is visualized in Figure 4. It is but this is in the case of the specimen cluded that additivating mortars with clear that for both salts, the reference additivated with borax considerably less crystallization modifiers during their specimens suffer considerable material than in the reference specimen. Both the production is a feasible method in order loss after 5 cycles. Contrarily, the addi- material loss and the visual observations Figure 4: Cumulative salt loss, comparison between reference and additivated specimens. to mitigate salt crystallization damage in tivated mortars show no or only minor show that both sodium chloride and so- porous building materials. material loss. The ferrocyanide stimula- dium sulfate have the potential to cau- tes efflorescence of the salt, i. e. crystalliz- se considerable damage in the reference ation outside the material, as does borax specimen. However, when the mortars amount was still present. These values 5. Outlook but to a lesser extent. This is consistent are additivated with modifiers, damage indicate that either the modifiers have le- with the hypotheses on damage reduc- does not occur or is considerably less. ached out (together with the salt efflore- The research presented in this paper tion proposed in the introduction. Figu- Two additivated specimens were des- scence) or that they are (partially) tightly shows the viability of using crystallizati- re 5 compares specimens contaminated alinated at the end of the test and the bound to the mortar structure. Modifier on modifiers to mitigate salt weathering with sodium chloride at the start and boron and iron content of the desalina- leaching is an important parameter to damage in porous building materials. end of the test. The reference specimen tion water was analysed via ICP (Induc- take into account in future applications, Although the proof-of-principle has been shows considerable surface loss at the tively Coupled Plasma) spectrometry. whereas the binding of the modifier shown on the laboratory scale, more re- end of the test. Contrarily, the specimen Approximately 10 wt% (borax) and 1 wt% could have implications for its effective- search is needed to develop the material with ferrocyanide shows no material (ferrocyanide) of the original modifier ness and working mechanism. into a commercial product, suitable for

104 105 S. J. C. Granneman et al. Mitigating salt damage in lime-based mortars with mixed-in crystallization modifiers

renovation or restoration works. Interes- terials. Journal of Cultural Heritage, 8: ting research paths to further develop 223–234, 2007. the mortar designed in this project are: 5 C. Rodriguez-Navarro and L. G. Benning. - Studying the effect of modifier in mort- Control of crystal nucleation and growth ars with different composition (e. g. ce- by additives. Elements, 9: 203–209, 2013. ment-based). 6 B. Lubelli, T. G. Nijland, R. P. J. van Hees, - Studying the speed of modifier leaching and A. Hacquebord. Effect of mixed in and if necessary developing possible so- crystallization inhibitor on resistance of lutions, e. g. encapsulation. lime-cement mortar against NaCl crystallization. Construction and Building - Assessing the effect of the identified Materials, 24: 2466–2472, 2010. modifiers on other salts and on salt mixtures and the possibility of combi- 7 S. J. C. Granneman, E. Ruiz-Agudo, B. ning different modifiers. Lubelli, R. P. J. van Hees, and C. Rodri- guez-Navarro. Study on effective mo- - Assessing the effectiveness of the de- difiers for damaging salts in mortar. veloped mortar (on test panels) in situ. In Proceedings of the 1st International Conference on Ageing of Materials and Structures, 2014. Figure 5: Comparison between reference (A/B) and specimens with mixed-in ferrocyanide (C/D), both contaminated Acknowledgements with sodium chloride. A/C show the specimens at the start of the test, B/D show the specimens ~ 15 minutes after 8 brushing after the 5th cycle. The reference specimen (B) shows sanding of the surface. The specimen with ferrocyanide S. J .C. Granneman, B. Lubelli, and R. P. J. shows no surface damage, only a large amount of efflorescence. This research has been financed by the van Hees. Mitigating salt crystallization Dutch IOP program on Self-Healing Ma- damage with mixed-in modifiers – a re- terials, under Grant number SHM012018. view. Manuscript in preparation.

9 S. J. C. Granneman, N. Shahidzadeh, B. References Lubelli, and R. P. J. van Hees. Effect of borax on the wetting properties and crys- 1 B. Lubelli, R. P. J. van Hees, and C. J. W. P. tallization behavior of sodium sulfate. Groot. Sodium chloride crystallization CrystEngComm, 19: 1106–1114, 2017. in a „salt transporting“ restoration plaster. Cement and Concrete Research, 36: 10 S. J. C. Granneman, B. Lubelli, and R. P. 1467–1474, 2006. J. van Hees. Characterization of lime mortar additivated with crystallization 2 C. Groot, R. van Hees, and T. Wijffels. modifiers. Manuscript submitted 07- Selection of plasters and renders for salt 2017. laden masonry substrates. Construction and Building Materials, 23: 1743–1750, 11 S. J .C. Granneman, B. Lubelli, and R. P. J. 2009. van Hees. Salt resistance of lime mortars additivated with crystallization mo- 3 C. Selwitz and E. Doehne. The evalua- difiers. Manuscript in preparation. tion of crystallization modifiers for controlling salt damage to limestone. 12 S. J. C. Granneman, B. Lubelli, and R. P. J. Journal of Cultural Heritage, 3: 205–216, van Hees. Mitigating salt damage in li- Figure 6: Comparison between reference (A/B) and specimens with mixed-in borax (C/D), both contaminated with sodium sulfate. A/C show the specimens at the start of the test, B/D show the specimens ~ 15 minutes after brushing 2002. me-based mortars with mixed-in crys- after the 5th cycle. The reference specimen (B) shows clear damage at the surface. The specimen with borax shows only tallization modifiers. In Proceedings of minor surface damage at the lower left corner (D). 4 B. Lubelli and R. P. J. van Hees. Effecti- the 4th WTA International PhD Sympo- veness of crystallization inhibitors in sium, 2017. preventing salt damage in building ma-

106 107 Studies for conservation issues SWBSS 2017 | 20-22 September 4th International Conference on Salt Weathering of Buildings and Stone Sculptures Efficiency of laboratory produced water repellent treatments on limestone

Efficiency of laboratory produced water gradual deposit of salts in a wall. The ma- of calcite for water, and therefore modify sonry acts as a filter system for impure its wettability. Ioannou et al.14 confirmed repellent treatments on limestone water into a structure, as the various so- the partial wettability of limestones to luble salts are drawn into the wall and water through a series of measurements 2-3 Cleopatra Charalambous and Ioannis Ioannou* then left behind. Upon changes in the of capillary absorption at different tem- Department of Civil and Environmental Engineering, University of Cyprus, Cyprus environmental conditions, salts can crys- peratures, using water and organic li- * [email protected] tallize either at the surface (efflorescen- quids. ce) or inside the pores (subflorescence) The most severe modification of cal- of a material.4-5 Crystallization of salts cite surfaces is due to the absorption of Abstract racteristics. The results suggest that all into a porous stone causes loss of cohe- carboxylic and especially fatty acids.15 treatments can permanently reduce the rence between the grain and the matrix. As quoted in the literature, the strongest A number of cultural and architectu- wettability of the stone under investiga- Weight loss, changes in the size of grains affinity for carbonate surfaces is shown ral heritage structures all over the world tion, without modifying its composition and pores, splitting and visible surface by medium-to-long chain fatty acids and are built with natural stone. Although or appearance. deterioration can therefore be produced carboxylate polymers.15 Consequently, this material is considered to be one of In order to investigate the durability by salt dissolution-crystallization cycles. surface treatments using fatty acids can the most durable geomaterials, many of the treated stone, wetting/drying cy- Thus, salt crystallization may modify the modify the wettability of calcium car- existing stone buildings and monu- cles were performed. The results provi- properties of the affected porous stone, bonate surfaces, without affecting their ments show clear evidence of decay and de strong evidence that treatment with leading to the reduction of the lifetime inherent composition. In this paper, the weathering. The deterioration of stone oleic acid positively affects the durability of a stone building or monument.6 efficiency of three laboratory produced is strongly related to the presence and of the stone under study. Consequently, The need to protect existing stonework water repellent surface treatments based movement of water within its pore net- the aforementioned surface treatment in buildings and cultural heritage sites on oleic acid is thoroughly investigated. work. Therefore, hydrophobic surface may be potentially used in practice to from decay and weathering mechanisms, treatments are usually adopted to pro- protect stone facades in buildings and such as salt crystallization, has led seve- tect existing or new stonework. Such tre- cultural heritage sites. ral scientists to use hydrophobic surface 2. Experimental Work atments, however, should not affect the treatments.7 Those treatments usually breathability of stone; else, there is a risk include alkyl silicone products (i. e. alkyl In the experimental study, three light of enhancing possible decay mechanis- Keywords: limestone, oleic acid, water siliconates, alkyl silanes, silicone resins), grey (according to Munsel Soil-Color ms, such as salt crystallization. repellency, wetting, drying water-based fluoralkylsiloxane, polydi- charts) freshly quarried Cypriot limes- Natural limestones appear to have a methyl siloxanes, phosphoric and poly- tones were used. They originate from degree of inherent water repellency. This maleic acid products.7-10 However, when the area of Agios Theodoros in Cyprus, has been confirmed through multiple 1. Introduction applied, they may restrict the breatha- which belongs to the Pachna Geological measurements of capillary absorption at bility of the treated material by suppres- Formation. Those stones are packsto- different temperatures, using water and The decay and weathering of historical sing stage I drying, thereby affecting its nes or poorly washed biosparites, with organic liquids. The measurements were stone masonries is one of the most com- microstructure by inducing pore clog- microsparry calcite and a small portion carried out on several building and deco- mon and severe problems the constructi- ging.4 Thus, suitable alternative surface of micrite as their intergrain cementing rative limestones, showing in each case on industry is facing all over the world. treatments need to be developed; these material.16 They also have a complex mi- an anomalously low water sorptivity. Although natural stone is considered to must confer hydrophobicity without af- neralogy; they show a rather high percen- This natural water repellency of limesto- be one of the most durable geomateri- fecting the breathability of the material.8 tage of calcite, with significant amounts nes was attributed to the presence of or- als, stone structures are susceptible to Limestones have an unpredictable be- of quartz, aragonite, zeolites (analcime), ganic contaminants, such as fatty acids, water-mediated decay processes, such as havior when water repellents are applied pyroxenes (augite), plagioclases (anorthi- in the pore network of the materials un- salt crystallization, induced by alternate to them.11 Furthermore, despite the fact te, andesine, orthoclase, albite) and clay der investigation. cycles of wetting and drying. that calcite surfaces are naturally hydro- minerals (chlorite, montmorillonite). Ta- In this paper, the natural water repel- Salt weathering is, in fact, considered philic12, many researchers13-14 state that ble 1 summarizes the mineralogical com- lency of Cypriot limestones is exploited to be one of the most important degra- when carbonate minerals are exposed position of the stone variety under study. to develop several water repellent surfa- dation mechanisms that a porous stone to natural environments, they acquire Initially, the sorptivity (S) of all three ce treatments, based on oleic acid. The may undergo at or near the Earth’s sur- organic contaminants, which induce to specimens was measured at different aforementioned laboratory produced face.1 The slow process of absorption of them a degree of natural resistance to temperatures using both water and or- treatments were applied on a Cypriot cal- water into a porous structure and its capillary water absorption. In fact, these ganic liquids (i. e. ethanol, 2-propanol, carenite with proven poor durability cha- subsequent evaporation may lead to the organic contaminants reduce the affinity n-heptane) in accordance with EN 1925.17

110 111 C. Charalambous and I. Ioannou Efficiency of laboratory produced water repellent treatments on limestone

Stone Petrological Colour Mineralogy 3. Results and Discussion variety family (Munsell Chart) (XRD) calcite (65-69%), quartz (6-7%), clinopyroxene 3.1. Capillary Absorption Measure- Agios 5Y 7/2 (4-5%), chlorite (4-5%), plagioclase (5-6%), ments before the Treatments Calcarenite Theodoros Light Grey K-feldspar (1-2%), aragonite (9-11%), montmorillonite and analcime (traces) The results of capillary absorption measurements before the treatments showed a linear correlation between the Table 1: General details and mineralogical composition of Agios Theodoros stone. cumulative absorption per unit surface area i and the square root of time t½, as expected. Consequently, the sorptivity S The results were plotted against (σ/η) ½, surfactant/plasticizer in renders applied (=i/t½) of each sample under study was where σ [Nm-1] is the surface tension and on limestone substrates.20 During the derived from the slopes of the respective η [Nsm-2] the viscosity of the wetting li- setting of the render, a reaction with cal- graphs. When the sorptivity values were quid at each temperature. From the gra- cium ions in limestones takes place and plotted against (σ/η)½, the data points fell phs and using equation (1), the so-called the oleate is transformed to a hydropho- into two groups, which lied on separate

intrinsic sorptivity (SI) and the water wet- bic metal soap. Because of this reaction, straight lines (Fig. 1 I, II, III). Despite the ting index (β) of each sample was estima- sodium oleate is considered a ‘Reactive fact that both the organic liquids and the ted.18 Hydrophobic Agent’. It is worth noting water sorptivity values increased linearly that sodium oleate shows no gelling ef- with (σ/η)½, the water data lied on a line fect, due to its high content of unsatura- with a lesser slope. This is in line with (1) ted fatty acids.20 The third treatment was previous relevant work14 and confirms performed using calcium oleate (i. e. the the partial wettability of limestone spe- calcium soap of oleic acid), produced by cimens, which is attributed to natural Following the initial sorptivity mea- mixing sodium oleate and calcium car- organic contamination, due to the pre- surements, the surface of each sample bonate in an aqueous solution. Once sence of a low-energy adlayer on the spe- was treated with a different in-house again, the concentration of calcium ole- cimen’s surface.14, 18 From the results of developed water repellent treatment, ai- ate solution was 0.5% w/w. The three so- Fig. 1, the water wetting indices of each ming to control its wettability. All treat- lutions were applied to the top surface specimen were calculated using equation ments were based on oleic acid (i. e. a fatty of the specimens by brushing. Following (1) (Table 2). acid with 18 carbon atoms in a chain). As this, the samples were allowed to dry at quoted in the literature19, when oleic acid room temperature. In order to observe is used as an organic additive, it can cont- the effect of the treatment, the sorptivi- 3.2. Capillary Absorption Measure- rol the nucleation and growth of calcium ty of the treated samples was again de- ments after the Treatments

carbonate (CaCO3), and it can modify the termined through capillary absorption wettability of its surface. In fact, back in experiments at different temperatures The results of the capillary absorpti- 2012, in a research carried out by Walker using water and organic liquids. on experiments after the application of et al.8, the use of oleic acid induced addi- The durability of the aforementioned the surface treatments are shown in Fi- tional hydrophobicity to the calcite sur- surface treatments was investigated by gure 2 for each specimen. A linear beha- face of the York Minister Cathedral. subjecting the test specimens to wetting/ vior between the sorptivity S and (σ/η)½ The first treatment included an etha- drying cycles. Wetting was performed is once again noted. The organic liquids nol solution of oleic acid (concentration using vacuum saturation, while drying line remains generally unchanged after 0.5% w/w). The second treatment was took place in an oven at 105oC, until all treatments (an indication that no si- performed using an aqueous solution of constant weight was reached. Ten cycles gnificant chemical or structural changes sodium oleate (concentration 0.5% w/w). of wetting/drying were performed. After have occurred in the test specimens14), Sodium oleate is the unsaturated metal each cycle, the sorptivity of the sample whereas the water line has an even lower soap of oleic acid; it has an equally high was determined through water capillary slope. This, strongly indicates that oleic Figure 1: Sorptivity values S versus (σ/η)1/2 for the three affinity for carbonated mineral surfaces absorption tests. limestones under study. (I): Agios Theodoros I, (II): Agios acid and its by-products adsorb well on and has been shown to act as a kind of Theodoros II, (III): Agios Theodoros III. the calcite surface of the samples under

112 113 C. Charalambous and I. Ioannou Efficiency of laboratory produced water repellent treatments on limestone

ce of the stone, which interacts in itself ly fell on a line of lesser slope, compared Specimen Water Wetting Indices Intrinsic with the oleate anions, thus producing a to the organic liquid data. This was attri- Treatment Sorptivity After surface After 10 cycles of calcium oleate product. Since the latter is buted to the presence of a hydrophobic -4 ½ Original (x10 mm ) treatment wetting/drying water insoluble, it remains on the CaCO3 natural organic contaminants adlayer surface, thus inducing further water re- below the specimen’s surface, inducing Oleic acid 4.83 0.26 0.004 0.03 pellency. a degree of inherent water repellency to Sodium oleate 4.74 0.19 0.10 0.04 Even in the case of oleic acid, where the stone. there appears to be an increase in the wa- The natural water repellency of the Calcium oleate 5.84 0.13 0.02 0.01 ter wetting index of the sample after the Agios Theodoros stone has been exploi- ten wetting cycles (Table 2), the surface ted through chemical modification of of the stone is still nearly water repellent the surface of three test specimens using Table 2: Water wetting indices of test specimens before (original) and after each treatment. at the end of the experimental procedure different treatments based on the use of (β=0.03). These results, therefore, provide oleic acid. All three surface treatments strong evidence that oleic acid and its (i. e. oleic acid, sodium oleate, calcium study and induce to them additional hy- by-products (sodium and calcium oleate) oleate) proved successful in further re- drophobicity. The latter is attributed to adsorb well on calcite surfaces, without ducing the water wetting index of the the deposition of the resulting Ca(C17H33 modifying their chemical/structural limestones under study. COO)2 onto the calcite (CaCO3) surface of composition or appearance (Figure 3). The efficiency of the aforementioned the test specimens. Hence, the aforementioned treatments treatments was further investigated th- may be adopted in conservation projects rough wetting/drying cycles. Data from for the protection of stone masonries. capillary absorption measurements af- 3.3. Capillary Absorption Measurements ter these cycles revealed that oleic acid after the Wetting/Drying Cycles and its salts and soaps adsorb very well 4. Conclusions on the calcite bearing surface of the li- The results of the soprtivity tests carri- mestone under study. Despite the fact ed out on each specimen after the com- The inherent water repellency of limes- that, there was an increase in the water pletion of the ten wetting/drying cycles tones originating from the Agios Theodo- wetting index of the sample treated with are also shown in Figure 2. Tests with ros area in Cyprus has been confirmed by oleic acid after wetting/drying, the stone pure organic liquids were further perfor- the experimental work presented in this retained its hydrophobicity. The other med after the completion of the wetting/ paper. When the samples were subjected two samples showed a further reduction drying procedure. to capillary absorption experiments with in their water wetting indices after ten From the results, it is evident that the water and organic liquids, at different cycles of wetting and drying. organic liquids line, and consequently temperatures, a significant differentia- This work has practical significance, the intrinsic sorptivity of the samples, tion in the respective S vs (σ/η)½ graphs since the durability of stone masonry is continues to remain unchanged. The re- was observed; the water data consistent- largely controlled by processes mediated sults also indicate that the samples were not affected by the wetting and drying cycles. In fact, in the cases of sodium and calcium oleate, there is a further significant reduction (ca. 50-60%) in the water wetting index of the samples after the wetting/drying procedure (Table 2). Whilst for sodium oleate this may be attributed to enhanced diffusion of the

treatment into the sample, for calcium Figure 2: Sorptivity values S versus (σ/η)1/2 for water and oleate the reason is not clear and needs organic liquids before and after all treatments. (○) water before treatments, (◊) organic liquids before and after all to be further investigated. treatments, (∆) water after water repellent surface tre- It is worth noting that sodium oleate atments, (□) water after 10 cycles of wetting and drying. (I): Agios Theodoros I, (II): Agios Theodoros II, (III): Agios Figure 3: Physical appearance of the stone surfaces before/after each treatment. A0: Untreated stone, A1: Oleic acid chemically adsorbs on the CaCO3 surfa- Theodoros III. treated stone, A2: Sodium oleate treated stone, A3: Calcium oleate treated stone

114 115 C. Charalambous and I. Ioannou Efficiency of laboratory produced water repellent treatments on limestone

by water, such as salt crystallization. The Bernabeu A. Ordonez S. Quantificati- 14 Ioannou I., Hoff W.D., Hall C. On the role results strongly suggest that the in-house on of salt weathering in porous stones of organic adlayers in the anomalous developed water repellent treatments using an experimental continuous par- water sorptivity of Lepine limestone. with oleic acid, sodium and calcium ole- tial immersion on method. Engineering Journal of Colloid and Interface Science ate not only manage to reduce the water Geology (59), 2001, 313-325. (279), 2004, 228-234. absorption of the treated stone, but they are also durable against wetting/drying 7 Charola A.E., Water Repellents and other 15 Zullig J.J., Morse J.W. Interaction of or- cycles. Therefore, the aforementioned “Protective” Treatments: A critical re- ganic acids with carbonate mineral sur- treatments may be potentially used in view. Proceedings of Hydrophobe III, faces in seawater and related solutions: conservation projects for the protection Hannover, 2001, 3-2001. I. Fatty acid adsorption, Geochimica et of stone facades in buildings and cultural Cosmochimica Acta (52), 1988, 1667- heritage sites. 8 Walker R., Wilson K., Lee A., Woodford 167. J., Grassian V., Baltusaitis J., Rubasinghe- ge G., Cibin G., Dent A. Preservation of 16 Modestou S., Theodoridou M., Fournari York Minister historic limestone by hy- R., Ioannou I. Physico-mechanical pro- drophobic surface coatings. Scientific perties and durability performance of Reports. (2:880), 2012, 1-5. natural building and decorative carbo- References nate stones from Cyprus. Geological 9 Thompson M., Wilkins S.J., Compton R.G., Society London, Special Publications 1 Benavente D., Garcia del Cura M.A., Fort Viles H.A., Polymer coatings to passivate (416), 2015, 145-162. R., Ordonez S. Durability estimation of calcite from acid attack: polyacrylic acid porous building stones from pore struc- and polyacrylonitrile. Journal of Colloid 17 EN 1925:1999 Natural stone test me- ture and strength. Engineering Geology and Interface Science (259), 2003, 338- thods – Determination of water absorp- (74), 2004, 113-127. 345. tion coefficient by capillarity, European Committee for Standardization 1999. 2 Benavente D., Martinez J., Cueto N., Gar- 10 Carmona P.M., Panas I., Svensson J.E., cia den Cura M.A. Salt weathering in Johansson L.G., Blanco M.T., Martinez 18 Taylor S., Hall C., Hoff W.D., Wilson M. dual-porosity building dolostones. En- S. Protective performance of two an- Partial wetting in capillary absorption gineering Geology (94), 2007, 215-226. ti-graffiti treatments towards sulfite by limestones. Journal of Colloid and and sulfate formation in SO2 polluted Interface Science (224), 2000, 351-357. 3 Rirsch E., Zhang Z. Rising damp in ma- model environment. Applied Surface sonry walls and the importance of Science (257), 2010, 852-856. 19 Zhang L., He R., Gu H. Oleic acid coating mortar properties. Construction and on the monodisperse magnetite nano- Building Materials (24), 2010, 1815-1820. 11 Charola A.E., Water repellent treat- particles. Applied Surface Science (253), ments for building stones: A practical 2006, 2611-2617. 4 Ioannou I., Hoff W.D. Water repellent in- overview. APT Bulletin (24), 1995, 10-17. fluence on salt crystallization in mason- 20 Stolz H.J. Oleochemicals – Important ry. Construction Materials (161), 2008, 12 Okayama T., Keller D.S., Luner P. The additives for building protection. ZKG 17-23. wetting of calcite surfaces. The Journal International, 2008. of Adhesion (63), 1997, 231-252. 5 Franzoni E., Bandini S., Graziani G. Ri- sing moisture, salt and electrokinetic 13 Thomas M., Clouse J.A., Longo J.M. Ad- effects in ancient masonries: From la- sorption of organic compounds on car- boratory testing to on-site monitoring. bonate minerals: 1. Model compounds Journal of Cultural Heritage (15), 2014, and their influence on mineral wetta- 112-120. bility. Chemical Geology (109), 1993, 201-213. 6 Benavente D., Garcia del Cura M.A.,

116 117 SWBSS 2017 | 20-22 September Environmental control for mitigating salt deterioration by sodium sulfate on Motomachi Stone Buddha 4th International Conference on Salt Weathering of Buildings and Stone Sculptures in Oita prefecture, Japan

Environmental control for mitigating salt deterioration mixed salt behaviour. However, it is too by sodium sulfate on Motomachi Stone Buddha in Oita complicated and controversial to discuss the matter in this paper. Therefore, this prefecture, Japan paper attempts to investigate the effects Kyoko Kiriyama1*, Soichiro Wakiya2, Nobumitsu Takatori1, Daisuke Ogura1, Masaru Abuku3 and of the improvements of the shelter at the Yohsei Kohdzuma2 Motomachi stone-cliff Buddha and aims 1 Kyoto University, Japan 2 Nara National Research Institute for Cultural Properties, Japan to consider methods to mitigate salt de- 3 Kindai University, Japan terioration, especially in sodium sulfate. * [email protected]

Abstract ter, the indoor temperature and relative 2. Background of the Motomachi humidity increased in winter. According stone-cliff Buddha This study aimed to develop an appro- to our experimental results, the deteri- Figure 1: Motomachi stone-cliff Buddha priate interior environment to mitigate oration of the Stone Buddha is reduced The Motomachi stone-cliff Buddha is the deterioration by sodium sulfate crys- highlighting the efficiency of the impro- located in Oita prefecture, Japan. It was tallization on Motomachi Stone Buddha vements. engraved onto a cliff of soft welded tuff in Oita prefecture of Japan. This Stone around the 11th to 12th centuries and was Buddha is carved on the fragile welded Keywords: salt deterioration, sodium sul- covered by a shelter which was built du- tuff and has been covered by a shelter for fate, shelter ring in the 20th century (Figure 1 and the preservation. On this site, salt crys- Figure 2). The entrance of the shelter fa- tallization, especially in winter, is a ma- ces east. Salt crystallisation especially jor deterioration factor of this Buddha. 1. Introduction in winter is the most important cause In previous studies, the crystallization of of deterioration. Powdering and scaling sodium sulfate and calcium sulfate were Salt crystallisation is one of the most occur at the surface of the stone. A pre- 1 identified by XRD. The temperature and common cause of stone deterioration. vious investigation identified thenardite, Figure 2: Exterior of the shelter relative humidity were measured both In the field of conservation , it is well mirabilite and gypsum by X-ray diffrac- indoor and outdoor. Water quantity ana- known that repeated cycles of dissoluti- tometry (XRD).8,9 Furthermore, a past lysis of the groundwater near the Stone on and recrystallization of salts inside a study showed that sulphate, calcium and Buddha has been conducted revealing a stone can provoke serious damage. Espe- sodium ions, that are the origins of then- higher solute content near the Buddha cially for sodium sulfate, with both hy- ardite, mirabilite and gypsum, are detec- than in the general groundwater in Ja- drated and anhydrous phases, is one of ted in the groundwater near the site. And pan. According to the temperature and the most dangerous salts for stone.2-5 The the content is higher than in the general humidity measured inside the shelter in solubility depends on temperature and groundwater in Japan.9 Although a drai- winter, it was suggested that the solubi- decreases as the temperature falls. The nage tunnel and well were built in order lity of sodium sulfate decreased greatly effects of temperature and relative humi- to lower the groundwater levels (Figure with dropping in interior temperature dity on in-situ salt weathering have been 3), an increase in the amount of salt was and phase change from mirabilite to investigated in literature.6,7 However, the reported after this construction. A past thenardite (sodium sulfate anhydrate) mechanism of in-situ salt weathering re- study also suggested that drop in insi- Figure 3: Cross sectional view caused the deterioration of the statue. mains controversial. de temperature in winter due to door- Hence, the shelter was improved to re- At the Motomachi stone-cliff Buddha opening of the shelter and water leaking duce ventilation frequency and block the to prevent collapse of the stone, poly- into the high-water-content stone could 3. Methods direct solar radiation in order to reduce mers have been used. But they have be causes of salt crystallisation at the Mo- evaporation and decreasing interior tem- strengthen the stone surface causing tomachi stone-cliff Buddha.9 Therefore, 3.1. Environmental research perature in winter. In this study, we have more degradation. A tunnel has been the shelter was improved by setting door conducted environmental research and dug to lower the groundwater levels and closer, double glass windows and boards In order to assess the effects on the salt crystallization experiments under limits the salt crystallization. However, to reduce ventilation frequency and to shelter improvements, the indoor tem- controlled temperature and relative hu- the issue of an ideal environment to res- block the direct solar radiation to the perature, relative humidity, ventilati- midity to assess the improvements of the train salt deterioration awaits further in- Buddha in November 2015 by the Oita- on frequency and local meteorological shelter. After improvements of the shel- vestigation. Many recent studies focus on city board of education. data have been measured before and

118 119 Environmental control for mitigating salt deterioration by sodium sulfate on Motomachi Stone Buddha K. Kiriyama et al. in Oita prefecture, Japan

3.2. Experiments the incubator. To prepare for samples C ter, which was replaced twice at an 1 hour and D, the other incubator was program- interval. The filter papers were dried at First, Aso soft welded tuff, which is med to 10°C and 75% humidity. Sample C 60°C until they reached constant weight. thesame stone of the Motomachi sto- was placed in the desiccator containing The percentage of damage was calculated ne-cliff Buddha, was selected for the ex- silica gel and sample D was placed in the by using equation (1). periment. The stone with the porosity of open desiccator. Both of these desiccators 30.4% and the density of 1.64(g/cm3) was were placed in the second incubator. The cut into cylinders (50 mm x 20 mm). Se- experiments consisted of 5 dissolution/ Percentage of damage =

cond, to saturate the samples in order to recrystallization cycles at intervals of (Mamount of residue/M drystone initial) (1) have less effect on the depth of salt dis- at least 4 days. At the beginning of the cycles, 20 ml distilled water was added Figure 4: Schematic representation of the inside with tribution at different relative humidity, location of data loggers all samples were immersed in distilled to saturate the samples. After 5 cycles, 4. Results and Discussion water for 3 days. Finally, the bottom face stone powder was gently removed from of the water-saturated samples were put the samples with a brush and collected. 4.1. Effects on the improvements of The residue was filtrated with 1 μm pore the shelter after the shelter improvements. Tempe- into a 20ml/10w%Na2SO4 solution under rature and relative humidity have been controlled temperature and relative hu- size quantitative filter papers to remove The doors closing periods are given in measured inside and outside the shelter midity (Table 1). the salt. The filter papers and the residue Figures 4a, b. Figure 4a shows the door with data loggers (HOBO Prov2) from The conditions of 20°C, 10°C, 75% RH were immersed in 50 ml of distilled wa- closing periods before improvements, November 2014 (Figure 4). The counts of and low RH were chosen to reproduce on the other hand, Figure 4b indicates door-opening/closing have been measu- field conditions in summer and winter in the state after the improvements. Before red with door-opening/closing sensors order to assess the effects of temperature a setting the door closers, the doors could (HOBO UX90 State Data Logger) from and humidity conditions on crystallizati- stay open for a few days because visitors February 2015 (Figure 4). A weather station of mirabilite and thenardite (Table 1). could keep opening and closing the doors on was installed in February 2015 which To prepare for samples A and B, an in- freely. After the improvements, the doors measured local meteorological data (tem- cubater was programmed to 20°C and became usually closed except during an perature, relative humidity, wind directi- 75% humidity. Sample A was enclosed operation periods of inside the shelter. on, wind speed, solar radiation and pres- in a desiccator containing silica gel, used Therefore, the results indicate that the sure). All measurements have been done to create a low humidity environment. ventilation frequency was reduced by the at 10 minutes intervals and the data were Sample B was placed in an open desic- improvements. collected roughly once in three months. cator. Both desiccators were placed into Figure 5 illustrates a phase diagram for sodium sulfate and temperature and relative humidity of daily average inside the shelter. The continuous lines indica- A B C D 4 b te the boundaries of the stable phases. Temperature (°C) 21.7 ± 1.0 21.0 ± 1.1 10.4 ± 0.2 10.8 ± 0.2 This diagram suggests that the interior environment before the improvements Relative Humidity 59.8 ± 14.5 77.2 ± 5.0 27.8 ± 29.8 73.1 ± 1.4 (November 2014 - October 2015) mainly varied in both thenardite and mirabilite

Table 1: Experimental conditions stable conditions. Previous studies showed the damage by sodium sulfate occurs because thenardite dissolution can produce solutions highly supersaturated Average temperature (°C) with respect to mirabilite, so that preci- Nov. 2014 - Feb. 2015 9.9 ± 2.8 pitation of this phase can lead to large crystallization pressures.5,10 Thus, it is Nov. 2015 - Feb. 2016 13.3 ± 3.2 important for the conservation of the Buddha that crystallization of sodium Figure 5: Door closing time a) before the improvements Table 2: Average indoor temperature before and after improvements in winter ns b)after the improvements sulfate should be reduced, cycles of crys-

120 121 Environmental control for mitigating salt deterioration by sodium sulfate on Motomachi Stone Buddha K. Kiriyama et al. in Oita prefecture, Japan

were confirmed on wood parts of the have been caused by decreasing interi- shelter by observation after the impro- or temperature and relative humidity vement. The high water content caused in winter, environmental control could by the rise in inside relative humidity not have been carried out because of the may be considered as the causes of in- poor airtight of the shelter. Then, the creasing gypsum and molds. These chan- shelter was improved in order to reduce ges after the improvement were caused a drop in temperature and humidity in by suppressing the ventilation of the winter. To evaluate the shelter impro- shelter throughout the year. Generally vements’ effect on the salt deterioration, speaking, winter in Japan is cold and dry, environmental research and observation the other hand summer is hot and hu- on were carried out. After these shelter mid. Therefore, it is necessary to consider improvements which led to reduction operation method of the shelter in each of ventilation frequency, it is considered season in the future. This could lead to that deterioration by powdering on the reducing molds or other microbiology at- stone surface was reduced. However, pre- tack caused by humid condition. dictable problems by humid condition such as new crystallization of gypsum and mold were detected after the improvement. Furthermore, in order to set the 4.2. Deterioration comparison by target value of the environmental cont- Figure 6: Phase diagram for sodium sulfate and temperature and relative humidity of daily average inside the shelter sodium sulfate rol, effects of temperature and humidity (based upon data from4) Blue dots are before improvements Temp. and RH, on the other hand white dots are after improvements. on salt deterioration were investigated The results of the experiments are pro- by salt crystallization experiment. The vided in Table 3. The rate of deterioration results of the experiments indicate that of samples are calculated by using equati- to reduce lowering relative humidity is A B C D on (1). The rate of deterioration of sample more important than to decrease a drop Rate of deterioration (%) 0.0034 ± 0.0009 0.0009 ± 6E-05 0.0035 ± 0.0026 0.0003 ± 8E-05 A and sample C were greater than sample in temperature. B and sample D. It is inferred that phase For future research, a target value of change to Thenardite may have an effect the interior environment in each season Table 3: Percentage of damage from equation (1) on the rate of deterioration. It is also pos- needs to be considered. Furthermore, by sible that Increasing in evaporation rate an appropriate operation of the shelter may promote further damage. As for the which increased the airtight and could tallization and dissolution of sodium sul- sodium sulfate. Consequently, the inte- difference of temperature (sample B and control environment more easily, mitiga- fate should be avoided and phase change rior environment has changed to reduce sample D), significant difference did not ting methods for not only salt crystalliz- to thenardite should be suppressed even crystallization of sodium sulfate. be detected. These result indicate that ation in winter but also other deteriora- if mirabilite crystallizes. Figure 5 indica- The amount of salt crystallization to reduce lowering relative humidity is tion problems of the Buddha should be tes that the interior environment after could not be measured quantitatively more important than to decrease drop considered continuously. the development of the shelter (Novem- because usual salt crystallized area in in temperature. Furthermore, the depth ber 2015 - October 2016) were mostly un- winter was covered with Japanese papers of salt crystallization affected by evapo- der solution of sodium sulfate stable pha- in order to remove salts. However, accor- ration rate did not be examined in this Acknowledgements se, even though sometimes in mirabilite ding to observations, the amount of salt study. Hence, this is a subject for future stable phase. This results shows that the crystallization in winter decreased ob- analysis. We are grateful to the Oita-city board improvements reduced the risk of preci- viously. Therefore, it is considered that of education for the image of stone-cliff pitation of thenardite. Furthermore, the deterioration by powdering on the stone Buddha (Figure5) and the groundwater indoor temperature before and after im- surface was reduced. On the contrary, on Conclusion table of the observation hole. provements in winter is shown in Table the other part of the stone surface, new This work was supported by JSPS KA- 2. The temperature in winter increased salt crystallization was confirmed after Salt crystallization, especially in win- KENHI Grant Number 26709043[Grant- 3.4°C after the renovation, which could the improvement. The salt was identified ter, is a major deterioration factor of this in-Aid for Young Scientists (A)] lead to inhibit decreasing solubility of as gypsum by XRD. Furthermore, molds Buddha. Although salt deterioration may

122 123 SWBSS 2017 | 20-22 September K. Kiriyama et al. 4th International Conference on Salt Weathering of Buildings and Stone Sculptures

References salt deterioration at buddha statue car- Numerical analysis on salt damage suppression of the ved onto a cliff at motomachi in oita 1 Goudie, A. and Viles, H., Salt Weathering prefecture of japan, Proceedings of the Buddha statue carved into the cliff by controlling the Hazards, John Wiley, Chichester, (1997). 13th international congress on the de- room temperature and humidity in the shelter terioration and conservation of stone : 1 1 2 3 4 2 Scherer, G. W., Stress from crystallizati- VOLUME II, Hughes, J. and Howind, T. Nobumitsu Takatori *, Daisuke Ogura , Soichiro Wakiya , Masaru Abuku , Kyoko Kiriyama and Yoshei Kohdzuma2 on of salt George, Cement and Concrete (eds.), University of the West of Scot- 1 Department of Architecture and Architectural Engineering, Kyoto University, Japan Research, (34), (2004), 1613-1624. land, (2016), 1163-1170. 2 National Institutes for Cultural Heritage Nara National Research Institute for Cultural Properties, Nara, Japan 3 Steiger, M. and Asmussen, S., Crystalli- 10 Tsui, N., Flatt, R. J. and Scherer, G. W., 3 Faculty of Architecture, Kindai University, Osaka, Japan 4 zation of sodium sulfate phases in po- Crystallization damage by sodium sul- Graduate School of Advanced Integrated Studies in Human Survivability, Kyoto University, Japan *[email protected] rous materials: The phase diagram Na- fate, Journal of Cultural Heritage, (4), 2SO4–H2O and the generation of stress, (2003), 109-115. Geochimica et Cosmochimica Acta, (72), Abstract 1. Research aim (2008), 4291-4306. Motomachi Sekibutsu is a Buddha sta- Motomachi Sekibutsu (Fig. 1) is a stone 4 Flatt, R. J., Salt damage in porous ma- tue that was carved into a cliff in Oita statue of Buddha that consists of tuff and terials: how high supersaturations are City, Japan, during the 11th or 12th cen- was carved into a cliff in Oita City, Japan, th th generated, Journal of Crystal Growth, tury. It was designated as a national his- during the 11 or 12 century. The statue (242), (2002), 435-454. toric site in 1934. The stone statue is consis constantly affected by the hygrother- tantly affected by the penetration of heat mal environment in the room and heat 5 Desarnaud, J. and Shahidzadeh, N., Im- and moisture into the cliff, and concerns and moisture transfer through the cliff pact of the Kinetics of Salt Crystallizati- have been raised about its deteriorati- (from which it cannot be separated). The- on on Stone Damage During Rewetting/ on. Various preservation measures have refore, the statue has suffered salt dama- Drying and Humidity Cycling, Journal been taken to prevent this; however, the ge, exfoliation and growth of mould and 1 of Applied Mechanics, (80), (2013), 6-8. main cause of deterioration, salt dama- bryophytes. Salt damage, caused by so- ge, has not been eliminated. Here we de- 6 Bionda, D., Methodology for the preven- velop a numerical analysis model to cal- tive conservation of sensitive monu- culate the heat and moisture behaviour ments : microclimate and salt activity in the statue and its shelter. Using this in a church, Proceedings of the 10th model, we reproduce the shelter’s hygro- International congress on deterioration thermal environment before and after and conservation of stone, Stockholm, renovation and evaluate it with respect (2004), 627-634. to damage caused by sodium sulphate. Our results show that the improvement 7 Zehnder, K. and Schoch, O., Efflorescen- in airtightness drastically contributes ce of mirabilite, epsomite and gypsum to decreasing the evaporation from the traced by automated monitoring on-si- statue and suppressing the salt phase ch- te, Journal of Cultural Heritage, (10), ange; thus, the renovation of the shelter (2009), 319-330. suppressed sodium sulphate salt damage to the statue. 8 Oita city board of education, Kunis- hiteishiseki Oitamotomachisekibutsu Keywords: conservation environment, Hozonsyurijigyo Hokokusho (in Japane- architectural environment, coupled heat se), oita, sohrinsha, (1996). and moisture transfer, sodium sulphate, phase change 9 Kiriyama, K., Wakiya, S., Takatori, N., Ogura, D., Abuku, M. and Kohdzuma, Y., The current state and the factors of Figure 1: Motomachi Sekibutsu

124 125 Numerical analysis on salt damage suppression of the Buddha statue carved into the cliff by controlling the room N. Takatori et al. temperature and humidity in the shelter

Figures 3: Variation of the measured temperature inside Figure 4: Variation of the measured relative humidity and outside the shelter inside and outside the shelter

4 show the inner and outer temperature 3. Method for Examining Salt Damage and relative humidity from March 2015 to November 2016. After the renovation, Salt damage to the porous statue ma- the daily fluctuation in room tempera- terial is caused by precipitated salt. The Figure 2: Shelter exterior ture decreased but the relative humidity porous body is destroyed by volume ex- maintained its high value. We measured pansion of salt, which is caused by crys- dium sulphate, has long been a concern. 2. Motomachi Sekibutsu Room the inside and outside concentrations of tallisation, thermal expansion and hyd- 4 To prevent these deteriorations, various Environment CO2 from October 3 to October 4, 2016, ration. To destroy the statue material, preservation measures have been under- and calculated the ventilation rate via the following processes must occur: i)

taken from 1986 to 1996 and again from 2.1. Shelter the decrease in CO2 concentration af- salt must precipitate in the material, and 2011 onward. These include constructing ter opening or closing the window. The ii) the crystal pressure of salt must be a shelter (Fig. 2), boring a tunnel behind At Motomachi Sekibutsu, the shelter ventilation rate was 0.21 times per hour high enough to break the material. Mo- the statue to reduce groundwater and was renovated twice, in 1995 and in 2015. when the windows were closed, and it tomachi Sekibutsu consists of a tuff., the renovating the shelter. Salt damage by In 1995, various construction works were was 9.9 times per hour when they were most prevalent damaging salt is sodium sodium sulphate is considered to be hal- done to improve thermal insulation, such opened. sulphate, and it is assumed that the pres- ted as a result of the shelter renovation as insulating the wall by a thermal insu- sure occurs when the precipitated salt ch- during November 2015.2 lation material and installing a windbre- Our aim is to suppress the further ak room at the entrance.1 In 2015, the degradation of the statue from salt da- renovations focused on salt damage due mage by controlling the hygrothermal to sodium sulphate and aimed to protect environment of the shelter that houses from solar radiation, improve the perfor- the Buddha statue. Previously, using cal- mance of thermal insulation and impro- culations of the heat and moisture behave the airtightness of the windows.2 Du- viour of the statue and considering the ring that renovation, thermal insulation shelter’s hygrothermal environment, we was improved, a door closure was moun- clarified that salt damage tends to occur ted, the windbreak room was mounted, at the knee of the statue under the room and the windows were covered with solar environment before the renovation in shading sheets. In this study, we refer to 2013.3 In this study, we focus on how that renovation as ‘the renovation’. the hygrothermal environment in the shelter is produced, and we reproduce it using numerical analysis. In addition, we 2.2. Room temperature, humidity and examine a method for coordination of ventilation rate the room environment, which suppresses the salt damage of the statue. We measured the hygrothermal environment inside and outside the shelter from November 2014. Figures 3 and Figure 5: Schematic of the numerical model of the cliff, including the statue

126 127 Numerical analysis on salt damage suppression of the Buddha statue carved into the cliff by controlling the room N. Takatori et al. temperature and humidity in the shelter

ture and humidity, which are boundary (3) conditions, were obtained from measured values (obtained from April 1, 2013 The boundary conditions for heat and to March 31, 2014 and January 1, 2016 to moisture, respectively, are: December 31, 2016) 7. The heat and moisture properties of each material were estimated from the literature.8, 9 The coefficients of heat and moisture were (4) 9.3 W/m2K and 2.85 × 10−8 kg/m2 Pa for the room/attic space, respectively, and 23.3 W/m2K and 1.14 × 10−7 kg/m2 Pa for the outside, respectively. Table 1 shows the calculation conditions. We aimed (5) to clarify how the room’s hygrothermal environment was changed by each reno- The boundary equations of the space are vation; therefore, we divided the renova- written as tion into the following factors to create our model. Heat balance Figure 6: Model for analysing the room environment i) Ventilation rate. Before the renovation, the ventilation rate between the room and the outside was 10.0 times per hour; however, after the renovation, it anges its phase from thenardite (Na SO ) amount of solar radiation by conside- (6) 2 4 was 0.2 times per hour. Furthermore, to mirabilite (Na SO .10H O).5 The salt ring the angle of inclination. The surface 2 4 2 the ventilation rate was 2.0 times per precipitates when the solubility decrea- area of the statue was obtained by mul- Moisture balance hour between the room and the outsi- ses with decreasing temperature and/or tiplying the length of each of the two-di- de and 0.1 times per hour between the water evaporates. mensional models by the depth, 10.8 m. room and the attic. The volumes of the room and the attic are 164.2 m2 and 100 m3, respectively. (7) ii) Thermal insulation performance of 4. Numerical Analysis We used the heat and moisture balance the wall and window. Before the reno- equations 6 to analyse the heat and mois- where Q is the heat, which includes ab- vation, the U-value (the average trans- sorbed solar radiation in the space, trans- 4.1. Methodology ture behaviour in the statue and wall. mission rate) of the wall was 4.55 W/ The equations are written as follows: mitted through the windows. m2K. After the renovation, the U-value was 1.23 W/m2K. We developed a numerical model to Heat balance calculate the heat and moisture behavi- 4.2. Analysis conditions iii) Solar radiation shielding. Before the our in the statue and shelter. The cliff, In this analysis, the outside tempera- renovation, the rate of solar trans- including the statue, is represented as a two-dimensional model, as used in our (1) previous study3 and shown in Fig. 5. The Case 0 Case 1 Case 2 Case 3 Case 4 Case 5 analytical model of the temperature and Moisture balance humidity in the shelter is shown in Fig. Ventilation rate A B A B B A 6. We divided the shelter into room space and attic space. The shelter includes three Thermal insulation A B B A B A kinds of walls: the insulated wall, the (2) wooden wall and the glass window. The Solar transmission A B B B A A walls are oriented along the northeast, where the relation between the chemical ‘A’ means after the renovation, and ‘B’ means before the renovation southeast and southwest directions. The potential of water µ and relative humidi- roof is inclined at 30°. We calculated the ty h is given by Table1: Analysis conditions

128 129 Numerical analysis on salt damage suppression of the Buddha statue carved into the cliff by controlling the room N. Takatori et al. temperature and humidity in the shelter

evaporation, and the amount of evaporation does not change significantly across all seasons. The decreased ventilation rate has the greatest influence on the evaporation from the statue, and the improved insulation performance has the least influence.

5.3. Phase change of sodium sulphate at the statue surface

At Motomachi Sekibutsu, salt damage can be seen at the knee of the statue (Fig. 11). In this section, we plot the calcula- Figure 9: Annual evaporation from the surface of the statue ted temperature and relative humidity Figure 7: Comparison in Case 0 Figure 8: Comparison in Case 0 at the surface of the statue on the phase

diagram of Na2SO4 and examine the pos- on and, in especially, contributes well in sibility of the resulting salt damage. Figu- mission of the window was 0.75.3 The red value, although there is a mismatch summer. The improvement in insulation re 12 shows the results of Case 1, Case 3 amount of solar radiation incident on over a short period. Those errors can be performance suppresses the evaporati- and Case 4, and Fig. 13 shows that of Case the statue was calculated by conside- mainly attributed to the constant ven- on from winter to spring, but there is a 1, Case 2 and Case 5. In Case 2 and Case ring the shape of the roof and the po- tilation rate throughout the year or the possibility of promoting the evaporation 5, which are decreased-ventilation-rate 3 sition of the sun. After the renovati- heat and moisture flux from the statue. from summer to autumn. The preventi- cases, the phase change from Na2SO4 to on, the rate of solar transmission was on of solar transmission suppresses the Na2SO4-10H2O rarely occurs because the 0.6 and the amount of solar radiation that reached the statue was zero. 5.2. Water evaporation from the statue

Using these conditions, we reproduced We examined the trends in salt preci- the room temperature and humidity in pitation for Cases 1–5. Here we assumed 2016 with Case 0 and reproduced from that the moisture flux from the surface April 2013 to March 2014 with Case 1, re- of the statue is equal to the evaporati- spectively. In Cases 2–5, we considered in on in the statue. Figure 9 shows the an- detail how each renovation affected the nual amount of water evaporation in the room environment. statue, and Fig. 10 shows the monthly amount of evaporation. Comparing Case 1, before the renovation, to Case 2, it can 5. Results and Discussion be seen that water evaporation is suppressed by the decreasing ventilation rate. Si- 5.1. Comparison of the measured and milar results are obtained because of the calculated values improvement in insulation performance and prevention of solar transmission. Fig. 7 and Fig. 8 show the measured These results show that after the renova- and calculated values of the temperature tion, evaporation and salt precipitation and relative humidity in the room for was drastically suppressed throughout Case 0 and Case 1. In both cases, the an- the year. Here we focus on the effects of nual fluctuation of the calculated value each renovation factor. The decreasing

agrees very well with that of the measu- ventilation rate suppresses the evaporati- Figure 10: Monthly average evaporation from the surface of the statue

130 131 Numerical analysis on salt damage suppression of the Buddha statue carved into the cliff by controlling the room N. Takatori et al. temperature and humidity in the shelter

6. Conclusion

In this study, we developed a numerical analysis model to calculate the heat and moisture behaviour in the statue and shelter. Using this model, we reproduced the shelter’s hygrothermal environment before and after the renovation and evaluated it with respect to sodium sulphate salt damage. The model was valid for recreating the hygrothermal environment in the shelter before and after the renovation. The renovation caused the evaporation from the statue to be suppressed by improving the airtightness and insulation performance and preventing solar radiation. The improvement in airtightness drastically decreased evaporation from the statue and suppressed the phase change of salt. Our results show that the renovation of the shelter suppressed sodium sulphate Figure 11: The knee of the statue salt damage to the statue. Figure 12: Phase diagrams for Case 1, Case 3 and Case 4

relative humidity is very high. However, in Case 1, Case 3 and Case 4, the temperature fluctuates around the low relati- References ve humidity, and phase change occurs easily. By focusing on the effect of each 1 Oita City Board of Education, 1996, “Ku- renovation factor, in Case 2, we see that nishiteishiseki Oita Motomachi Seki- the temperature fluctuates around the butsu Hozonsyurijigyo Hokokusho (in high humidity and the fluctuation of re- Japanese),” Oita, sohrinsha. lative humidity is smaller in Case 2 than in Case 1 before the renovation. 2 Oita City Board of Education, 2016, “Ku- The temperature and humidity in Case nishiteishiseki Oita Motomachi Seki- 3 are approximately equivalent to tho- butsu Hozonsyurijigyo Hokokusho” (in se in Case 1, but from winter to spring, Japanese). the relative humidity is higher in Case 3 than in Case 1. Therefore, the phase 3 N. Takatori, D. Ogura, S. Wakiya, M. Ab- change from Na2SO4-10H2O to Na2SO4 is uku, K. Kiriyama: Numerical analysis of restrained during these seasons. Consi- the influence of hygrothermal variation dering the solar transmission, the tempe- on salt weathering of a Buddha statue rature is lower and the relative humidity carved into a cliff–Study on the conser- is higher in Case 4 than in Case 1 th- vation of a stone Buddha carved into a roughout the year. These results suggest cliff at Motomachi PART1-, Journal of Figure 13: Phase diagrams for Case 1, Case 2 and Case 5 that the improvement of airtightness Environmental Engineering (Transac- strongly contributes to the suppression tions of Architectural Institute of Ja- of salt damage to the statue. pan), Vol. 733, pp. 215-225, 2017.

132 133 SWBSS 2017 | 20-22 September N. Takatori et al. 4th International Conference on Salt Weathering of Buildings and Stone Sculptures

4 A. S. Goudie, H. A. Viles: Salt Weathering Management of sodium sulfate damage to Hazards, Wiley, 1997. polychrome stone and buildings 5 R. J. Flatt: Salt damage in porous materials: How high supersaturations are ge- David Thickett* and Bethan Stanley nerated, Journal of Crystal Growth, Vol. English Heritage 242, pp. 435-454, 2001. *[email protected]

6 M. Matsumoto: Energy conservation in heating cooling ventilating building: Abstract system. Inspection of the environmental Heat and mass transfer techniques and data when material fell, shows similar alternatives (ed. Hoeogendoorn C.J. and English Heritage holds large collections depression of the RH below the theoreti- Afgan N.H.), Washington: Hemisphere of polychrome stone. Often the poly- cal values for sodium sulfate. Pub. Corp., pp. 1-45, 1978. chromy only remains as small fragments, loosely adhered to the stone surface with Keywords: Sodium sulfate, polychromy, 7 Japan Meteorological Agency, 2017, refer any original binders having converted to acoustic emission, image analysis, pre- to the past weather data, . Thetford Tomb fragments are amongst the most significant part of the collecti- 8 AIJ: Air and Moisture Transfer Through on. Salt analyses indicated very high con- 1. Introduction New and Retrofitted Insulated Envelo- centrations of sodium sulfate present, pe Parts (Hamtie), Final Report, Vol. 3, (up to 1.5% by mass of the stone). Consi- English Heritage holds large collec- TASK 3: Material Properties, AIJ, 2001. dering the very fragile nature of the re- tions of polychrome stone. Often the maining polychromy and the aggressive polychromy only remains as small frag- 9 M. K. Kumaran: Thermal and Moisture nature of this salt, strict preventive con- ments loosely adhered. Such material Transport Property Database for Com- servation is needed to ensure survival of is extremely susceptible to salt damage mon Building and Insulating Materials, the information. Theoretically, keeping and evidence of polychromy can easily Final Report from ASHRAE Research the RH below the thenardite transition be lost. Sodium sulfate is an extremely Project 1018-RP, 2002. line (the RH is temperature dependant) damaging salt and used in several salt should avoid any damaging salt transi- crystallisation tests.1, 2 It undergoes an tions. Monitoring with acoustic emission over 300% expansion when converting in the storage environment, provided a from the anhydrite, thenardite to the direct tracing of salt transitions. When decahydrate, mirabilite. The RH that this plotted against temperature, it became conversion occurs at is strongly tempe- clear that transitions were occurring at rature dependant.3 Controlling tempera- between 3 and 7% lower RH than expec- ture and RH can prevent the damage. Ho- ted from theory. Tests with pure sodium wever, the exact parameters are needed sulfate powder coincided with the the- to design effective control. oretical values. The most likely reason The sixteenth century limestone Thet- is the effect of the pore structure, par- ford tomb fragments are amongst the ticularly fine pores. The values are con- most significant part of the collection. sistent for a single piece of stone and They formed the focus of a major recent vary between pieces. A good correlation historical research project.4 The tomb was observed between acoustic emission was destroyed during the English Refor- events and small pieces appearing on the mation (1530s) and the small (less than imaging plate of a prototype particle de- 24 cm) pieces were exposed to the ele- position analyser. Loss of material from ments for almost half a century before two interior building surfaces was moni- acquisition into the National collection. tored using a similar automated camera In these circumstances it is highly likely

134 135 D. Thickett and B. Stanley Management of Sodium Sulfate Damage to Polychrome Stone and Buildings

that the original organic binder present 1000 kHz, 150 kHz resonant), clamped to Ion concentration (% mass/mass of dry stone) with the pigment has converted to oxala- the stone and with couplant present as Object te. This makes the pigment stone bond well. Test blocks (6 cm by 6 cm by 6 cm) number chloride nitrate sulfate sodium potassium calcium extremely weak. of Caenstone impregnated to give appro- Two techniques have been used to di- ximately 1 and 0.2% sodium sulfate by 78101879 0.022 0.020 1.354 0.485 0.004 0.054 rectly monitor salt transitions or materi- weight, were exposed to increasing RH. 78101845 0.015 0.014 1.335 0.478 0.001 0.043 al loss caused by them. Acoustic emission The RH was generated in a polycarbonate allows a direct measurement of certain chamber with a commercial Rh generato- 78101881 0.008 0.009 1.251 0.449 0.002 0.046 phenomena. It has been used to moni- re, (Preservatech MiniOne unit). Slyglide tor sodium sulfate transitions5 and sto- couplant gel was found to be required 78101843 0.008 0.009 1.247 0.448 0.000 0.032 ne decay.6 The data can be of sufficient to achieve sufficient sensitivity for the 78101880 0.012 0.011 1.224 0.433 0.003 0.041 time resolution to allow direct correlati- Woodwatch sensors with the 0.2% sulfate on with environmental data. Automated blocks. The Woodwatch system genera- image analysis of a glass plate was used ted signals for the 1% sulfate blocks, but Table 1: Ion concentrations extracted from drillings of Thetford Tomb limestone to monitor material falling from the car- this was not deemed sensitive enough. ved surfaces of the pieces. Various other Two cycles with the 0.2% sulfate blocks analytical techniques were used to deter- generated loss from their surfaces. More The temperature and RH was measured found to reliably detect particles down mine the nature of the lost material. of the Woodwatch units were available to with a Rotronic Hygrolog logger. The RH to 3μm diameter.9 The system works well the project, although the use of couplant at which significant acoustic emission oc- in dark or very low light surroundings. is not ideal. Removal of the couplant was curred was taken to indicate the transiti- This was facilitated by running the expe- 2. Methods and Materials investigated. Acetone was found to re- on. Acoustic emission sensors were atta- riments in closed cupboards. Any fall of move all visible residues of the Slyglide ched to the five pieces for twelve months material onto the plate was determined Salt analyses were undertaken with from the Caenstone. Staining with iodine in their storage environment. The tem- by subtracting sequential JPEG images 0.4mm drilled samples from break sur- vapour indicated no remaining organic perature and RH was monitored beside using Image J software. A particle appe- faces. The first 2 mm of the sample from material on the stone surface or in the each block with a Rotronic Hygrolog D aring in the difference image was assu- the surface was discarded and sampling interior (the blocks were cut). logger with hygrclip probe, calibrated med to have been lost, in the 30 minute continued for another 10 mm after that. A set of preliminary experiments were with UK National Measurement Accredit- time interval from the stone surface. The The sample volume was over thirty times also undertaken to determine the effect ation Service traceable standards. Triaxi- position of the particles lost from the po- the maximum grain volume of the fine of the other small concentrations of ions al shock loggers (MSR145) were attached lychrome stone and falling onto the glass grained limestone determined by eye. present on the acoustic emission detec- to two pieces to assess shock events that plate was noted. They were collected with The samples were dried at 110 °C, and tion of the hydration transition. Salts could be detected with the sensors and, a chemically sharpened needle under extracted with 18.2MOhm/cm water. The (sodium, calcium and potassium sulfa- potentially mis-interpreted, as salt ac- magnification every 3 months and analy- filtered solutions were analysed with a te, sodium nitrate and chloride) were tivity. Acoustic emission monitoring of sed with a Renishaw System 1000 Ramas- Dionex DX600 ion chromatograph with; mixed by grinding to obtain a compo- wooden objects has been limited by large cope confocal Raman system, 632.8 nm AS14A column and an eluent of 8mM so- sition close to the soluble extracts from amounts of noise.8 laser, Nicolet Inspect FTIR microscope dium carbonate and 1mM sodium bicar- the stone object with highest percentage Further tests were undertaken with and Eagle III XRF system. bonate for anions and CS12 column with of non sulfate and sodium ions (object two pieces of stone (4.2 by 2.3 by 1.8 and A second set of monitoring was carri- a 1mM methane sulfonic acid eluent of number 78101879), hence referred to as 2.2 by 4.5 by 2.0 cm). The acoustic emissi- ed out with a sodium sulfate laden group for cations. The solution ion concentra- impure sodium sulfate. Water was added on, T, RH (using a calibrated Meaco sys- of brick casemates at Fort Brockhurst, on tions were converted to percentages by to form a solution and this was dried at tem with Rotronic hydroclip probes) and the South coast of England. The prototy- mass based on the dry mass of stone. 60ºC in a Bakelite top. When dry the top vibration were monitored. Each stone pe image analysis system was placed un- Five pieces were analysed with acoustic and impure sodium sulfate was placed was placed above a prototype image ana- der a mortar join exhibiting powdering emission. Initially, the performance of on top of a WD acoustic emission sensor. lysis system, such that any material lost, and loss. Salts had been observed mainly two instruments was compared. A Phy- These were placed in an environmental fell onto a glass plate. A digital camera to act at the mortar and not the brick sur- sical Acoustics Pocket AE system was chamber and the temperature set to 15, was focused through magnification onto faces. The T and RH were measured with used with WD sensors (1-1000kHz, non 20, 25 and 30ºC. At each temperature the plate and recorded images with a cir- a Meaco system. The casemates are dark resonant) with and without Slyglide cou- the RH was increased from 5% below cumferential LED lighting system every when not in use. The increase in percen- plant. A Hanwell Woodwatch was also the pure sodium sulfate transition RH at 30 minutes. The side LED lighting and tage coverage of the glass plate was mea- assessed with the supplied sensor (1- 0.2% intervals using glycerol solutions.7 camera/magnification combination were sured with image analysis (Image J) every

136 137 D. Thickett and B. Stanley Management of Sodium Sulfate Damage to Polychrome Stone and Buildings

30 minutes. Readings were discarded for 3. Results sodium sulfate (labelled impure). They of the powder samples were pigments 72 hours after the room had been acces- are coincident within the experimental (hematite, carbon black, calcite), over sed due to dust raised and deposited. The salt analyses for the Thetford poly- parameters (1% RH). The fourth, higher 62% of the powder samples were sodium Previous experiments had shown over chrome stone are shown in Table 1. temperature point was not plotted, to sulfate anhydrite (the phase information 99.9% of the small amount of dust sett- Errors are in the order of 3% of the cal- allow better reading of the graph, it also being lost due to the time delay before led in this period. No activities took place culated value. The errors were calculated was coincident with the pure transition analysis) and the remainder, limestone during the monitoring period that would from the reproduceability of multiple value. powder. The Thetford limestone included raise large amounts of dust. standard injections on the ion chromato- Acoustic emission is occurring a few approximately 2% iron, which XRF distin- A basement store room in Rangers graph, the calibration graph 10 and the percent RH below the theoretical transi- guished from calcite used as a pigment, House, London was losing plaster from stated error in the measuring balance tion line in all instances. The stated ac- that had no detectable iron present. several areas of the walls. Salt analysis, and measuring pipettes used to prepare curacy for the probes is 0.03ºC and 0.8%. Two reasons for this observed RH de- extracted both from the losses and from the solutions. The major anion is sulfate, All RH values were within 1% during the pression were postulated. Slow ingress of drillings indicated sodium sulfate was which makes up over 97% of the anions three point calibrations. The RH is plot- RH and T into the stone, could retard the the predominant salt present in the wall present. The levels are very high, over ted to one decimal place. The digital out- transition behind the prevailing ambient and in one area this made up over 99% of 1.2% by mass of the stone. Sodium is put from the probe give a much higher conditions. The presence of fine porosity the soluble ions detected. The automated the predominant cation, making up over precision but was not used with no way could reduce the critical RH of the transi- image analysis system was placed below 88% of the cations present. to check the calibration to that level of action through the Kelvin effect.11 This has an area on an internal wall’ to capture The temperature and RH recorded curacy. The traceable standards are only been postulated for sodium sulfate, but loss and the room temperature and RH when acoustic emission events were ob- valid to one decimal place. The depression this is the first experimental evidence.12 measured with a calibrated Meaco sys- served, are shown in Figure 1. The mira- is consistent for a particular piece of sto- Examination of the T/RH traces before tem. Access to the store was restricted for balite to thenardite transition for pure ne. Fragment 78101843 showed a value of the observed transitions indicated a lar- the monitoring period, so that re-depo- sodium sulfate (pure) is marked on the approximately 6%. Fragments; 78101879, ge proportion, over 25% where if this was siting dust did not cause significant in- figure. Figure 1 also shows three of the 78101881 and 781018800 showed appro- due to a lag in heating or RH transmissi- terference. four points determined for the impure ximately 4% and 78101845 showed 3%. on, then the transition would have been The store in which the monitoring took expected to be observed higher than the place had temperatures mainly between theoretical values, due to the environ- 15 and 24ºC during the 7 month measu- mental data around those instances. rement period. Lower temperatures were The monitoring experiments at Fort experienced in the last 60 days. Brockhurst and Rangers House showed The number of particles lost from sto- powder falling from the mortar and plas- ne 78101843 are shown in Figure 2. ter, coinciding with 4 and 6% RH below Powder was observed falling from the the thenardite to mirabilite transition objects, generally coincidentally with line. acoustic emission events detected (within the same 30 minute measurement slot). Three instances were observed with 4. Discussion and Conclusions acoustic emission, but no powder detected, generally with lower numbers of It has been shown that the hydration acoustic emission events detected. Two transitions of sodium sulfate in some po- instances of powder were observed with rous matrices can be lowered by several no acoustic emission. Examination of percent RH from the expected equilibri- the vibration logger data indicated high um values. This has also been shown to shock levels coincident with two events. relate to the amount of loss observed. This was probably caused by unusual ac- Care needs to be exercised when inter- tivity related to transferring files in the preting both acoustic emission and ma- library above the room used for the expe- terial loss data. Several other phenome- riments. Similar data was produce with na can cause acoustic emission, and its 78101879. The Raman and FTIR micros- use for monitoring wooden objects has Figure 1: Temperature and RH values at which acoustic emission was observed. cope analysis showed approximately 30% been impacted by significant noise. 8 Ca-

138 139 D. Thickett and B. Stanley Management of Sodium Sulfate Damage to Polychrome Stone and Buildings

reful control of the environment reduced stone. For each location, four values are this in these three instances. When salt shown, those calculated from the equili- expansion occurs, loss of material may brium line and those, 3, 4 and 6% below. not be instantaneous and loose, hardly In some instances, (Beeston store 4% bonded material is often observed, on and St Augustines Abbey 6%), the number surfaces. This could fall at a later date. of transitions increases dramatically, due Small numbers of events were detected to the nature of the environment. It was with acoustic emission and no powder proposed to move the Thetford material loss and two instances of powder loss from its existing store to a newly created with no acoustic emission due to large store. The existing store environment shock events. However, the overall excel- was assessed using damage functions de- lent correlation between the two sets of veloped from this work and compared to results gives a reasonable degree of con- the anticipated new store environment.13 fidence, that the events detected fit the interpretation presented. This effect has important ramifica- tions for designing environments to control such phenomena. Display of some of the Thetford pieces was achieved in References a dehumidified showcase, controlling the RH below 60% RH using a Minicli- 1 RILEM, Essais recommandees pour l’alk- ma unit. This value was determined by teration des pierres et evaluer l’efficacite Figure 2: Acoustic emission and particle loss observed from polychrome limestone fragment 78101843 the research presented and knowledge des methodes de traitement, Materiaux of the minimum temperature likely to et Constructions (17-75), 1980, 216-220. be encountered. Without the research, a value of 70% would probably have been 2 Lazzarini, L. and Laurenzi-Tabasso, M., used, reducing the conditioning load, but Il Restauro della Pietra, Cedam, Padua potentially allowing further damage to 1986. occur. The showcase was designed with a flat dark metal panel below the polychro- 3 Bionda, D., Modelling indoor climate me objects on display so any loss would and salt behaviour in historical buil- be readily visible. This was examined dings, PhD Thesis, Swiss Federal Institu- every two years with magnification and te of Technology, Zurich, 2006. any suspect particles analysed, as described previously. No pigment or limestone 4 Lindley, P.G., Representing Re-Formati- particles were detected during eight ye- on: Reconstructing Renaissance Monu- ars on display. Depending on the situa- ments, http://representingreformation. tion, it is often not possible to achieve net/ accessed 25 May 2017 sufficient environmental control to totally stop an effect and the result is often 5 Grossi, C.M., Esbert, Rosa Maria, Suarez expressed in a calculation of number of del Rio, L.M., Montoto, Modesto, Lau- transitions. With the often complex en- renzo-Tabasso, M., Acoustic emission vironments in heritage buildings, even monitoring to study sodium sulphate small changes in parameters can affect crystallization in monumental porous the number of transitions observed. Fi- carbonate stones, Studies in Conserva- gure 3 shows the number of events where tion, (42), 1997, 115-125. the transition line was crossed from the 6 Figure 3: Number of thenardite to mirabilite transitions calculated from environmental data of English Heritage loca- thenardite, into the mirabalite region at Esbert, R.M, Grossi, C., Suarex del Rio, tions with polychrome stone. English Heritage sites with polychrome L.M., Calleja, L., Ordaz, J. and Montoto,

140 141 SWBSS 2017 | 20-22 September D. Thickett and B. Stanley 4th International Conference on Salt Weathering of Buildings and Stone Sculptures

M., Acoustic emission generated in tre- Conservation of marble artifacts by phosphate ated stones during loading. In Sixth In- ternational Congress on Deterioration treatments: influence of gypsum contamination and Conservation of Stone Proceedings, ed J.Ciabach, Turon, 1988, 403-410. Enrico Sassoni1*, Gabriela Graziani1, Elisa Franzoni1 and George W. Scherer2 1 Dept. of Civil, Chemical, Environmental and Materials Engineering (DICAM), University of Bologna, Italy 7 Miner C.S. and Dalton N.N., Glycerol, 2 Dept. of Civil and Environmental Engineering (CEE), Princeton University, USA Reinhold, New York, 1953. *[email protected]

8 Jakieła, Sławomir, J. Bratasz, Ł. and Ko- Abstract duction in gypsum solubility in rain is złowski, R., Acoustic Emission for Tra- expected. cing the Evolution of Damage in Woo- The use of ammonium phosphate solu- den Objects, Studies in Conservation tions has proven to be very promising for Keywords: marble, gypsum, black crusts, (52-1), 2007, 101-109. protection and conservation of marble. hydroxyapatite, protection However, all the studies carried out so far 9 Bowden, D. and Brimblecombe, P., Moni- have been performed on uncontamina- 1. Introduction toring dust at Ickworth House with the ted marble. Unfortunately, this is rarely dust-bug, Views, (42), (2005), 25–27. the case in the field, because marble arti- Aqueous solutions of diammonium

facts exposed outdoors are often affected hydrogen phosphate (DAP, (NH4)2HPO4)) 10 ISO11095 Linear calibration using refe- by sulfation, i. e. formation of a gypsum have proven highly promising for pro- rence materials, 1996. crust on the surface. Because gypsum is tection and consolidation of marble.1-4 much more soluble than calcite, the out- Thanks to the reaction between the phos- 11 Camuffo, D., Microclimate for cultural come of the ammonium phosphate tre- phate solution (also containing a cal- heritage, Elsevier, Amsterdam, 1998. atment is expected to be sensibly altered cium source) and the substrate, new by the presence of gypsum. Therefore, in calcium phosphate (CaP) phases are 12 Charola, A. E., and Weber. J. The hydrati- this study the nature and morphology of formed 5. These new phases are able to on-dehydration mechanism of sodium the new calcium phosphate phases for- improve marble resistance to dissoluti- sulphate. In Seventh International Con- med by reacting gypsum with aqueous on in rain (thanks to their lower solubi- gress on the Deterioration and Conser- solutions of diammonium hydrogen pho- lity than calcite)2, 5 and marble cohesion vation of Stone, Laboratorio Nacional sphate (DAP) were investigated. In par- (thanks to their bonding action at grain de Engenharia Civil, Lisbon, 1992,. ticular, the effect of DAP concentration, boundaries).1, 4 Ideally, the new calcium 581–590. ethanol addition (aimed at reducing gyp- phosphate should be hydroxyapatite

sum solubility), and pH were explored. (HAP, Ca10(PO4)6(OH)2), which is the least 13 Xavier-Rowe, A., Newman C., Stanley The result is that phase formation can be soluble CaP phase in aqueous solutions B., Thickett D., and Pereira Pardo L. A controlled by suitably tuning the above at pH > 4.5, 6 However, depending on the new beginning for English Heritage’s mentioned parameters. Phases with low reaction conditions (e.g., pH 3 or addi- archaeological and architectural stored solubility (such as tricalcium phosphate tion of external calcium sources 5), dif- collections. In ICOM-CC 17th Triennial and hydroxyapatite) can be obtained by ferent CaP phases may form alongside Conference Preprints, Melbourne, ed. increasing the ethanol concentration, the HAP, such as octacalcium phosphate

J. Bridgland, International Council of DAP concentration or the pH. However, (OCP, Ca 8(HPO4)2(PO4)4∙5H2O), β-tricalcium

Museums, Paris, 2014, art. 1517, 10 pp. their formation is associated with diffu- phosphate (β-TCP, β-Ca3(PO4)2), brushite

(ISBN 978-92-9012-410-8). sed cracking, likely because of excessive (CaHPO4∙2H2O), monocalcium phosphate

growth of the new phases. Among the in- monohydrate (MCPM, Ca(HPO4)2∙H2O) vestigated formulations, treatment with and monocalcium phosphate anhydrous 1 a 0. M DAP solution with 30 vol % etha- (MCPA, Ca(HPO4)2). These phases have nol at pH=8 seems to be the most suita- sensibly different solubility in water 1, ble one, as it leads to formation of brus- hence the exact nature of the new CaP hite (about 30 times less soluble than phases is fundamental for success of the gypsum), without cracking, so that a re- treatment.

142 143 E. Sassoni et al. How Conservation of marble artifacts by phosphate treatments: influence of gypsum contamination

As pointed out by some recent appli- in the new calcium phosphate layer was 2.2. Treatments um hydroxide was investigated for in- cations of the DAP-based treatment to observed.8, 9 creasing DAP concentrations (0.05, 0.1 some real artifacts1,7, the possible presen- The present study is aimed at investi- Samples were treated by full immer- and 0.2 M) and a given concentration ce of contaminants on the surface to be gating and optimizing the nature and sion for 24 hours in 200 ml of aqueous of calcium ions (corresponding to a 30 treated can have a significant impact on the morphology of the new CaP phases solutions of DAP with different formula- vol% ethanol addition). the nature of the new calcium phosphate formed starting from gypsum, by promo- tions, designed to investigate the effects phases formed after treatment. In fact, in ting the formation of phases with low so- of the following parameters: After immersion in the solutions for 24 the field marble is frequently contamina- lubility and preventing cracking. To this hours, the samples were extracted and ted with gypsum, resulting from marble aim, the influence of several parameters 1) Effect of ethanol (EtOH) concentrati- abundantly rinsed with water to remove sulfation induced by the high concent- (DAP concentration, pH, and addition of on. Because gypsum solubility in an unreacted DAP and ammonium sulfate 8-10 rations of sulphur dioxide in the atmo- ethanol to reduce gypsum solubility) was aqueous solution can be reduced by (that is expected as a by-product ). 11 sphere in the past decades. An example investigated. adding ethanol to the solution , increasing ethanol additions to water (0, of marble affected by surface sulfation is 10, 30 and 50 vol %) were explored to 2.3. Characterization illustrated in Figure 1. Gypsum formed study the influence of decreasing calci- on the surface of marble elements can- 2. Materials and methods um ion concentrations, for a given DAP The mineralogical composition of the not always be completely removed before concentration (0.1 M). new CaP was determined by grazing in- marble treatment. Because gypsum has 2.1. Samples cidence X-ray diffraction (GID, Bruker a much higher solubility in water than 2) Effect of DAP concentration. Increasing D8 Discover X-Ray Diffractometer, Cu calcite ( 2.4 and 0.014 g/l, respectively), To investigate the effect of gypsum on ~ ~ DAP concentrations (0.05, 0.1 and 0.5 anode, incidence radiation θ=0.5°, detec- the treatment outcome is expected to the nature of the new CaP phases, a sim- M) were explored to study the influ- tor scan range 2θ=3-37°). This technique be sensibly affected by the presence of plified system was considered. Tests were ence of phosphate ion concentration detects the composition of surface layers 1 gypsum. carried out on specimens of gypsum pas- for a given calcium ion concentration without interference from the substrate, A few studies have investigated the ef- tes, produced by mixing bassanite with (determined by an ethanol addition of thanks to the reduced penetration depth fects of treating gypsum with aqueous water (water/bassanite weight ratio 0.5). 30 vol%). of the incoming X-rays. 7-10 3 solutions of DAP. Depending on the Prisms with 4×4×16 cm size were cast The morphology of the new phases was reaction conditions (e. g. DAP concentra- and then, after hardening, cubes with 3) Effect of pH. Because higher pH is assessed by observing the samples (after tion, pH, duration), phases with low solu- 1 cm edge length were sawn and used for expected to favor formation of HAP coating with carbon) using an environ- bility (such as β-TCP, OCP, HAP) and more the tests. instead of brushite 12, the effect of in- mental scanning electron microscope soluble phases (such as brushite) have creasing the pH to 10 using ammoni- (FEI Quanta 200 FEG ESEM). been found. 7-10 However, when high DAP concentrations were used (3.0 M 9 and 3.8 M 8), the formation of cracks and fissures

Figure 1: Example of marble decoration affected by sulfation (Monumental Cemetery in Bologna, Italy, XIX century). Beneath the gypsum surface layer (also incorporating some particulate matter), marble exhibits grain disaggregation Figure 2: CaP phases formed by reaction with a 0.1 M DAP solution with increasing ethanol content at pH=8

144 145 E. Sassoni et al. How Conservation of marble artifacts by phosphate treatments: influence of gypsum contamination

3. Results and discussion Increasing the pH from 8 to 10 had a strong impact on the composition of the The composition and morphology of new phases (Figure 4). At DAP concentra- the new CaP phases formed by reacting tions of 0.05 and 0.1 M, alongside brushi- gypsum with a 0.1 M DAP solution with te (the only phase formed at pH=8) also increasing amounts of ethanol are repor- β-TCP and HAP were formed. Further in- ted in Figure 2. While fewer phases were creasing DAP concentration to 0.2 M (the formed with no or low ethanol addition, highest concentration that did not lead abundant new phases were found for 30 to immediate precipitation at pH=10), and 50 vol% ethanol additions. The new only -TCP and HAP were found. However, phases were identified by GID as brushite in all cases where the pH was increased and β-TCP, respectively. While in the case up to 10, diffused cracking occurred, of β-TCP (50 vol % ethanol addition) dif- because of excessive growth of the new fused cracking was observed, no cracks CaP phases. Cracking also led to flaking were visible in the case of brushite (30 and detachment from the substrate, vol % ethanol addition). Therefore, the which of course is undesired.

Figure 4: CaP phases formed by reaction with a DAP solution with increasing concentration at fixed 30 vol% ethanol content and pH=10

is thought to be the fact that brushite cial. On the contrary, even if less soluble formation is kinetically favored, because phases such as β-TCP and HAP could be - brushite contains HPO4 ions (which are obtained, their formation was always as- the main species originated from DAP sociated with diffused cracking, which 3- dissociation), whereas HAP contains PO4 would be detrimental for the treatment ions (which are only a minor fraction). success. Therefore, the formulation invol- Accordingly, it has recently been pro- ving 0.1 M DAP in 30 vol % ethanol at pH 8 posed that precipitation of CaP phases (leading to brushite and no cracking) ap- follows the Ostwald’s rule, i.e. the phase pears as the most promising one, among with the fastest precipitation rate (e.g. those investigated in this study. Further brushite) is preferentially formed, even if tests are in progress to assess its ability it is not the most stable phase (e.g. HAP) 9. to diminish the water solubility of trea- To verify this hypothesis, two solutions ted gypsum. In addition to the conser-

of CaCl2 and DAP were prepared with vation of sulfated marble, the same for- Ca/P molar ratios of 10:6 (corresponding mulation is expected to be suitable also Figure 3: CaP phases formed by reaction with a DAP solution with increasing concentration at fixed 30 vol% ethanol content at pH=8 to that of HAP) and 1:1 (corresponding to for the conservation of gypsum stuccoes. that of brushite). By powder XRD it was Further tests are in progress to assess the found that very similar results were ob- ability of this treatment to reduce the latter ethanol concentration was selected The change in composition of the new tained in the two cases, i.e. formation of water solubility of gypsum stuccoes and in the prosecution of the study. CaP phases, passing from brushite to brushite and a minor amount of HAP, increase their mechanical properties. When higher and lower DAP concent- β-TCP, when either the DAP or the etha- independently of the initial Ca/P ratio. rations were explored for a fixed 30 vol% nol concentrations were increased might The hypothesis that brushite formation ethanol addition, the new phases repor- seem counterintuitive, because in both is kinetically favored over that of HAP, 4. Conclusions ted in Figure 3 were obtained. With a DAP cases the Ca/P ratio was diminished. In basically independently of the Ca/P ra- concentration of 0.05 M, brushite was fact, a higher Ca/P ratio in the solution tio in the starting solution, seems hence In this study, gypsum samples were formed, with no visible cracking. On the would be expected to favor the forma- confirmed. treated with different aqueous solutions contrary, when the DAP concentration tion of phases with a higher Ca/P ratio In any case, brushite is about 27 times of DAP and the composition and morpho- was increased up to 0.5 M, β-TCP formed (hence, β-TCP instead of brushite). Howe- less soluble in water than gypsum 1, hen- logy of the new CaP phases were investi- but heavy cracking occurred. ver, this was not the case and the reason ce its formation is expected to be benefi- gated. The effects of ethanol addition to

146 147 E. Sassoni et al. How Conservation of marble artifacts by phosphate treatments: influence of gypsum contamination

the DAP solution (aimed at diminishing application to historic artifact, Mater consolidant, Constr Build Mater (143) gypsum solubility), DAP concentration, Design (88) (2015) 1145-1157. (2017) 298-311. and solution pH were investigated. The results of the study indicate that it is 2 Naidu S., Blair J., Scherer G. W., Acid-re- 10 Sassoni E., Graziani G., Scherer G.W., possible to obtain phases with very low sistant coatings on marble, J Am Ceram Franzoni E., Preliminary study on solubility (such as β-TCP and HAP) by Soc (99) (2016) 3421-3428. the use of ammonium phosphate for increasing the DAP concentration, by in- the conservation of marble-imitating 3 creasing the EtOH concentration or by in- Graziani G., Sassoni E., Franzoni E., Sche- gypsum-stuccoes, In: Papayianni I., Ste- creasing the solution pH. However, in all rer G. W., Hydroxyapatite coatings for fanidou M., Pachta V. (Eds), Proceedings these cases, the new CaP phases exhibi- marble protection: Optimization of cal- of the 4th Historic Mortars Conference ted diffused cracking, presumably becau- cite covering and acid resistance, Appl HMC2016, Santorini (GR), 10-12 Octo- se of excessive growth of the new phases. Surf Sci (368) (2016) 241-257. ber 2016, p. 391-398. For this reason, the formulation invol- 4 Sassoni E., Graziani G., Franzoni E., Sche- ving 0.1 M DAP in 30 vol % ethanol at 11 Gomis V., Saquete M.D., García-Cano J., rer G. W., Some recent findings on marb- pH 8 (leading to formation of uncracked CaSO4 solubility in water–ethanol mix- le conservation by aqueous solutions of brushite) seems like the most promising tures in the presence of sodium chlo- diammonium hydrogen phosphate, MRS one. Indeed, as brushite is much less so- ride at 25 °C. Application to a reverse Advances, DOI: 10.1557/adv.2017.45. luble than gypsum, a benefit in terms of osmosis process, Fluid Phase Equilibr reduction in dissolution in rain is expec- 5 Naidu S., Scherer G. W., Nucleation, (360) (2013) 248–252. ted. Treatment with the same solution growth and evolution of calcium phos- 12 with the aim of forming brushite is also phate films on calcite, J Colloid Interf Sci Eliaz N., Metoki N., Calcium Phosphate expected to be beneficial in the case of (435) (2014) 128-137. Bioceramics: A Review of Their History, gypsum stuccoes, both in terms of reduc- Structure, Properties, Coating Techno- tion of solubility in rain and increase in 6 Sassoni E., Naidu S., Scherer G. W., The logies and Biomedical Applications, mechanical properties. Relevant experi- use of hydroxyapatite as a new inorga- Materials (10) (2017) 334. mental tests are in progress. nic consolidant for damaged carbonate stones, J Cult Herit (12) (2011) 346-355.

Acknowledgements 7 Ma X., Balonis M., Pasco H., Toumazou M., Counts D., Kakoulli I., Evaluation of This project has received funding from hydroxyapatite effects for the consoli- the European Union’s Horizon 2020 rese- dation of a Hellenistic-Roman rock-cut arch and innovation programme under chamber tomb at Athienou-Malloura in the Marie Sklodowska-Curie grant agree- Cyprus, Constr Build Mater (150) (2017) ment No 655239 (HAP4MARBLE project, 333–344. „Multi-functionalization of hydroxyapatite for restoration and preventive con- 8 Snethlage R., Gruber C., Tucic V., Wendler servation of marble artworks“). E., Transforming gypsum into calcium phosphate – the better way to preserve lime paint layers on natural stone? In: Mimoso JM & Delgado Rodrigues J, Sto- ne consolidation in Cultural Heritage (2008) 1-13. References 9 Molina E., Rueda-Quero L., Benavente D., 1 Sassoni E., Graziani G., Franzoni E., Re- Burgos-Cara A., Ruiz-Agudo E., Cultrone pair of sugaring marble by ammonium G., Gypsum crust as a source of calcium phosphate: comparison with ethyl sili- for the consolidation of carbonate sto- cate and ammonium oxalate and pilot nes using a calcium phosphate-based

148 149 SWBSS 2017 | 20-22 September 4th International Conference on Salt Weathering of Buildings and Stone Sculptures Electrode placement during electro-desalination of NaCl contaminated sandstone – simulating treatment of carved stones

Electrode placement during electro-desalination of At both metallic electrodes there are carving needs to be clarified. The present pH changes due to electrolysis: paper is an experimental investigation, NaCl contaminated sandstone – simulating treatment of where T-shaped stone samples with a carved stones At the anode: known and relatively even distribution + - H2O → 2H + ½ O2 (g) + 2e (1) of NaCl are desalinated from one set of Lisbeth M. Ottosen* and Lovisa C. H. Andersson electrodes. Department of Civil Engineering, Technical University of Denmark, Lyngby, Denmark At the cathode: * [email protected] - - 2H2O + 2e → 2OH + H2 (g) (2) 2. Materials and methods Abstract 1. Introduction As seen from (1) and (2) pH decreases at the anode and increases at the cathode. 2.1. Stone for the experimental work Carved stone sculptures and orna- Salt induced decay of historic carved It is necessary to neutralize the pH chan- ments can be severely damaged by salt stone, sculptural or ornamental, is a ma- ges to prevent severe pH changes of the The experimental work was conducted 1,2 induced decay. Often the irregular surfa- jor cause for loss of important cultural stone. The work by underlined the im- with Gotland Sandstone, which is a stone ces are decomposed, and the artwork is heritage. Often the carvings making up portance of avoiding stone acidification, type often carved in Denmark. The grey lost. The present paper is an experimen- the surface pattern are lost first, and thus as in experiments without pH neutraliz- calcitic Gotland sandstone is composed tal investigation on the possibility for these constitutes the most fragile parts. ation; the stones were severely damaged of about 60 wt% quartz grains, cemented using electro-desalination for treatment Poulticing is a group of used and discus- next to the anode. Use of a calcite rich together by 7-10 wt% calcite, and with of stone with irregular shape with only sed techniques for removal of salts from clay poultice offers neutralization of the lower amounts of clay minerals, micas, acidification from the anode.3 The calcite two electrodes. The used Gotland sands- stone monuments. In these methods, a feldspar minerals, pyrite, and glanco- buffers the acid and the clay gives worka- tones were contaminated by NaCl in the poultice is attached to all outer surfa- nite.11 The porosity is unusually large, bility, so the poultice can have optimal laboratory. Due to the relatively good ho- ces of the monument and the idea is to around 15 % by volume.11 This type of contact to the stone surface during the mogeneity in initial salt concentration transport the salts from the stone into stone was chosen as it is commonly used treatment. obtained in this way, interpretation of the poultice by diffusion and/or advecti- and relatively homogeneous. T- shaped Electro-desalination has been tested the ED process were direct. Stones with on. However, in case of fragile carvings, stone samples for the experiments were an up-side-down T-shape formed the core successfully in laboratory scale in diffe- poulticing can be difficult to apply, as all cut from the same stone block, and of the investigation. Electro-desalination rent types of sandstones: Posta and Cotta these parts can further decompose by had the same size, see figure 1a. To have experiments were made with different sandstones4, Gotland Sandstone2, Nexø the physical contact. This paper is a la- a uniform distribution of the salt prior duration to follow the progress. Success- sandstone5, and granite.6 In fact no limit boratory investigation on the use of elec- to the experiments, the stone samples ful desalination of the whole stone piece has been seen in stone type. Also success- tro-desalination (ED) to remove salt from were contaminated with the salt in the was obtained, showing that also parts ful removal of chlorides4, nitrates7 and a sandstone with a T-shaped geometry, 8 lab. The stone pieces were dried at 105°C not being placed directly between the sulphates have been gained. These re- where the electrodes are placed in poul- and vacuum saturated by 80 g/l solution electrodes were desalinated. This is im- sults were obtained with stone segments tice around the simulated carving. of NaCl in a desiccator prior to the ED ex- portant in case of salt damaged carved with plane surfaces. The present work fo- ED is based on application of an electric periments. stones, where the most fragile parts thus cuses on the use of ED to desalinate car- can be desalinated without physically DC potential gradient to the salt conta- ved stones. placing electrodes on them. The Cl remo- minated stone. In the electric field, ions Two works9,10 have previously focused 2.2. Electro-desalination experiments val rate was higher in the areas closest in the pore solution are transported by on ED of carved stones. Both these ex- to the electrodes and slowest in the part, electromigration towards the electrode perimental works were carried out with Electrode compartments as seen in which was not placed directly between of opposite polarity, and hereby the ions stones, which had an uneven salt distri- the electrodes. This is important to in- from the damaging salts are transported bution, and both works confirmed that it figure 1b. A frame was folded in thin corporate in the monitoring program to out from the stone. The electrodes are was possible to extract salts from the ele- plastic and jointed with tape to fit the decide when a desalination action is fi- placed externally on the stone surface in vated carvings of the stone, which were horizontal part of the stone at each side nished. electrode compartments with poultice, not placed directly between the electro- of the attribute. The frames were filled and the ions from the salts concentra- des. The uneven salt distribution in these with poultice; a mixture of kaolinite and 3 Keywords: electro-desalination, sandsto- te in the poultice during the treatment. stones, made interpretation of the result CaCO3 with an initial water content of ne, carved stone, NaCl When the poultices are removed after difficult. Further both investigations 54%. Inert platinum coated electrode the desalination, the ions of the salts are only included one experiment each, and meshes were placed at the top of each removed with them. thus the progress in desalination of the electrode compartment. The sandstone

150 151 L. M. Ottosen and L. C. H. Andersson Electrode placement during electro-desalination of NaCl contaminated sandstone – simulating treatment of carved stones

a b Water content (%) pH Cl (mg/kg)

REF 10.4 ± 0.2 8.8 ± 0.06 4290 ± 530

ED1 9.9 ± 0.6 9.0 ± 0.9 4050 ± 1060

ED4 9.9 ± 0.3 10.8 ± 0.9 1580 ± 760

ED6 9.1 ± 1.1 11.3 ± 1.0 650 ± 500

ED13 3.0 ± 0.2 11.3 ± 0,9 21 ± 4 Figure 1: (a) The dimensions of the T-shaped stones and (b) The stone with electrodes.

Table 2: Water content, pH and chloride concentration (average ± standard deviation) in the stone segments at the end of the REF and ED experiments and electrode compartments were wrap- The water content in the segments was ped in plastic film to hinder evaporation. calculated as weight loss after drying at A reference experiment (REF) was made 105°C. The dried segments were grinded 3.2. Comparison of REF and ED1 with no applied current, but electrode in a mechanical mortar. Following, 10 g experiments compartments were placed at the stone powder was suspended in 25 ml distilled Figure 3a and b shows respectively the exactly as in the ED experiments. The water and agitated for 24 h. The samples resulting Cl concentrations in the diffe- duration of the REF experiment was 7 settled for 10 min and pH was measu- rent stone segments of the REF stone with days. Four ED experiments were made red. The samples were filtered through poultice but without an applied electric differing only in duration: 1, 4, 6 and 13 0.45 μm filter. The Cl concentrations were field and in the corresponding ED1 expe- weeks (ED1, ED4, ED6 and ED13). A cons- analysed by ion chromatography (IC, Di- riment with poultice and applied electric tant current of 10 mA was applied to onex DX-120). field, both after 7 days. the electrodes all through these experi- The poultices were changed every 7 day The REF stone shows lower Cl concent- ments. Table 1 shows the dry weight of except from ED13, where the last poul- ration in the segments, which had direct the different stones, the water content af- tice was used in 3 weeks. The poultice Figure 2: Segmentation and numbers after REF and ED physical contact with the poultice (seg- ter the vacuum saturation and the dura- was weighed, and Cl concentration, water experiments ments 3a, 3b, 3e and 3f, figure ).2 The Cl tion in weeks of the experiments. content and pH were measured using the concentrations in the two REF poultices After the REF and ED experiments, the same methods as for the stone segments. 3. Results and discussion were 3600 and 4200 mg/kg. As the Cl con- stones were segmented with hammer centration was non-measurable in the and chisel into 16 segments (Figure 2). 3.1. Overall results poultice before the experiment, the Cl was transported from stone to poultice Table 2 shows the overall results from during the 7 days of the experiment. In Weight (g) Initial water content (%) Duration (weeks) the ED experiments as average values the same period, the water content in the (± standard deviation) of water content, poultice decreased from 53.6% to 47.3%. REF 1528 12.0 1 (no current) pH, Cl concentration. The water contaent In addition, the average water content in was maintained at the same level when the stone decreased from 12.0% (table 1) ED1 1513 12.0 1 changing the poultice every week, but to 10.4% (table 2). Thus, the wrapping of not if the same poultice was used for 3 the stone and poultice did not fully hin- ED4 1536 11.6 4 weeks as was the case for the last poul- der evaporation. The possible transport ED6 1504 12.1 6 tice in ED13. The pH increased when the mechanisms for Cl from stone to poultice duration of ED was prolonged due to elec- are diffusion and advection. Calculating ED13 1481 11.9 13 trolysis at the cathode, which was not the pore water concentration of Cl in the buffered sufficiently in the poultice. The two materials based on stone concentra- Cl concentration decreases to a very low tion and water content, the pore water Table 1: REF and ED experiments. Dry weight of stone, initial water content after vacuum saturation and duration of experiment level in ED13. concentration in the stone (41 g/l in aver-

152 153 L. M. Ottosen and L. C. H. Andersson Electrode placement during electro-desalination of NaCl contaminated sandstone – simulating treatment of carved stones

a lower than the concentration in the poultices in the REF experiment, respectively.

3.3. Cl removal from stone over time during ED experiments

Figures 4a, b and c show the Cl concentrations in the different segments (figu- b re 2) of the ED experiments over time. Together these figures illustrate how the desalination progresses in the stone works. The Cl concentration decreased over time in all segments and a very low level of below 30 mg Cl/kg was obtained in every segment in ED13. Thus a very important conclusion can be drawn. It is possible fully to desalinate such attribute Figure 3: Chloride concentrations in the segment of the (a) REF stone and (b) and in the segment of the ED1 by ED. experiment after 7 days [mg/kg] The removal rate differed in the different segments. Figure 4a show that the Cl removal was fastesr in the segments age) is much higher than in the poultice in row 2 than in row 1, i. e. the desali- (8.3 g/l), and thus the concentration gra- nation occurred from the row closest to dient between the two allows a continu- the electrodes first and later to the upper ed diffusion. Whether the drying of the row. The same pattern is seen when com- poultice and stone transport water from paring the desalination of row 3 (figure stone to poultice cannot be assessed on 4b) and row 4 (4c) when e. g. comparing basis of the current data, however, this the Cl concentrations after 6 weeks in the transport must be considered limited, two figures. In row 3 the concentrations and thus advection is not considered as were less than about 500 mg/kg and in being of major importance. row 4, the concentrations were all higher Due to the applied electric field, the Cl than 500 mg/kg. That row 3 is desalina- concentration pattern in the ED1 stone ted before row 4 confirms the pattern differs from the REF (figure 3b). As Cl- found for ED of bricks in3, when the elec- electromigrates from the cathode end trodes were placed at the same side of the towards the anode, the concentration in bricks. Further, it is evident that the Cl the stone segments, which were in direct concentration decreased fastest from the contact with the cathode poultice (3e and end of the stone closest to the cathode 3f) were lower than the concentration in (figures 4a and b), e. g. 3f and 4f had. 3a and 3b (in the anode end). The concen- From the Cl concentration and weight tration in 3d and 2d are clearly decreased of poultice measured every week (when as well. The latter shows that the electric the poultices were changed) the accumu- field lines distributed into the attribute lated Cl removal over time can be calcu- already during the first week of ED. The Cl lated. Figure 5 shows the Cl removed into concentrations in the anode and cathode the anode poultice. The Cl removal in the poultices were 7670 mg/kg (cathode) and four ED experiments progresses very si- Figure 4: Cl concentration in the different stone segments as function of ED duration: (a) in the attribute, segments in 250 mg/kg (anode), which are higher and milar and as discussed in section 3.2, the row 1 and 2 (b) segments in row 3 and (c) segments in row 4.

154 155 L. M. Ottosen and L. C. H. Andersson Electrode placement during electro-desalination of NaCl contaminated sandstone – simulating treatment of carved stones

removal into the anode poultice is about desalination must be followed also by sed to a very low overall level during whi- vil Engineering. Kgs. Lyngby, Denmark, double the concentration in the REF ex- other means to ensure full desalination le this part was desalinated. Thus it can- June 20-23, 455-460. periment during 1 week. The total amount of Cl into the anode not be assessed whether the desalination During the first 3-4 weeks, the accumu- poultice was 3300 mg Cl (figure 5) and for is finished only on basis of the removal 6 Feijoo, J., Nóvoa, X.R., Rivas, T.; Mosque- lated Cl removal in the ED experiments comparison, only 65 mg Cl was removed rate into the poultice at the electrodes. ra, M.J., Taboada, J., Montojo, C., Carrera, increases almost linearly with duration. into the cathode poultice. This under- F., Granite desalination using electro- Between 4 and 9 weeks, the removed lines the effect of the applied electric migration. Influence of type of granite amount of Cl decreases every week and field. and saline contaminant. Journal of Cul- after 10 weeks the removal is very low. tural Heritage (14), 2013, 365–376. This pattern corresponds well to the results in figure 4, where it is seen that af- 4. Conclusions References 7 Skibsted, G., Ottosen, L.M., Jensen, P.E., ter 6 weeks, the concentration is lowered Paz-Garcia, J.M., Electrochemical desali- significantly in every segment. However, This investigation shows that ED has 1 Herinckx S, Vanhellemont Y, Hendrickx nation of bricks - Experimental and mo- combining the two experiments also potential to desalinate carved stone, as R, Roels S, De Clercq H., 2011, Salt remo- deling. Electrochimica Acta, (181), 2015, show, that it is important to continue the the salts can be removed from parts, val from stone building materials using 24-30. ED of the carved object after the point which are not directly between the an electric field. In: Proceedings from where the Cl removal rate into the poul- electrodes. In the present investigation, the international conference on salt we- 8 Ottosen, L.M., Electro-desalination of tice decreases. Evaluating the ED only on T-shaped stones were fully desalinated athering on building and stone sculptu- sulfate contaminated carbonaceous basis of figure 5 could mislead to the con- using two electrodes from an initial res, I. Iannou & M. Theodoridou (eds.), sandstone – risk for salt induced decay clusion that the desalination was suffi- concentration of about 4300 mg Cl/kg to Cyprus 19-22 October, 357-364. during the process. In Science and Art: cient after 6-7 weeks, but figure 4a shows, about less than 34 mg Cl/kg in all parts of A Future for Stone : Proceedings of the that at this point the Cl concentration is the stone. The part not situated directly 2 SKIBSTED, G., Ottosen, L.M., Jensen, P.E., 13th International Congress on the De- still rather high in segments 1a and 1b. between the electrodes was desalinated Electrochemical desalination of limes- terioration and Conservation of Stone. Thus in the case of ED treatment of sculp- with the slowest rate, and the Cl amount tone spiked with Na2SO4 – importance Eds. John J. Hughes, Torsten Howind, vol. tures and ornaments, the progress of the removed into the anode poultice decrea- of buffering anode produced acid. Paper 2, 2016, 897-904. E in Matrix changes and side effects induced by electrokinetic treatment of 9 Ottosen, L.M., Skibsted, G., Præstholm, porous and particulate materials. PhD T., Electrodesalination of sandstones thesis, 2013, Technical University of with irregular shapes and uneven dis- Denmark. tribution of salts. Proceedings from SWBSS2014 3rd International Conferen- 3Rörig-Dalgaard, I., Preservation of ma- ce on Salt Weathering of Buildings and sonry with electrokinetics – with focus Stone Sculptures. Ed. H. De Clercq, 2014, on desalination of murals. PhD The- p 405-420. sis. Department of Civil Engineering, Technical University of Denmark, 2009. 10 Feijoo, J., Matyscák, O., Ottosen, L.M., Rivas, T., Novoa, X.R., Electrokinetic 4 Ottosen, L.M., Christensen, I.V., Electro- desalination of protruded areas of sto- kinetic desalination of sandstones for ne avoiding the direct contact with NaCl removal - Test of different clay electrodes. Materials and Structures, poultices at the electrodes. Electrochi- (50:82), 2017, 2-15. mica Acta. (86), 2012, 192– 202. 11 Nord, A.G., Tronner, K., Effect of acid 5 Petersen, G., Ottosen, L.M., Jensen, P.E., rain on sandstone: The Royal Palace (2010) The possibility for using electro- and the Riddarholm Church, Stock- kinetics for desalination of sandstone holm. Water Air and Soil Pollution, (85), with low porosity. Proceedings from 8th 1995, 2719-2724. fib International PhD Symposium in Ci- Figure 5: Accumulated Cl amount removed into the anode poultice during the REF and ED experiments.

156 157 SWBSS 2017 | 20-22 September 4th International Conference on Salt Weathering of Buildings and Stone Sculptures How not to bother salts while grouting

How not to bother salts while grouting of moisture and salt in porous materials ly lower compared to the water amount by evanescent field dielectrometry, 4th suggested in the technical data sheets Chiara Pasian1*, Francesca Piqué2, Cristiano Riminesi3 and Albert Jornet2 International Conference on Salt Weathe- of PLM A and LEDAN RI.STAT BASE B. In 1 Department of Conservation and Built Heritage, Faculty for the Built Environment, University of ring of Buildings and Stone Sculptures, the suspension medium to prepare the Malta, Msida, Malta 2 Institute Materials and Constructions, University of Applied Sciences and Arts of Southern Switzer- Potsdam 20-22.9.2017), and on site, at the grouts, the amount of water was progres- land, Lugano, Switzerland San Vincenzo Oratory in Pazzallo, Tessin sively decreased. The three suspension 3 Institute for the Conservation and Valorization of the Cultural Heritage (ICVBC), National Research (CH), on decorated plasters afflicted by media selected were: 100% water (‘cont- Council, Florence, Italy * [email protected] severe problems of delamination associ- rol’ representing the traditional grouts ated with high amounts of soluble salts. preparation), 50% water: 50% ethanol, Abstract Keywords: injection grouting, water, alco- 25% water: 75% ethanol.4 PLM A and hol, dielectrometry, SUSI© LEDAN RI.STAT BASE B required the ad- The objective of this research work was 2. Materials and mixtures dition of a filler (quartz sand, inert and to assess if the injection of water-reduced non-porous) when mixed with water-et- grouts (water-ethanol-based) mobilises 1. Introduction and research aims Three injection grouts were evaluated: hanol to obtain a cohesive grout with no soluble salts to a lesser extent compared two hydraulic lime-based pre-mixed com- shrinkage.4 to the injection of conventional water-ba- Injection grouting is an intervention mercial grouts (PLM A CTS and LEDAN Set grout performance and working sed grouts. Ethanol was used as a partial aiming to stabilise and re-adhere del- RI.STAT BASE B TecnoEdile Toscana) and properties of the water-reduced grouts substitute for water in grouts prepara- aminated plaster/render introducing a one slaked lime-based grout designed in were tested in the laboratory following tion, being a poorer solvent for soluble compatible adhesive material with bul- our laboratory (called in this paper grout international standards and tests speci- salts when compared to water. Three in- king properties1, 2. Injection grouts are C).4 Grout C was formulated with the fically designed for injection grouts.7 The jection grouts were evaluated; for each suspensions composed of a binder, agg- following components and proportions grouts mixed with water-ethanol showed grout individually, the performance of regates, additives and a suspension medi- in parts by volume (pt/V): 1 pt/V slaked adequate shrinkage, porosity, cohesion, mixtures prepared with 100% water was um, which is typically water. In lime- and lime putty, 1.5 pt/V quartz sand (Ø < 740 and adhesion in replicas of delaminated compared to the performance of the hydraulic lime-based injection grouts, μm), 1 pt/V pozzolan (pozzolana flegrea), plaster, as well as good injectability, and same grout prepared with water-ethanol. water is required for chemical setting as 0.6 pt/V ammonium carbonate, and 0.1 therefore they are potentially suitable for To assess the different performance in a well as to improve injectability, however pt/V plasticiser (Sika Viscocrete-2S, Sika implementation on site.4 salt loaded system, the grouts were tes- an excessive amount of water may lead AG) 4. Ammonium carbonate, which nated into replicas of delaminated plaster to bleeding or segregation3, shrinkage turally decomposes releasing carbon intentionally contaminated with NaCl. and thus grout failure. During grouting dioxide, was added to provide carbon 3. Soluble salts movement – Assessment The salt movement was followed by eva- water is introduced in high amounts in dioxide to help carbonation in the defi- in the laboratory nescent field dielectrometry using the the porous materials of the wall pain- ciency of air (void in depth). The slaked SUSI© instrument. Grouts with reduced ting and/or historic structure (during lime used in this research is composed of In order to assess the mobilisation and

water content were also tested on site, pre-wetting and by grouting itself), po- ca. 50% Ca(OH)2 – 50% water (in weight), solubilisation of salts while grouting, at the San Vincenzo Oratory in Pazzallo, tentially jeopardising water-sensitive ori- determined gravimetrically. replicas of delaminated plaster intentio- Tessin (CH), on decorated plasters afflic- ginal materials, solubilising soluble salts, The grouting mixtures were prepared nally contaminated with NaCl were pre- ted by severe delamination associated and leading to their movimentation and using as suspension medium water (con- pared. The effects of the injection into with high amounts of soluble salts. Two damaging re-crystallisation. Water-re- trol) or water-ethanol mixed in different such replicas of water-ethanol grouts adjoining areas, comparable in condition duced injection grouts were designed proportions. Alcohols are less effective were compared with those of water-ba- and in salt content, were stabilised with and studied to reduce these risks. solvents for ionic substances such as so- sed grouts. the same grout differing only in the na- The objectives of this research include luble salts when compared to water, as ture of the suspension medium used (wa- the evaluation of performance of desi- they have the polar hydroxyl group (– ter-ethanol vs. water). Both in the labo- gned grouts with reduced water content OH) and a non-polar chain. Ethanol was 3.1. Materials and methods ratory and on site, it was experimentally in relation to the solubilisation and mo- selected as a partial substitute for water verified that water-ethanol-based grouts vimentation of soluble salts in contami- because of its miscibility with water and As ethanol is a poorer solvent for solub- caused a significant reduction of salts nated plasters. The performance of the because it is a poor solvent for salts.5, 6 le salts when compared to water5, 6, the mobilisation, bothering salts to a lower grouts was assessed in the laboratory The amount of suspension medium to injection of water-ethanol-based grouts extent compared to a typical water-based through evanescent field dielectromet- prepare the grouts (mL suspension me- should solubilise salts in the delamina- grouting intervention. ry using SUSI© instrument* (C. Riminesi dium/g dry pre-mixed grout or g slaked ted plaster to a lesser extent compared to and R. Olmi, Diagnostics and monitoring lime) used for this research is significant- the injection of water-based grouts. This

* Acronym for Sensore per la misura di Umidità e Salinità Integrato, integrated sensor for measuring humidity and salinity. 158 159 C. Pasian et al. How not to bother salts while grouting

was verified in the laboratory with SUSI©* 3.2. Calibration – Determination and 1) PLM100-0 (100% water) vs. PLM25-75 water : 75% ethanol solution for grouts system (acronym for Sensore per la misura correction of ethanol contribution (25% water : 75% ethanol) PLM25-75, LEDAN25-75 and C25-75). Mea- di Umidità e Salinità Integrato, integra- surements of MC and SI were performed ted sensor for measuring humidity and SUSI© typically works in presence of 2) LEDAN100-0 (100% water) vs. LE- with SUSI© immediately after pre-wet- © DAN25-75 (25% water : 75% ethanol) salinity). SUSI is an instrument based on water to determine MC and SI. In this re- ting. Injection of the grout was then per- Evanescent Field Dielectrometry (EFD), search the authors wanted to assess the 3) C100-0 (100% water) vs. C25-75 (25% formed, one grout in each replica (Fig. 1). which is a technique deriving from die- solubilisation of salts in presence of a water : 75% ethanol) After the injection, measurements of lectric spectroscopy.8 SUSI© works in the water-ethanol grout, therefore a new ca- MC and SI were performed (Fig. 2): imme- microwave range (1-1.5 GHz), it is porta- libration of the instrument was needed. Before the injection, measurements diately after the injection; at 10 minutes ble and allows real-time non-invasive Samples (10 cm x 10 cm and 2 cm thick) with SUSI© of MC and SI were performed interval within the first hour, after 2, 3 measurements of sub-surface moisture made of the same plaster of the replicas on the dry replicas. The replicas were pla- and 4 hours, after 24, 48 and 72 hours. In content (MC) and the detection of soluble (fine plaster, slaked lime : sand 1:2 pt/V) ced in a vertical position to simulate the this way, variations of MC and SI could be salts (SI, i. e. salinity index). The values of were prepared. 9 plaster samples were injection into a wall. Before injecting the followed over time for each replica. MC and SI can be separately and indepen- prepared with no soluble salts and 9 ana- grout, replicas were pre-wetted. Pre-wet- dently measured by the instrument. The ting is performed to clear the cavity and logous samples were prepared with plas- 3.4. Discussion moisture content (MC) of porous materi- ter contaminated with 0.5% NaCl (0.5% of to wet the internal surfaces of the void, als such as plaster is measured due to the the weight of the sand used to prepare with the aim of reducing the absorpti- Graphs in Fig. 3a-f show the variation dielectric contrast between water and the plaster and 0.7% of the total weight of on of the liquid contained in the grout of MC and SI over time for each replica. the host material (plaster).9 The water 4 the plaster). The samples were immersed mix and reduce grout shrinkage. The For each grout typology (respectively content which can be measured ranges in three different solutions: 100% water, pre-wetting liquid employed for each PLM, LEDAN and grout C) it is visible that between 0 and 20% (water volume frac- 50% water : 50% ethanol and 25% water : replica was the one used to prepare each MC is much higher for grouts prepared tion); the range of salinity index varies 75% ethanol. In each solution 3 non-con- grout respectively (100% water for grouts with 100% water (controls) compared to between 1 and 10 from no-salt to satura- taminated samples and 3 contaminated PLM100-0, LEDAN100-0 and C100-0; 25% grouts prepared with water-ethanol (Fig. ted salt conditions.10 SI is a measure of samples were immersed. The samples 3a-c-e). This was predictable, as MC indi- the salt ions present in solution in the were kept lifted from the bottom of the cates the water volume fraction in the host material. Replicas of delaminated plaster were container with the solution through host material (plaster). prepared (in stratigraphy: brick, coarse glass rods; the samples were immersed More interesting are the results con- plaster composed of slaked lime : sand in the solution up to the height of 1 cm cerning SI (Fig. 3b-d-f); for all grouts the 1:3 pt/V, fine plaster composed of slaked of their total thickness. mixture prepared with 25% water : 75% lime : sand 1:2 pt/V). The delamination For every sample, MC and SI were mea- ethanol (red in the graphs) has a much (void) was produced between the coarse sure at t0 (dry samples); after the samples lower SI compared to the control prepa- and the fine plaster, using the following were placed in the solutions, for each red just with water (black in the graphs). procedure: the coarse plaster was ap- sample MC and SI were monitored at 30 This because a 25% water : 75% ethanol plied on a brick tile; on this plaster, in seconds interval until complete saturati- solution solubilises salts to a lesser ex- the middle of the tile, small cylinders of on. The time for complete imbibition was tent compared to water.5, 6 Data regarding Figure 1: Injection of grout PLM25-75 into the replica of recorded for each sample. A calibration of SI are presented in Table 1: dry-ice (solid CO2) were stacked together delaminated plaster in the shape of a truncated pyramid; the the instrument taking into consideration second plaster layer (fine plaster) was the contribution of ethanol, with and wi- Grout Range of SI over time applied on the dry-ice cylinders. As the thout soluble salts, was performed. dry-ice sublimes, an empty void between PLM100-0 2.0 – 8.5 the two plaster layers is created.4 The fine PLM25-75 2.2 – 4.0 plaster was contaminated with 0.5% NaCl 3.3. Injection of water-reduced grouts LEDAN100-0 2.3 – 6.5 (0.5% of the weight of the sand used to into contaminated replicas prepare the plaster and 0.7% of the total LEDAN25-75 1.0 – 3.8 weight of the plaster). Six replicas of delaminated plaster (see C100-0 2.3 – 5.5 The grouts chosen for the experiment Section 3.1) were prepared.4 The following were the grouts with the least amount of grouts were injected, one for each replica: C25-75 0.0 – 1.8 water (25% water : 75% ethanol) vs. the Figure 2: SUSI© measurements on a replica after 48 hours controls (100% water). from the injection Table 1: Range of salinity index (SI) over time of the replicas

* Pat. CNR n. 1626 – FI 2004A000187. 160 161 C. Pasian et al. How not to bother salts while grouting

The maximum value of SI after the in- which used to be a cowshed and which jection of PLM25-75 is less than half com- was thus a vehicle for nitrates. pared to the maximum value after the The pilaster located on the north wall injection of the control (PLM100-0). The of the Oratory presented delamination maximum value of SI after the injection between the brick support and the coarse of LEDAN25-75 is about 2/5 lower com- plaster (Fig. 4) affecting an area of appro- pared to the maximum value after the ximately 30 x 80 cm and 3 cm thick. Dif- injection of the control (LEDAN100-0). fuse salt efflorescence was present both Values referred to grout C25-75 are par- inside the delamination void and on the ticularly low (maximum value of SI 1.8): plaster surface; the salts present were the maximum value measured for C25- identified as manly composed of nitro- 75 is about 1/3 of the maximum value kalite (KNO3 – equilibrium relative hu- measured for the control (C100-0). midity 94.6% at 20°C 11) (A. Kueng, SUPSI, To increase wetting and reduce the internal report). overall provision of water, pre-wetting is This area was considered suitable for often performed by conservators with a testing and comparing the effects of resolution of water and alcohol4, typically duced water grouts with those of regular followed by the injection of a water-ba- grouts. sed grout. The graphs reporting the measures performed with SUSI© show that the injection of grouts prepared with 25% water : 75% ethanol (after pre-wetting with the same solvent mixture) solubilises salts in the replicas to a lower extent compared to the injection of the same grouts prepared with 100% water.

4. On site application of grouts with reduced water content

Grouts with reduced water content were tested on site, at the San Vincenzo Oratory in Pazzallo, Tessin (CH). The Ora- tory, decorated with wall paintings and stuccoes and dated to the XVIII c., was abandoned in the last decades and it is affected by severe problems of delamination associated with high amounts of soluble salts. The north wall of the Ora- tory was the most affected by problems deriving from soluble salts and high humidity (RH constantly around 80% with 19-21°C T in the months during which the research was carried out; environ- Figure 3: Graphs showing MC (3a PLM, 3c LEDAN, 3e grout C) and SI (3b PLM, 3d LEDAN, 3f grout C) over time, from t0 (dry replica) to 72 hours after the injection. The first point of measure on the graph is performed on the dry replica, the mental monitoring by A. Kueng, SUPSI). Figure 4: Delaminated pilaster in San Vincenzo Oratory, second point immediately after pre-wetting, the third point immediately after the grout injection; measures are then The north wall is adjacent to a building north wall taken at 10 minutes interval within the first hour, after 2, 3 and 4 hours, after 24, 48 and 72 hours

162 163 C. Pasian et al. How not to bother salts while grouting

4.1. Objectives, materials and prelimi- 40 cm above the ground. The salts distri- ter the injection over time to follow the nary testing bution in a wall is extremely heteroge- grouts setting (every 5 minutes for the neous both on the surface and in depth. first hour; 2, 3, 4 hours after the injecti- The objective of the application on site In particular, in the case of capillary ri- on; 24 and 48 hours after the injection). was to verify that also in a real case the sing damp, ions are distributed in the In order to better assess the positioning injection of a grout prepared with wa- wall in different concentrations also ac- of the grouts in the delamination (Fig. 6), ter-ethanol mobilises salts to a lesser ex- cording to the height from the ground11; the grouts were both slightly heated in tent compared to the injection of a grout therefore it is preferable to compare a water bath prior to injection. In Fig. 6a prepared just with water. two areas at the same height from the PLM100-0 is more clearly visible compa- A piece of plaster, already detached in- ground. red to PLM25-75 (Fig. 6b) because it was side the delamination, was extracted; its Before performing the intervention, possible to heat it longer (PLM25-75 is et- porosity was measured in the laboratory crystallised salts were removed from hanol-based: it could not be heated long (standard SIA 262/1:2003 12) as well as its the surface (area of 600 cm2) with a dry because its suspension medium is more compressive strength (standard UNI EN brush and weighed (0.26 g). Salt efflore- volatile than water). It was verified that 1015/11 13). The porosity and the compres- scence was evenly distributed on the sur- the positioning of two grouts injected in the delamination was kept separate by the foam rubber, as planned.

σc (N/mm2) n (%) tot. por. UE (%) capill. por. LP (%) air pores compr. strength Figure 6a: Thermoimage after the injection of PLM100-0 4.3. Assessment Original plaster 35.2 28.5 6.7 3.24 After two months, the salt crystallisa- PLM25-75 46.4 39.7 6.7 3.57 tion was assessed on the surface of the Figure 5: The two areas of the pilaster where PLM100- pilaster where the intervention was per- 0 (control, 100% water, fully blue) and PLM25-75 (25% Table 2: Porosity and compressive strength of the original plaster and grout PLM25-75 water : 75% ethanol; a quarter blue) were injected formed. Salt efflorescence was removed with a dry brush in the two separate but equivalent areas (where grout PLM100-0 and where grout PLM25-75 were injec- sive strength of the original plaster were face. Salts were also removed with a dry ted) and separately weighed. In the area compared with those of the grouts with brush inside the delamination aided by where the water-ethanol-based grout reduced water content, tested in the labo- a mini vacuum cleaner, which allowed PLM25-75 was injected the salt crystal- ratory with the same standard procedu- removing both salts and dust inside the lisation was about 40% less (in weight, res. The injection grout with the compres- void. 0.12 g) compared to the area where the sive strength most similar to that of the The selected area was divided vertical- water-based grout PLM100-0 was injected original plaster was PLM prepared with ly by placing in the middle of the delami- Figure 6a: Thermoimage after the injection of PLM100-0 (0.21 g). 25% water : 75% ethanol (PLM25-75) (see nation a piece of foam rubber (3 cm x 20 Table 2). PLM25-75 has a higher porosity cm, 3 cm thick as the delamination itself) compared to that the original plaster (see to create a barrier. In this way the two 5. Final discussion and conclusions Table 2); this is considered good because grouts (PLM100-0 and PLM25-75) could the porosity of the injection grout should be injected in the two separate chambers Injection grouts with reduced water be similar or higher to that of the origi- created in the delamination area at the content were evaluated, comparing the nal material 1. same height (Fig. 5). water-ethanol-based grout (25% water The intervention was followed with a : 75% ethanol) with the correspondent thermocamera (Agema 570 Thermovi- control (100% water). In the present rese- 4.2. Intervention sion PRO). The two different areas were arch their performance in terms of salts monitored before the intervention, du- mobilisation was assessed: The lowest part of the delaminated pi- ring the pre-wetting (performed for both laster, an area of 30 cm width x 20 cm areas with a solution 25% water : 75% - in the laboratory with SUSI© on replicas height (delamination 3 cm thick) was ethanol), during the injection respecti- of delaminated plaster contaminated chosen for the test. This area was about vely of PLM100-0 and PLM25-75, and af- Figure 6b: Thermoimage after the injection of PLM25-75) with NaCl

164 165 C. Pasian et al. How not to bother salts while grouting

- on site in a limited area of intervention References 7 B. Biçer-Şimşir & L. Rainer, Evaluation of affected by delamination and severe so- Lime-Based Hydraulic Injection Grouts luble salts problems 1 I. Griffin, Pozzolanas as Additives for for the Conservation of Architectural Grouts: An Investigation of Their Wor- Surfaces – A Manual of Laboratory and In the laboratory, all the replicas injec- king Properties and Performance Cha- Field Test Methods, Los Angeles: The ted with the water-ethanol-based grout racteristics, Studies in Conservation 49 Getty Conservation Institute, 2013. showed a much lower salinity index (1) (2004) 23-34. (SI) compared to replicas injected with 8 R. Olmi, M. Bini, A. Ignesti, S. Priori, C. the correspondent water-based grout 2 S. Rickerby, L. Shekede, Fan Zaixuan, Riminesi, & A. Felici, Diagnostics and (control). This verified because a lower Tang Wei, Qiao Hai, Yang Jinjian, F. Pi- monitoring of frescoes using evane- amount of salts was brought in solution qué, Development and Testing of the scent-field dielectrometry, Measure- by the water-ethanol grout suspension Grouting and Soluble-Salts Reduction ment Science and Technology, 17:8 medium compared to the 100% water Treatments of Cave 85 Wall Paintings, (2006) 2281. grout suspension medium. The data of in: Conservation of Ancient Sites on the SI measured with SUSI© prove that wa- Silk Road. Proceedings of the Second In- 9 R. Olmi & C. Riminesi, Study of water ter-ethanol-based grouts solubilise salts ternational Conference on the Conser- mass transfer dynamics in frescoes by to a much lesser extent compared to wa- vation of Grotto Sites, Mogao Grottoes, dielectric spectroscopy, Il Nuovo cimen- ter-based grouts. Dunhuang, People’s Republic of China, to della Società Italiana di Fisica C 31:3 In the limited area treated on site, the June 28-July 3, 2004, Ed. N. Agnew. Get- (2008) 389. injection of a water-ethanol-based grout ty Conservation Institute, Los Angeles, led to a much lower salts re-crystallisa- 2010, pp. 471-479. 10 N.Proietti, D. Capitani, V. Di Tullio, R. tion on the plaster surface compared to Olmi, S. Priori, C. Riminesi, A. Sanso- the injection of the water-based control. 3 B. Biçer Şimşir, I. Griffin, B. Palazzo-Ber- netti, F. Tasso & E. Rosina, MOdihMA Water-reduced grouts (25% water : 75% tholon, L. Rainer, Lime-based injection at Sforza Castle in Milano: Innovative ethanol) proved to be suitable for site grouts for the conservation of architec- Techniques for Moisture Detection in implementation with the advantage of tural surfaces, Reviews in Conservation Historical Masonry. In Built Heritage: bothering salts to a lower extent compa- 10 (2009) 3-17. Monitoring Conservation Management red to a typical water-based stabilisation 2015, pp. 187-197, Springer Internatio- intervention. 4 C. Pasian, F. Piqué, A. Jornet, Non-struc- nal Publishing. tural injection grouts with reduced water content: Changes induced by the 11 K. Zehnder & A. Arnold, Monitoring Acknowledgements partial substitution of water with alco- wall paintings affected by soluble salts, hol, Studies in Conservation 62:1 (2017) in: The Conservation of Wall Paintings, We would like to thank Dr Christian 43-54. ed. S. Cather, The Getty Conservation Paglia, director of the Institute Materials Institute, Marina del Rey, 1991, 103-108. and Constructions (SUPSI), allowing the 5 S.P. Pinho, E.A. Macedo, Solubility of conduction of the tests at the IMC. Many NaCl, NaBr, and KCl in water, methanol, 12 SIA 262/1 2003, Perméabilité à l’eau, thanks to Andreas Kueng (IMC, SUPSI), ethanol, and their mixed solvents, Jour- Construction en Béton–Spécifications who performed the salts analyses and nal of Chemical & Engineering Data 50 complémentaires, Société suisse des in- the environmental monitoring in San (2050) 29-32. génieurs et des architectes, Zurich. Vincenzo Oratory. Thanks are also due to Prof Tiziano Teruzzi (IMC, SUPSI) for his 6 A. Kolker, J. de Pablo, Thermodynamic 13 UNI EN 1015/11 1999, Methods of Test help in the use of the thermocamera. Modelling of the Solubility of Salts in for Mortar for Masonry–Part 11, Deter- Mixed Aqueous-Organic Solvents, In- mination of Flexural and Compressive dustrial & Engineering Chemistry Rese- Strength of Hardened Mortar, Milano. arch 35 (1996) 228-233.

166 167 SWBSS 2017 | 20-22 September 4th International Conference on Salt Weathering of Buildings and Stone Sculptures Moisture Transport during Poultice Application

Moisture transport during poultice application fect other parts. Nevertheless the questi- In the study presented here we show on occurs to which depth that argument the time resolved moisture development is valid. Especially in cases where no or in three different substrate materials tre- Christoph Franzen1*, Martin Aulitzky2 and Stephan Pfefferkorn2 1 Institut für Diagnostik und Konservierung an Denkmalen in Sachsen und Sachsen-Anhalt e. V., minor evaporation happens and the poul- ated with five different poultice types. 2 Dresden, Germany tices are wet all the time of application 2 Fakultät Bauingenieurwesen/Architektur, Hochschule für Technik und Wirtschaft Dresden, the cross section where the effects of dif- Germany fusion is effective is not clear. However, 2. Materials and methods *[email protected] we focus here on drying poultices and address their limitations by formulating 2.2. Building material to desalinate Abstract the water. With time water evaporates at the requirements: The poultice material the surface, which is the poultice surface. should be free of salts. The same applies Three different important building Water movement is the key parameter The water in the porous material moves to the water to prepare the poultice. Thus materials were chosen for the experi- in poultice application for desalination. to the poultice where it evaporates. The the use of demineralised water is recom- mental work. Two varieties of natural Water is the eluent and transport me- dissolved salt ions are transported in the mended. The ready to use wet poultice sandstone material of Elbe Sandstone: dium for the salts to be extracted from poultice. When the water evaporates the shall be workable on site. The material Postaer Sandstone and Cottaer Sands- the porous substrate. While from the hy- salts precipitate. The salts are eliminated is either applied by trowel or spittle, or tone and one type of brick was used for pothetical point of view the dynamical from the object by removal of the poul- is sputtered, than in terms of viscosity the tests. In table 1 some data of those processes of water ingress, dissolution tice. - Thus there are three moisture pa- it has to fulfil the requirements of the porous building materials is presented, of salts, transportation and precipitation thways during poultice application to be spraying device. The poultice has to adhe- detailed material description is given by.4 seems to be well known, measured data regarded which happen partially in par- re to the surface to the object. The adhe- on the water quantities at specific stages allel and change their magnitude during rence has to stand all the time of applica- of the process are scarce. Especially, time the process. There is the ingress of water tion, as lost of contact does terminate the 2.3. Poultice material approximations about the change in from the poultice into the substrate and function. On the other hand the poultice main movement direction of the water in the substrate, and then there is the should be removable from the surface ea- Five commercially available poultice are unidentified. By experimental data reflow from the substrate to the poul- sily and free of residues. The poultice has materials from four trade specific pro- we give indications to the total amount tice and finally the evaporation from the to bring the optimal amount of water to ducers were chosen: RE and RA are from of water used with different poultice poultice. Evaporation, also to be regarded the system, enough to get salts in solu- building material companies, RI, for fine compositions and the development of as a driving force, acts from the very be- tion but on the same time not flush the grained materials, and RII, for coarse the water ratio during the process until ginning, although the first situation to salts deeper into the material. Finally the grained, from a producer specialises on equilibrium dryness is reached. In con- be regarded is the water ingress from the poultice has to catch the salts. To reach restoration material and material A is clusion we suggest increasing the ratio wet poultice into the dry substrate due all these demands nearly all poultices for pure cellulose (BC200). The material ty- of water and probably the poultice layer to concentration balancing and capillary desalination nowadays are multi compo- pes were characterised optically, by XRD thickness from cycle to cycle to increase suction. Effects of gravitational flow will nent systems. There is material to carry and sieve measurements, the workability the effective cross section stepwise. not be discussed here, but should be ta- the water, clays to secure adherence and of the pasted material was noted, as well ken into consideration much more in fu- workability, and matrix components for as the development during application. Keywords: desalination, water mo- ture. stability, minimising the shrinking and Furthermore on prisms of the hardened vement, salt transport, poultice material Of course in practical execution of that providing a pore structure.1, 3 poultice strength measurements were simple and effective idea some more executed. effects do happen, sometimes limiting the efficiency of the application. First to 1. Introduction mention is that the procedure acts on the surface of salt contaminated porous Material/sample code Postaer Sandstone/P Cottaer Sandstone/C Brick/Z Desalination by poultice applications material. Thus it is to be assumed, that it Total pore volume is a widely applied technique in restora- does not effect to salts sitting deeper in 21 22 - tion.1 The general idea is to apply tem- the material. What could be a drawback Vol. (%) porarily an additional layer on the salt could also be regarded as advantage. In W-Value (kgm2h-1/2) 9.12 5.61 4.94 contaminated porous material. That the very most cases the salts are concen- poultice layer brings water to and into trated near to the surface, thus poultices the substrate material. Salts dissolve in act where the problem is and so not af- Table 1: Three important building materials were chosen for experiments

168 169 C. Franzen et al. Moisture Transport during Poultice Application

2.4. Methods, sampling procedure te of the system was visually inspected, the poultice removed for visual inspec- poultice/ Wc Wp shrink Comp. Flex. -2 Emoddyn To win insight in the state of water mo- tion, moisture and salt determination, code [-] [Lm ] [mm/m] Strength Strength vement at eight pre-defined time stages and the substrate block gravimetrically dummies were destructively sampled. inspected, drill powder sampled in 12 RE 0,322 7.0 21.1 1.63 0.79 2.95 Consequently for each of the five poul- stages of 1 cm depth and analysed for tice materials eight dummies from three moisture content6 and salt content by RA 0,600 7.7 21.6 0.68 0.41 0.34 substrate materials were prepared. The electrical conductivity. Finally the block natural sandstone was cut into blocks was chopped in stone masons fashion for R I 0,475 8,7 72.0 nd 1.65 3.13 of 10x15x10 cm, two bricks pieces were further visual inspection of the ink drop. glued together to win comparable sur- R II 0,475 6,6 18.7 1.16 0.50 1.03 faces for the experiments of 11.3x15 cm. The block sides were sealed, the surfaces 3. Results and Discussion A 4,500 14.8 52.1 nd 0.06 nd contaminated with NaCl-salt spray. All dummies were marked with a 50 µl drop 3.1. Poultice material of ink.5 Samples were taken in increasing Table 2: Five commercially available poultice materials time intervals after 4 hours, 7 hours, 1 day, 2 days, 3 days, 6 days, 2 weeks, and Generally looking quite similar with 3 weeks. At each sampling stage the sta- naked eye you could notice in dry raw poultice material RE sand, clay and cellulose, RA contents lesser sand but foam glass spheres, R I looks like pure powder and R II appears as powder with some sand. Table 2 shows significant material data of the poultices, from mass related water content (Wc) as recommended by the producers and the resulting mass of water provided to the wall (Wp). The strength data was determined on samples comparable to the end stage of the desalination action. Obviously there are major differences in the poultice performances due to divers material compositions. Figure 3 Figure 1: Chopping of the block shows their size fractions with standard sieves. When it comes to the creation of specific pore sizes for effective desalination as recommended by7 we suggest to define adapted mesh combinations for the future. Of course the cellulose content in all mixtures hampers the sieving. For displaying the results (diagramms in Figure 3) giving the sieve fraction in mass could be misleading due to different densities e.g. sand and foam glass. Reference to volume lead to other difficulties. Figure 4 gives the XRD patterns of the four smallest mesh fractions, where the

Figure 2: Ingress of ink drop clay is to be expected. Here significant Figure 3: Size fractionated poultice materials (A – cellulose not displayed)

170 171 C. Franzen et al. Moisture Transport during Poultice Application

material differences become obvious as happens in most cases after 4 h or 7 h, so- sometimes bentonite, sometimes kaoli- metimes 24 h. Despite the lowest W-value nite is used in the industrial produced laboratory determined for brick the wa- material mixtures. Thus as the clays mater uptake in the very first (4) hours is nipulate the workablity, effect the work always higher than into the sandstones. of later cleaning und influence the final The data provides the willingness of the strength we could not yet decide if it is poultice to release water to the system: the type of clay or the amount or the to- A>>RA≥RE>>RI≥RII. In that very first 4 tal sieve line, which leads with RI to tho- hours the sandstones under all poultices se problematic results of high shrinkage, take always about 20-30% of the water which got also visible through the expe- equal to the minimum of what the brick riments. does take. There the pore system of the substrate is the limiting factor of water uptake. Taking into account the expe- 3.2. Water distribution vs. time rimental procedure we conclude, that the graviational effects of water flow in The customer has to find out, which those partially supersaturated water sy- poultice material works best in his sub- tems have to be taken into account much ject on the given substrate. Wendler8 sug- more. gested a classifiing parameter to evaluate When the poultice gets dryer than the the effectivity of different poultice mix- subtrate the general water movement is tures by the time the water movement is returned at least mathematically. Never- changed. As to be seen in Figure 5 (first theless in parallel the water front in the column) it is evaluated when the wetted substrate moves on. That is to be seen in st nd substrate begins to loose weight. That Figure 5 2nd column, where the water con- Figure 5: 1 column: water contents in experiment: total, poultice, substrate and mount evaporated vs time; 2 : depth resolved water content; 3rd: conductivity results

tent in the substrate is shown in all depth 3.3. Water related effects at all investigated time stages. In brick material most poulitces affect a depth The results of the conductivity measu- about 4 to 5 cm. In Postaer sandstone RE rements are heterogenious. Most proba- an RA reach about 7 cm while R I and R II bly the concentration of contamination on the other hand affect 4 cm which par- was too low, nevertheless trends are to tially can be explained by a problematic be seen. In the first hours in all cases performance during the application. the salt peak from the very top is flus- In spite of the lower W-value the Cot- hed into about 3 cm depth. After 1 day taer sandstone it tends to take up more it broadens and, of course, decreases. A water than the Postaer Sandstone. It is low, decreased restitic contamination in important to note, that Cottaer sand- all the profile where water took access stone in many cases has not completely to is visible until the very end of inves- dried out after 21 days. Pure cellulose tigation. Under poultice material which carrys about double of the water into the looses eager contact during the applica- system compared to the poultice mixtu- tion, most probably due to high shrinki- res. With such a cellulose treatment Post- ge and low adherence to the substrate, a aer sandstone can be wetted totally down top peak and a tail into the materials is to 12 cm depth. visible. Also the ink drop test has limited interpretation potential. The ink was practi- Figure 4: XRD pattern of four poultice materials, smallest mesh fraction

172 173 C. Franzen et al. Moisture Transport during Poultice Application

cally invisible in the brick, showed a cone binations to offer workable poultices. Re- 3 WTA Guideline 3-13-18/E, Salt reduction in Postaer sandstone. In Cottaer sands- fering to the amount of water provided of porous building material with poul- tone it was hardly visible but tended to by the pultice for the desalination the tices. WTA Publications, Munich, in pre- be vanished on the top and was most materials have minor differences, whi- paration, 2018. intensive in about 2 cm depth. Also in le pure cellulose is able to really flood the poultice material the ink was solely a system. The performance of the poul- 4 Grunert S, Der Elbsandstein: Vorkom- visible in the pure cellulose material AC. tices already when simple material data men, Verwendung, Eigenschaften Elbe However, we recommend to use much like shrinkige measurement is taken into Sandstone: deposits, use, properties- more than 50 µl ink for contamination account, is for some of the comercially GeologicaSaxonica. J Cent Eur Geol for possible future experiments. available products problematic. The ad- 52(53), 2007, 3–22. vantages and disadvantages of specific clays taken for the mixtures are not well 5 Lenz R, Systemoptimierte Salzminde- 3.3. Water related effects understood. Here much more investiga- rungskompressen, Material – Wirkung tion on one and but also information, Tagungsbeitrag Retrospektive & Pers- probably the concentration of contami- what kind is implemented, is needed. pektive(n) in der Wandmalereirestauri- nation was too low, nevertheless trends The study shows that the moisturing erung Werkstattgespräch 2016, in prep., are to be seen. In the first hours in all of the substrate happens within the first 2017. cases the salt peak from the very top is hours of poultice application. Depth of flushed into about 3 cm depth. After 1 several centimeters are reached, a distri- 6 WTA, 2002, WTA-Merkblatt 2-11-02/D: day it broadens and, of course, decreases. bution of salts from the top into the stone Messung der Feuchte von mineralischen A low, decreased restitic contamination at that stage is unavoidable. Water depth Baustoffen. München: Fraunhofer IRB in all the profile where water took access penetration seems solely partially predi- Verlag, 2006. to is visible until the very end of inves- cable and connected to the W-value of the tigation. Under poultice material which substrate. Gravitation effects and teh wa- 7 Franzen, C., Analytische Begleitung von looses eager contact during the applica- ter release properties of the poultice are Salzreduzierungsmaßnahmen, in: Pra- tion, most probably due to high shrinki- effects much more to be regarded. Thus xisorientierte Forschung in der Denk- ge and low adherence to the substrate, a for successively progressive maintaining malpflege – 10 Jahre IDK-, Hrsg.: Institut top peak and a tail into the materials is deeper zones it is adviseable to start with für Diagnostik und Konservierung an visible. thinner (less water) poultices in the be- Denkmalen in Sachsen und Sachsen-An- Also the ink drop test has limited inter- ginning and increase the layer thickness halt e.V., 2006, 31 – 40. pretation potential. The ink was practi- for cycle to cycle to increase the effective cally invisible in the brick, showed a cone cross section stepwise. 7 Pel, L., Sawdy, A. & Voronina, V., Physical in Postaer sandstone. In Cottaer sands- principles and efficiency of salt extrac- tone it was hardly visible but tended to tion by poulticing. Journal of Cultural be vanished on the top and was most Heritage 11(1), 2010, p 59-67. intensive in about 2 cm depth. Also in the poultice material the ink was solely References 8 Wendler E., Möglichkeiten und Grenzen visible in the pure cellulose material AC. einer Salzminderung durch Kompres- 1 However, we recommend to use much HERITAGE, A. et al., Current use of poul- sen: Materialoptimierung und aktuelle more than 50 µl ink for contamination tices in the conservation of monuments. Fallstudien an Naturstein, Ziegel, Putz for possible future experiments. In: A. HERITAGE, A. HERITAGE & F. ZEZ- und Wandmalerei, Natursteinsanierung ZA, (eds). Desalination of Historic Buil- Stuttgart, 2007, S. 29-38. dings, Stone and Wall Paintings. Lon- 4. Conclusion don: Archetype Publications Ltd, 2013.

In the study five different poultice ma- 2 WTA Guideline 3-13-01/E Non-destruc- terials were characterised and tested on tive desalination of natural stones and three substrate materials. The commerci- other porous materials with poultices. al produceres use different material com- WTA Publications, Munich, 2005.

174 175 SWBSS 2017 | 20-22 September The application of hydroxyapatite-based treatments to salt-bearing porous limestones: A study on sodium sulphate- 4th International Conference on Salt Weathering of Buildings and Stone Sculptures contaminated Lecce Stone

The application of hydroxyapatite-based treat- with causing weathering, salts inside the or soluble phases.4, 5 Stone samples were stone may interfere with consolidants’ contaminated by sodium sulphate, which ments to salt-bearing porous limestones: A study penetration and hardening reactions, is one of the most widespread and noxi- thus possibly having a negative impact ous among the salts that can be found on sodium sulphate-contaminated Lecce Stone on the treatments’ outcome: for this rea- on site, then were treated with two diffe- Gabriela Graziani1, 2*, Enrico Sassoni1, George W. Scherer3 and Elisa Franzoni1 son, desalination procedures are normal- rent HAP-based formulations, previously 1 Department of Civil, Chemical, Environmental and Materials Engineering (DICAM), University of ly performed prior to consolidation. Ho- developed by the authors for limestone Bologna, Italy wever, even after desalination some salts and marble consolidation.1, 2, 3, 6 The ef- 2 Laboratory of NanoBiotechnology (NaBi), Rizzoli Orthopaedic Institute, Bologna, Italy 3 Department of Civil and Environmental Engineering (CEE), Princeton University, USA may remain inside the stone, interfering fects of the treatments on phase formati- *[email protected] with the success of consolidation. There- on were analyzed and compared to those fore, the effects of the application of con- obtained by ethyl silicate, currently the solidants on salt-bearing stones needs to most used product for the consolidation be investigated. of limestone and, in particular, of Lecce Abstract comparison to ethyl silicate was deter- Treatments based on hydroxyapatite Stone. mined, as it is currently the most used (HAP) (produced by treating the stone Salt weathering is among the most se- consolidant for this lithotype. Finally, with an aqueous solution of diammoni- vere phenomena affecting porous limes- the durability of the most promising um hydrogen phosphate, DAP) have been 2. Materials and Methods tone, often leading to the loss of precious treatment to further salt crystallization proposed for limestone consolidation, material from historical sculptures and was evaluated, again in comparison with exhibiting remarkable efficacy, compa- 2.1. Materials building façades. Together with exerting ethyl silicate. tibility and durability.1, 2 However, tests a weathering action, salts can also ham- Promising results were obtained, as carried out so far mainly focused on Tests were carried out on Lecce Stone per the success of consolidation treat- the presence of sodium sulphate did not uncontaminated samples, which is ra- (LS), which has non-uniform microstruc- ments, by interfering with consolidants’ prevent the HAP- based consolidants rely the case on site. Salt-related issues ture and mechanical properties, even penetration and/or curing reactions. from uniformly distributing in the sto- could arise for real specimens because within a single stone slab. Cores 5 cm in On site, application of consolidants on ne, sealing cracks and forming insolub- not only HAP but also metastable calci- height and 2 cm in diameter were used non-contaminated stone is rarely feasible phosphate phases. Interestingly, the um phosphate (CaP) phases and, possibly, for all the tests, except for the evaluation le, especially in the case of porous limes- application on non-desalinated stone re- phosphate salts could form, depending of the durability after consolidation, for tones; for this reason, testing application vealed that the nature and the amount on the substrate contamination3; moreo- which 5x5x5 cm3 cubes were used, given of consolidants on salt laden stone is of of the phases that form as a result of the ver, HAP might easily incorporate foreign the low resistance of the stone to salt particular relevance to guarantee their treatment are dependent on the amount ions, possibly altering its crystallinity weathering. All cores and cubes were cut successful application in the field. of salts in the stone. Finally, most promi- and solubility.4, 5 from one quarry slab. In this paper, two hydroxyapatite sing HAP-based formulation was found In the literature, the presence of fo- Samples were contaminated by cy- (HAP)-based treatments have been ap- to enhance the stone’s durability towards reign ions has been found to promote cles of sodium sulphate crystallization plied to Lecce Stone, a very porous or- further salt crystallization, also compa- HAP growth, but also to cause the for- (samples “SALT”), performed as follows: ganogenic limestone, highly susceptible red to ethyl silicate. mation of soluble phases, depending on to salt weathering. The two treatments substrate composition, nature of conta- 1) Wetting phase (24 h): samples were im- differ for DAP concentration, applicati- Keywords: salt weathering, cultural heri- minants and surface roughness.6 For this mersed to a depth of about 1 cm in a on procedure and presence/absence of tage, calcium phosphates, limestone reason, the investigation of application 14 wt% solution of sodium sulphate

ethanol in the formulation. Prior to tre- of a hydroxyapatite-based treatment on decahydrate (Na2SO4•10H 2O) in deioni- ating, specimens were subjected to salt salt-laden stone is of particular interest. zed water; crystallization cycles in a sodium sulpha- 1. Introduction In this study, we studied the effects te solution, to cause salt contamination of applying HAP-based treatments on 2) Drying phase (22 h): samples were and induce weathering. They were then Salt weathering is among the main salt-contaminated Lecce stone, a porous dried in ventilated oven at 50°C; desalinated, purposely leaving a percen- causes of stone deterioration, especially organogenic limestone characteristic of 2- Baroque Architecture of southern Italy. 3) Cooling (1h) and measuring of weight tage of salts in the stone (SO4 after des- for porous limestone, that can undergo alination ~0.15-0.3 wt%). Phase formation severe damage, such as flaking, scaling This stone contains significant amounts and dynamic elastic modulus (total: 2 h). and distribution as a result of different and pulverization, all leading to the loss of silica, aluminum, phosphorus and ma- contamination levels were investigated, of significant amounts of material and of gnesium, which could also have an im- Two cycles were performed on the and the efficacy of the treatments in architectural/sculptural details. Together pact on the formation of hydroxyapatite cores, 5 on the cubes. The number and

176 177 The application of hydroxyapatite-based treatments to salt-bearing porous limestones: A study on sodium sulphate- G. Graziani et al. contaminated Lecce Stone

duration of the cycles were selected ba- water:poultice ratio 1:6). Limewater 2.4. Evaluation of the treatments’ effects treatment, which mitigates the impact of sed on preliminary tests, with the aim of poultice is applied with the aim of pro- the high heterogeneity of the stone on contaminating the stone with a suitable viding further calcium ions to boost Phase formation was examined by FT- the evaluation of the treatments’ perfor- amount of salts, without causing rup- HAP formation and to remove possible IR. For a better understanding of the ef- mance. To have a preliminary evaluation ture, and conserving a morphology sui- soluble phosphate remaining in the fect of the presence of salts on the HAP of the efficacy of the treatments, dyna- table for mechanical testing. Salt content stone after treatment; based treatments, FT-IR was performed mic elastic modulus was also measured as a result of the cycles was measured by on heated (HEAT), desalinated (DESALT) on samples artificially weathered by he- ion chromatography (IC). 2) Treatment “0.1M”: A 0.1 M DAP solu- and non-desalinated (SALT) samples, to ating (instead of salt weathering), so as To reproduce the conditions on site, tion also containing 0.1 mM CaCl2 highlight the impact of different levels to determine their effects on uniformly where stone is normally desalinated be- and 10 vol% ethanol is applied, follo- of salt contamination. FT-IR was also per- weathered and uncontaminated stone. fore the application of the consolidant wed by the application of a 0.1 M DAP formed on ES-treated samples to verify For the durability tests, after desali- (even if a fraction of salts usually re- + 0.1 mM CaCl2 solution without the whether the presence of salts interferes nation and treatment, 4 stone cubes for mains in the stone), part of the cores and addition of ethanol. Both solutions are with ES curing. each condition were subjected to 10 addi- all the cubes were partially desalinated applied by brushing (10 strokes) 7. This Phase morphology and distribution tional salt crystallization cycles, perfor- after contamination. This also allowed procedure was developed for the con- were evaluated by SEM/EDS, to under- med exactly as during the contaminati- evaluation of the impact of different le- solidation of marble, for which it was stand whether the presence of salts pre- on phase. The effects of the cycles were vels of salt contamination on phase for- found that ethanol promotes HAP for- vents consolidants from penetrating evaluated by measuring weight loss and mation (samples “DESALT”). Desalination mation (and thus better efficacy) with deeply into the stone and to evaluate the dynamic elastic modulus after each cycle was performed by applying a deionized lower amounts of precursors. treatments’ capability of sealing cracks. and after 10 cycles and desalination. water poultice (water:cellulouse pulp ra- Mechanical properties after treatments tio 1:5, thickness ~1 cm). The poultice was Ethyl silicate (ES) was also tested for were evaluated in terms of dynamic ela- kept sealed for 24 hours to allow solubili- comparison’s sake, as it is currently the stic modulus (5 samples were taken for 3. Results and Discussion zation of the salts, then left to dry in con- most used product for consolidation of each condition). Dynamic elastic modu- tact with the samples to allow advection Lecce Stone. For ES treatment a commer- lus was selected as it is non destructive Morphology and composition of Lecce of the salts towards the external surface. cial formulation (Estel 1000, CTS, Italy, and allows testing of the same samples Stone was investigated by SEM/EDS (Figu- Residual salt content was evaluated by IC composed of 75 wt% ethyl silicate and before and after salt contamination and re 1). Untreated stone consists of a cement performed on the ground stone after salt 25 wt% white spirit D40, also containing extraction and filtering. The investigati- 1 wt% dibutyltin dilaurate as a catalyst) on of phase formation and of mechanical was applied by brushing (10 brush stro- properties after treatment was also car- kes). After treatment, samples were left ried out on uncontaminated, uniformly to cure for 1 month in room conditions weathered samples. For this purpose, a prior to testing. A deionized water poul- batch of stone cores (samples “HEAT”) tice was applied to ethyl silicate-treated was artificially weathered by heating at cubes before performing durability cy- 400°C for 1h, according to a procedure cles (water:poultice ratio 5:1, poultice previously developed by the authors for kept on the samples for 48 h), to remove promoting cracks formation in the stone.1, 2 possible hydrophobicity linked to residual ethoxy groups, that could interfere 8 2.2. Treating procedure with the uptake of saline solution. The poultice was not applied to ES-treated Two HAP-based treatments were tested cores to avoid desalination and because that were previously developed for the they did not need to be subjected to any consolidation and protection of limesto- characterization test involving absorpti- ne and marble, respectively: on of water or saline solution. Stone cores were treated on the whole 1) Treatment “3M”: A 3 M DAP solution external surface, while cubes were trea- is applied by brushing (10 strokes) ted on one face perpendicular to the bed- followed by the application of a lime- ding planes. water poultice (1.7 g/l Ca(OH)2, lime- Figure 1: SEM images and EDS spectra of untreated Lecce Stone

178 179 The application of hydroxyapatite-based treatments to salt-bearing porous limestones: A study on sodium sulphate- G. Graziani et al. contaminated Lecce Stone

mainly constituted by calcite with traces about 0.15-0.30 wt% of salts remain in of phosphorous, magnesium, alumini- the stone, depending on the specimen um, iron and silicon, that envelops fos- and the initial contamination, their resi- sil shells only composed of calcium car- dual presence being confirmed by FT-IR bonate (Figure 1). To have a preliminary and EDS spectra (Figure 2 and 3). Interes- evaluation of the treatments’ efficacy tingly, salt content after the application prior to application on contaminated sto- of treatment 3M significantly decreases = ne, they were first applied to samples ar- (average SO4 content =0.0244 wt%), indi- tificially weathered by heating, that also cating that the application of a limewa- exhibit more uniform weathering (Table ter poultice allows for some desalination 1). Significant efficacy is obtained for the of the stone, which can be considered an application on heated stone, especially advantage of this treatment. Salt content for treatment 3M that causes higher in- in 0.1M and ES samples, instead, remains creases in Ed compared to ethyl silicate. substantially unaltered in all the exa- Remarkably, both ES and 3M fully restore mined areas of the samples.

Ed [GPa]

After heating Before heating After heating (“HEAT”) ΔEd ΔEd and consolidation

3M 10.5±0.6 7.7±0.4 -27% 12.5±0.8 +62%

0.1M 10.6±1.2 7.8±0.6 -26% 9.1±0.7 +17%

ES 12.8±0.5 9.3±0.2 -27% 13.0±1.1 +40%

Table 2: Dynamic elastic modulus on samples artificially weathered by heating. Values for stone before heating are different between the different batches of samples because of the heterogeneity of the stone.

the initial dynamic elastic modulus of Phase formation after treatments is re- unweathered stone. A lower increase is ported in Figure 2. FT-IR spectra confirm obtained by treatment 0.1M, which ho- the presence of residual sulphate in des- wever can still be considered promising, alinated stone. No formation of soluble particularly in view of the much lower calcium phosphates phases nor phospha- amounts of precursors used. te salts or other soluble compounds is Lecce Stone exhibits a very low re- detected for any of the treated speci- sistance to salt weathering, significant- mens. For sample 3M the application on ly variable between different slabs and non-desalinated (“SALT”), compared to even different areas of the same slab. desalinated stone (“DESALT”), leads to a For this reason, two salt crystallization higher formation of reaction products. cycles were selected to achieve salt con- However, given the position of the bands, tamination without causing rupture in it seems that octacalcium phosphate the stone. As a result of the cycles, about (OCP) is the phase that forms in the pre- = 1.43 wt% sulphates (expressed as SO4 ) sence of higher amounts of salt (samples are deposited inside the stone, the total “SALT”), while HAP forms in its absence amount being quite variable between and for lower levels of contamination different specimens. After desalination, (samples “HEAT” and “DESALT”).9 Figure 2: FT-IR spectra on untreated and treated specimens, for different levels of salt contamination

180 181 The application of hydroxyapatite-based treatments to salt-bearing porous limestones: A study on sodium sulphate- G. Graziani et al. contaminated Lecce Stone

In general, the effect of salts could be due suggesting that HAP nucleation could be to the fact that their presence and the inhibited by ions contained in the stone damage they cause affect surface rough- matrix (probably magnesium, which is ness, specific surface and the presence of known to be an HAP growth inhibitor 4, 5). sites available for nucleation. This indica- For Lecce stone, therefore, the presence tes that the level of salt contamination of trace elements in the stone seems to has an impact on phase formation; howe- have a stronger impact on HAP formation ver, despite OCP being a metastable pha- than sodium sulphate has. se (eventually converting in HAP), it is far Consistently, also in sample 0.1M, less soluble than calcite, hence its forma- phosphate formation is favoured on the tion is not considered a drawback of the shells, indicating that, even with ethanol treatment. The absence of any band cor- in the solution phase, formation is lower responding to sodium sulphate, in agree- in ion-doped areas of the stone. The for- ment with IC, confirms that salts were mation of calcium phosphates is gene- removed by the application of the lime- rally lower compared to 3M and it is qui- water poultice during the 3M treatment. te non-uniform (in some areas P is not Treatment 0.1M leads to the formation detected by EDS). However, because the of HAP (see the presence of the phospha- stone itself contains phosphorous in va- te stretching bands at 600 and 560 cm-1, riable concentrations, it is not possible to and the shift in the main band with detect the formation of phosphate pha- the appearance of a shoulder at ≈1100 ses unless a large amount forms. The tre- cm-1), but a significantly lower amount atment, however, is effective in sealing of phosphate phases forms compared cracks, also in depth in the sample, again to 3M, in contrast to what was reported indicating that the presence of salts does for marble.7 However, because the stone not prevent the penetration of the con- itself contains phosphate fractions, the solidant. The ability of the HAP-based lower phase formation complicates the consolidants to seal cracks is reflected in exact discrimination between HAP and their ability to fully restore the dynamic other calcium phosphate phases. Data elastic modulus of untreated stone (see for SALT-0.1M treatment are not reported Table 2). The lower phase formation in here, as no significant differences can be 0.1M treatment compared to 3M, howe- assessed between the spectra for SALT ver, combined with the lower increase of and DESALT samples. Sample ES exhibits dynamic elastic modulus, indicates that a broad band in the 1040-1100 cm-1 area, the formulation needs further optimi- probably resulting from the overlapping zation for being applied to Lecce Stone, Figure 3: SEM images of treated samples’ cross sections. In figure, the red square indicates significant formation of of the bands at 1037 cm-1 (Lecce Stone), and that a higher concentration of DAP phosphate phases in correspondence of one shell, the arrows indicate a crack that remains unselaed after ethyl silicate -1 9 treatment. 1080 cm (Si-O-Si of hydrolyzed ES) and should be used, given the high porosity 1104 cm-1 (sodium sulphate). No bands at of the stone (experiments are in pro- 1170 cm-1 are present, indicative of Si-O-C.11 gress). However, because HAP is found The formation of higher amounts of higher specific surface and the different Phase morphology and distribution in to form and seal cracks also in depth in calcium phosphates in the presence of surface texture of the samples. Conver- treated samples was evaluated by SEM/ the sample, it seems that ethanol has a higher amounts of salts could be due to sely, the formation of OCP instead of EDS. In sample 3M, phosphate phases positive impact on HAP formation. SEM increased surface area and roughness HAP is possibly due to the fact that dif- were detected both on the surface and images on ethyl silicate treated samples (increase in surface roughness being cau- ferent ions (including Na) can exert op- in depth in the samples, indicating that indicate abundant phase formation, but sed by salt damaging the stone), similar posite impact on phase formation, and the presence of salts does not prevent a also reveal the presence of unsealed to what was found in 10 for ammonium on crystallinity and solubility of reaction uniform penetration of the consolidant. cracks that were not noticed in HAP-ba- oxalate treated samples, where more products, so not only the ions themselves Interestingly, phosphate formation is sed treatments. For these samples, EDS calcium oxalate was found to form on but also the relative amount of each ion higher adjacent to and inside the shells, analysis did not provide meaningful in- non-desalinated samples because of the can have an effect on phase formation 4, 5. which exhibit no ion-contamination, formation, as a high and very variable

182 183 The application of hydroxyapatite-based treatments to salt-bearing porous limestones: A study on sodium sulphate- G. Graziani et al. contaminated Lecce Stone

Ed [GPa] pability of increasing stone durability phosphate formation. The 0.1M treat- Before salt After 2 cycles and towards further weathering cycles was ment, however, still guarantees sea- ΔEd After consolidation ΔEd weathering cycles desalination evaluated (results in Table 3). Untrea- ling of cracks even in the depth of the ted specimens were also considered for sample, which suggests that optimiz- 3M 16.5±0.3 16.2±0.6 -2% 17.8±0.3 +10% comparison’s sake. As can be seen in Ta- ation of this treatment is worth pur- ble 3,3M samples experience the lowest suing; 0.1M 10.9±0.3 10.8±0.2 ≈ 11.4±0.3 +6% decrease in dynamic elastic modulus after the cycles, thus indicating that this 5)´Treatment 3M is promising as it in- ES 11.1±0.3 10.7±0.6 -4% 13.9±0.4 +30% treatment has the best durability. This creases resistance of the stone to also indicates that, despite apparent low further salt weathering, whereas ethyl Table 2: Dynamic elastic modulus of samples subjected to salt crystallization cycles efficacy obtained for the cores (caused silicate treatment does not. by limited weathering level of the specimens), still this treatment is promising Ed [GPa] as it enhances the stone durability towards salt weathering better than ethyl Before salt weathering cycles After 10 cycles and desalination ΔEd silicate does. 5. References Untreated 18.1±0.4 14.2±0.3 -22% 4. Conclusions 1 Sassoni E., Graziani G., Franzoni E., An 3M 18.1±0.4 16.0±1.1 -12% innovative phosphate-based consoli- The application of two HAP-based tre- dant for limestone. Part 1: Effectiveness ES 17.4±0.3 13.6±0.7 -22% atments was tested on Lecce Stone cont- and compatibility in comparison with aminated by sodium sulphate. The follo- ethyl silicate, Construction and Buil- wing conclusions can be derived: Table 3: Durability of treated samples in terms of dynamic elastic modulus ding Materials (102), (2016), 918-930. 1) The presence of salt in the stone does 2 Sassoni E., Graziani G., Franzoni E., An not prevent penetration of the HAP-ba- amount of silica is contained in the stone because the stone was not severely da- innovative phosphate-based consoli- sed consolidants nor cause formation itself. maged by weathering cycles, which were dant for limestone. Part 2. Construction of phosphate salts instead of calcium The efficacy of the treatments was aimed at investigating the effects of salt and Building Materials (102), (2016), 931- phosphates; evaluated by dynamic elastic modulus contamination rather than causing sig- 942. (see Table 2). Weathering resulted main- nificant damage to the stone. However, 2) Depending on the amount of salts con- ly in the pulverization and consequent because phosphate formation was found 3 tained in the stone, either HAP or OCP Sassoni E., Naidu S., Scherer G. W., The detachment of the external layers of the to occur also deep in the samples (Fi- are obtained, both suitable for consoli- use of hydroxyapatite as a new inorga- samples, but did not cause significant gure 3) and because only HAP and OCP dation of limestone; nic consolidant for damaged carbona- decrease in the elastic modulus of the and no other CaP phases were detected te stones, Journal of Cultural Heritage, non-detached parts. None of the HAP-ba- by FT-IR (Figure 2), it can be concluded 3) Increasing levels of sodium sulphate (12), (2011), 346-355. sed treatments provides dramatic increa- that no pore occlusion occurred, which affect the nature of the phases that ses in dynamic elastic modulus compa- would prevent the consolidant from pe- form, but promote the formation of 4 Supova M., Substituted hydroxyapatites red to the values after weathering. The netrating deeply into the stone, and no higher amounts of calcium phospha- for biomedical applications: A review, increase is higher for samples treated by formation of metastable phases took tes. Instead, other ions contained in Ceramics International, (41), (2015), ethyl silicate, which, however, exhibit the place as a result of salts contamination, the stone (possibly magnesium) affect 9203-9231. highest amount of residual salts in the which were the main concerns regarding the nucleation of HAP, leading to a pores, which could cause some pore oc- the application on salt weathered stone. nonuniform coverage of the substrate, 5 Boanini E., Gazzano M., Bigi A., Ionic clusion resulting in increases in Ed not This implies that higher increases in me- and to the preferential formation of substitutions in calcium phosphates ascribable to a consolidating effect. Ho- chanical properties can be obtained for HAP in uncontaminated shells. synthesized at low temperature, Acta wever, the values after treatment exceed more severely damaged samples (experi- Biomaterialia (6), (2010), 1882-1894. those of the unweathered specimens, ments are in progress). 4) Comparing the 0.1M and 3M treat- thus indicating that low increases in To better investigate the effect of the ments, the latter exhibits better ef- 6 Sassoni E., Graziani G., Franzoni E., Re- dynamic elastic modulus are registered 3M treatment compared to ES, their ca- ficacy, as it results in more calcium pair of sugaring marble by ammonium

184 185 G. Graziani et al.

phosphate: comparison with ethyl silicate and ammonium oxalate and pilot application to historic artifact, Materi- als and Design 88 (2015) 1145-1157.

7 Sassoni E., Graziani G., Franzoni E., Scherer G.W., Some Recent Findings On Marble Conservation By Aqueous Solutions Of Diammonium Hydrogen Phosphate, MRS Advances, DOI: 10.1557/ adv.2017.45.

8 Franzoni, E.; Graziani, G.; Sassoni, E., TEOS-based treatments for stone consolidation: acceleration of hydrolysis–condensation reactions by poulticing, Jour- nal of Sol-gel Science and Technology, (74), (2015), 398 – 405.

9 Tao J., FTIR and Raman Studies of Struc- ture and Bonding in Mineral and Or- ganic–Mineral Composites, Methods in Enzymology,(532) (2013), 533 – 556.

10 Fort R., Alvarez de Buergo M., Go- mez-Heras M., Vazquez Calco C., He- ritage, weathering and conservation, Proceeedings of the International conference on heritage, weathering and conservation, 21-24 June 2006, Madrid, Spain.

11 Rubio F., Rubio J., Oteo J.L., A FT-IR study of the hydrolysis of tetraethylortosilica- te [ES], Spectroscopy Letters, (3), (1998), 199–219.

186 Salt reduction SWBSS 2017 | 20-22 September 4th International Conference on Salt Weathering of Buildings and Stone Sculptures Evaluation of desalination and restoration methods applied in Petra (Jordan)

Evaluation of desalination and restoration methods applied in Petra (Jordan)

Wanja Wedekind* and Helge Fischer Applied Conservation Science (ACS), Goettingen/Berlin, Germany * [email protected]

Abstract on mortar, restoration procedures, control of success One of the major causes for the deterioration of historical monuments made 1. Introduction up of sandstones can be attributed to the circulation and evaporation of salt la- The ancient city of Petra is situated den fluids percolating through the rock south of the Dead Sea in the Kingdom of material, an aspect that has been eit- Jordan. It has gained international reco- her overlooked or neglected in previous gnition when it was declared a UNESCO restoration attempts, often with dama- World Heritage Site in 1985 for its unique, ging consequences. Particularly the lack approximately 650 relatively well preser- of consideration for the importance of a ved facades carved out of the sandstone quantitative analysis of the salt content bedrock. This magnificent monumental of the rock, has led to adverse effects in rock architecture, however, is in danger of severe and accelerating disintegrati- subsequent restoration attempts. Figure 1: a) Distribution of alveolar weathering and traces where the rainwater drains at façade no. 826. b) Tafoni 3 In the course of restoration of antique on due to various forms weathering and with sampling point of the crust and drill core holes. c) Cascade-like water stains after a rainfall. d) Idealized cross tombs no. 825 and 826 in the world he- neglect. section of the sampling area. e) Electrical conductivity at depth in the drill core samples. ritage site of Petra the role of salt in the In order to halt or slow down the ob- weathering process was fully taken into servable decay of these monumental al Highness Prince Hassan bin Talal, the 2. Environmental conditions account and the restoration process exe- structures and preserve them for future then Crown Prince of the kingdom. cuted accordingly. generations, a project was devised and Scientific investigation and research 2.1. Climate With regard to the desalination two executed aiming at establishing a Jordani- continued intermittently from the year different methods were applied: the poul- an institution capable of independently 2000 to 2016 by the University of Goet- Petra is located in a semiarid climatic tice method and the sprinkling method. handling all aspects of the restoration tingen (GZG), the University of Applied zone with a mean annual rainfall of 190 The degree of desalination was in both effort on a continuous and permanent Arts and Sciences of Hildesheim (HAWK), mm.1 Heavy rainfalls may occur during cases intermittently measured until the basis. The result was the establishment and the private institution Applied Con- the winter months, whereas the winters concentration of the rock was evaluated of an institute that is now known as the servation Science (ACS). This work focu- are cold and wet. Due to the vicinity to to be low enough to initiate the restorati- “Conservation and Restoration Center in sed primarily on the role of salt (NaCl) in the Mediterranean Sea lying some hund- on process with application of the selec- Petra (CARCIP)”. The project was managed the weathering process, its impact on the red kilometres to the west and winds ted restoration mortar. Its condition was by one of the authors of this paper, fun- behaviour of restoration mortars, as well blowing primarily from that direction a then probed by long term successive in- ded by the German government and exe- as the development of efficient and safe high load of soluble elements are cont- spections. cuted within the framework of the “Ger- desalination procedures. ained in the rainwater .2 This study discusses the results and man International Cooperation Agency Particularly on the basis of these con- limits of both procedures and the (GIZ)” and the “Department of Antiqui- tinuing activities, new methods in the techniques and methods applied in the ties (DOA) of the Jordanian government conservation effort, as well as new resto- course of the restoration of the two mo- in the years 1993 to 2002. From its start 2.2. Rock material ration materials were devised and imple- numents. the project furthermore secured the sup- mented by the authors and presented in port of the “Bavarian State Conservation The tomb facades of Petra were chisel- this paper. Keywords: Petra, sandstone, salt weathe- Office (BLfD). It was for most of its dura- led out of the reddish sandstones of the ring, desalination procedures, restorati- tion also under the patronage of His Roy- Cambrian Umm Ishrin Formation. In

190 191 W. Wedekind and H. Fischer Evaluation of desalination and restoration methods applied in Petra (Jordan)

detail colours vary from reddish-brown reaching the surface. This process cons- the surface during heavy rainfalls. (Figu- decided to start its use on monument no. to yellowish-brown, gray or white. The titutes the most important factor for the res 2c, 6b) 825. brown colour is attributed to the deposi- destruction of built and rock-cut archi- There, however, surprisingly it soon 6, 7, 8, 9, 10, 11 tion of limonite (FeO.OH.nH2O) in the po- tecture in Petra. turned out to respond in a rather un- res during formation of the sandstones, The resulting formation of tafoni com- 4. Description of the tombs no. 825 and expected and unfavourable way. The while the reddish colours stem from the prise locally limited, as well as comple- 826 material started to crumble, sometimes

precipitation of hematite (Fe2O3) flakes tely deteriorated areas, their size ranging already a short time after application. Ca- in the pores. The different lithological from more centimetres to of several me- The tombs no. 825 and 826 are located reful evaluation of the problem led to the layers of the sandstone are medium- to ters in diameter (Figure 1b). This kind of beside each other along the southern tip conclusion that the rapid disintegration fine-grained and show porosities ranging extreme deterioration is generally sur- of the so-called “Kings Wall” (Figure 2 b). of the mortar was due to salt contamina- from 4.2 % to 20.6 %. Average values are rounded by mostly undamaged rock. Tomb no. 825 is on the left and is also tion of the rock onto which the mortar 13 %.3 In contrast, the limonite rich varie- known as the tomb of the 14 graves (Fi- was applied. ties show a low porosity of 7%.4 3.2.Weathering processes gure 2 b). It was evidently built after the This can also easily be explained by construction of the water pipes in the the fact, that Silica-sols are colloidal so- second century within the rock. It is 19 Comparative examinations of the lutions of SiO2 in water that are negati- 3. Weathering processes and agents sandstone monuments of Petra have meters high and 9 meters wide. vely charged.12 The negative charges are shown that two varieties of tafoni de- Tomb no. 826 has a double tin frieze balanced and stabilized with a defined 3.1. Weathering in arid and semi-arid velopment can be distinguished. Dama- (Figure 2b). It is of the Assyrian type and amount of positively charged ions, na- environments ged spots evolve either as horizontal or 21 m tall. The maximum width of the pe- mely Na+. After application of the mort- vertical tafoni (Figure 2). destal part is 12 meters, whereas the top ar, the Na-ions migrate from the rock Areas in arid or semi-arid climatic zo- Horizontal tafoni are usually found part is 10 meters wide. The monument’s into the mortar leading to a destabiliz- nes like Petra show an increasing amount above the microporous limonite layers actual ground level is invisible because ation of the silica-sol binder. (Figure 9b). of salt deposition.5 Therefore, weathering which in turn are frequently observed of submersion in rubble. Sometimes even salt efflorescence appe- due to salts plays a crucial role. parallel to sedimentation layers. This is ared on the surface of the freshly applied Weathering phenomena observed in to be expected due to the blocking and mortar. In consequence, collapsing of the Cambrian sandstones in Petra are accumulation of seeping aquifer water 4.1. The monuments in relation to en- the mortar took place (Figure 9 a). It thus mainly caused by salt crystallization on above the less permeable limonite layers. countered weathering processes turned out that salt reduction was of pri- the retreating surface of the rock, due to Vertical tafoni develop left and right of me importance in any further attempt to evaporation of percolating saline water areas affected by rainwater flowing down Tomb no. 825 is affected by seeping apply the chosen mortar. aquifer water moving through the massive rock, whereas tomb no. 826 is affected by run off water. Therefore, at tomb 825 5.2. Salt extraction by application of horizontal oriented alveoles are created, poultices whereas at monument no. 826 mainly vertical ones are observed (Figure 2). The negative results experienced with the repair mortar on monument 825 stressed the necessity to devise a method 5. Evaluation of restoration methods to effectively desalinate the sandstone and materials before its renewed application poultices. One method to reduce the salt load was 5.1. The “Silica sol repair mortar” to chisel off strongly salted areas.13 The method chosen, however, was salt extrac- The mortar chosen for restoration on tion by application of poultices. In first monument 825 was a silica-sol mortar, trials, clay based poultices were used. based on good experiences made by the These turned out to be unsuitable as they BLfD on several sites under restoration in clinged to the suface and upon removal Bavaria, where it was applied on sandsto- would even sometimes destroy delicate nes. After promising results of its appli- Nabatean chisel marks. They were finally Figure 2: The treated monuments. a) Mapping by the CARCIP team. b) Monument no. 825 and 826 and c) mapping of monument no. 826. cation in a Nabatean quarry the project replaced by poultices developed by the

192 193 W. Wedekind and H. Fischer Evaluation of desalination and restoration methods applied in Petra (Jordan)

conservator Egon Kaiser and one of the it was a valuable learning experience. It authors. These consisted of a wet mix- should also be stressed in this connecti- ture of cellulose and washed sand in a on that the main focus of the project was proportion of 1:5, and in general, showed to train local Jordanian staff in its ability a good workability as well as promising to manage and master a wide range of results in the beginning. The moist poul- techniques employed in the conservati- tices were applied in square patches on on and restoration effort, in accordance the salt contaminated areas and were re- with the highest international standards, moved after drying (see Figure 4d). including aspects such as planning, do- However, even after the application cumentation, assessment procedures, of several cycles of poultice application site preparation, the use of heavy and salt reduction turned out to still be in- specialized equipment, as well as the sufficient, as the silica sol repair mortar mastering of very specialized skills like would not bind as expected in some al- surveying, core drilling or the use of ana- veolar areas (Figure 9b). lytical tools. Obviously the efficiency of the desali- Towards the end of the project, howe- nation process needed to be further im- ver, the authors concluded that a new proved. In order to find a proper soluti- method for the efficient desalination of on, a systematic quantitative analysis of the monuments in Petra had to be de- the progress of desalination after each veloped and that the approach chosen so individual cycle of poultice application far and still supported by the project ad- was undertaken in defined sample areas. visor 14. Though possible alternative me- Thus samples from the poultices were thods had already been suggested by the taken after each cycle, dissolved in dis- authors, these had been outright rejected tilled water and their relative salt cont- by the project advisors. ent measured by electrical conductivity (Figure 6a). But even after up to 17 cycles of poultice application an adequate im- 5.3. The sprinkling method provement of the behaviour of the silica sol mortar could not be observed. The After finalizing restoration monument results thus indicated that the chosen 825, work was shifted to the adjacent mo- method of salt extraction was unfeasib- nument no. 826, which showed rather le, not only with respect of its efficien- different types of degradation and thus cy but also for the sheer enormity of the presented new challenges. Like before restoration effort, considering the size of the different weathering forms were the monuments. The poultice application mapped, quantified and documented method thus might work well on smaller, with other relevant observations. Ad- isolated objects, that are cut off from the ditionally electrical conductivity mea- water cycle encountered in objects that surements were carried out on selected exposed to geological processes. surface areas using a portable measuring Nevertheless, the restoration of monu- device (type: protimeter). Thereupon ment 825 was finalized and considered a number of samples were taken from a success, keeping in mind that on large defined areas in selected alveolae. First restored areas of its surface the salt load samples of crust material, then samples was rather small, no damage was done of poultices applied to the very same Figure 3: Procedure of desalination by the sprinkling to the monument and that all in all the areas (Figure 4e). Finally two drill cores method. a) Tafoni no. 9 at monument no. 826. b) The installed sprinkling device, sprinkling and final result was satisfactory, though at a were taken from these areas and also in- collection of the eluate. c) Measurement of the very high price that stood in no relation vestigated (Figures 2a, d, e). Different ions collected eluate. d) Application of the poultice and e) sampling of the poultice. to the success. At least it can be said that within the stone material of the drill co-

194 195 W. Wedekind and H. Fischer Evaluation of desalination and restoration methods applied in Petra (Jordan)

res were identified in different sections indicating no significant further salt ex- Salt content of Salt content in the first Salt content in the Number of by ion-chromatography. traction. Tafoni the crust (g) two poultices (g) eluate by sprinkling (g) collected litres The totally new approach in the resto- The spray pressure applied to the sto- ration effort here, however, rests in the ne during the sprinkling process is very 1 0.33 0.43 - - introduction desalination method de- low. In addition, this is a particularly vised and implemented by the authors. suitable method in situations where the 2 0.04 19.5 100.4 68 This approach was inspired by previous prevailing climate is very warm due to observations on the development of ver- the rapid evaporation rates there. Thus in 3 0.25 13.3 91.3 105 tical tafoni only adjacent to the areas Petra the water soaked up by the sands- that were marked by run off water, a tones normally evaporates within a few 4 0.36 0.4 - - process that had also been observed on days. After each subsequent washing some rare occasions of heavy rainfall. cycle the amount of salt extracted from 5 0.47 17 10.7 73 This clearly indicated that rainwater was- the surface was, as expected, observed to hes out salt accumulations on and under continuously decrease. Eventually salt 6 2.81 25 145.6 220 the rock surface. This observation then concentrations decrease to a value, whe- led to the development of a method, that re mortars can be safely applied. 7 0.89 20.5 114.1 145 mimicked the natural process of desali- Measurements on the drill cores ta- 8 0.37 18.3 396.7 349 nation. Thus a device was put into ope- ken before and after desalination give a ration that would sprinkle water under quantitative indication of the amount of 9 0.17 20.7 133.7 145 controlled conditions onto the surface of salt dissolved from the rock. the salt affected surfaces. This procedure It should be mentioned in this connec- was then labelled the sprinkling method. tion, however, that in spite of the fact, Table 1: Evaluation of the near-surface salt contamination and desalination at monument no. 826 By this method water is sprayed onto that salt contamination is efficiently re- the stone wall surface through fine nozz- duced in areas treated by this method, les (Figure 3b). At the start of this proce- side effects could be observed in some dure water is predominantly absorbed instances in the immediate vicinity to by the porous stone surface through the not treated adjacent areas. Obviously capillary forces. Water absorption is de- some of the salty solutions would migra- pendent upon the transport properties te towards the fringe of the treated are, of the rock and these are essentially con- a side effect that was actually observed. trolled by the pore space properties, like Though the effects are minor, we are porosity and pore radii distribution and aware that they are of concern. It must are a time-dependent process.15 be stated here, however, that such effects The excess water not absorbed by the cannot be precluded in situations, where stone runs off the treatment area and is the treated rock is part of a huge geolo- collected at the bottom in 1l containers gical entity. (Figure 3c). Electrical conductivity of the It may be added here, that similar side eluate of each sample container is then effects were also encountered at the frin- measured thus giving an indication of ges of poultices that were used in the ear- the amount and concentration of the dis- ly stages of the project for salt extraction. solved substances (Figure 3c). After the sprinkling a poultice of the same composition as used in previous applications 6. Results was applied onto the treated area (Figure 3d) in order to detect if additional salt It was realized already at an early stage would be extracted that way. A sample of application of the sprinkling method,

taken therefrom (Figure 3e) was dissol- that it was far more efficient, economical Figure 4: a) Ongoing damage progress of restored areas treated with silica-sol mortars during a period of 15 years on ved and tested by electrical conductivity, and less time consuming than the me- monument 825. b) Sample of a dry-slaked lime mortar just after application in 2006 and c) after 10 years.

196 197 W. Wedekind and H. Fischer Evaluation of desalination and restoration methods applied in Petra (Jordan)

thods previously employed. While five years after restoration only chener Geowissenschaftliche Beiträge the ‘Silk Tomb’(Petra, Jordan) and their In order to demonstrate the effec- a few damages in the restored areas (Value 4) Wissenschaftsverlag Mainz, possible source. In: Ioannou, I., Theodo- tiveness of desalination by use of the could be observed on both monuments, Aachen, 2005. ridou, M. Salt Weathering on Buildings sprinkling method, data collected from around 10 years later almost all mortar and Stone Sculptures (SWBSS), Limas- monument 826 are shown as an example applications were affected by crumbling 4 Wedekind W, Weathering and Conserva- sol, 2011, 81-88. in Table 1. on monument 825 (Figure 4a). tion of monuments constructed from The table indicates that the highest As can be seen on monument no. 826, tuff and sandstone in different environ- 11 Heinrichs K, Diagnosis of weathering degree of desalination by the sprinkling only very slight signs of degradation ment, PhD-thesis, Goettingen, 2016. damage on rock-cut monuments in Pe- method was encountered in tafoni 8, the could be observed 10 years after the contra, Jordan. Environmental Geology, vo- lowest one in tafoni 5. In total around servation treatment. (Figure 4a). 5 Wellmann HW, Wilson AT, Salt weathe- lume 56, issue 3, 2008, 643-675. 1000 g of solvent material was extracted Interestingly, a lime mortar newly dering, neglected geological erosive agent from all tafoni areas by sprinkling. Tafo- vised by one of the authors also exhibits in coastal and arid environments. Na- 12 Snethlage R, Testing and Evaluation ni 1 and 4 only show a very low salt con- no signs of alteration after 10 years (Figu- ture (205), 1965, 1097-1098. of Stone Repair Materials in Petra. In: tent within the first two poultices, and res 4b and c), indicating that there is also Kuehlenthal Mn Fischer H (eds.) Petra therefore were not treated by sprinkling. room for the development of alternative 6 Wedekind W, Ruedrich J, Salt-weathe- – The Restoration of the Rockcut Tomb However, from tafoni 2, 3, 6, 7 and 9 an the repair mortars. Results of this poten- ring, conservation techniques and stra- Facades, Arbeitsheft Bayer Landesdenk- amount of solvent material of around tially also far more economic effort will tegies to protect the rock cut facades in malamt f Denkmalpflege, Munich, 100 g to 150 g were extracted. be presented in a separate paper. Petra/Jordan. In: Fort R., Álvarez de Bu- 2000, 184-190. ergo M., Gomez-Heras M. & Vazquez-Cal- vo C. (eds.). Heritage, Weathering and 13 Kuelenthal M, Guidelines and Proce- 7. Summary and conclusions Conservation. Taylor & Francis, London, dures for the Restoration of the Monu- 2006, 261-268. ments in Petra. In: Kuehlenthal M, Fi- The sprinkling method for desalinating scher H (eds.) Petra – The Restoration of natural stone in Petra proved to be a tool References 7 Abd El-Hady M. The deterioration of Nu- the Rockcut Tomb Facades, Arbeitsheft that can effectively and easily reduce the bian sandstone blocks in the ptolemaic Bayer Landesdenkmalamt f Denkmal- salt load of areas affected by salt conta- 1 Wedekind, W, Jordan - Petra. In Truscott temples in Upper Egypt. Proceedings of pflege, Munich, 2000, 87-88. mination, while the quantity of salt ext- M, Petzet, M, Ziesemer J, Heritage at Risk the 9th International Congress on Dete- racted can be continuously monitored at – ICOMOS World Report 2004/2005 on rioration and Conservation of Stone. Ve- 14 Kuehlenthal M, Bavarian State Conser- the same time. Quantitative analysis of Monuments and Sites in Danger, Mu- nice, June 19-24. Elsevier, 2000, volume vation Office, Assesment and restora- the degree of desalination also provides nich, 2005. 2, 783 – 792. tion concept of Monument 826, intern better insights into the weathering pro- project report, CARCIP, 2000. cesses of the individual buildings. 2 Kuehlenthal M, Fischer H (2000) Petra, 8 Bala’awi, F, Wind speed and salt simula- Taking into account the particular con- Die Restaurierung der Grabfassaden, tion tests: Towards a more comprehen- 15 Wittmann FH, Feuchtigkeitstransport ditions encountered in Petra, where e.g. The Restoration of the Rockcut Tomb Fa- sive approach. In: Salt weathering on in porösen Werkstoffen des Bauwesens. the extended presence of a scaffold struc- cades, Munich, 2000. buildings and stone sculptures (SWBSS): In: Goretzki L. (ed.): Verfahren zum ture in front of a monument can be an proceedings from the international con- Entsalzen von Naturstein, Mauerwerk issue, the use of the more efficient and 1 Gaudi A, Viles H, Salt Weathering Ha- ference, The National museum, Copen- und Putz. Feiburg, 1996, 6-16. faster sprinkling method for desalinati- zards. John Wiley & Sons, 1997. hagen, Denmark, 22-24 October 2008, on constitutes an additional advantage. 41-50. Fifteen years after the application of si- 2 Al-Khashman, O.A., Jaradat, A.Q., Sala- lica-sol-mortar in areas treated this way, meh, E. Five-year monitoring study of 9 Eklund S, (2008) Stone weathering in the the mortars can still be seen to exhibit chemical characteristics of wet atmo- monastic building complex on Moun- almost no sign of degradation, in cont- spheric precipitation in the southern tain of St. Aaron in Petra, Jordan. Master rast to the observations made on monu- region of Jordan. Environ Monit Assess of Arts Thesis. University of Helsinki. ment 825 treated with the previous me- (185), 2013, 5715–5727 thod as shown in Figure 4. 10 Gomez-Heras M, Lopez-Arce P, Bala‘awi A faster degradation of the repair 3 Heinrichs K, Diagnose der Verwitte- F, Vazquez-Calvo C, Fort R, Ishakat F, Al- mortar is clearly visible on monument rungsschäden an den Felsmonumenten varez de Buergo M, Allawneh F, Charac- no. 825, as compared to monument 826. der antiken Stadt Petra / Jordanien, Aa- terisation of salt combinations found at

198 199 SWBSS 2017 | 20-22 September 4th International Conference on Salt Weathering of Buildings and Stone Sculptures Salt extraction by poulticing in the archaeological site of Coudenberg

Salt extraction by poulticing in the archaeological salt damage is seen at certain heights or The density, the porosity and the pore in specific regions, with mainly the first size distribution of the dry poultice ma- site of Coudenberg centimetre of a substrate being affected. terials, brick and mortar are determined Lowering the salt content near the surfa- by mercury intrusion porosimetry (MIP, Sebastiaan Godts*, Roald Hayen and Hilde De Clercq ce and diminishing daily changes in RH Autopore IV, Micromeritics). From the Royal Institute for Cultural Heritage (KIK-IRPA), Brussels, Belgium both in intensity and recurrence will mi- results the theoretical properties of the *[email protected] tigate salt damage, contributing to a sus- poultice, that is, the advective or diffusi- tainable preservation and an appropriate ve properties are determined. presentation for visitors. The evaluation of the poultice effi- Abstract 1. Introduction While the removal of salts from mo- ciency and the salt extraction/migration vable stone objects by means of the water processes is based on a comparison of A common method for the extraction The archaeological site of Coudenberg, bath method is quite well known, in-situ the evolution of the salt content in the of salts is the use of poultice materials. the last material evidence of the former salt extraction of brick masonry by me- depth of the masonry and in the poul- Despite extensive scientific research, the Palace of Brussels, is situated at the he- ans of poultices is often carried out on tice, before and after salt extraction tre- salt extraction process in practice often art of Belgium’s capital underneath the the basis of general assumptions, and in atment. The total amount of ions, inclu- remains empiric. This paper aims to Royal Square and its 19th century buil- most cases without control of the effi- ding or excluding equimolar amounts 2+ 2- further the understanding of a salt ext- dings. The project is the culmination of ciency or the effect in the depth of the of Ca and SO4 (reflecting the gypsum raction by poulticing in practical experi- twenty-five years of archaeological work, substrate. This paper goes deeper into content), in kg.m-2 (per sample depth) are ments. Three types of poultice materials which have progressively enabled diffe- the practice of salt extraction on brick considered during the evaluation. The are tested on brick masonry in 12 different areas of the site to become accessib- masonry using three different poultice salt content at a certain sampling depth rent locations in the subterranean archa- le to the public. Not all the materials of types, focusing on the influence of some in the masonry (brick and mortar) is de- eological site of Coudenberg in Brussels. the site have been preserved in the same of the execution parameters on the salt termined as follows: The tested poultice materials have diffe- condition and the project faces consider- extraction efficiency.

rent compositions including kaolin clay, able challenges in both conservation and φbrick · xbrick · d · Abrick + φmortar · xmortar · d · Amortar sand and/or cellulose fibres, theoretically presentation considering the relatively (I) favouring advection and/or diffusion. It recent exposure from burial.1 Major wa- 2. Materials and methods

is shown that for an accurate interpreta- ter infiltrations and salt damage are now With φbrick and φmortar, respectively the tion of the results it is crucial to evaluate visible in a large part of the site. 2.1. Sampling, salt analysis and efficien- mass density (specific density) of the -3 the salt content in the poultice and to a The presence of salts in the subterrane- cy evaluation brick and mortar (in kg•m ), xbrick and

depth of at least 20 cm in the different an site is essentially caused by water and xmortar, respectively the total content of materials of the substrate, while conside- ion transport that occurred over an ex- Before and after each poultice applica- soluble ions in each material, excluding 2+ 2- ring the different densities and surface tended period of time. During this time tion samples are taken from the mortar equimolar amounts of Ca and SO4 (in area of each material. Furthermore, it is salts accumulated unevenly at the evapo- and brick at different depths. Samples wt%), d, the sampling depth (in meter) shown that the presence of gypsum over- ration front, near the surface at certain are obtained by powder drilling (Ø8 and Abrick and Amortar, the proportio- shadows the results, hence the exclusion heights and preferentially in the materi- mm) in intermediate steps, 0-1, 1-3, 3-5, nal surface area (in m²/m² wall surface) of 2+ 2- of equimolair contents of Ca and SO4 is als with the smallest pores. Damage phe- 5-10, 10-15 and 15-20 cm, to obtain an in- respectively brick and mortar. The total essential for the evaluation. Finally, the nomena are naturally explained by the depth moisture and salt profile. For each amount of salts in the sampled volume is results of these experiments reveal that accumulation of salts and the occurren- sample, the actual and the hygroscopic obtained by adding the amount of salts regardless of the different procedures, ce of crystallisation cycles. These cycles moisture contents are determined gra- found at every sampling depth for both such as, the poultice type, pre-wetting or are promoted by changes in relative hu- vimetrically. The quantity of anions (Cl−, brick and mortar. The result is expressed − 2− + + 2+ application time, salts are being redistri- midity (RH). Daily RH changes typically NO3 , and SO4 ) and cations (Na , K , Ca , as the amount of salts per unit of surface buted into the wall rather than extracted. influence the first centimetres and -di and Mg2+) of the filtered aqueous extract area (kg•m-2). The comparison of the salt minish rapidly in depth of the substrate, is measured by ion chromatography (IC, content in the substrate after treatment Keywords: salt extraction, desalination, with crystallisation cycles occurring less Metrohm). The theoretical amount of with the initial salt content indicates the 2- poultice, historic masonry, archaeologi- frequently further in the substrate from carbonates (CO3 ) is determined from the amount of salts which have disappeared cal site daily, weekly, monthly to seasonally. The excess of analysed cations. The results out of the sampled volume and which amount of cycles that influence the first are presented as kg•m-2 (per sampled either have been absorbed by the poul- centimetres can easily run up to more depth), based on the specific density tice or either migrated away within the than a 100 cycles per year.2 Hence, visible (drilled powder) of the dry mortar/brick. substrate to a greater depth or outside of

200 201 S. Godts et al. Salt extraction by poulticing in the archaeological site of Coudenberg

cally favour a diffusive flow towards the masonry is pre-wetted with a known poultice (Figure 1). The poultice contains amount (0.5 2 or 4 l•m-²) of deminera- cellulose fibres (Arbocel PZ8 1,4 mm). lized water by spraying. Pre-wetting is The poultice is mixed with water (78 wt% carried out with intermediate steps to H2O for 22 wt% of dry poultice material) prevent runoff and allow maximal ab- and contains 0,2% biocide. sorption. In the case of the application of poultice KS, the masonry is not pre-wet- Poultice KS is based on a poultice de- ted. All poultices (KCS, C and KS) are ap- scribed and recommended by Bourgès plied to the masonry with a spatula in & Vergès Belmin.5,6 This poultice has a thickness of approximately 1 cm on a enough small and large pores compared surface of approximately 1 m2. Poultice to the substrate to theoretically favour C is immediately covered with a water/ both an advective and a diffusive flow vapour impermeable foil after applica- towards the poultice (Figure 1). The poul- tion. The different poultice application tice contains kaolin clay (IMERYS quali- conditions are based on experiences and ty China Clay Speswhite) and calibrated traditions found in literature and discus- sand (Sibelco CEN196 sand fraction 0,5-1 sions with practitioners. At the end of the mm), 1:5 by weight (0.8:1 by volume). The experiments samples are taken from the

poultice is mixed with water (20 wt% H2O poultice material and from the substrate for 80 wt% of dry poultice material). as described earlier. For the purpose of

Figure 1: Cumulative pore volume distribution (vol%) of the dried poultice materials KCS (0), C (Δ) and KS (□) compared this paper the results of the experiments to the avg. pore volume distribution of the brick and mortar from 7 locations (◊). The three different poultices (KCS, C are averaged per poultice type. and KS) are tested on brick masonry in a total of 12 different locations (1-12). the sampled area. As only the salts absor- 2.2. Poultice materials and application Locations 1, 6, 10 and 11 are free stan- 3. Results bed by the poultice are effectively extrac- procedures ding walls suffering from rising damp. ted, the amount of salts disappeared out Locations 2, 3, 7 and 8 are located on In total 3 poultice types were tested of the sampled volume is compared to The following poultice materials have vertical earth retaining walls, while loca- in 12 locations with the results of each the amount of salts detected within the been tested: tions 4, 5, 9 and 12 are situated on vaults experiment showing considerable varia- poultice (kg•m-2), determined as follows: Poultice KCS (Saltpull type 1, Rewah) that are presumably in contact with tions, making the interpretation rather is a poultice based on the findings de- earth. The main salts, excluding gypsum, challenging. However, several important 3, 4 φpoultice · xpoultice · t scribed by Lubelli & van Hees and was found within the walls are identified as results can be described looking at the (II) selected for the purpose of this project as sodium chloride (NaCl) in locations 1, 2, individual test results (not shown) and it has enough small pores compared to 4, 5, 6, 8, 9, 10 and 12, together with cal- at the averaged results per poultice type. the substrate to theoretically favour an cium nitrate (Ca(NO ) ) in locations 4, 5, Although tests were carried out with With φpoultice the mass density (bulk 3 2 -3 advective flow towards the poultice (Fi- 8 and 9. Sodium sulfate (NaSO ) is found different amounts of water for pre-wet- density) of the poultice (in kg•m ), xpoultice 4 the total content of soluble ions in the gure 1). The poultice contains cellulose in locations 3, 7 and 11, presumably in ting the surface, no significant influence poultice, excluding equimolar amounts fibres (Arbocel PZ8 1,4 mm), kaolin clay combination with smaller amounts of so- could be derived. Exceptionally, an ir- 2+ 2- (IMERYS quality China Clay Speswhite) dium carbonate (Na CO ), as theoretically regular migration of salts was recorded of Ca and SO4 (in wt%) and t the total 2 3 thickness of the poultice (0.01 m). The ef- and calibrated sand (Sibelco CEN196 sand derived from the excess of sodium ions to unknown locations in the substrate ficiency of the salt extraction process is fraction 0,5-1 mm) (1:2:1 by weight). The in these locations. when pre-wetting the surface. Similar determined as the amount of salts that poultice is mixed with water (61 wt% Poultices KCS, C and KS are applied to exceptions were derived when the subst- have been absorbed by the poultice in H2O for 39 wt% of dry poultice material) the salt contaminated masonry in res- rate showed high moisture contents befo- comparison to the initial salt content in and contains 0,2% biocide. pectively 5 (1-5), 4 (6-9) and 3 locations re the experiments; water added during the masonry, expressed in %. Poultice C (Cellulose Saltpull, Rewah) (10-12), and removed after respectively the pre-wetting procedure or from the is a commonly and traditionally used 11, 19 and 47 days. Before each poultice poultice itself provoked in such a case poultice in the restoration practice and application samples are taken from the an oversaturation of the bedding mortar was selected as it has enough large pores substrate (as described earlier). Before which allowed water and salts to migrate compared to the substrate to theoreti- application of poultices KCS and C the from the mortar towards the brick. Whe-

202 203 S. Godts et al. Salt extraction by poulticing in the archaeological site of Coudenberg

re the moisture content was lower before ce, as compared to the total amount of the experiments this phenomenon was salts in the substrate, the poultice tre- naturally seen reversed as the mortar in atment is not so efficient, certainly in all locations has smaller pores than the areas where salts might migrate back to brick, and hence tends to absorb rather the surface. When considering the first than expel its pore water (and salts). cm of the substrate (0-1 cm) the results are easily misinterpreted as an efficiency of 54 and 35% is obtained. When looking 3.1. Poultice KCS at the results, excluding gypsum, a general decrease of the salt content in the The average results of the five experi- wall is seen up to a depth of 10 cm. Some ments with poultice KCS are shown in of these salts will have migrated to the figure 2. The amount of salts, excluding poultice, however, the rest most probably

gypsum (CaSO4), that have disappeared diffused into the depth of the substrate from the masonry (0-20 cm) after the as an increase of the salt content is obser- poultice application is 4% of the total salt ved from 15 cm onwards. content before application, whereas in Figure 3: Poultice C, avg. of 4 locations (6-9). Left: Results of the avg. ion concentration (kg•m-2, per cm), detected in the case the presence of gypsum is inclu- the masonry, while taking the respective surface areas of the brick and mortar into account: before (0, black line) and after (◊, blue line) excl. CaSO4, and again before (Δ, red line) and after (□, green line) incl. CaSO4. The vertical bars on the ded an increase of the salt content with 3.2. Poultice C right show respectively the avg. amount of salts deposited in the poultice (dotted and orange outlined column, kg•m-2,

14% is observed. The salt content, either total thickness of the poultice: 1 cm) and the avg. amount of salts, excl. CaSO4, that have disappeared from the substrate (full and blue outlined column, kg•m-2, total of sampled depth 0-20 cm) and idem, incl. CaSO (striped and green excluding or including gypsum, that has The average results of the four experi- 4 outlined column, kg•m-2, total of sampled depth 0-20 cm). migrated towards the poultice amounts ments with poultice C are shown in figure to only 6 and 4%, respectively, of the to- 3. The amount of salts, either excluding tal salt content before application. Hen- or including gypsum, that has disappeared from the masonry (0-20 cm) after in figure 4. The amount of salts, either the poultice application is respectively excluding or including gypsum, that has 18 and 24%, while the amount of salts disappeared from the masonry (0-20 cm) that migrated towards the poultice is in after the poultice application is respec- both cases only 1% of the total salt cont- tively 14 and 27%, while the amount of ent before application. When considering salts that migrated towards the poultice the first cm of the substrate (0-1 cm) the is respectively only 4 and 3% of the total results would indicate an efficiency of salt content before application. Taking 8 and 5%. Although the efficiency con- into account the first cm of the subst- cerning the extraction is slightly lower rate (0-1 cm) the calculated efficiency is compared to poultice KCS, the salt con- instead 73 and 25%. A general decrease tent in the substrate, excluding gypsum, of the salt content, excluding gypsum, has decreased significantly further from is seen up to a depth of 20 cm of which the surface up to a depth of 15 cm. Whi- only a small amount have migrated to le practically none of these salts have the poultice. migrated to the poultice, approximately 17% of them have disappeared, and the- refor most likely migrated to lesser salt 4. Conclusions contaminated areas and into the depth (beyond 20 cm) of the substrate. The results and interpretation of the extraction process show important dif- Figure 2: Poultice KCS, avg. of 5 locations (1-5). Left: Results of the avg. ion concentration (kg•m-2, per cm), detected in the masonry, while taking the respective surface areas of the brick and mortar into account: before (0, black line) and ferences when considering the gypsum

after (◊, blue line) excl. CaSO4, and again before (Δ, red line) and after (□, green line) incl. CaSO4. The vertical bars on 3.3. Poultice KS content and different sampling depths -2 the right show respectively the avg. amount of salts deposited in the poultice (dotted and orange column, kg•m ) and in the substrate. An inaccurate interpre- the avg. amount of salts, excl. CaSO , that have disappeared from the substrate (full and blue column, kg•m-2, total of 4 The average results of the three ex- tation is highly possible when including sampled depth 0-20 cm) and idem, incl. CaSO4 (striped and green column). For the latter the negative value indicates an average increase of the salt content over the sampled volume. periments with poultice KS are shown gypsum in the calculations as it is often

204 205 S. Godts et al. Salt extraction by poulticing in the archaeological site of Coudenberg

will have a limited lifespan. Besides the Int Conference on Salt Weathering on ease of use of poultice C and a signifi- Buildings and Stone Sculpture, ed. J.S. cant difference in cost concerning the Albertsen, Copenhagen 2008. raw materials, both poultices KCS and KS leave a white haze behind on the sub- 6 Bourgès, A. and Vergès Belmin, V., per- strate, caused by the kaolin clay, which sonal communication during the 13th raises concerns regarding the aestheti- International Congress on the Deterio- cal changes and loss of material during ration and Conservation of Stone, Glas- the subsequent cleaning of the surface, gow 2016. increasingly so when considering a repeated treatment.

References Figure 4: Poultice KS, avg. of 3 locations (10-12). Left: Results of the avg. ion concentration (kg•m-2, per cm), detected in the masonry, while taking the respective surface areas of the brick and mortar into account: before (0, black line) and 1 Williams, T., Coudenberg, Brussels, Con- after (◊, blue line) excl. CaSO4, and again before (Δ, red line) and after (□, green line) incl. CaSO4. The vertical bars on the right show respectively the avg. amount of salts deposited in the poultice (dotted and orange column, kg•m-2) and servation and Management of Archaeo- -2 the avg. amount of salts, excl. CaSO4, that have disappeared from the substrate (full and blue column, kg•m , total of logical Sites (CMAS), (16-1), (2014), 1-3. sampled depth 0-20 cm) and idem, incl. CaSO (striped and green column). 4 http://dx.doi.org/10.1179/135050331 4Z.00000000074.

unevenly distributed in the substrate dure on the migration processes further 2 Godts, S., Hayen, R and De Clercq, H., In- and generally occurs in much higher con- into the depth of the substrate forces an vestigating salt decay of stone materials centrations as compared to other salts. evaluation of the salt content up to grea- related to the environment, a case study As shown in the results these aspects, ter depths. A general rule applies in that in the St. James church in Liège, Belgi- in combination with its low solubility salts tend to migrate to further depths um, Studies in Conservation, (2016) DOI: (~2.0–2.5 g/l at 25 °C), overshadow the and to areas with lower salt contents. 10.1080/00393630.2016.1236997. overall extraction process. When com- The overall reduction of the salt cont- paring the salt content in the poultice ent in the substrate to a depth of 20 cm 3 Lubelli, B., van Hees, R.P.J. and De Clercq, to the total salt content in the substrate is 4 and 14% for the poultices KCS and KS H, Fine tuning of desalination poultices, (from 0-20 cm depth), the average results (favouring advection). Although the ext- try-outs in practice, in proceedings of of the poultice applications indicate a raction of salts seems the least effective the 2nd Int Conference on Salt Weathe- rather low efficiency for the tested poul- for poultice C, favouring diffusion, it is ring on Buildings and Stone Sculptures, tices, KCS, C and KS, with respectively clear that the salt content in the sampled eds. Ioannou, I. & Theodoridou, M., Li- an efficiency of only 6, 1 and 4% of salts depth to 20 cm has decreased with 18%. massol 2011, 381-388. extracted. When considering only the Although hardly any salts are extracted, first centimetre of the substrate, extrac- the results of this poultice show that the 4 Lubelli, B. and van Hees, R.P.J., Desalina- tion efficiency values of 54, 8 and 73% salt content is sufficiently low to prevent tion of masonry structures: Fine tuning are obtained, which evidence the ease of further salt damage at the surface, at of pore size distribution of poultices to misinterpretation of the salt extraction least for the time being. substrate properties, Journal of Cultural efficiency. Although the extraction of Surprisingly poultice C is the preferred Heritage, (11), (2011), 10-18. salts at the surface is of utmost import- poultice in this case, the decision is es- ance, as it is the area mainly influenced sentially based on practical concerns and 5 Bourgès, A. and Vergès Belmin, V., Com- by daily RH cycles and, hence, most likely its repeatability. As long as the subterra- parison and optimization of five desali- subjected to severe damage, the signifi- nean site is subject to water infiltrations nation systems on inner walls of Saint cant influence of the poulticing proce- it can be assumed that any treatment Philibert Church in Dijon, France. 1st

206 207 SWBSS 2017 | 20-22 September 4th International Conference on Salt Weathering of Buildings and Stone Sculptures Desalination of Cotta type Elbe sandstone with adapted poultices: Optimization of poultice mixtures

Desalination of Cotta type Elbe sandstone with a b adapted poultices: Optimization of poultice mixtures

Julia Maitschke1 and Heiner Siedel2* 1 University of Applied Sciences Potsdam, Dept. of Conservation and Restoration, Potsdam, Germany 2 TU Dresden, Institute of Geotechnical Engineering, Chair of Applied Geology, Dresden, Germany *[email protected]

Abstract Keywords: salt reduction, desalination, Cotta type Elbe sandstone, poultice mix- Cotta type Elbe sandstone has been tures frequently used as construction material and ornamental stone on buildings over centuries. Desalination of this sandstone Figure 1: Thin section image of Cotta type sandstone (Nicols II) with pores coloured by blue dye (a) and typical pore 1. Introduction size distribution obtained by MIP (b) (often performed with cellulose poultices in restoration practice) is ineffective in Salt reduction or “desalination” with many cases due to its high amount of poultices is a well-known and widely fine pores. used technique in conservation of na- the laboratory have already been trans- often used by restorers in practice is A mixture of cellulose, kaolin and tural stone buildings and sculptures.1 ferred to conservation practice4 or tried compared to the one designed with res- sand (1:2:1 by weight, CKS_121) with a Recent investigations into the transport out in the laboratory for local important pect to pore size distribution of the Cotta high portion of fine pores was applied phenomena of salt and water led to bet- building stones. 5 type sandstone. for poultice desalination on artificial- ter general understanding of the interac- From a practitioner’s point of view pro- ly salt-loaded specimens of Cotta type tion of poultice and substrate properties perties of the poultice material like ad- sandstone. Moreover, another mixture of during the desalination process.2-3 As a hesion to the substrate, shrinkage while 2. Materials and methods the same components (CKS_128 with cel- result, in current discussions of the poul- drying, and the removal of the poultice lulose-kaolin-sand 1:2:8 by weight) and tice materials main attention is drawn to from the stone surface without remnants 2.1. Cotta type Elbe sandstone and test two poultices that are frequently used the role of an appropriate pore size dis- after drying are of importance beside ef- specimens in restoration practice (ready-made Raja- tribution of the poultice with respect to ficiency. Higher clay contents in the poul- sil (RAJ) poultice and Arbocel® CC1000/ the substrate material to be desalinated. tice do not only provide the appropriate Cotta type sandstone is a variety of BWW40 (ARB) mixture) were applied for Analyses of the pore size distribution portion of fine pores needed for efficient the Cretaceous Elbe sandstone which comparison. The pore size distributions of the substrate by mercury intrusion desalination of stones with high contents has been used for building purposes in and the structures of all poultices were porosimetry (MIP) can give useful hints of smaller pores but also change the pro- East Germany and especially in Saxony characterized by MIP. The results showed towards the pore size properties needed perties of the poultice towards higher over centuries. The sandstone is a fi- highest amounts of extracted salt for the for the poultice to achieve an efficient shrinkage and therefore worse adhesion ne-grained, clay-bearing quartz arenite poultices with clay and higher portions desalination. Generally, the wet poultice to the substrate. Moreover, very small with kaolinite and illite contents. The of sand (CKS_128, RAJ). The CKS_121 should provide a portion of pores coar- clay particles tend to adhere to the stone average diameter of grains is 0.12 mm. poultice, although fitting best with re- ser then those in the substrate to offer surface and in the pore structure of the The sandstone structure in thin section gard to pore size distribution, shows high enough water to the stone material in substrate after drying and removal of the and its pore size distribution (total poro- shrinkage due to the high amount of kao- the wetting phase on one hand. On the poultice layer. To match both the prob- sity ca. 20 vol.%) is shown in Figure 1. As lin, which leads to loss of adhesion to the other hand another portion of pores in lems of salt transport from the stone into can be seen from Figure 1b, the majority substrate during the desalination pro- the poultice finer then those in the stone the poultice and the practical workability of pores is concentrated between 0.1 and cess. The results clearly demonstrate that substrate is needed to extract the water of poultices for desalination, compromi- 10 µm. Since the material is vulnerable in desalination practice of stones with with dissolved salts from the stone by ad- ses with respect to the clay content are to salt attack, desalination is a standard fine pores compromises have to be found vection in the drying phase.2 Appropriate needed in most cases. The study presents procedure in the course of most of the between the competitive parameters of structures can be designed in mixtures investigations carried out to optimize a restoration measures planned for histo- “ideal” pore structure and low shrinkage of cellulose, clay and sand by varying poultice for desalination of a fine-pored, ric objects made of Cotta type sandstone. of the poultice, both influenced by clay the proportion of the respective compo- widely used building and sculptural sto- To investigate the efficiency of desali- contents in the mixture. nents.3 These principles established in ne. Moreover the efficiency of poultices nation, stone specimens of unweathered

208 209 J. Maitschke and H. Siedel Desalination of Cotta type Elbe sandstone with adapted poultices: Optimization of poultice mixtures

sealing also the bottom side with epoxy a Material Sand Kaolin Cellulose resin, the specimens were dried at room conditions over 14 days. Profile analy- Ottendorf-Okrilla Caminau (Kaminauer Arbocel® BWW 40 Origin ses of the total salt content on random (Euroquarz GmbH) Kaolinwerk GmbH) (Kremer) samples showed a comparable salt distri- SiO2 46.6% bution within the test specimens, similar Chemical SiO2 98% Al2O3 35.9% Cellulose 99.5% to that found in monuments (Figure 2c). omposition Al2O3 1,15% K2O 2.4%

2.2. Poultice materials Table 1: Material parameters of the poultice components in own mixtures (source of data: data sheets of the producers)

According to the pore size distribution of the Cotta type sandstone (Figure 1b) b and the results presented in3 a cellulose-kaolin-sand (CKS) 1:2:1 mixture (by weight) with pore sizes in the same range was selected for the starting experiments. Kaolin and sand came from local deposits (Caminau and Ottendorf-Okril- la, respectively) but had grain size distributions very similar to those used in 3 (cf. Figure 3). The chemical composition of the Caminau kaolin (Table 3) differs so- Figure 3: Grain size distribution of the used sand (a) and kaolin (b) components (red line = kaoline Caminau, black line mewhat from the ideal kaolinite (46.5% c kaoline Polwhite™ C for comparance) SiO2 and 39.5% Al2O3) but is similar to

that of the Polwhithe™ C material (SiO2 3 47%, Al2O3 37%) used in. The difference application of 1 cm thick poultices on wa- ved from the sandstone surface for ana- to ideal kaolinite in case of Caminau kao- ter saturated specimens (15x20x3 cm) of lyses of the extracted salt amounts. lin might be explained by the presence Cotta type sandstone. Since the shrinka- of small quantities of illite/mica, as inge was remarkable high for the CKS_121

dicated by 2.4% K2O. The mineralogical mixture, a mixture CKS_128 of the same 2.3. Analytical methods composition of Caminau washed kaolin components, but with higher amounts of is described by contents of 80-90% kaoli- sand was additionally investigated. MIP measurements for characteriza- nite and 10-20% illte/mica, but no swel- The selected different poultice sys- tion of the pore size distribution of dry ling minerals like montmorillonite or tems (Table 2) were applied to the arti- poultices were carried out with Porotec mixed layers 6. For comparison, two poul- ficially salt-loaded Cotta type sandstone Pascal 140 and Pascal 440 systems. tices that are often used in restoration specimens described in 2.1 in the follo- For the assessment of the quantity Figure 2: Dimension of the stone specimens used for desalination experiments (a, b) and analytical control of practice were involved in the tests: a pure wing way: First, all stone surfaces were of salt moved from the stone substra- the salt distribution on two random samples (c) cellulose poultice (mixture of Arbocel pre-wetted with 25 ml of demineralised tes into the poultice after desalination, BC 1000 and BWV 40 1:2.1, 5) and a rea- water. Afterwards, the wet poultices were parts of the dry poultices (area 13x18 cm, dy-made poultice currently provided on applied with a thickness of 1 cm. Each thickness 1 cm each) were put in plastic Cotta type sandstone in the dimension the market (produced by Rajasil; mixture type of poultice was applied to 3 sandsto- bottles, and 1 litre of demineralised wa- of 15x20x10 cm with the largest areas cut of Arbocel®, Ca bentonite, Poraver® and ne specimens. The specimens treated this ter was added. The bottles were moved perpendicular to the bedding were pre- quartz sand) were applied. Different poul- way were then stored in the workshop (in in a shaking apparatus for 12 hours and pared (Figure 2a, b). After sealing the spe- tice mixtures were characterized with re- upright position) for 3 weeks at 18 °C/42% afterwards placed in the laboratory 30- cimens at the 4 smaller sides with epoxy spect to their pore size distri bution as r. H. Observations of visual changes (eff- 80 hours for total sedimentation of the resin, they were loaded with 3% sodium well as to their workability and shrinka- lorescence, loss of adherence, cracks) du- solids at the bottom of the bottle. 500 ml sulphate solution by capillary suction ge behavior. The latter was determined ring the drying process were registered. of the clear solution were then pipet- parallel to bedding over 7 days. After at the beginning of the experiments by After 3 weeks, the poultices were remo- ted from every bottle and filtered. The

210 211 J. Maitschke and H. Siedel Desalination of Cotta type Elbe sandstone with adapted poultices: Optimization of poultice mixtures

Water – dry Dry weight of poul- Shrinkage after 3. Results tice is expected to provide much water to Mixture Material solid ratio tice (per 100 cm2, drying, length the substrate but not to actively extract (weight portions) (by weight) 1 cm thick) (g) (%)/width (%) 3.1. Characterization of poultice salt solution from the stone by advecti- Arbocel® properties on in dryer state. The Rajasil ready-made BC 1000 : BWW (cellulose) 4.5* 20.5 1.3/3.0 poultice and the CKS_128 mixture have 5 40 = 1 : 2.1 poultice (ARB) MIP investigations gave insight into both pores in the range of the Cotta type Ready-made mix- the structure and pore size distribution sandstone and above. Rajasil poultice ture of Ca bentonite, of the applied poultice mixtures. The re- 0.6** 87.5 0/0 (RAJ) cellulose, sand and sults are displayed in Table 3 and Figure Poraver® 4. As can be seen from Table 3, the distri- 3.2. Observations during the poultice Cellulose (Arbocel® bution of pores in the CKS_121 poultice treatment CKS_121 (own BWW 40) : kaolin : fits best to the pore size distribution of 0.8*** 86.3 4.0/6.0 mixture) sand (0.5 - 1 mm) the Cotta type sandstone with respect After the application (Figure 5a, b) the = 1:2:1 to the theoretical optimum transport poultices were controlled visually every 2, 3 Cellulose (Arbocel® conditions. The pore size range of the day during the entire desalination cyc- CKS_128 (own BWW 40): kaolin : Arbocel® poultice lies above that one of le of three weeks. Under the specimens 0.39*** 126.1 0/0 mixture) sand (0.5 - 1 mm) the Cotta type stone; i. e. the wet poul- with the Arbocel® poultice, which was = 1:2:8

6 3 *taken from **as recommended by the producer ***comparable to a b Table 2: Material parameters of the applied poultice mixtures

Main range of Maximum of Total pore Median pore Material pore diameters pore diameter volume (%) diameter (µm) (µm) distribution (µm) (from MIP)

ARB 22.05 5 - 30 22 84.67

RAJ 12.38 1 - 60 50 59.29

CKS_121 1.25 0.1 - 5 1.8 51.75

CKS_128 6.74 0.4 - 30 17 41.91 c d

Cotta stone 1.51 0.1 - 6 5 21.99

Table 3: Data obtained from MIP measurements for different poultice materials

solvent was evaporated in a dry box at stepwise from the surface to depths of 60 °C. The respective evaporate was weig- 0-1, 1-2 and 2-3 cm. 2.5 g of each drill hed, and the salt amount extracted per powder sample were eluted with 50 ml of cm2 could be calculated. Moreover, two demineralised water for 24 hours. After- profiles of drill powder for salt analysis wards, the electrical conductivity of the (from the upper and the lower part) were solution was measured with a Hanna In- taken from each sandstone specimen be- struments conductivity meter HI 991300. fore and after desalination, respectively. Figure 4: MIP diagrams of the applied poultices: a) pure Arbocel® poultice, b) Rajasil ready-made mixture, c) cellulo- The samples of drill powder were taken se-kaolinite-sand 1:2:1 mixture (CKS_121), d) cellulose-kaolinite-sand 1:2:8 mixture (CKS_128)

212 213 J. Maitschke and H. Siedel Desalination of Cotta type Elbe sandstone with adapted poultices: Optimization of poultice mixtures

and showed salt efflorescence behind the (108.1 g/m2) was obtained from the Raja- a b poultice in these areas. Two weeks after sil ready-made mixture experiment. It is the application, the Arbocel® poultices closely followed by the CKS_128 mixture had lost contact in the upper parts (dis- with 74.6 g/m2 salt extracted on avera- tance between poultice and substrate up ge, with only small differences between to 2 mm). The CKS_121 mixture showed the three specimens (72.6-76.1 g/m2). The now visible shrinking and loss of adhe- CKS_121 mixture extracted only 61.0 g/ sion also at the bottom of the stone spe- m2 salt on average, whereas the worst re- cimens. Nearly no further changes could sult (47.0 g/m2 extracted salt on average) be observed on the other poultices, apart was found for the Arbocel® mixture. Al- from RAJ, were in some places efflore- though the latter could be expected from scences on the rim lead to minimal loss theoretical considerations of salt and of poultice material. moisture transport in the poultice and the sandstone substrate (cf. 3.1 and 2, 3), the low extraction rate for the CKS_121 3.3. Salt contents in the poultices after mixture is astonishing with regard to the desalination appropriate pore size distribution. Howe- ver, this might be explained by its early As can be seen in Figure 6, different loss of adhesion to the stone substrate amounts of salts were extracted from the (see 3.2). stone specimens by different poultice mixtures under the same drying conditions. The highest average value for all 3.4. Efficiency of desalination still very wet after one day, some dro- three specimens investigated for each c plets of water were found that had come mixture (79.6 g/m2) as well as the hig- The efficiency of desalination was cal- out during the first day after the appli- hest absolute value for a single specimen culated for the different profile depths of cation. In contrast, the CKS_128 mixture already showed first small cracks on the surface after one day. On the third day, the upper parts of the CKS_121 poultice started to visibly loose adhesion to the substrate. On the fourth day, rims of efflorescing salts appeared around the poultices, apart from the CKS_121 and Arbocel® mixtures (which were the mixtures with the highest water contents, cf. Table 1). After one week, the upper parts of all CKS_121 poultices had lost adhesion; one of them had a distance up to 9 mm to the stone surface (Figure 5c). Mo- reover, they showed single cracks with a width > 1.8 mm. Cracks were also found on the surface of the CKS_128 poultices, but no loss of adhesion to the substrate was detected here. The Rajasil ready-made mixture showed only few small cracks Figure 5: a) Pre-wetting of the stone surface, b) applica- and still a good adhesion to the substra- tion of the CKS_121 poultice on the pre-wetted stone specimen, c) loss of adhesion on the CKS_121 mixtures te. The Arbocel® poultice mixtures star- Figure 6: Salt amounts extracted from the salt-loaded Cotta type sandstone specimens with different poultice mixtures one week after application ted to loose adhesion in the upper parts during a 3 week’s desalination cycle

214 215 J. Maitschke and H. Siedel Desalination of Cotta type Elbe sandstone with adapted poultices: Optimization of poultice mixtures

the sandstone specimens in the stone be- markable increase in salt content (nega- (Table 2). Therefore another poultice with tely dried to moisture equilibrium with fore and after desalination by using the tive efficiency) in the outermost profile cellulose, kaolin and a higher amount the surrounding air might lead to con- electric conductivities measured after section (0-1 cm). The efficiency in deeper of sand (1:2:8 by weight, shrinkage = 0) centration of the activated salts near the extraction of salts from the drill powders sections (1-2, 2-3 cm) is comparable to was additionally tested in the desalinati- stone surface (cf. Figure 7 c), which could and the following equation: this of RAJ and CKS_128 (between 5 and on experiment on artificially salt-loaded be dangerous to the stone substrate. Al- 35% for ARB and between 5 and 40% for stone specimens, which also involved a ternatively, more frequently repeated

Efficiency [%] = (ecb - eca ) x 100 / ecb (1) CKS_121). customary Rajasil (RAJ) poultice and an short application periods of only a few Arbocel® mixture (ARB) often used by days (until adhesion gets lost) with the

with ecb = electric conductivity [µS/ restorers. The results showed compara- CKS_121 mixture could be tested. This

cm] before and eca = electric conducti- 4. Discussion and conclusions ble amounts of extracted salts (75-80 g/ might save time but will increase costs vity [µS/cm] after the poultice cycle. As m2) for RAJ and CKS_128 poultices, whe- for material and personnel. More practi- can be seen from Figure 7, the Rajasil According to theoretical considerations reas the CKS_121 (61 g/m2) and the ARB cal experiments are needed to work out ready-made mixture and the CKS_128 on transport of salts from the stone sub- mixture (47 g/m2) brought worse results. strategies balanced between appropriate mixture show reductions of salt content strate to the poultice by advection2, 3, an Visual observations showed an early loss poultice materials, workability and costs. in nearly all sections of the investigated appropriate cellulose-kaolin-sand poul- of adhesion of both CKS_121 and ARB, profiles. Efficiency is between 10 and 40% tice mixture with high kaolin content leading to the concentration of salts ac- (RAJ) and between 5 and 35% (CKS_128) (cellulose-kaolin-sand = 1:2:1 by weight) tivated by wetting beneath the stone sur- Acknowledgements in the majority of the investigated profile was designed for Cotta type Elbe sands- face (0-1 cm depth, Figure 7). Higher salt sections. In contrast, the Arbocel® mix- tone. However, this poultice showed high contents in the CKS_121 poultice com- Thanks are due to Annett Willomit- ture and the CKS_121 mixture show a re- shrinkage due to the high clay content pared to ARB indicate that the mixture zer (Institute of Construction Materials, with high clay content must be very acti- TU Dresden) for performance of the MIP ve in the short period immediately after measurements and to Frank Hoferick a b application until loss of adhesion. This is (Zwingerbauhütte Dresden) for fruitful in accordance with4, where good results discussions and practical help. were obtained on materials comparable to Cotta type sandstone with respect to their pore size distribution by repeated treatments with CKS_121 mixtures and very short desalination cycles of only 4-5 days. However, the long-term activity References (and the total efficiency) of the mixtures with higher sand contents and no shrin- 1 Vergès-Belmin, V. & Siedel, H.: Desalina- kage is much better, although their pore tion of masonries and and monumental size distributions do not fit best the the- sculptures by poulticing: a review. - Int. c d oretical demands for the pore structure Journal for Restoration of Buildings and of the Cotta type sandstone. The results Monuments 11 (2005) 6, 391-407. clearly demonstrate that in desalination practice of stones with fine pores com- 2 Pel, L. Sawdy, A. & Voronina, V.: Physi- promises have to be found between the cal principles and efficiency of salt ex- competitive parameters of “ideal” pore traction by poulticing. - J. Cult. Her. 11 structure and low shrinkage of the poul- (2010), 59-67. tice, both influenced by clay contents in the mixture. Even if the shrinking of 3 Lubelli, B. & van Hees, R.P.J.: Desalinati- kaolin clay is significantly lower than on of masonry structures: Fine tuning that of bentonite, all poultice mixtures of pore size distribution of poultices to with high clay contents will shrink while substrate properties. - J. Cult. Her. 11 Figure 7: Efficiency of desalination displayed for different profile depths in each of the three sandstone specimens for a) drying and therefore raise problems with (2010), 10-18. Arbocel® mixture, b) Rajasil ready-made mixture, c) CKS_121 mixture and d) CKS_128 mixture. Positive values indicate salt reduction, negative ones increase of salt content after desalination, related to the original salt content in every adhesion. Loss of adhesion before the specimen (u = upper part, l = lower part of the stone specimen) system substrate-poultice has comple-

216 217 SWBSS 2017 | 20-22 September J. Maitschke and H. Siedel 4th International Conference on Salt Weathering of Buildings and Stone Sculptures

4 Lubelli, B., van Hees, R.P.J. & De Clerq, H.: Tracing back the origins of sodium sulfate formation on lime- Fine tuning of desalination poultices: stone as a consequence of a cleaning campaign: the case study try-outs in practice. - In: Ioannou, I. & Theodoridou, M. (eds.): Proc. 2nd SWBSS on Charité and Espérance sculptures of Chartres cathedral Conference, Limassol 2011, 381-386. Sara Benkhalifa2, Véronique Vergès-Belmin1*, Olivier Rolland3 and Lise Leroux1 1 Laboratoire de Recherche des Monuments Historiques, CRC-LRMH-CNRS USR 3224, France 5 Kröner, S., Mañas Alcaide, B. & Mas-Bar- 2 Conservatrice - restauratrice de sculptures, France 3 berà, X.: Influence of substrate pore size Conservateur - restaurateur de sculptures, France *[email protected] distribution, poultice type, and application technique on the desalination of Abstract 1. Introduction medium-pore stones. - Studies in Con- servation 61 (2016) 5, 286-296. In 2012, five years after cleaning by At Chartres cathedral, four limestone Mora paste, sodium sulfate related de- sculptures representing the virtues alle- 5 Störr, M.: Die Kaolinlagerstätten der terioration was noticed on two XVIIIth gories Charity, Esperance, Faith and Hu- Deutschen Demokratischen Republik. - century limestone sculptures of Chart- mility and originating from its last Jube Schriftenr. Geol. Wiss. 18 (1983), 1-226. res cathedral. This paper presents the re- built between 1765 and 1770, are display- sults of the diagnostic study and of the ed on the ground floor of the southern 6 Bourgès, A. & Vergès-Belmin, V.: Appli- conservation intervention performed on Tower. The statues were separated into cation of fresh mortar tests to poultices these sculptures, with a focus on trials two groups after the dismantling of the used for the desalination of historical to determine the salt phases responsib- rood-screen in 1866. Charity and Espe- masonry. - Materials and Structures 44 le for their deterioration. The methodo- rance were fixed to a wall located on the (2011), 1233-1240. logy chosen includes mineralogical and ground floor at Chartres Musée des Beaux chemical analyses of samples collected Arts, in a situation sheltered from rain on the sculptures during the diagnostic but exposed to outdoor environment, phase and after kaolin-based poultice and probably in contact with the ground desalination (XRD, quantification of so- (Figure 1). Faith and Humility were placed luble salts, SEM-EDS), but also mineralo- outdoors, some meters above the ground, gical analyses of the salts extracted from against a monumental doorway at Hotel the poultices. Dieu and later at Chambre de Commerce. Mora paste chemicals (and in parti- No restoration is documented between cular EDTA disodium salt) are probably 1866 and the 2000’s. responsible for the quick deterioration In 2007, the four sculptures reintegra- of the sculptures through complex and ted the cathedral, where they were pla- unidentified chemical pathways : at least ced on pedestals ca.1 m height, back to two of the mineral phases found in this the walls but without any contact with case study are not present in the JCPDS the masonries (Figure 1). Before being data base on mineral phases. It is propo- reinstalled, Charity and Esperance were sed to set up an inventory of unidentified cleaned using “Mora” paste, while Faith phases found in case studies similar to and Humility did not undergo any in- this one, in order to gather the knowled- tervention, apart the application of a li- ge spread in different institutions, and mewash, clearly visible to the naked eye later to build scientific projects on the on their surface. It is believed that this topic of EDTA-related deterioration. intervention was performed to give the two sculptures an homogeneous colour, Keywords: thenardite, EDTA, Mora, AB57, as this paint layer covers pollution black cleaning, desalination, limestone crusts. In 2012, five years after their restoration, strong scaling, granular disintegrati-

218 219 Tracing back the origins of sodium sulfate formation on limestone as a consequence of a cleaning campaign: the case study S. Benkhalifa et al. on Charité and Espérance sculptures of Chartres cathedral

on and salt efflorescence were noticed on Charity and Esperance sculptures (Figure 2). This led Irène Jourd’heuil, the curator responsible for their preservation to ask for a condition survey, aiming at understanding the reasons for their quick degradation, and at setting up a conservation project. This paper presents the results of the diagnostic study and conservation of these sculptures, with a focus on trials to determine the salt phases responsible for their deterioration.1, 2

Figure 3: Charity (three images on the left) and Esperance (three images on the right): distribution of stone losses due to granular disintegration + scaling (blue spots), and to mechanical impacts (red lines). Sampling locations are shown 2. Analytical methodology as red spots.

The nature and quarry area of the stone were identified through macroscopic in found on the sculpture Esperance was intrusion porosimetry (MIP) revealed situ observations and observation, under analyzed in ATR mode with a Perkin-El- that the stone has a unimodal pore size binocular lens, of centimetric samples mer SP 100 IRTF spectrometer fitted with distribution, centered on 1µm. Its porosi- which were compared with samples a SPOTLIGHT 400 microscope. ty, as measured by MIP, is equal to 20%. from the LRMH collection of 6000 sto- The conservation intervention consis- Many millimetric relicts of a paint lay- ne samples originating from 700 French ted essentially into five successive appli- er and of a clayish material were found quarries and 500 monuments. The pore cations of poultices based on 0.5 – 1 mm on the sculptures. SEM-EDS analyses per- size distribution was determined with a sand (Aqua gravel from Aquasand), kao- formed on the paint relicts revealed that Micromeritics mercury intrusion poro- lin (Speswhite from Imerys Mineral) and barium sulfate is present locally, sugge- simeter. Scales and efflorescence collec- micron-size calcite (SoCal31 from Solvay). sting that the paint layer might contain ted on the sculptures were characterized The first set of poultices removed from barite as a pigment. with a powder diffractometer Brucker D8 each sculpture was collected and put in The mineralogical characterization of advance, fitted with a copper anticatho- distilled water (6 to 9 Kg / 12.5 L). After the clayish material by X-ray diffracti- de. Some surface samples were analyzed setting of the poultice material, the su- on revealed the presence of palygorski- on a JEOL JSM T30, fitted with an Oxford pernatant was collected, filtered and left te (Mg5Si8O20(OH).2,8H2O) as the main energy dispersive spectrometer (SEM- to evaporate, using a fan to accelerate the phase, quartz (SiO2) and calcite (CaCO3). EDS). A fragment of the poultice relicts process. Pizza-like light brown evaporates This suggests that the material is an atta- were obtained after one month. Crystals pulgite poultice, most probably relictual of different shapes were collected in the from the 2007 cleaning campaign. These evaporates and characterized by X-ray data do not seem at a first sight in line diffraction with the testimony of the conservator who worked on the sculptures. He indicated that he had applied “MORA” pads 3. Condition survey and analytical to clean them, after “many unsuccessful results trials”. He indicated that he had applied “MORA paste” to clean them, after “many The sculptures are 1.5 life size and unsuccessful trials”. The original MORA are made of a fined grained light beige paste recipe, also known as “AB 57 paste”, and very homogeneous oo-pelletoïdal was proposed in 1972 by Pr. Paulo Mora Figure 1: Charity and Esperance sculptures: presentation Figure 2: Face of the statue Espérance: blackening in and micritic limestone originating from and his wife Laura Mora from Instituto at Chartres Musée des beaux Arts in 1998 (left), and into 1998 (detail of figure 1), scaling and granular disintegra- the cathedral in 2012 (right) tion in 2012. quarries near Tonnerre, France. Mercury central per il Restauro, Rome, for clea-

220 221 Tracing back the origins of sodium sulfate formation on limestone as a consequence of a cleaning campaign: the case study S. Benkhalifa et al. on Charité and Espérance sculptures of Chartres cathedral

ning stone buildings. It comprised Water ted: salt efflorescence, granular disinteg- Depth Chloride Nitrate Sulfate Sodium Potas- Calcium Location Sample cc.1000; Sodium Bicarbonate g.50; EDTA ration and scaling developed on a high (cm) (%) (%) (%) (%) sium (%) (%) (disodium salt) g.25; Desogen, a quater- proportion of Esperance and Charity sur- nary ammonium salt, cc.10; and carboxy- face (Figure 3). E4 0.0-0.5 0,07 0.06 0,50 0,59 0,01 0,69 methylcellulose (CMC ) g.60. No clay com- All surface samples collected in deteri- Left arm E5 4.0-5.0 0,01 0,01 0,08 0,04 0,01 0,72 ponent entered in the original recipe. orated parts of the sculptures contained

However, its formulation has consider- calcite (CaCO3) and sodium sulfate crys- E7 0.0-0.5 0,09 0,24 0,22 0,64 0,03 0,83

ably changed in France since the original tallized as thenardite (Na2SO4) (Figure 4). Left foot publication by Mora.3 For instance on Some samples contained also gypsum E8 4.0-5.0 0,06 0,15 0,06 0,08 0,02 0,73 Amiens cathedral in 1995, cellulose pow- as a minor phase. Another crystallized der (instead of CMC) impregnated with a phase was put in evidence on the diffrac- Table 1: Results of the soluble salts quantification at two places of the sculpture “Esperance” Excessive ionic contents water solution of biammonium carbona- tograms, but unfortunately it could not are indicated by bold letters. See Figure 3 for samples location. te was used as a first step of sculptures be identified (Figure 4 and Table 2). Stone cleaning. CMC gels are quite difficult to powder samples were collected at diffe- sence of thenardite). At 4-5 cm depth, sul- into solution together with salt related remove from carved substrates; the ope- rent depths through drilling, and their fates quantities are below 0.1% on both sources of calcium during water extrac- ration often necessitates much water to anionic and cationic contents were mea- sampling locations, suggesting that the tion. These results suggest, considering 4 be performed. This is probably the rea- sured after salt extraction , following the bulk of the sculpture may be considered the location of the sample where nitra- son why other water retentive agents recommendation of the Italian Standard as not contaminated by this ionic species.4 tes and chlorides were found, that du- such as cellulose powder or clays, have Normal 13/83 : dosaggio dei Sali solubili Nitrates contents are rather high both ring its stay on the ground at Chartres replaced CMC gels in a number of work- (Table 1). close to the surface and in depth in the Musée des Beaux Arts from 1866 to 2006, shops in France. It is thus possible that Excessive sodium and sulfate quanti- sampling location located close to the the sculpture has been contaminated the conservator in 2007 considered using ties are put in evidence in the water ext- foot of the sculpture (E7-E8). by ground salts through capillary rise. a modified MORA paste based on clay to racts of the samples collected at 0-0,5 cm The same tendency, to a lower extent, The most probable kick-off event that clean the sculptures. depth (Table 1). This analytical result is in is observed for chlorides contents. Calci- determined the strong thenardite rela- The sculptures were strongly deteriora- compliance with mineralogical data (pre- um contents cannot be interpreted with ted decay seems to be the 2007 cleaning confidence as some of the calcium carbo- campaign, since no heavy stone loss can nate of the substrate most probably got be observed on the picture taken before

Figure 4: Diffractogram of a typical efflorescence sample collected on the sculpture Esperance. Thenardite (red), Figure 5: Results of the soluble salts quantification on the sculpture “Esperance. Location of sampling to be found on gypsum (green) and calcite (blue) are present; most peaks cannot be attributed to an identified mineral phase. Figure 3 (ref. “desal”)

222 223 Tracing back the origins of sodium sulfate formation on limestone as a consequence of a cleaning campaign: the case study S. Benkhalifa et al. on Charité and Espérance sculptures of Chartres cathedral

the 2007 restoration. Interestingly sca- poultice removal, but the overall results ling and granular disintegration have ap- were considered positive, as salt contami- peared preferentially on areas where bla- nation was strongly reduced by the con- ckening was present before 2006 (Figure servation intervention (Figure 4). Kaolin 3), suggesting that some of the chemicals relicts adhering to the surface were gent- used for cleaning purposes may be res- ly removed from the surface using a soft ponsible for the salt decay. This may be brush. Tonnerre limestone having a very put in relation with the comments of the light creamy colour, the slight and local conservator in 2007: “MORA paste” was whitening due to the presence of minute used to clean the sculptures, after “many proportions of kaolin left after brushing unsuccessful trials”. did not impair the overall aspect of the Considering the nature and depth of sculptures. the contamination on the one hand, and The poultice desalination was evalua- the pore size distribution of the substra- ted on the basis of quantification of so- te on the other hand, it was considered luble ions from zero to 1 cm depth.7 It a reasonable solution to desalinate the appears that the main ions sodium and sculptures with kaolin-based poultices.5, 6 sulfates have clearly decreased while chlorides contents, relatively high before desalination, remained unchanged as a 4. Poultice desalination result of poulticing.

After preliminary trials with poultices containing mixtures of various 5. Salts extracted from the sculptures proportions of kaolin, cellulose powder, Figure 6: General aspect of the evaporates, and overview of the sampling. micron-sized calcite and sand, the follo- The salts extracted from the two sculp- wing recipes were selected: Charity recei- tures looked pretty much the same (Fi- ved five applications of a kaolin:sand:- gure 6): a light brown colour, and a ring Phases water 1:5:1,5 (vol) poultice. Esperance shape sequence of crystals. Samples col- Sample thenardite gypsum unident. unident. received one application of a mixture kao- lected from the outer to the inner ring calcic EDTA halite NaCl Na SO CaSO .2H O phase b phase a lin:nanocalcite:sand:water 0,7:0,3:5:1,5 were characterized by X-ray diffraction 2 4 4 2 (vol), followed by four applications of the (Table 2). Charity 1 ++++ + + ++ (+) ++ mixture kaolin:sand:water 1:5:1,5 (vol) The evaporates contain mainly then- Charity 2 ++ ++ tr ++++ ++ tr on the body parts of the sculpture, while ardite. The other major phase is an units face received five applications of the identified phase (ref “unident phase b”), Charity 3 ++++ - - - - - mixture kaolin:nanocalcite:sand:water different from the unidentified phase Charity 4a +++ ++ tr ++ + - 0,7:0,3:5:1,5 (vol). The reason for a limited (ref unident phase “a”) found on surfa- Charity 4b ++ ++ - +++ + - application of the mixture containing ce samples collected on the sculptures nanocalcite was that this poultice disin- before the desalination intervention. Charity 5 ++ + ++ +++ (+) +++ tegrated too easily on removal, inducing Three other phases are locally abundant: Esperance 1 ++++ - - + +++ - a strong dust production (health issues gypsum and halite, and the unidentified and dust deposition on artworks around), phase “a”. Calcic EDTA is found in four Esperance 2 +++ ++ - ++++ +++ - while the kaolin/sand mixtures detached samples. Interestingly, the sample cont- Esperance 3 ++++ + ? ++ + - more easily into big and coherent pieces. aining the higher amount of calcic EDTA The addition of nanocalcite was a trial to also contains high amounts of the two Esperance 4 +++ ++ - ++ ++ + reduce the stone stripping problems du- unidentified phases. The sample Charity Esperance 5 ++++ + ? ++ ++ + ring the removal of dried poultices. 3 was collected on big flat quadrangular Esperance +++ - ? - (+) +++ Kaolin based poultices being much white crystals visible in the two evapo- efflorescence adhesive, a number of millimetric sca- rates, suggesting that they are thenardi- les were stripped from the substrate on te (or rather mirabilite dehydrated into Table 2 : Overview of the phases found in the evaporates.

224 225 Tracing back the origins of sodium sulfate formation on limestone as a consequence of a cleaning campaign: the case study S. Benkhalifa et al. on Charité and Espérance sculptures of Chartres cathedral

thenardite). Other crystal shapes cannot case study, gypsum more probably ori- is also most probably linked to the fact Mora paste chemicals (and in particu-

be attributed clearly to a single phase, al- ginates from SO2 atmospheric pollution, that the chemicals could penetrate the lar EDTA disodium salt) are probably re- though it is clear that the cauliflower-like as plaster of Paris repairs are quite locali- stone down to 1cm instead of staying at sponsible for the quick deterioration of brown crystals (ref. Charity 2 and Espe- zed and absent close to the places where its very surface: In the original Mora for- the sculptures through complex and still rance 2) in both evaporates are essenti- samples were taken. mula the chemicals were kept by CMC, unidentified chemical pathways: at least ally made of the unidentified phase “b”. Sodium sulfate formation is for us re- while in this case study attapulgite was two of the mineral phases found in this Halite and gypsum were probably pre- lated to the 2007 cleaning campaign, in used instead of CMC. Attapulgite abili- case study are not present in the JCPDS sent in the sculptures before the 2007 particular the use of a modified Mora ty to retain water is far lower: this clay data base on mineral phases. The litera- cleaning campaign: halite often origi- paste. Sodium is present in the Mora easily changes from non-plastic to fluid ture on stone conservation being very nates from capillary rise in places where formula as di-sodium EDTA salt, and as over water addition, and in its fluid state, poor with that respect, we propose to bodies have been buried1, which is gene- sodium bicarbonate. Sulfates were pre- it cannot hold water. We believe that in build an international data base on mi- rally the case around churches. Gypsum sent as gypsum. Although we are not such a condition, the Mora paste chemi- neral phases formed as a result of stone contaminates very often limestone able to determine the exact sequence of cals solubilized in water may have more cleaning, and due to the use of EDTA. As sculptures which have been exposed to reactions that took place into the stone easily penetrated the stone. a first step, an inventory of stone conser- industrial pollution, but can also origina- as a result of Mora paste application, we The unidentified phase “a” found in vation issues related to the use of EDTA is te from repairs: in France, plaster of Pa- think that its chemical cocktail is res- efflorescence may also have played a role to be completed. ris (calcium hemihydrate) has been used ponsible for the formation of thenardite in the deterioration process. The uniden- a lot for this kind of use. In the present on the artworks. The strong deterioration tified phase b has not been detected in efflorescence, but appears to be a major Acknowledgements phase in the salts extracted by the poul- d(nm) d(nm) tice: this phase either was already pre- The authors wish to thank Irène Unidentified Unidentified Unidentified Unidentified sent in the sculptures and could not be Jourd’heuil, Curator at Chartres Cathed- phase b phase a phase b phase a identified because absent in the deteri- ral for her constant support during this - 75287 - 95659 orated parts we sampled, or was formed project, and Susana Ramirez from ERM 14. - 10283 - 66108 via some chemical reaction between the laboratory for her reactivity and profes- salts present in the sculptures, the relicts sionalism concerning the salt analyses. 42659 60388 12. 77839 - of Mora components and the kaolin-ba- 05. - 35667 11. - 87934 sed desalination poultice. - 31276 10. - 12769 - 26957 09. - 76787 6. Conclusion References - 01822 - 80737 1 08. - 53432 This case study of a sodium sulfate re- Benkhalifa S.. Rapport d’étude pour - 66869 - 45963 lated deterioration following the applica- quatre Vertus, par Berruer 1765-1770, tion of chemical products is interesting cathédrale de Chartres, Chartres, Pierre 07. 59332 39003 17940 37221 as it points out one of the sources of a calcaire. S. Benkhalifa, restauratrice de - 98898 - 33351 quite frequent secondary effect of sto- sculptures, 36 rue Damrémont, 75018 04. 36658 67528 - 17657 ne cleaning: the formation of secondary Paris. Report, 2012, 56p. salts, i.e. salts not present originally nor 06. 05857 58439 - 12826 applied during conservation, and far 2 Benkhalifa S. Rapport d’intervention - 52884 - 10238 more noxious than the original ones. dessalement après tests et recherche de - 26473 - 06840 The newly formed salts may modify the composition de compresse de dessalement adaptée ; Vertus en pierre calcaire - 92234 equilibrium relative humidity of the Unidentified phase a: found in samples collected on the salts present in the substrate, or simply de Berruer : la Charité, l’Espérance, l’Hu- sculptures before desalination and in the salt mixture 87233 73648 extracted from the desalination poultices be more dangerous that the original ones milité , la Foi, 1765-1770. Cathédrale de 03. 54945 65018 (gypsum trapping soiling particles). Here Chartres ; S. Benkhalifa, restauratrice de Unidentified phase b: found in the salt mixture extracted from the desalination poultices 50377 61350 the worst of all has formed: sodium sul- sculptures, 36 rue Damrémont, 75018 fate. Paris. Report, 2014, 17p. 41063 54237 Table 3: Inter-reticular distances of unidentified phases

226 227 SWBSS 2017 | 20-22 September S. Benkhalifa et al. 4th International Conference on Salt Weathering of Buildings and Stone Sculptures

Granite and schist masonry desalination by 3 Mora P., Mora L., Une méthode d‘élimi- nation des incrustations sur les pier- poultices at Jacobine Church in Morlaix, France res calcaires et les peintures murales, ICOM, In „Réunion du Comité de l‘ICOM Barbara Brunet-Imbault*, Benjamin Reidiboym and Clément Guinamard pour la conservation, Madrid,2-7 octob- Studiolo: engineering office for cultural heritage conservation and restoration re 1972“, 4p. * [email protected]

4 Morin C., Analyses quantitative de sels extractibles à l’eau. Rapport ERM Abstract including poultice formulation and des- 12249CM342, 2012, 11 pp. alination practice (number of poultices The Jacobine Church of Morlaix in applications, wall surface preliminary 5 Lubelli B., van Hees R.P.J., Desalination France has been deconsecrated in the preparation). Granite cut stone and schist of masonry structures: Fine tuning of XIXth century. Afterwards, it has been blocks masonries have distinct proper- pore size distribution of poultices to used as a stable and then turned into a ties, so desalination practice must be ad- substrate properties. Journal of Cultu- market which included salt storage for justed to obtain an optimized efficiency, ral heritage, Vol 11, Issue 1, Jan–March food preservation. Moreover, the church including a different poultice formulati- 2010, p. 10–18 is built downhill of a significant slope on and a different masonry preparation. next to the Jarlot river. The church, built Nevertheless, the desalination of such 6 Bourgès A., Vergès-Belmin V., Compa- with granite and schist, presents import- contaminated masonries is a challenge rison and optimization of five desali- ant salt resurgences, granular disintegra- to preserve lime plaster from future de- nation systems on the inner walls of tions and moisture areas on joint mort- gradations. Saint-Philibert church in Dijon, France. ars. The project of the city is to restore SWBSS proceedings, Copenhagen, 2008, the church and turn it into an exhibition Keywords: desalination, practice, poul- p. 29-40. space included in a new museum. Accor- tice, efficiency ding to this project, the architects want 7 Ramirez S., Morin C., Analyses quantita- to apply new lime plasters on the walls, tive de sels extractibles à l’eau. Rapport which is not possible because of the high 1. Introduction ERM 17103SR139, 2017, 4pp. levels of salts. In order to evaluate the salt contamination, samples have been The church is built next to the Jaco- 8 Arnold A., Zehnder K., Salt weathering taken in different heights and depths in bine convent of Morlaix, turned into a on monuments (Bari): Grafo, In „ the stones and mortars. Contamination pro- museum in 1887. In the restructuration conservation of monuments in the Me- files have been obtained and poultices project of the museum, the church will diterranean Basin: proceedings / of the based on cellulose fibers, clays and fine be integrated in the museum and turned 1st International Symposium, Bari, 7-10 quartz sand have been formulated. Sto- into an exhibition room. This church June 1989 ; scientific editor Fulvio Zez- ne and poultice porosimetries have been has a complex story, as it’s been used as za“, p. 31-58. measured in order to adjust the poultice a stable and a market with salt storage. formulation. The goal of this adjustment Moreover, the natural slope above the was to obtain the appropriate pore size northern elevation and the proximity distributions to improve the moisture of the Jarlot river foster capillary rise in transport mechanisms between stones the masonries. During the XXth century, and poultices. Three poultices formula- cement plasters have been applied on all tions containing variable ratios of cel- the interior walls of the church, made of lulose, sand and clays have been selec- schist blocks; only the granite cut stones ted. Test areas included join removal as have been preserved from plastering. Ce- much as possible and vacuum cleaning ment has also been used for the pillars of the wall surface before the first poul- and arch’s joints. tice application. These on-site trials allowed us to study desalination efficiency,

228 229 B. Brunet-Imbault et al. Granite and schist masonry desalination by poultices at Jacobine Church in Morlaix, France

2. Materials and methods

2.1. Building stones and degradation schist and patterns granit stone joint mortar schist and joint mortar The church’s masonries are constituted of granite and schist. Salt efflorescences, granit stone granular disintegration and moist areas granit stone can be observed at several places on the stones and on the cement plasters.

2.2. Sampling Figure 2: Location map of the samples. The nature of the analysed materials is indicated for each location

The sampling for the contamination diagnosis before the desalination tests had med on square samples of each poultices Formulations are indicated for dry to be representative of the different ma- formulations, in order to optimise the components. The best rheology is ob- terials in place (granite and schist stones, poultices in terms of desalination effi- tained by adding 30g of water for 100g of mortar) and of the different degradation ciency. The open porosity was measured dry components. patterns. In a first approach, we focused according to the NF EN 1936 standard. sampling on the most impressive pat- The mercury porosimetry was perfor- terns in order to analyse the feasibility of med with a porosimeter Micrometrics 2.4.2. Masonry preparation the desalination process in these areas, Autopore IV 9500 V1.09 at mercury pres- especially in the capillary rise zones. sures ranging from 0.034 to 206 MPa. Two surfaces were defined to perform Stones and mortar powders were Furthermore, water absorption of gra- the desalination tests: granite mason- sampled by drilling at two heights nite and shist stones was determined in ry (zone 1) and schist masonry (zone 2). (around 20cm high and 2m high) and situ using the sponge test method. Cement plaster was firstly removed from 3 depths (0-5, 10-15 and 30-40mm).; 30 the schist masonry. Each zone were sepa- powder samples were analysed before rated in two sub-zones according to the desalination tests for analysis of the so- 2.4. Desalination tests surface preparation: luble salts : sodium, potassium, chlorides, sulfates and nitrates were quantified. 2.4.1. Poultice formulations - type a methodology: surface brushed Three formulations of poultices were and cleaned with vacuum cleaner and 2.3. Analysis performed with cellulose fibers, silica joint removal up to 4 to 6cm depth sand and attapulgite clay. To obtain poul- Salt analyses were performed by io- tices permitting advection on stone with - type b methodology: joint removal up nic chromatography by the BPE labora- a very low porosity, poultices with a high to 2cm depth tory with the Thermo Scientific Dionex ratio of clay were studied. Aquion. Results correspond to the mass percentage of these ions inside the sample. Samples were prepared by aqueous BWW 40 Cellulose 0-2mm Sand Attapulgite clay extraction according to the standard NF Poultice T1 6 volumes 2 volumes 2 volumes EN 16455. Results are obtained with a relative precision of 5%. Poultice T2 3 volumes 3 volumes 4 volumes Porosity and porometry measurement were performed on cores of granite and Poultice T3 3 volumes 2 volumes 5 volumes schist stones sampled on site. The same Figure 1: Overviews and details of salt efflorescences and moisture areas on granite and cement plaster covering measurements have been also perfor- the schist masonry Table 1: Poultice formulations

230 231 B. Brunet-Imbault et al. Granite and schist masonry desalination by poultices at Jacobine Church in Morlaix, France

3. Results while sodium and potassium does not really decrease down to 4cm depth. Schist 3.1. Salt contamination contamination is extremely important, especially in the capillary rise area, and The results of the soluble salt analy- mortars are also strongly contaminated. ses are shown in figure 3. The french thresholds for stone conservation are the following ones : 3.2. Porosity and porometry of stones and poultices - Chloride < 0.10% - Sulfate < 0.10% if it’s associated with The results of the porosimetric mea- sodium or potassium, <5% if associated surements are presented in table 3. The with calcium, poultice porosity significantly decreases - Nitrates < 0.5% when clay is added 80% of the schist po- - Sodium <0.04% if associated with chlo- res are smaller than 1µm while for the ride granite stone, only 38% of the pores are - Potassium <0.05% if associated with smaller than 1µm, and 53% have a diame- chloride or 0.20% if associated with ter /radius ranging from 1 to 10µm. Poro- nitrates metry of the poultices indicates that the clay clearly plays a role in the presence In the granite stone, chlorides and sul- of very small pores (<0.1µm). The volume fates are mainly present in subsurface f of this category of pores increases when while sodium and potassium does not re- the clay proportion increases in the poul- ally decrease down to 4cm depth. Schist tice recipe. On contrary, the use of more contamination is extremely important, cellulose fibers increases the proportion especially in the capillary rise area, and of pores bigger than 10µm. Porometry mortars are also strongly contaminated. curves indicate that the poultices havea In the granite stone, chlorides and sul- category of pore space smaller thatn the fates are mainly present in subsurface f one of the granite and schist stones.

Schist M1 Granite M2 Poultice T1 Poultice T2 Poultice T3

Developed pore 1.671 0.205 15.1 20.5 30.9 surface (m2/g)

Porosity (%) 2.59 3.07 55.6 46.5 43.9 Figure 3: Ion quantifications on granites, schists and mortars before desalination Pore grades

Access radius (µm) Porosity of the different pore grades (%) 2.4.3. Poultice applications Poultices were removed after 3, 6 and 8 days respectively for the 1st, 2nd and 3rd 0.01 µm – 0.1 µm 59.9 11.5 17.4 31.5 41.6 Desalination poultices were applied applications. Very high levels of rain and 0.1 µm – 1µm 19.5 27.4 5.03 9.52 10.2 without previous water imbibition of humidity at that time slowed down the nd rd the wall. Poultice was applied by projec- poultices drying, so the 2 and 3 appli- 1µm – 10µm 10.1 53.2 65.7 54.1 45.3 tion then smoothed with a trowel. The cations remained in place longer. Even protocol consists in 3 consecutive appli- after 8 days, poultices were not totally 10µm – 100µm 10.5 7.9 11.8 4.91 2.86 cations for each zones and methodolo- dry.

gy so 4 poultices were applied 3 times. Table 3: Porosity and porometry of the poultices

232 233 B. Brunet-Imbault et al. Granite and schist masonry desalination by poultices at Jacobine Church in Morlaix, France

We can conclude that, the 3 poultice for- thodologies all over the desalination pro- mulations may permit advective extrac- cess. On granite, the desalination with tion on these granite and schist stones. methodology “a” may be considered as According to the proportion of volume satisfactory as chloride, nitrate and sul- of the different pore categories, T2 for- fate rates reach values below recommen- mulation poultice is the most suitable ded thresholds. With the b methodology, formulation in relation with the granite sulfates, remain over the thresholds. On pore size distribution while T3 formula- schist stones, results indicate a very im- tion poultice is the most suitable for the portant reduction of the sulfate rates in schist in place. both cases, with a and b methodology. Tables 6 and 7 show that the b methodology allows a better extraction of 3.3. Water rising measures anions and sodium while sulfates and sodiums remain higher with the a me- Sponge tests were performed on grani- thodology. te and schist stones. The results indicate Tables 8 and 9 indicating results on that water absorption of the shist stone mortars shows a significant efficiency is 44% more important than the one of for chloride and nitrates but sulfates, granite. sodium and potassium remain over thresholds for the mortars of granite ma- Figure 4: Pore size distribution curves of poultices formulations sonry (zone1). For the mortars of schist 3.4. Poultice efficiency masonry (zone 2), all ions except sulfates decrease under thresholds. The following tables indicate the results obtained on the granite and schist zones with the type a b desalination me-

Figure 5: Pore size distribution curves of granite and schist Table 4: Poultice efficiency on granite in zone 1 with type a methodology

234 235 B. Brunet-Imbault et al. Granite and schist masonry desalination by poultices at Jacobine Church in Morlaix, France

Table 7: Poultice efficiency on schist in zone 2 with type b methodology Table 5: Poultice efficiency on granite in zone 1 with type b methodology

Table 6: Poultice efficiency on schist in zone 2 with type a methodology Table 8: Poultice efficiency on mortar in zone 1 with type b methodology

236 237 B. Brunet-Imbault et al. Granite and schist masonry desalination by poultices at Jacobine Church in Morlaix, France

hing, cleaning with a vacuum cleaner Bibliography and deep joint removal has a significant effect to the desalination efficiency. Re- Bourgès A. and Vergès-Belmin V., Appli- sults show that in case of a large amount cation of fresh mortar tests to poultices of mortar, like in schist masonry, deep used for the desalination of historical joint removal induces poultice craqueles masonry. Materials and structures, vol while in granite cut stone masonry, with 44, n°7, 2011, 1233-1240. thin joints, deep joint removal fosters salt extraction. In both cases, for schist Bourgès A. and Vergès-Belmin V., Com- blocks masonry and granite cut stone parison and optimization of five des- masonry, salt extraction on mortar is not alination systems on inner walls of satisfying due to the high initial conta- Saint-Philibert Church in Dijon, France. mination of the mortars and the micro- Salt Weathering on Buildings and Stone structure with large pores. Sculptures: 22-24 October 2008, The Na- This study shows the feasibility of the tional Museum Copenhagen, Denmark: poultice adjustment to the stone porome- [Proceedings from the International Con- try, even with a very thin porous micro- ference], 2008, 29-40. structure. Granite cut stone and schist blocks masonries of this church have De Clercq H., Godts S., Debailleux L.,

Table 9: Poultice efficiency on mortar in zone 2 with type b methodology distinct properties; this study shows that Vanhellemont Y., Vanwynsberghe N., desalination practice must be adjusted to derammelaere L. and de Swael V., Electro- obtain an optimized efficiency, including phoresis as a tool to remove salts from 4. Result analysis Concerning the mortar contamination, different poultice formulations and dif- stone building materials – Results from results show an efficiency for schist ma- ferent masonry preparations. Deep joint lab experiments and on-site application. Initial contamination histograms are sonry with two T3 poultice applications, removal fosters salt extraction, avoiding International congress on the deteriora- representative of an important and glo- especially for the cations, but sulfates co- their diffusion, but this deep joint remo- tion and conservation of stone – Glasgow bal contamination, especially schists mes up with the 3rd application, which val leads to the cracking of the poultice, 2016, 289-297. which are strongly contaminated due to is significant of a stock of sulfates deep because of the large mortar microstruc- their thin sheet microstructure. in the stone. On the granite masonry, ture inducing a high water capillary ri- Lubelli B. and Van Hees R., Desalinati- Tests for granite desalination indicate T2 poultice applications doesn’t permit sing. Large microstructural mortar used on of masonry structures : Fine tuning a better efficiency of the T2 poultice after good extraction. Anions and cations re- with very thin microstructural schists of pore size distribution of poultices to previous brushing and cleaning with a main in levels over thresholds. complexifies the poultice adjustment. In substrate properties. Journal of Cultural vacuum cleaner and deep joint removal; this case study, we focused on the salt ex- Heritage 11 (1), 2010, 10-18. anions are all under thresholds. Two T2 traction in the schists due to their very poultice applications enable significant 5. Discussion high rate of contamination but the adjus- salt extraction on granites but the 3rd apted poultice microstructure for the very plication causes a cations increase, may- Results are significant of good poultice thin schist pores enhances the water ca- be because repeated applications bring to efficiency. The adaptation of the poultice pillary action in the large mortar pores, the surface some very deep salts. porometry to the different stone poro- leading to this cracking phenomenon. For the schist masonry desalination, metries allows to perform efficient salt In conclusion, this study permitted to a single T3 poultice application after extractions, even for schist porometry, obtain a significant salt contamination opening the joints for only 1cm gives the which is very thin. In this particular case, decrease. Nevertheless, desalinate such best results. The test including brushing, we obtain astonishing results. Neverthe- contaminated masonries is a challenge cleaning and deep joint removal doesn’t less, sodium and potassium remain over to preserve durable lime plaster in the permit anion reduction under threshold. threshold, which can, with the sulfate future. This could be caused by large poultice rates, provide salt cristallisations, even cracks in case of deep joint removal. The if anions are very low in several cases poultice dries way faster on the joint re- after desalination. The preliminary pre- moval zones and cracks appear rapidly. paration of the masonry including brus-

238 239 Case studies SWBSS 2017 | 20-22 September Salt content of dust and its impact on the wall paintings of the church St. Georg at the UNESCO World Heritage site 4th International Conference on Salt Weathering of Buildings and Stone Sculptures Monastic Island of Reichenau in Germany

Salt content of dust and its impact on the wall paintings 1. Introduction horizontal and inclined surfaces in nave, crypt and outside, mostly from window of the church St. Georg at the UNESCO World Heritage site In co-operation with the State Of- sills. Sampling intervals were monthly or Monastic Island of Reichenau in Germany fice for Monument Conservation Baden longer. Figure 1 shows the sampling sites. Württemberg, the Materials Testing Ins- The samples were analysed by optical Jürgen Frick*, Manuela Reichert and Harald Garrecht titute University of Stuttgart (MPA) ini- micros-copy and partly by ion chromato- Materials Testing Institute, University of Stuttgart, Germany tiated a national research project funded graphy and x-ray diffraction. Passive dust * [email protected] by the German Federal Envi-ronmental sampling on boron or carbon substrates Foundation (DBU). The Project has star- and fine dust measurements with an op- Abstract different areas in the nave and the crypt. ted in 2015 with the aim of the identi- tical particle counter (OPC, Grimm 1.109) The paper identifies possible sources and fication of anthropogenic risks and pre- were performed as well. St. Georg is one of the three Romanes- analyses the harmfulness in relation to ventive mitigation measures to reduce que churches on the island of Reichenau the indoor climate by salt mixture si- the environmental stress. One part was at Lake Constance, Germany, built bet- mulation with ECOS/RUNSALT. The data the collection and analysis of dust and 3. Results ween the 9th and 11th century. UNESCO in- were compared with drilling samples of fine-dust samples from time to time. Ad- scribed the monastic island of Reichenau walls and passive dust sampling data. ditional microbial volatile organic com- 3.1. Dust accumulation

in the World Heritage List in 2000. St. Ge- An experimental analysis by dynamic pounds (MVOC) sampling, fine-dust, CO2 org has meticulously restored wall pain- vapour sorption (DVS) of dust samples is and climate monitoring and passive dust The main series of dust samples was tings which are exposed to a very humid planned in near fu-ture. sampling was performed. See1, 2, 3 for an taken at the window sills of crypt and indoor environment. Anthropogenic ris- overview. This paper will focus on the arcade windows of the west apsis. The ks and preventive mitigation measures Keywords: Salts in dust, climate assess- analysis of dust and fine-dust samples accumulation differs with season with to reduce the environmental stress were ment, DVS, source identification, salt and the possible impact on the salt con- maximum values in autumn, see Figure identified within a research project. One mixture analysis by ECOS/RUNSALT tent on the wall paintings and walls in 1. Additional samples from outside (crypt aspect of research was the impact of salt relation to the indoor climate. and cellar window sills) and nave (south in dust collected from time to time from and north upper window sills) were taken. Table 1 lists the average values of 2. Experimental dust accumu-lation for all sites. The average daily accumulation of 6.1 Dust samples were collected by vacuum to 6.8 mg/(m2d) is comparable for nave cleaning on a quartz filter from several (west apsis) and crypt, except for the

west west crypt crypt crypt crypt crypt apsis apsis south north east south north Period north south exterior exterior

Average within period respectively in mg/(m2d)

6.7.15-25.7.16 6,1 3,6 6,2 6,5 6,8 - -

25.7.16-2.9.16 - - - - - 17 47,7

nave south wall nave north wall

west east west east

13.10.15-15.3.16 7,6 - 3,9 -

15.10.15-16.3.16 - 3,7 - 2,9

Figure 1: Sampling sites at St. Georg at the isle of Reichenau. Table 1: Average values of dust accumulation for all sampled sites. The respective periods are given.

242 243 Salt content of dust and its impact on the wall paintings of the church St. Georg at the UNESCO World Heritage site J. Frick et al. Monastic Island of Reichenau in Germany

crypt south window with 3.6 mg/(m2d). 3.2. Salt analyses (apsis south window sill) are given. A gra- analysis from the crypt with additional The exterior samples (only one sampling phical interpretation of these results is occurrence of niter. A sample from the period) showed higher values of 17.0 and The salt content of the dust samples shown in Figure 2. The first samples from apsis con-tained calcium oxalate hydrate. 47.7 mg/(m2d). At the north and south was analysed by ion chromatography. In July 2015 had a large collection time Due to the lime plaster of the sills and wall of the nave an increase from east to Table 2 the results for the two main se- compared to the following samples, but walls it was not clear if the amount of west is visible (Table 1). ries in crypt (east window sill) and nave the amount was comparable for most of calcium is due to the substrate. Additi- the ions. Therefore within a short period onally, some crystallized salts from the of one or two months the salts were accu- substrate could be in the collected dust mulated again. samples. Therefore the fine dust collec- In the crypt main ions were potas- ted in the filters of the optical particle sium, nitrate, as well as sodium, chlo- counters (OPC) and dust samples from ride and sulphate, whereas in the apsis quasi neutral surfaces (organ, crypt altar) higher concentrations of ammonia and were collected and analysed. The results phosphate occurred. Additional samples are shown in Table 3 together with two from window sills in the nave (north and samples from the outside window sills of south wall) showed nearly comparable the crypt. Due to the low amount of dust results as the apsis sills (see Table 4 in 1), (1.2 to 1.7 mg) from the OPC filters, these but with higher amounts of ammoni- values are less accurate. um (around 0.2 to 0.9 mass-%). Especial- On the filters high amounts of sodium, ly samples from the north wall showed ammonium, sulphate and nitrite occur- high amounts of nitrate (up to 0.9 mass- red compared to the main series (Table %) and oxalate (up to 0.3 mass-%). 2). The amounts of calcium, potassium, Some dust samples from crypt and ap- magnesium and nitrate are comparable sis were analysed by x-ray diffrac-tion. As to the samples from the crypt series. Th- expected calcite, gypsum and quartz oc- erefore salts from aerosol are a possible Figure 1: Dust accumulation of the main series in crypt (window sills) and nave (west apsis arcade windows sills). curred in both samples and some unspe- source. cific mixed crystals. Figure 3 shows an

Figure 2: Salt analyses by ion chromatography of the two main series of dust collection. Figure 3: X-ray diffractometry of a dust sample from the crypt east window sill.

244 245 Salt content of dust and its impact on the wall paintings of the church St. Georg at the UNESCO World Heritage site J. Frick et al. Monastic Island of Reichenau in Germany

+ + + + + + + + + + Positon/ Period/ Date Na NH4 K Mg2 Ca2 Positon/ Period/ Date Na NH4 K Mg2 Ca2 mass-% mass-% mass-% mass-% mass-% mass-% mass-% mass-% mass-% mass-% Crypt east OPC-Filter 06-07-15 1.07 < 0.01 11.30 0.33 2.64 Crypt 1.9 3.1 1.2 0.3 3.3 06-07-15 – 11-09-15 0.58 0.02 1.19 0.19 2.07 01-09-16 – 20 10-16 11-09-15 – 15-10-15 0.55 0.01 1.16 0.18 1.87 Apsis north L 1.4 2.1 0.8 0.2 2.2 17-12-15 – 13-01-16 0.93 0.02 1.90 0.15 0.90 01-09-16 – 20-10-16 Apsis north D 13-01-16 – 15-03-16 0.50 < 0.01 2.07 0.14 1.19 2.7 1.6 2.2 0.2 1.8 01-09-16 – 20-02-17 Apsis south Organ 06-07-15 0.17 0.43 0.27 0.20 1.28 1.6 2.4 1.0 0.2 2.4 01-09-16 – 20-10-16 06-07-15 – 11-09-15 0.18 0.03 0.34 0.21 0.83 Extra sites 11-09-15 – 15-10-15 0.18 0.02 0.22 0.15 0.81 Organ 0.79 < 0.01 0.21 0.07 1.29 17-12-15 – 13-01-16 0.18 0.21 0.17 0.07 0.77 02-09-16 – 24-03-17 13-01-16 – 15-03-16 0.31 < 0.01 0.35 0.06 0.86 Crypt altar 0.34 < 0.01 0.30 0.09 1.68 24-03-17 Positon/ Period/ Date Cl- NO - NO - PO 3- SO 2- C O 2- Crypt south outside 2 3 4 4 2 4 0.16 0.50 0.13 0.05 0.64 02-09-16 mass-% mass-% mass-% mass-% mass-% Crypt north outside Crypt east 0.19 0.76 0.27 0.06 2.15 02-09-16 06-07-15 0.93 < 0.01 9.78 n.d. 2.52 n.d. 06-07-15 – 11-09-15 1.12 < 0.01 2.64 0.08 2.38 n.d. 11-09-15 – 15-10-15 0.82 0.01 2.70 0.04 2.13 - - - 3- 2- 2- n.d. Positon/ Period/ Date Cl NO2 NO3 PO4 SO4 C2O4 17-12-15 – 13-01-16 0.71 < 0.01 4.10 < 0.01 1.47 n.d. mass-% mass-% mass-% mass-% mass-% 13-01-16 – 15-03-16 1.00 n.d. 3.64 0.01 1.94 0.04 OPC-Filter Apsis south Crypt 0.4 0.6 3.9 0.1 10.2 0.9 06-07-15 0.15 0.02 0.07 0.19 0.49 n.d. 01-09-16 – 20 10-16 Apsis north L 06-07-15 – 11-09-15 0.19 0.15 0.43 0.47 0.59 n.d. 0.3 0.2 2.1 0.1 6.8 0.5 01-09-16 – 20-10-16 11-09-15 – 15-10-15 0.19 < 0.01 0.53 0.27 0.51 n.d. Apsis north D 17-12-15 – 13-01-16 0.19 0.09 0.04 0.04 0.69 n.d. 0.2 0.2 2.4 0.0 6.3 0.0 01-09-16 – 20-02-17 13-01-16 – 15-03-16 0.09 n.d. 0.29 0.09 0.37 0.16 Organ 0.2 0.5 2.6 0.1 8.5 0.6 01-09-16 – 20-10-16 Table 2: Salt analyses by ion chromatography of the two main series of dust collec-tion. Top: Cations. Bottom: Anions. The end of period or the date represents the sampling date. Extra sites Organ 0.06 n.d. 0.21 0.02 0.57 0.10 The quasi neutral sites are comparable therefore some dissolution of salts could 02-09-16 – 24-03-17 to their counterparts in nave (organ and occur. Crypt altar 0.58 n.d. 1.77 < 0.01 2.52 0.02 apsis) and crypt (altar and east window 24-03-17 sill). Therefore crystallised salts from the Crypt south outside 0.09 0.02 0.03 0.22 0.26 < 0.01 walls could not be the only contribution 4. Discussion 02-09-16 to the salt content. The outside samples Crypt north outside 0.07 0.17 0.03 0.25 0.30 < 0.01 are comparable to the apsis samples with The discussion will focus on the risk 02-09-16 amounts of ammonium and phosphate potential for the precious wall paintings but less sulphate and chloride. In general and possible sources of the salts. Table 3: Salt analyses by ion chromatography of dust samples from OPC filters, quasi neutral sites and from outside. these sites are not protected from rain; Top: Cations. Bottom: Anions.

246 247 Salt content of dust and its impact on the wall paintings of the church St. Georg at the UNESCO World Heritage site J. Frick et al. Monastic Island of Reichenau in Germany

4.1. Risk potential 2) Treatment of nitrite, ammonium and therefore the calcium content was re- As for the crypt sample the salts are phosphate: In case of the crypt samples duced accordingly. mostly dissolved. Phase changes oc-cur- The measured salt content of the dust the amount could be neglected. In case red for the three salt mixtures mirabilite, is relevant and due to the accumula-tion of the apsis sample this was not pos- 4) The last step was the removal of aphthitalite and picromerite. of around 6 mg/(m2d) it is a continuous sible. In a first step nitrite was added gypsum from the data due to the low The simulations show that salt so- source for salts. The indoor climate is to nitrate which keeps the charge and solubility. The result is a set of data lutions occur in the crypt as well as in relatively humid therefore a solution of the particle number but changes the with either calcium (apsis, crypt drill the nave which could penetrate into the salts is possible which could be collected mass. The next step was the calculati- dust) or sulphate (crypt) set to zero. porous walls. Therefore there is a risk of by the porous wall structure. To estima- on of the charge sum of am-monium continuous salt accumulation in the who- te the risk poten-tial a simulation of the and phosphate because ammonium le church. The risk is higher in the crypt possible salt mixtures was performed phosphate is one of the main salts for Figure 4 shows the deliquescence hu- due to the higher salt contents in the with the soft-ware ECOS/RUNSALT.4 Two these ions. In our case the amount midities for the different simulated salt dust there. Two samples from nave and samples from crypt and apsis of known of ammonium was higher therefore mixtures for the sampling period from crypt were treated by dynamic vapour th th sampling periods and a third sample of nitrate and sulphate (as well cations of 6 July 2015 to 11 September 2015 in sorption (DVS), but the subsequent analy- drill dust (0 to 0.2 cm depth) from the major ammonium salts) were reduced the crypt in comparison with the climate sis by ion chromatography is in prepara- east wall of the crypt near to window sill accordingly. The results are sets of data of the same period. It could be clear- tion. This will give information about the 5 (analysis taken from ) were chosen. RUN- ECOS/RUNSALT compatible ions. ly seen that in most of the time all salts viability of simulated data. SALT works with a limited number of in the dust are in so-lution and a lot of

ions and gypsum should be removed in phase changes occurred for KNO3 (Niter). advance due to the low solubility. There- 3) Adjustment of charge sum to zero. Car- The drill dust sample which represents 4.2.Salt sources fore the fol-lowing steps were performed bonate could not be detect-ed by ion the wall surface would have some phase to generate ion balanced data for the dif- chromatography because for the anion changes for hexahydrite in that period. The relatively high amounts of salts in ferent samples: analysis a hydro-gen carbonate/ car- In Figure 5 the results of simulati- the filter of the OPCs indicate that aero- bonate solution is used as eluent. The on for the apsis sample are shown. The sols could contribute to the salt content 1) Calculation of mole data out of the assump-tion was that the charge dif- sampling period was from 17th December in the dust. In the region of Baden-Würt- analyses. ference is mainly due to calcium and 2015 to 13th January 2016. temberg6 fine dust collected outside

Figure 4: Comparison of simulated deliquescence humidities with climate data of the dust sampling period (6th July Figure 5: Comparison of simulated deliquescence humidities with climate data of the dust sampling period (17th 2015 to 11th September 2015) in the crypt. December 2015 to 13th January 2016) in the ap-sis.

248 249 Salt content of dust and its impact on the wall paintings of the church St. Georg at the UNESCO World Heritage site J. Frick et al. Monastic Island of Reichenau in Germany

at different sites and periods was exa- hall serves as sluice, only when the front Geologie, Rohstoffe und Bergbau (LGRB) ditions, as described in Price (2000). mined. Ammonium salts (nitrates and door is closed the groups enter the nave im Regierungspräsidium Freiburg”, Ro- http://science.sdf-eu.org/runsalt/ sulphates) occurred in percentages from by a second door. But due to the high bert Lung, Corinna Luz, tourism office Price, C. A. (Ed.), 2000. An expert che- 25% (high traffic street in Stuttgart) to interest as UNESCO world heritage the and catholic congregation of Reichenau. mical model for determining the en- 48% (Mann-heim urban background) and church has large amounts of visitors. Ty- vironmental conditions needed to pre- 51% (Black forest rural clean area). The pical dust from visitors is textile and bio- vent salt damage in po-rous materials. sources are secondary aerosols formed logical particles as well as dust from the European Commission Research Report No 11, (Pro-tection and Conservation of from gaseous NOx and ammonia in the surrounding area from the shoes. The lo- atmosphere. In 1 the deposition rates in wer amounts of salts in the dust samples European Cultural Heritage). Archetype St. George were compared with data from of the nave compared to the crypt (Table References Publications, London. a sheltered outside position (clock tower 2) might have their ex-planation in the 5 of Christchurch in Mainz).7 The outside additional dilution by non-mineral dust 1 Reichert, M.; Frick, J.; Scheuvens, D.; Zöldföldi, J., “Zerstörungsfreie und mini- rates are 5 to 100 higher than in St. Geor- from visitor traffic. Zapf, J.M.; Weinzierl, W., ”Raumluft Mo- malinvasive Untersuchungen zu Feuch- ge which is not amazing due to the traffic nitoring in St. Georg auf der Reichenau: te- und Salzbelastungen in der Krypta surrounded site. This are hints that sec- Leichtflüchtige organische Verbindun- St. Georg auf der Reichenau“. In “UN- ondary aerosols are one of the sources 5. Conclusion gen, Staubpartikel und deren Quellen“. ESCO-Weltkulturerbe Reichenau – Die Wand-malereien in der Kirche St. Ge- for salts in the dust. In “UNESCO-Weltkulturerbe Reichenau org als Schlüssel zu einer nachhaltigen Another source could be salts from The salt content in the dust contributes – Die Wandmalereien in der Kirche St. Denkmalpflege“, ed.: Jakobs, D. Garrecht, agricultural use. The Reichenau island to the salt content of the walls due to so- Georg als Schlüssel zu einer nachhal- H., Arbeitsheft 33 Regie-rungspräsidium is intensively used by agriculture. In [1] lution processes according to the climate tigen Denkmal-pflege“, ed.: Jakobs, D. Stuttgart Landesamt für Denkmalpfle- a calculation was performed by one of conditions. Due to the con-tinuous accu- Garrecht, H., Arbeitsheft 33 Regier-ungsge, Fraunhofer IRB-Verlag, Stuttgart, the authors (Weinzierl) which showed mulation of dust (around 6 mg/(m2d), see präsidium Stuttgart Landesamt für 2017, pp. 111 – 124. that nitrates from agriculture could con- Table 1) it could increase the salt cont- Denkmalpflege, Fraunhofer IRB-Verlag, tribute only to a minor part. The calcu- ent in the walls over time and could be Stuttgart, 2017, pp. 157 – 176. 6 Ahrens, D.; Anke, K.; Drechsler, S.; Gro- lated amount of nitrate of 0.013 mass-% harmful to the precious wall-paintings. 2 Zapf, J.M., “Analyse von Feinstaub-Depo- mes, B.; Holst, J.; Lumpp, R; Sähn, E.; is small compared to inside values and The ECOS/RUNSALT simulations show sitionsproben aus der Kir-che St. Georg Scheu-Hachtel, H.; Scholz, W., “Einfluss- only comparable to the outdoor samples the poten-tial risks for the walls. Main (Reichenau) mittels Rasterelektronen- größen auf die zeitliche und räumliche (Table 2 and 3). On the other hand the sources of the salts in the dust are aero- mikroskopie“. Bachelor thesis, Institut Struktur der Feinstaubkonzentratio- loose bound ground contains fine silt sols and input from visitor traffic. In the für Angewandte Geowissenschaften, TU nen“. Re-port LUBW Landesanstalt für which could easily blow off by wind eros- crypt some recrystallized salts from the Darmstadt 2017. Umwelt, Messungen und Naturschutz ion. Sili-cate rich particles are one of the walls could contribute as well. Baden-Württemberg, 2007, chapter major compounds detected with passive It was suggested to keep the regula- 3 Weinbruch S.; Scheuvens, D., “Möglichkei- 3.10. Online available at: http://www4. sampling.1, 2 tions for the visitor traffic in the sum- ten und Herausforderun-gen der Einzel- lubw.baden-wuerttemberg.de/servlet/ The higher amount of nitrates and sul- mer period. Further research is needed to partikelanalyse in denkmalgeschützten is/36153/. phates in dust samples from the crypt estimate the accumulation of dust by the Innenräumen“. In “UNESCO-Weltkultu- 7 (Table 2) might be due to the uptake of automatic window opening for climatisa- rerbe Reichenau – Die Wandmalereien Bundschuh, P.; Auras, M.; Kirchner, crystallised salts from the wall dur-ing tion. It is planned to evaluate sedimenta- in der Kirche St. Georg als Schlüssel zu D.; Scheuvens, D.; Seelos, K., “Expo- sampling. There is as well a difference on tion rates and flow conditions. einer nachhaltigen Denkmalpfle-ge“, sitionsprogramm zur Wirkung ver- quasi neutral surfaces (organ and crypt Acknowledgements ed.: Jakobs, D. Garrecht, H., Arbeitsheft kehrsbedingter Immissionen auf Na- altar) visible, but on wall samples the dif- We thank the “Deutsche Bundestif- 33 Regierungspräsidi-um Stuttgart Lan- tursteinoberflächen“. In “Baudenkmäler ference is bigger (compare Table 3). tung Umwelt” (German Federal Envi-ron- desamt für Denkmalpflege, Fraunhofer unter dem Einfluss verkehrsbedingter There are several ways for aerosols to mental Foundation – DBU) for funding. IRB-Verlag, Stuttgart, 2017, pp. 177 – 180. Immissionen“, ed.: Institut für Steinkon- enter the church. One is the auto-matic We greatly acknowledge the help and servierung e.V., Bayerisches Landesamt opening of windows to control the cli- support of our project and cooperation 4 ECOS/RUNSALT 2005. RUNSALT (Copy- für Denkmalpflege; IFS Bericht 49/2015, mate in nave and crypt. Another source partners “Landesamt für Denkmalpfle- right (c) 2002-2005 by Davide Bionda, pp. 53 – 77. is the visitor traffic (only the nave could ge im Regierungspräsidium-Stuttgart”, Ph.D.) is a graphical user interface to be visited). During the summer period Institute of Applied Ge-osciences (Group the ECOS thermodynamic model for (May to September) the visitor traffic is of Environmental Mineralogy) Technical the prediction of the behaviour of salt restricted to guided tours. The entrance University of Darmstadt, “Landesamt für mix-tures under changing climate con-

250 251 SWBSS 2017 | 20-22 September 4th International Conference on Salt Weathering of Buildings and Stone Sculptures Investigation of Salts Souces at the Karadjordje’s Gate on the Belgrade Fortress

Investigation of salts souces at the Karadjordje’s appear as a result of the interaction of sed material used for restoration had si- aerosol pollutants with certain minerals, gnificant influence on stone degradation Gate on the Belgrade Fortress as in the case of gypsum12, or may ori- that occurred after the restoration was ginate from mortar in contact with the carried out. Maja Franković1*, Nevenka Novaković2, Suzana Erić3, Predrag Vulić3 and Vesna Matović3 stone, or even from the stone itself.13 1 Central Institute for Conservation in Belgrade, Serbia Other important sources of salts inclu- 2 Cultural Heritage Preservation Institute of Belgrade, Serbia 14 3 University of Belgrade, Faculty of Mining and Geology, Belgrade, Serbia de biological activity and polychrome, 2. Site characteristics and historical *[email protected] cleaning and conservation/restoration background of the Karadjordje’s Gate treatment.15 Abstract their degradation. Monuments are ex- During restoration of heritage buil- The Karadjordje’s Gate is located in posed to climate change, pollution, use dings, mortars are frequently used for the central and oldest part of the urban The Karadjordje’s gate is a monument of demands, lack of maintenance, as well as the repointing of joints or for the ‘‘pla- environment of Belgrade - the capital of culture, part of the historical complex inappropriate conservation treatments. stic’’ repair of stone. They consist of a bin- Serbia. The climate in this area is tempe- of the Belgrade Fortress. The gate, dated Incompatibility in phisical and chemical der, aggregates and sometimes additives rate continental: average annual tempe- from 1740 to 1791, is made of autoch- properties of building materials, in com- or adjuvants.16 Plastic repair mortars are rature 11.7 oC; average annual rainfall thonous limestone of Miocene age. After bination with environmental factors, of- often subdivided based on their binders, 669 mm/year; relative humidity 69.5% years of exposure to environmental con- ten causes damage of the built material. as repair mortars with cement, lime, or (data from the Republic Hydro-Meteoro- ditions and different anthropogenic in- A significant part of the damage of a combination of both.17 Interpreting the logical Service of Serbia). fluences, the stone blocks showed a wide monuments is due to salt crystallization philosophical and ethical guidelines of The gate, dated from 1740 to 1791, was range of decay forms. The gate was sub- in the pores of stones and bricks.1 The both the Venice Charter and the Nara Do- not used for a long time – plans from the ject to a restoration campaign in 2007 cristallization or hidration pressure of a cument, an ideal repair mortar for natu- end of the 18th century already depict it using cement based materials. Today, the particular salt or, in short „presence of ral stone should be durable enough, but walled up, with no bridge. It was put to gate shows renewed signs of degradation: salts“ are often attributes of deteriorati- self-sacrificing on the long run.18 Port- use again only after World War II, during detachment of “artificial stone” used for on of materials.2 The deteriorating effect land cement was often used in the past the rearrangement of Kalemegdan park restoration; scaling and disaggregation of salts is grouped under the detachment for repairing stone monuments. Due to and the fortress. Today, Karadjordje’s gate of the original stone blocks. Salts efflore- category of the ICOMOS-ISCS Glossary3 the poor compatibility and low porosi- lies in the second communication line scence is present around the joints of stone but deterioration induced by salts forms ty of Portland cement compared to the approaching Belgrade fortress, connec- blocks in the upper part of the gate and a continuum between the granular disin- stone types, use of this paste can have ting it with a part of the town along Sava under the reconstructed rosettes. tegration and scaling, delamination, and negative consequences. It forms low-per- River. Characterization of salts was carried out blistering patterns.4 Many mechanisms meable to the impermeable zone, which The gate consists of frontal (south) fa- by SEM-EDS and XRPD analyses. Results are involved in the deterioration of po- prevents circulation of moisture through cade (investigated area in this paper), showed the presence of the following rous materials by salts. For damage to the more permeable original materials central aisle with two side rooms for salts: syngenite, gypsum, thenardite, dar- occur, salts must move into and within and produces a accumulation of water guards and north facades built of bricks apskite, bassanite, niter, aphthitalite and porous bodies, a process that requires in the more weathered original stone and stone. Frontal facade consists of dou- witzkeite. The paper concludes that there the presence of water or moisture.2 and to its freezing during winters.15 The ble stone arches and it is finished with is an influence of restoration materials to Some of main control factors of salt authors report that the Portland cement decorative cornice (Fig. 1). Facade is built salt contamination and to the decay of induced damages are: the rock fabric ele- paste could also be a source of the wa- entirely of limestone blocks with sim- stone and “artificial stone”. ments, especially porous network5, 6, the ter-soluble salts that contribute to the ple decorations representing two sime- type of salts and their amount7, the cli- rapid disintegration of the more recently trical rosettes. The gate was subject to Keywords: limestone, salt, restoration matic conditions, especially evaporation inserted slabs. the restoration campaign carried out in materials, decay rate, water supply rate and mechanical In the case of the Karađorđe‘s gate, on 2007. The cleaning was done by water jet resistance of the material.8, 9 which this study was carried out, salts and restoration of damaged stone blocks Salts may accumulate in natural stone have emerged after the restoration tre- was carried out, without prior consolida- 1. Introduction over time in several ways. Some of the atment in which cement was used for tion of the stone, with so called „artificial primary sources of salt contamination restoration of stone blocks and joints. stone“. Joints were restored with Portland Construction materials in monuments include capillary uptake of ground and The study was carried out with the ob- cement mortar. of culture are subject to a number of en- surface water, interaction between buil- jective to characterize present salts and vironmental factors that, acting together ding materials10, or deposition of acidity to determine their possible sources. The in different combinations, influence from the atmosphere.11 Salts may also main hypothesis was that the cement ba-

252 253 M. Franković et al. Investigation of Salts Souces at the Karadjordje’s Gate on the Belgrade Fortress

patterns were obtained from 4 to 70o 2θ The prevailing allochemical com- using CuKα1,2 radiation with a scan of pounds in all types are fossil-skeletal 5o/min. fragments of red algae (Lithotamni- um ramisissimum; family Corallinace- ae) (Fig. 2b). The remains are micrite in 4. Results composition with a well-developed and prominent mesh texture consisting of 4.1. Characteristics of built materials micron-sized rectangle chambers. Fossil remains of some other species (e. g., bryo- Petrographic analyses of natural stones zoans, gastropods and different forami- imply that the most dominant material nifer species) were also recognised. They built into the structure are autochtho- are recrystallised and filled with sparry nous, allochemical limestones, Miocene calcite. The algal biosparrudite is extre- in age. Textural features and micro- and mely porous and weakly consolidated macro-faunal characteristics of the li- rock with sparite bounds with numerous mestones identify them as Grainstone vesicles, mesopores, rounded cavities microfacies and Algal rudstone micro- and channels but fine/capillary pores, facies. These two microfacies types are too. Grainstone microfacies is moderate limestones mineralogically composed of porous rock with pores in range up to calcite (CaCO ), with very small amounts 0.1 mm. Water absorption of limestones Figure 1: The view of the Karadjordje’s Gate in Belgrade (Serbia) and position of the taken salt samples 3 of silici-clastic input. The limestones are used to build Belgrade fortress is about whitish, pale yellow or beige in colour 15%, while the compressive strength vary (Fig. 2a). from 4.5 to 13 MPa.19 3. Methodology stone were taken from the monument. The “artificial stone” used for restora- Petrographic analyses of the built stone tion purposes was composed of 2 parts Salt samples were taken from stone were performed on thin sections using ground limestone aggregate, and binder blocks built in exterior walls of the Ka- a Leica DMLSP microscope for polarised that consisted of 1 part white cement and radjordje’s Gate in November 2016. The light that was connected to a Leica DC 0.1 parts slaked lime. Mineral pigments main criteria for the sampling positions 300 digital camera. Classification of built were added in order to match the colour were: significant quantities of salt in the stone was done according to textural of the original stone. Water absorption form of efflorescence on the surface and characteristics. Determination of the we- of the artificial stone was 1.6%, its com- the originality of the blocks (natural and athering forms types was done according pressive strength at 28 days is 43.3 MPa.19 artificial) where salts appear. In order to to ICOMOS-ISCS glossary on stone deteri- Portland cement mortar was used for 3 analyse of salt composition by X-ray pow- oration patterns. restoration of the joints. der diffraction (XRPD) and SEM/EDS, five SEM-EDS was performed using a JEOL samples of salts were taken from the sub- JSM-6610LV scanning electron microsco- strate in following order: S1 – beneath pe connected to an X-Max energy disper- 4.2. The decay forms of built stone the broken rosette made of “artificial sto- sive spectrometer to identify the mor- ne”; S2 – from the surface of the cement phology and chemical composition of Mapping of the Gate’s façades and re- mortar joint above restored stone block; the mineral phases present in the salts. gistration of decay forms of stone degra- S3 – beneath a scale on the original stone The samples were covered with gold and dation after the restoration was carried block; S4 – from contact zone between ro- carbon using a BALTEC-SCD-005 sputter out. Damage occurs on the stone blocks sette of “artificial stone” and original sto- coating device, and the results were re- at the upper part of the south façade. The ne substrate; S5 – beneath a scale located corded under high vacuum conditions. most dominant decay forms are scaling on the original stone block surrounded X-ray diffraction analyses were used and disaggregation of the original stone by blocks reconstructed with “artificial to determine the phase compositions of blocks, as well as detachment of “artifi- stone” (Fig. 1). powdered salts and powdered limestone. cial stone” used for restoration from the In order to examine the built stone, The XRPD was performed using a Rigaku Figure 2: The macroscopic appearance (a) and photo- rosettes (Fig. 3a, b). Salts deposits occur micrograph (XPL) of built Algal rudstone from the structurally different samples of natural Smartlab diffractometer. The diffraction Karadjordje’s Gate around the joints on the stone blocks

254 255 M. Franković et al. Investigation of Salts Souces at the Karadjordje’s Gate on the Belgrade Fortress

gates and variable content (4.5-20.7%) ordinarily follows syngenite in the all samples. Although thenardite is present to a lesser extent (6.5-13.4%), its presence in the form of tabular aggregates or bipyramidal crystals, is evident on the left side of the façade. Darapskite and niter (crust-like aggregates) occur up to 4% in contents. Aphthitalite occurs only in sample S1 as trigonal crystals ranging

Figure 3: The damage forms on stone blocks. a) detach- from 3 to 10 μm in content of 8.3%, whe- ment of “artificial stone”; b) scaling and disaggregation reas basanite is present on right side of of the original stone blocks; c-d) efflorescence on mortar joints and surface of artificial stone outer façade (3-5%).

in the upper part of the gate, as well as 5. Conclusion beneath the broken parts of rosette or beneath the scales on the original stone The study of damage type and salts de- (Fig. 3c, d). posits on stone blocks on the Karadjord- je‘s gate show that there are several factors which contribute to decay of restored 4. Identification and distribution of stone blocks as well as to the occurrence present salts of salts. First, cement mortars used for joint filling and “artificial stone” possess XRPD and SEM-EDS analyses of the different properties from the original samples showed the presence of the follo- substrate. Its incompatibility is reflected wing salts: syngenite, gypsum, thenardi- in significantly lower water absorption te, darapskite, bassanite, niter, aphthitali- and greater compressive strength com- te and witzkeite (Table 1, Fig. 4, 5). paring to original limestone substrate. Calcite prevails in the all samples but Orginal soft stone blocks easily absorb its presence origins of substrate. Synge- and release water. That indicates that the nite occurs in prismatic or rarely elong- contact zone between the original poro- ated aggregates as dominant salt in all us limestone and cement mortar used samples but vary in contents (8.5-13.4%). for restoration represents hygric imper- Gypsum in the form of tabular aggre- meable zone. Longer moisture retention

Sample/stone Salt content (%) type G S T N D B A W S1/a-o 4 11 13 3 3 / 8 1 S2/a 11 9 7 4 1 / / / S3/o 4 14 7 4 3 / / / S4/a-o 21 11 / / 2 3 / / S5/o 14 13 / / / 5 / / Figure 4: X-ray diffraction patterns of the salts formed on built limestones of Karadjordje’s Gate with respective semi Legend: a – artificial stone; o – orginal limestone; G – gypsum; S – syngenite; T – thenardite; N – niter; D – darapskite; quantitative analyses results B – bassanite; A – aphthitalite; W – witzkeite

Table 1: Composition of investigated salt samples

256 257 M. Franković et al. Investigation of Salts Souces at the Karadjordje’s Gate on the Belgrade Fortress

the properties of the „artificial stone“ do 7 Ruiz-Agudo E, Mees F, Jacobs P, Rodri- not match those of the original, especi- guez-Navarro C The role of saline so- ally its hygric and mechanical proper- lution properties on porous limestone ties, and if the Portland cement is used salt weathering by magnesium and so- in restoration, damage is likely to occur dium sulfates. Environ Geol (52) (2007) and it will lead to the deterioration of the 269–281. very stone that is meant to be protected. 8 Goudie AS, Parker AG Experimental simulation of rapid rock block disintegration by sodium chloride in a foggy coastal desert. J Arid Environ (40) (1998) 347–355. References 9 Diaz Gonçalves T.C., Salt crystallizati- 1 Goudie A, Viles H Salt weathering ha- on in plastered or rendered walls, PhD zards. Wiley, Chichester (1997). thesis, Universidade Técnica de Lisboa Instituto Superior Técnico, (2007) 19-22. 2 Charola E.A. Salt in the deterioration of porous materials: an overview, JAIC (39), 10 Price C.A., Stone conservation: An over- (2000) 327-343. view of current research. Santa Moni- ca, Calif.: Getty Conservation Institute, 3 Vergès-Belmin V. et al, ICOMOS-ISCS J. Paul Getty Trust. (1996) 7-9. Illustrated glossary on stone deterioration patterns, Monuments and Sites 11 Steiger M., Charola A.E., Sterflinger K., XV, Ateliers 30 Impression, Champigny/ Weathering and Deterioration. In: Sie- Marne (2008). gesmund, Siegfried, Snethlage, Rolf (Eds.), Stone in Architecture Properties, 4 Snethlage R, Wendler E Moisture cycles Durability, Springer (2014) 265. and sandstone degradation. In: Baer NS, Snethlage R (eds) Saving our architec- 12 Halsey, D.P., Dews, S.J., Mitchell, tural heritage: conservation of histo- D.J., Harris, F.C., Real time measure- ric stone structures, Wiley, Chichester, ments of sandstone deterioration: a (1997) 7–24. microcatchment study. Build. Environ. (30) (1995) 411–417. 5 Thomachot-Schneider C., Gommeaux Figure 5: SEI images with EDS spectrums of salts from the analyzed samples S1-S5 M., Fronteau G., Oguchi C.T., Eyssautier 13 McKinley, J.M., Curran, J.M., Turking- S., Kartheuser B., A comparison of the ton, A.V., Gypsum formation in non-cal- in repair mortars can cause its chemi- The thick earthen mound that covers properties and salt weathering suscep- careous building sandstone: a case cal dissolution which could be potential the top part of the gate retains moisture tibility of natural and reconstituted study of Scrabo sandstone. Earth Surf. source of sulphates. Water retention also which is drawn inside by capillary action stones of the Orval Abbey (Belgium), Process. Landf (26), (2001) 869–875. causes physical damage due to freeze/ and evaporates on the surface of stone Environ Earth Sci (63) (2011) 1447–1461, thaw cycles. Gate‘s architectural features blocks. The water percolation from the DOI 10.1007/s12665-010-0743-8. 14 Gómez-Heras, M., Benavente, D., Álva- and its positioning within the rampart mound could also be the source of salts rez de Buergo, M., Fort, R., Soluble salt also contribute to described degradati- originating from grass fertilizers or de- 6 Benavente D., Martínez-Martínez J., minerals from pigeon droppings as on mechanism. Slightly indrawn south composition of organic matter, such as Cueto N., García-del-Cura M.A., Salt we- potential contributors to the decay of facade creates semi-sheltered area at the niter, present in three of the samples. athering in dual-porosity building do- stone based cultural heritage. Eur. J. upper part and prevents stone blocks to This case study demonstrates the im- lostones, Engineering Geology 94 (2007) Mineral. (16) (2004) 505–509. be washed by rain, which is pogodna en- portance of compatibility of materials 215–226. vironment for the forming of gypsum. used in restoration with the original. If

258 259 SWBSS 2017 | 20-22 September M. Franković et al. 4th International Conference on Salt Weathering of Buildings and Stone Sculptures

15 Přikryl, R., Novotná, M., Přikrylová Investigation and examination of a degraded Egyptian Weishauptová, Z., Št’astná, A., Physical and mechanical properties of the re- painted limestone relief from Tell Hebua (Sinai) paired sandstone ashlars in the facing 1 2 3 4 5 masonry of the Charles bridge in Pra- Eid Mertah *, Moamen Othman , Mohamed Abdelrahman , Mohamed Fatoh and S. Connor 1 Conservator of archaeological materials, Conservation Centre, Egyptian Museum of Cairo, Ministry gue (Czech Republic) and an analytical of Antiquities, Egypt study for the causes of its rapid decay. 2 Director of conservation Centre, Egyptian Museum of Cairo, Ministry of Antiquities, Egypt Environ. Earth Sci. (63) (2011) 1623– 3 Director of Scientific Research, Projects Sector, Ministry of Antiquities, Egypt 1639. 4 Chief of stone Conservators, Conservation Department, Egyptian Museum of Cairo, Ministry of Antiquities, Egypt 5 Research Fellow at The Metropolitan Museum of Art, New York 16 Isebaert A., Van Parys L., Cnudde V., * [email protected] Composition and compatibility requirements of mineral repair mortars for stone – A review, Construction and Abstract ancient “khetem of Tjaru”2, about 4 km Building Materials (59) Elsevier (2014) from the city of al-Qantara East, on the 39–50. This paper presents a study bas-relief East Bank of the Suez Canal, about 50 km from the New Kingdom (1550-1070 BC), north of Ismailia. 17 Feilden BM., Conservation of Historic which was found in Sinai in 2009 and Recent excavation led by Mohamed Buildings. 3rd ed. Oxford: Architectu- came the Egyptian Museum, Cairo, in a Abd El-Maksoud for the Supreme Coun- ral Press, Elsevier (2003). quite challenging condition. The interest cil of Antiquities revealed an impressive of this piece lies in the danger that the defence system, consisting of a series of 18 Schueremans L et al., Characterization salt causes to the block’s surface, and par- mud brick fortresses, protecting temples, of repair mortars for the assessment ticularly to the pigments of its decorati- administrative buildings and domestic of their compatibility in restoration on. The block is sculpted on two opposite structures, dating back from the New projects: research and practice. Constr sides, which both need to be consolida- Kingdom to the Late Period (ca. 1550-332 Build Mater 25 (2011) 4338–50 http:// ted, in order to be safely displayed in the BC). These fortified cities were intended dx.doi.org/10.1016/j.conbuildmat. museum’s galleries. The authors describe to protect Egypt’s North-Eastern border. the investigation carried out to identify The site of Tell Hebua II, which seems 19 Matović V., Condition report of the built the types of salts and to select the pos- to have been used for military purposes stone in the King‘s gate of the Belgrade sible techniques to remove them and pre- essentially in the New Kingdom, continu- fortress, Institute for the protection of serve the surface and pigments. ed to be occupied in later times. Indeed, monuments of culture, Belgrade (2007) the excavators found this relief, together (unpublished document). Keywords: limestone relief, Sinai, salt we- with several other temple blocks decora- athering, crystallized salts, stabilization. ted and inscribed, reused in the casing of later tombs set within earlier structures. According to the continuity of the orna- 1. History mentation and the orientation of their inscriptions, these blocks most proba- The piece which is the object of this stu- bly belonged to a tripartite sanctuary, dy is made of limestone. The decoration the foundations of which could not be is on two opposing sides. Its dimensions identified so far. These blocks attest two are: Height. 107; Width. 114; Thickness. phases of decoration: the first one during 27 cm. It is currently in display in the the reign of Thutmosis II (1482-1480 BC) gallery in the Egyptian Museum, Cairo, and a second during that of Ramesses II with the inventory number JE 100014 (1279-1213 BC). (= SR 4/15988). The decoration of the studied relief The block was found in a New Kingdom presents a particular interest, since it military site, discovered in North-West displays offering scenes on either of its Sinai on April 22, 2009, Tell Hebua II,1 the opposing sides, sculpted at two different

260 261 E. Mertah et al. Investigation and examination of a Degraded Egyptian Painted Limestone Relief from Tell Hebua (Sinai)

times: on one side, carved in raised re- Due to the quality of its reliefs and its lief3, Thutmosis II receives life from the poor state of preservation, the piece was god of war Montu, while on the other brought to the Egyptian Museum, short- side, which is carved in sunken relief, ly after its discovery. Ramesses II makes an offering of bread to the god of earth Geb. Very few repre- sentations of Thutmosis II are known so 2. Conservation state far, which makes this relief, of very high quality, a precious testimony of his reign. The block JE 100014 was not found in Furthermore, this block is evidence of its original setting, but reused as the clo- the restoration and reuse of stone struc- sing of the entrance to a tomb cut into tures during the reign of Ramesses II in the rock. Since it was then buried under Figure 3: [left image] Crystallized salts in the upper area of the Ramesses II’s side, [right image] Crystallized salts in the the fortress and its temple, two centuries soil and surrounded by mud, it was in lower area of the same side. after its building. relatively wet and closed environment

for more than two thousand years. The bably intensified this process of salt da- photographs taken at the time of its dis- mage since salts were moving to the sur- covery still show a good state of preserva- face of the block as drying occurred, and tion of the upper side of the piece, which therefore accelerated the deterioration of was that of Thutmosis II’s offering scene.4 its sculpted and painted decoration. According to the available records, and The majority of the painted areas beca- to the information provided by Gharib me weak and needed urgent consolidati- Sonbol, chief of the restoration team du- on. In many zones, crystallized salts have ring the excavation, the lower face of the appeared below the layer of colour, which block, showing Ramesses II, had already might lead to its loss in the future. This suffered salt damage when the block was situation is particularly visible on the discovered. The removal of the limestone side which contains the scene of Rames- block from its wet atmosphere to a much ses II (which was placed face down, when dryer one, in the Egyptian Museum, pro- the piece was reused as a slab for closing

Figure 1: Slab JE 100014 in its current display in the Egyptian Museum – face 1 [left image] (offering scene of Ramesses II) and face 2 [right image] (offering scene of Thutmosis II).

Figure 4: The circle shows the area where Microballoons filler was used to complete the missing parts of the pigment Figure 2: Slab JE 100014 when discovered as closing for the entrance of a tomb layer. This intervention should be reduced in order to become more discreet and suitable for the rest the surface.

262 263 E. Mertah et al. Investigation and examination of a Degraded Egyptian Painted Limestone Relief from Tell Hebua (Sinai)

the tomb). On that side indeed, crystalli- infrared spectroscopy [FTIR] Analysis). zed salts are not limited to the painted They revealed that the painting was pre- layer, but also invade the whole surface, viously restored; partly at least on the ar- especially in the upper and lower areas, chaeological site itself, according to the where the inscription is carved. information provided by the restoration Tests and analyses were carried out on team. As the analysis demonstrated (see the painting (visual examination, exa- below, Fig. 7), it appeared that Paraloid mination using magnifying glasses and B72 had been used in the consolidation optical light microscope, portable X-ray process, and that microspheres known fluorescence [PXRF], Fourier transform also as microballoons filler (i. e. a glass

Figure 8: Digital light microscope images taken on the Thutmosis II’s side. A shows the crystallized salts inside the red colour; B shows the crystallized salts above and below the pigments, which led to the fall of some parts of the colour’s layer; on that photograph, we can see the flacking between the painted layer and the original surface. The crystallized salts inside the blue colour are visible on image C, while on D are shown the crystallized salts inside the eye of the figure of the king. E and F show the cracks and the micro-cracks on the surface of the relief. [A, C, D, E, F] 50x; [B] 250x.

Figure 5: Spots where Digital light microscope images Figure 7: Side of the offering scene of Thutmosis III, on fibre based on Silica [Si]) was also used to techniques carried out to identify the were taken on the side of the block which shows the the slab JE 100014. The photo shows the areas where offering scene of Ramesses II. Digital light microscope images were taken. complete some parts, probably after the salts emerging from the block and dama- relief entered the Egyptian Museum. ging its surface, and the nature and the The block needs a re-restoration, a sta- state of preservation of the pigments on bilization of its fragile areas and a study the decorated surface. of the possible ways to remove crystallized salts from the weak parts, and to preserve the painted layer. 3.1. Optical light microscope (O.L.M.) In the framework of our study, different techniques, described in the follo- A portative optical light microscope wing parts, have been used to identify was used in order to reveal details such the causes of rapid deterioration, salt as the porosity, grain morphology and weathering, which may have partially micro-cracks. A series of twelve spots occurred in the external environment (six on each side) were analysed (see fig. 5 where the relief was lying before its dis- and 7). The used device (Dino-Lite Digital covery, but which has continued, since Microscope) allows taking high quality the relief has been installed in a poorly photos (from 50 to 250x), even without controlled interior environment, in the direct contact with the surface of the ob- galleries of the Egyptian Museum. ject – the fragile state of preservation of the relief JE 100014 required a distance of a few millimetres to be kept. This first 3. Analytical Procedure examination allowed determination of Figure 6: Digital light microscope images on Ramesses II’s side of the slab. Images A and B show the crystallized salts the different lithotypes of the block: it on the surface of the relief. C and D show the crystallized salts above and below the pigments, which already led to This section discusses the results of provided information about the dama- the fall of some parts of the colour’s layer. On image E the crystallized salts are captured inside the red colour, while F consists in a photograph of the crystallized salts above the blue colour. [A, B, E] 250x; [C, D, F] 50x. the different analyses and investigation ged layers, such as the sequence of layers,

264 265 E. Mertah et al. Investigation and examination of a Degraded Egyptian Painted Limestone Relief from Tell Hebua (Sinai)

the particle size, as well as the colour and SN 177), which we used to determine the No. Elements Results texture of these layers. It also showed the specific elements that are present within 1 Ca, Cl cracks and the micro-cracks on the surfa- the pigments and in the stone block itself The stone itself is limestone ce of the relief. (including the salts). (1) Side with 2 Ca, Cl (calcium carbonate). Furthermore, the microscope exami- We selected a series of twenty-two Ramesses II’s 3 Ca, Cl As revealed by the analysis, the side (1) suffers nation revealed that the salts are not points (fifteen on the Ramesside size and offering scene 4 Ca, Cl from a widely spread crystallization of chlorine only above the layer of colour, but also eight on the Thutmoside one) in order to 5 Ca, Cl salts on the surface. under it and inside it and thus putting it cover all the different areas of the sur- (2) Side with 16 Ca, S The side (2), however, contains sulphur, and may in particular danger (see fig. 6 and 8). face of the block, including the different Thutmosis II’s 17 Ca, S therefore contain calcium sulphate salts. painted layers, as well as the various offering scene parts covered with salt. 18 Ca, S 3.2. Portable X-Ray Fluorescence [PXRF] The analysis revealed a widely spread crystallization of chlorine salts on Ra- Table 1: Results of the [PXRF] analysis on the stone itself X-Ray Fluorescence is widely used messes II’s side of the slab, and what ap- in the field of archaeometry due to its pears to be calcium sulphate salts on the non-destructive properties, its high sen- other side, due to the presence of sulphur No. Elements Results sitivity and its applicability to a wide ran- (see table 1). Both types of salt are also 9 Ca, Cl, Si, Fe, S (1) Side with The PXRF revealed the presence of Iron ge of situations. The Egyptian Museum’s present within the red pigment (made of Ramesses II’s 10 Ca, Cl, Si, Fe, S lab owns a portable device (Elio Device hematite, see Table 2), as well as in the (Fe) inside all the red pigment, which means offering scene 11 Ca, Cl, Si, Fe, S blue and green pigments (both acquired that it consisted of red hematite (iron oxide 19 Ca, Cl, Fe, S hematite Fe O ). from copper firing process, see Tables 3 2 3 (2) Side with and 4). 20 Ca, Cl, Fe, S The presence of chlorine and calcium sulphate Thutmosis II’s On the side (1), depicting Ramesses II 21 Ca, Cl, Si, Fe, S is due to the crystallization of salts inside the offering scene pigments (as well as under and above them). (see figure 9-10), two analyses were made 23 Ca, Cl, Fe (14 and 15) on what seemed to be white colour. This revealed itself to be, through Table 2: Results of the [PXRF] analysis on the red colour the PXRF, not to be a pigment but the microballoons used during the previous restoration (see above). No. Elements Results

6 Ca, Cl, Si, Cu, S The presence of chlorine and calcium sulphate 3.3. Fourier Transform Infrared (1) Side with inside the green pigment is due, like in the case Spectroscopy [FTIR] analysis Ramesses II’s 7 Ca, Cl, Si, Cu, S of the red one, to the crystallization of salts (see offering scene above). However, the colour was here acquired A sample of the red colour which had 8 Ca, Cl, Si, Cu, S from copper. already fallen from the surface of the

block was sent to the laboratory of the Table 3: Results of the [PXRF] analysis on the Green colour (only on side 1) Ministry of Antiquities for a FTIR analysis, which allowed identifying an adhesive applied on the pigment layer, in order No. Elements Results to fix it during the previous restoration: Paraloid B72. (1) Side with 12 Ca, Cl, Si, S, Cu The elements are similar to those observed Ramesses II’s previously. Indeed, green and blue colours are offering scene 13 Ca, Cl, Si, S, Cu both acquired from a firing process of copper. 4. Preliminary conclusion and salt- (2) Side with The presence of chlorine and calcium sulphate reduction treatment Thutmosis II’s 22 Ca, Cl, Si, S, Cu provides again the nature of the crystallized salt inside the pigment. Figures 9 and 10: Slab JE 100014, with numbers showing offering scene the areas where PXRF analysis was carried out – face 1 [Fig. 9] (offering scene of Ramesses II) and face 2 [Fig. 10] This article discusses the investigation (offering scene of Thutmosis II). carried out on a limestone relief found in Table 4: Results of the [PXRF] analysis on the Blue colour

266 267 E. Mertah et al. Investigation and examination of a Degraded Egyptian Painted Limestone Relief from Tell Hebua (Sinai)

a wet environment ten years ago, current- An examination with a Portable X-Ray degree of solubility, in order to choose maximum of possibilities, the relative ly in display in the Egyptian Museum, Fluorescence [PXRF] tool established the the most appropriate method for remo- humidity and temperature in the gallery Cairo (inv. JE 100014). This block has been nature of the different elements from ving them to prevent further desinte- would help prevent further salt damage; subject to different deterioration pro- which the stone and pigments were com- gration of the stone due to their active this is of course quite difficult in the cur- cesses, and needs urgent consolidation posed. The whole surface of the block and presence. Some salts revealed themselves rent conditions of display, but will hope- and restoration. Atmosphere variations its painted decoration were found to suf- to be of calcium sulphate, which is very fully be improved in the near future. during the years following the block’s fer from a crystallization of chlorine and soluble and will be easily removable, but removal from its original environment calcium sulphate salts. some other salts might be of unsoluble caused salt crystallization and numerous Finally, a Fourier Transform Infrared calcite, which would therefore require a internal and external stresses. Spectroscopy [FTIR] analysis applied on a mechanical cleaning. The chlorides iden- Our team carried out various sample of red pigment revealed the pre- tified on the side of Ramesses II (see ta- techniques of analysis, in order to esta- sence of adhesive Paraloid B72, applied ble 1), they may show a variable level of References blish the different causes of damage and on the pigment layer in order to fix it du- solubility (either high solubility, if they to identify the most suitable methods of ring a previous restoration. consist of calcium or copper chlorides, 1 Concerning the discovery of the relief, restoration. First of all, before carrying These different methods demonstrate or low solubility in the case of sodium see Abd El-Maksoud M. and Valbelle D., out any conservation work, the block was that the weathered surfaces of the block chloride), and therefore require further “Tell Héboua II: Rapport préliminaire carefully examined on all its sides. Each have undergone continual warming and investigation.5 sur le décor et l’épigraphie des éléments of the alterations and degradations was cooling cycles, due to the semi-dry atmo- Before any new restoration, we will architectoniques découverts au cours mapped, documented and registered in sphere of Tell Hebua. The surfaces now first proceed to some de-restoration of des campagnes 2008-2009 dans la zone the conservation records. exposed are enduring physical disinte- the piece. The microballoon filler, ap- centrale du khétem de Tjarou”, Revue A first investigation with the Optical gration, as assessed from macroscopic plied at the arrival of the block in the d’Égyptologie 62 (2011), p. 1-39. light microscope (O.L.M.) allowed identi- visual analysis. The block suffers from museum, should be reduced to a smaller fication of the nature of the stone and the several deterioration of its structural area, in order to produce a more suitab- About the site of Tell Hebua, its defen- different lithotypes present in the block, coherence: cracks, microcracks, cavities, le and sympathetic aesthetic appearance system and its region, see: Abdel and drawing out a map of all cracks and loosing of material, resulting of needles ce. In other situations, a de-restoration, Maksoud M. and Valbelle D., “Tell Hé- micro-cracks. It revealed the presence of of salts between grains, and accumulati- although desirable, will probably be dif- boua-Tjarou: l’apport de l’épigraphie”, salts not only above, but also inside and on of different kinds of dirt. ficult: indeed, the presence of the mate- Revue d’Égyptologie 56 (2005), p. 1-44; beneath the layers of colour, on the deco- We are currently in the process of iden- rials used in early restoration treatment Id., Tell Heboua (1981-1991). Enquête rated surface of the reliefs. tifying the nature of the salts and their will have an impact on the ways of un- archéologique sur la Deuxième Pério- dertaking future consolidation work and de Intermédiaire et le Nouvel Empire must be taken into account. The Paraloid à l’extrémité orientale du Delta, Paris, B72 applied on all the painted surfaces 1998. Abdel Maksoud M., “Projet de forms a film, covering large areas, which sauvetage des sites antiques du Nord prevents the salts reaching the surface Sinaï”, Discussions in Egyptology 24 of the block during their migration. This (1992), p. 7-12; Oren E., “Migdol: A New creates a detachment between the stone Fortress on the Edge of the Eastern Nile and the painted layer, which is likely to Delta”, BASOR 156 (1984), p. 7-4; Red- continue to increase. ford, D., “Report on the 1993 and 1997 Due to the friable nature of the stone, Seasons at Tel Qedwa”, Journal of the and the fragility of the pigments6 each American Research Center in Egypt 35 consolidation and conservation inter- (1998), p. 55. vention will be performed according to the needs of each of the individual pro- 2 Concerning the name of the site in blematic zones, after thoroughly testing the antiquity and the nature of fron- all possible procedures. tier fortresses, see Somaglino C., “Les In order to prevent further degrada- ‘portes’ de l’Égypte de l’Ancien Empi- tion of the relief, some precautions will re à l’Époque Saïte”, Égypte, Afrique & be necessary to keep it on display in the Orient 59 (2010), p. 10; Id., Du magasin Figure 11: Spectrum FTIR of the sample of red pigment Egyptian Museum. Controlling, to the au poste-frontière dans l’Égypte ancien-

268 269 SWBSS 2017 | 20-22 September E. Mertah et al. 4th International Conference on Salt Weathering of Buildings and Stone Sculptures

ne : étude lexicographique du vocable (2015), available online 06 November Technique for transportation of stone sculptures khetem, (unpublished Phd dissertation, 2015 (http://www.e-conservation.org/ Paris IV University, under the direction issue-3/52-Egyptian-Limestone-Stele); damaged by salt crystallization of D. Valbelle, 2010, forthcoming in the Doehne E., “Salt Weathering: A Selective Ifao series. Review”, Geological Society Special Pub- Vinka Marinković lication. Natural Stone, Weathering Phe- Croatian Conservatian Institute, Croatia 3 On the side of Thutmosis II, the figures nomena, Conservation Strategies and * [email protected] and the hieroglyphs of are all in rai- Case Studies 205 (2002), p. 51-64. sed relief, except for a few signs in the middle of the inscription, which are in Abstract pel of St. John (The Chapel of the Blessed sunken relief. These signs, which give Giovanni Orsini), created by the masters the name of “Montu, lord of Thebes”, The paper presents a technique for Nikola Firentinac and Ivan Duknović in were most probably originally in raised handling and transportation of stone the period from 1461 to 1497. The cha- relief too, but were erased during the sculptures damaged by salt crystalliz- pel was restored and cleaned in between iconoclastic phase of the Amarnian pe- ation. A limestone sculpture from the 2000 and 2002 as part of the American riod (1353-1334 BC), and then restored, Cathedral of St. Lawrence in Trogir was foundation project - Venetian Heritage. necessarily in sunken relief, probably in studied and pre-consolidated with cy- The Croatian Conservation Institute star- the time of Ramesses II, when the king clododecane. First, test were performed ted the monitoring and preliminary in- commissioned a new ornamentation of in the lab. On 5 glass slides were applied vestigations of the chapel in 2016 becau- the temple. Cf. Abdel Maksoud M. and powder of the Seget stone and then con- se some adverse changes were obsereved Valbelle D. 2011 (cf. supra). solidated with 50% solution of CCD in on the surface of sculptures and stone Shellsol T, 50% solution of CCD in white interior. In a few locations, it was pos- 4 Our thanks go to Gharib Sonbol, who spirit, 50% solution of CCD in toluene, sible to notice high levels of powdering, directed the team of restoration on the with holt melted CCD and hot melted sanding and sugaring of the stone, which site during the excavation, and who CCD with cotton gauze. After the tests was caused by salt crystallization. kindly provided us information about were performed, a technique of pre-con- The Annunciation group (sculptures of the context of discovery and the state of solidation with melted cyclododecane the Angel and the Virgin) positioned on preservation of the block in 2009, and and facing tissue was used in situ on the the triumphal arch of the chapel was in allowed us to use the photos taken du- sculpture. Practical uses of the technique a terrible state. The bases and lower back ring the mission, in this article. Thanks were considered in the paper. zones of both sculptures were extremely are also due to Bianca Madden, for her damaged, and were in unstable condi- coments and kind revision of English Keywords: cyclododecane, stone, pre-con- tions with a high loss of the substrate. writing. solidation, salts. First analyses of samples from the background zones of both sculptures showed 5 Nord A.G., “Effloresecence salts on we- presence of the following salts: chlorides athered building stone in Sweden”, in 1. Introduction and nitrates, in concentrations ranging Geologiska foreningens I stockholm from 0,13-0,17%. Both sculptures were forhandlingar 114/4 (1992), p. 223-229: The Cathedral of St. Lawrence in Trogir made of local limestone (called Seget). “The solubility of calcium sulphate salt (Croatia) is a Roman Catholic triple-na- is (2,4g/l at 20°C and PH7) is more than ve basilica constructed in the Romanes- 200 times greater than that of calcite, que-Gothic style. Trogir Cathedral was 2. Methods and materials but the solubility of sulfates is much constructed over a period of four cen- less than that of chlorides.” turies (1200-1598), and it illustrates all The presence of water soluble salts in the the successions of styles in that time in samples was determined by chemical 6 For similar cases and valuable sour- Dalmatia. It is one of the most important analyses and UV/VIS spectroscopy with ces of inspiration for salt treatment, monuments in Croatia, and was decla- a Perkin Elmer Lambda 25 instrument. see: Mirabootalebi R., “Conservation red a World Heritage Site by UNESCO in Compositional salt analysis was per- of A Highly Degraded Egyptian Limes- 1987. The integral part of the cathedral formed using X-ray diffraction method tone Stele”, e-conservation Journal 3 is the richly decorated Renaissance Cha- with a Philips Vertical X-ray Goniometer

270 271 V. Marinković Technique for Transportation of Stone Sculptures Damaged by Salt Crystallization

(X-Pert type). Water content measure- CDD can be applied as a melt or as a so- ment of the stone showed that the water lution [Burton, 2009]. In order to choose coefficient values were within the refe- an appropriate technique and method of rence values for the same type of stone application, some test were performed in collected from the Seget quarry. the laboratory prior using it on the sculp- Considering the extremely damaged sta- tures of Trogir. On glass slides we applied te of the sculptures and their original powder of the Seget stone, and then we position (5m above the floor level), we applied 50% solution of CDD in Shellsol decided to dismantle them and carry out T (glass slide 1, Fig. 1), a 50 % solution of the conservation work of desalination in CDD in white spirit (glass slide 2, Fig. 1), the workshop. a 50 % solution of CDD in toluene (glass Before the dismantling and transportati- slide 3, Fig. 1), hot melted CDD (glass slide on of the sculptures it was necessary to 4, Fig. 1) and hot melted CDD with cotton pre-consolidate the most unstable pow- gauze (glass slide 5, Fig. 1). dery areas of the sculptures. Cyclodo- After a visual observation, it was possib- Figure 2: Glass slides with Seget powder and different types of application of CDD after fingernail scratching test (from decane was chosen as consolidant. le to conclude that the hot melted CDD left to right: 50% solution of CDD in Shellsol T; 50 % solution of CDD in white spirit; 50 % solution of CDD in toluene; hot Cyclododecane (abbreviated CDD) is in both cases formed a dense film, and melted CDD; hot melted CDD with cotton gauze) a waxy 12-carbon cyclic hydrocarbon the stone powder was consolidated and (C12H24) that sublimes from a solid to packed (the particles of powder are in- gas at room temperature.1 This has made terconnected). When CDD was dissolved test was unstable, and the consolidation lost. Other negative changes were not re- it an appropriate material in conservati- in solvent, the layer was less dense and effect was less effective. corded on the surface of the stone. on as a temporary consolidant, adhesive the powder was less consolidated and pa- After the preliminary laboratory tests, or barrier layer during the last few deca- cked. the technique of pre-consolidation with des.2 When the object is highly damaged After a fingernail scratching test (after melted CDD and facing tissue was used in 4. Disscusion and contaminated with soluble salts, and 24 hours) it was possible to conclude situ. CDD was reactivated with heat and when it needs to be transported, CDD is that the powder treated with hot melted applied on the surface of the sculpture of Only practical uses of the technique were an appropriate solution. One of the main CDD (in both cases, glass slide 4-5 with the Angel using a brush. Simultaneously, considered in the paper. The investiga- reasons is that harmful soluble salts are hot melted CCD and hot melted CCD with on the surface we applied two layers of tions were limited because of technical not locked in the material. Some success- gauze, Fig. 2) was stable and still packed. cotton gauze. After the pre-consolidati- and financial issues. Future analytical ful cases have already been reported in Powder treated with CDD dissolved in the on, the sculpture was wrapped in foil and research should provide answers to the literature.3 solution (glass slide 1-3, Fig. 2) after the lowered to the ground using a crane. The following questions: facing tissue did not detach in 10 days in the lab environment (approximately 20˚C - depth of the penetration of CCD on the with moderate airflow). CDD was reacti- Seget stone, vated with a heating gun and removed. - sublimation rates, - negative aspects in the presence of harmful soluble salts. 3. Results

Laboratory and field tests of pre-consoli- 5. Conclusion dation, handling and transportation of stone sculptures damaged by salt crys- Cyclododecan (CDD) has been used in- tallization showed good results when the creasingly as a temporary consolidant technique with hot melted cyclododeca- for a variety of weak or fragile mate- ne and facing tissue was used. The sculp- rials. This contribution focuses on the ture of the Angel from Trogir Cathedral practical use of cyclododecane as a tem- was successfully transported to the work- porary consolidant in cases of handling Figure 1: Glass slides with Seget powder and different types of application of CDD (from left to right: 50% solution of shop using the aforementioned method. and transportation of stone sculptures CDD in Shellsol T; 50 % solution of CDD in white spirit; 50 % solution of CDD in toluene; hot melted CDD; hot melted CDD with cotton gauze) A minimum of unstable material was damaged by salt crystallization. Melted

272 273 SWBSS 2017 | 20-22 September V. Marinković 4th International Conference on Salt Weathering of Buildings and Stone Sculptures

CDD, melted CCD with facing tissue and Investigation and conservation of salt damaged epitaphs saturated solutions of consolidant in white spirit, shelsoll T and toluene were in the church of Werben (Saxony-Anhalt, Germany) applied on glass slides with stone powder. By visual observation and using a Steffen Laue*, Dörte Poerschke and Benjamin Hübner fingernail scratching test, it was possible University of Applied Sciences Potsdam, Department of Conservation and Restoration, Germany to conclude that the best effect of conso- *[email protected] lidation was obtained during the application of hot melt CDD. After the lab tests, the technique of pre-consolidation with Abstract 1. Introduction melted cyclododecane and facing tissue was used in situ. The extremely damaged The case study of the St. Johannis Werben is located in the northern sculpture of the Angel from Trogir Cathe- church in Werben (Saxony-Anhalt) is one part of Saxony-Anhalt at the river Elbe dral was consolidated with melted CDD of those instances where conservators in the administrative district of Stendal and cotton gauze. After the consolidati- and restorers working in the field of buil- between Magdeburg and Wittenberge. on, it was transported to the workshop ding heritage have to find sustainable Because of the proximity to the river, the of the Croatian Conservation Institute. solutions for a cultural heritage site af- town was and is continuously threatened During the intervention, a minimum of fected by high humidity levels and salt by floods, especially in the past when no unstable material was lost. content for centuries. In the church of adequate flood prevention was in place. Werben, the ground floor of the steeple St. Johannis church was built with houses a remarkable collection of pre- bricks in the 15th century but includes ol- Acknowledgements dominantly polychrome epitaphs from der parts from the Romanesque period 1. The project is supported by the Ministry the 16th – 18th centuries. These epitaphs The church consists of a three axial nave of Culture of the Republic of Croatia. reflect in an unique way the story of im- with the choir in the East and a huge portant citizens of Werben and provide steeple in the West (Fig. 1). In this pa- an extensive overview of the design and per, we focus on the ground floor of the techniques used in the manufacturing steeple in the western part of the church of such monuments through various epochs. During an inventory of the epi- References taphs, several damages were attributed to salts and the high moisture content in 1 Watters, C. “Cyclododecane: A Closer the ground floor of the steeple. Salt ana- Look at the Practical Issues”, Anatolian lyses, climate measurements and calcu- Archaeological Studies (XVI), (2007), lations with the computer model ECOS/ 195–204. RUNSALT were used to understand the damage processes of the epitaphs. The 2 Burton, D. J: An investigation into the results provide the basis for applying efficacy of cyclododecane as a volatile appropriate conservation methods and exclusion layer providing isolation of materials pertaining to the room and the an un-saturated surface during consoli- polychromic epitaphs. As a case study dation, Conservation Studies, 2009. one epitaph was removed and preserved in the workshop of the conservation and 3 Cleere, D. C.: „Cyclododecane re-investi- restoration department in Potsdam. gated - An experimental study on using cyclododecane to secure a stable cera- Keywords: Werben, polychromic epitaph, mic surfaces prior to transportation“, salt damage, climate, ECOS/RUNSALT, Conservation News (94), (2005), 26-28. conservation concept

Figure 1: St. Johannis church in Werben, southwestern part with steeple

274 275 S. Laue et al. Investigation and conservation of salt damaged epitaphs in the church of Werben (Saxony-Anhalt, Germany)

Today, each epitaph stands on a ca. 40 cm Permanent measurement of room tem- high pedestal made of bricks that are perature, surface temperature and rela- plastered (Fig. 3). The epitaphs are fixated tive humidity was implemented using by iron anchors to the back. Additionally, RH/T sensors and a data logger. bricks with mortars were applied irregu- To investigate under which climatic larly to support the backward connecti- conditions new salt crystallization ta- on of the epitaphs to the walls. Probably kes place and what kind of damage they later grey cement mortars were used to produce, a monitoring process was im- reinforce the connection of the epitaphs plemented: Climate measurements were to the back. executed combined with the periodic ob- In an inventory of the epitaphs made servation of the crystallized salts at de- by students and in the M.A. thesis by fined surface areas.3 Dörte Poerschke2, materials and dama- The observations were compared using ges of the epitaphs were determined and the computer model ECOS/RUNSALT mapped showing advanced degradation in order to assess under which climate of the stones and colored surfaces obvi- condition salts are crystallizing in the ously caused by moisture, salt crystalli- ground floor of the steeple.4, 5 zation and microbiological impact. Thus, The reduction of the salt ion content the aim of the investigation was to un- through cellulose poultices from the epi- Figure 2: Ground map of St. Johannis 1 containing the ground floor of the steeple (TE) derstand the damage processes in the taph of Leonhardt Kämpfe were verified ground floor and to find solutions for the by taking 5 x 5 cm sized poultice samples conservation and restoration of the epi- from the surface, extracting the ions by with a floor area of 9,20 m x 4,20 m and taphs as well as the masonry. deionized water and measuring the elec- a height of 4,82 m in the zenith of the trical conductivity which correlates to barrel-vault. Today, the ground floor can the total amount of extracted salt ions. only be reached through a wooden door 2. Analytical approach from the nave (Fig. 2). The masonry of the ground floor consists of 1,50 m – 1,80 m To find an appropriate conservation 3. Damage thick walls made of bricks and mortar concept for the ground floor of the stee- that today, contain only one window to ple as well as for the epitaphs the follo- In the ground floor of the steeple a the south leading to a moist and dark wing investigations were completed. high rate of damage in bricks and plaster room. The walls were plastered with a During the inventory the epitaphs and due to moisture and salts was discerned. lime mortar that is now partly damaged the damage of the walls and epitaphs At the masonry the plaster shows de- by moisture and salts. were documented. Salt efflorescences teriorations like lacunas, cavities, bulges, Since 1868, the ground floor has been and crusts were sampled from different and partly chalking6 whereas bricks part- used as a depository for 18 stone epitaphs surfaces and analyzed by polarizing ly decay by flaking and granular disinte- from the 16th – 18th centuries. These epi- microscopy, microchemistry and XRD gration. taphs reflect in a unique way the story of (Siemens D5005, CuKα-radiation) to de- In sum, the strongest damages are vi- important citizens of Werben and provide tekt the crystallizing salts. sible up to a height of approximately an extensive overview of the design and Drill powder samples were taken to 1.5 m. Between heights of about 1.5 and techniques used in manufacturing the- investigate the distribution of salt ions 3 m the damage is considerable but scat- se monuments through various epochs. concentration in two vertical profiles tered and above 3 m strong damages of The epitaphs are partly polychrome and and into the depth at the southern and the plaster mainly show at the western gilded and show the wealthy members of western wall. 1 g of sample material was walls including salt efflorescences. citizenry and clergy of the town (Fig. 3). It mixed with 100 ml distilled water for On the epitaphs the loss of paint by fla- is not quite clear where the epitaphs were about 24 hours and then filtered before king and peeling7 is visible as well as the + + 2+ 2+ - - located before 1868 but it has to be assu- the analyses of Na , K , Ca , Mg , Cl , NO3 disintegration of the sandstone itself. med that they were placed somewhere in Figure 3: polychrome epitaphs at the southern wall and and SO 2- were completed using ion chro- Salt crystallization on the epitaphs can the only window in the ground floor, in the middle the 4 or around the church and the cemetery. epitaph of Leonhardt Kämpfe matography (Dionex, IC90). be observed in areas where the epitaphs

276 277 S. Laue et al. Investigation and conservation of salt damaged epitaphs in the church of Werben (Saxony-Anhalt, Germany)

are connected with the walls by bricks or of calcite to gypsum and the low solubi- sample height (cm) depth (cm) Ion concentration (wt%) mortar indicating the transport of ions lity of gypsum. Cl- NO - SO 2- Na+ K+ Mg2+ Ca2+ through the pore system of the bricks 3 4 and mortar into the epitaphs. - Nitrate, sodium and potassium are dis- 1 90 0- 1 0,09 0,48 0,57 0,09 0,05 0,07 0,37 tributed in all heights and depths espe- 2 90 1- 2 0,17 0,87 0,16 0,14 0,13 0,11 0,24 cially above 2.0 m of height. The con- 3 90 2- 4 0,16 0,84 0,10 0,16 0,18 0,09 0,26 Salts centration of sodium compared with 4 140 0- 1 0,09 0,58 0,33 0,16 0,13 0,08 0,19 potassium is mainly twice as much, 5 140 1- 2 0,12 0,66 0,07 0,21 0,18 0,06 0,10 On the wall surfaces, crystallized salts which was a little surprise at first becau- 6 140 2- 4 0,12 0,52 0,04 0,22 0,22 0,04 0,12 consist of niter (KNO3), epsomite (Mg- se no crystallized sodium salts could be . . 7 140 4- 7 0,08 0,31 0,04 0,16 0,18 0,03 0,10 SO4 7H2O) hexahydrite (MgSO4 6H2O) and detected on the ground floor surfaces. . 8 200 0- 1 0,08 0,64 0,34 0,27 0,21 0,03 0,15 gypsum (CaSO4 2H2O). The distribution of salt ions was analy- - Slightly increased amounts of magnesi- 9 200 1- 2 0,19 1,11 0,27 0,49 0,32 0,06 0,11 sed in two vertical profiles at the western um can be found in the lower parts of 10 200 2- 4 0,18 0,94 0,21 0,41 0,28 0,05 0,10 and at the southern wall. As an example, the wall which correlates with the mag- 11 200 4- 7 0,17 0,76 0,13 0,38 0,28 0,05 0,10 Fig. 4 demonstrates the distribution of nesium sulfates found in this region of 12 300 0- 1 0,11 0,65 1,06 0,28 0,16 0,04 0,51 ions at the western wall in the heights the walls. 13 370 0- 1 0,16 0,62 0,37 0,24 0,09 0,07 0,23 of 0.9, 1.4, 2.0, 3.0, 3.7 m. Table 1 shows the ion content (in weight percent) at Although hardly any tendency of en- Table 1: Salt ion distribution in the western wall expressed in weight percent (wt%) each height and depth at the same wall. richment of salt ions can be observed Although the distribution of salt ions in in different heights and depths, there is both walls is rather complex, some gene- no clear explanation why the salt ions ral tendencies can be noted: spread in the wall as pointed out. The sible explanation for the accumulation of ween 75% and 95% resulting in a perma- ions might have been transported partly water and ions in higher regions of the nently moist environment. - Sulphate and calcium are always accu- by rising damp but not into the height wall, but there is no evidence for that. mulated in the first centimeter of each of 3.7 m. The infiltration of water through Nevertheless, it can be concluded that height, probably due to the conversion defects in the masonry could be a pos- the ground floor of the steeple is affected Simulation of the crystallization of by an accumulation and complex distri- salts using ECOS/RUNSALT bution of salt ions in the masonry consis- + + 2+ 2+ - ting mainly of Na , K , Ca , Mg , NO3 and ECOS/RUNSALT is a computer model 2- SO4 . to predict the crystallization behavior of salts based on the thermodynamic model ECOS (Environmental Control of Climate Salts).4, 5 For this purpose, a set of representative salt ion contents is needed de- Permanent climate measurements of rived from the first cm of the surface of room temperatures, surface temperatu- a monument. Then, the data have to be res and relative humidity in the ground simplified (even ionic balance, removal floor started in 2013. Fig. 5 shows the of gypsum) before the model is used to daily average temperature and relative calculate which soluble salts will crystal- humidity from December 2013 until June lize under predetermined environmental 2014 demonstrating that the relative hu- conditions like temperature and relative midity in the ground floor consistently humidity. Thus, the program makes it lies between 80 and 90% both in summer possible to predict under which environ- and in winter. After evaluation of the cli- mental conditions salts will probably da- mate measurement results of two years mage the surfaces of a monument. it can be concluded that the temperature As an example, in Fig. 6 the data of in the ground floor varies between 4°C sample 12 are calculated with ECOS/ Figure 4: Salt ion distribution in the western wall expressed in Equivalentconcentration (mEq/kg) and 18°C and the relative humidity bet- RUNSALT assuming a room temperature

278 279 S. Laue et al. Investigation and conservation of salt damaged epitaphs in the church of Werben (Saxony-Anhalt, Germany)

with regard to the crystallization process air exchange with the outdoor climate of salts presuming that crystallization resulting in this quite stable humid en- mostly happens below a relative humidi- vironment. The environmental condition ty of 70% with much damage to the walls keeps most of the analyzed salt ions in and epitaphs. Thus, the calculations give solution, but enables biological growth. an important hint for conservation mea- With higher air exchange rates between sures: To avoid further damages in the the climate in the ground floor and the ground floor through salt crystallization, outdoor climate the relative humidity le- the relative humidity should not fall be- vel would probably be lower than 70% at low 70% of relative humidity. certain times resulting in more damage due to salt crystallization. On the other hand, the high relative Conclusion of the weathering situation humidity levels produce damages to the painted epitaphs resulting especially in In the ground floor of the steeple, the the degradation of the binder 8. The paint masonry and epitaphs suffer from high layers of the epitaphs show in parts ext- Figure 5: Daily average values of room temperatures (scale right) and relative humidity (scale left) in the ground floor moisture content with a relative humi- reme damages like flaking. Keeping the dity of predominantly above 80%. Mois- epitaphs permanently in the ground ture sources are probably rising damp floor, the polychromy of the epitaphs and infiltration from outside over the would be slowly destroyed forever. years. Only one door to the nave, which Based on the climate situation in the is mostly closed, and one little window ground floor today, it can be concluded to the south do not allow for enough that the high relative humidity levels are

Figure 6: Image of ECOS/RUNSALT after calculation of sample 12 at 10°C showing the crystallized salt phases between 50% and 90% of relative humidity

of 10°C. It can be seen that the crystalliz- and magnesium sulfates and nitrates. In ation of salts starts below 86% of relative this area of the wall, efflorescences of ni-

humidity with the salt mirabilite (Na- ter (KNO3), but no mirabilite were deter-

SO4.10H2O). Furthermore the chart de- mined. monstrates that the main crystallization Further ECOS/RUNSALT calculations of salts will be below a relative humidity with data from other surface areas in of 70% mainly consisting of the sodium the ground floor yielded similar results Figure 7: Epitaph of Leonhardt Kämpfe in Werben before (left) and after conservation (right)

280 281 S. Laue et al. Investigation and conservation of salt damaged epitaphs in the church of Werben (Saxony-Anhalt, Germany)

beneficial regarding salt damage but pro- the workshop. The front was still cover- nis in Werben. The epitaphs as well as the References mote the decomposition of the paint lay- ed with cyclododecane; thus, no further masonry suffer from extreme humidity ers on the epitaphs as well as biological damages or crystallization of salts to the and high salt contents. The ventilation 1 Dehio, G., Handbuch der deutschen growth everywhere in the room. polychromy were expected. of the room would decrease the relative Kunstdenkmäler, Der Bezirk Magde- Additionally, a salt ion reduction of the humidity which would cause more salt burg, Berlin, (1975). epitaph was realized by exchanging the damages as demonstrated by the calcula- Conservation of the epitaph of poultice weekly until the salt ion load in tions with ECOS/RUNSALT. Therefore, we 2 Poerschke, D., Schadensanalyse und Leonhardt Kämpfe the poultice was significantly reduced recommend to remove the epitaphs from Auswertung des Schadenspotentials im verified by electrical conductivity mea- the ground floor, and preserve them af- Turmerdgeschoss der St. Johanniskirche After the awareness of the weathering surements. Selected poultice samples ter salt reduction as shown in this paper. in Werben.- unplublished Masterthesis situation in the ground floor, the decisi- were analysed by ion chromatography to Of course, after a salt reduction and the at the University of Applied Sciences on was made that one epitaph should be determine which soluble ions could have completion of preservation measures the Potsdam, (2013). removed and transported to the work- been extracted. The results proved that best solution for the epitaphs would be shop in Potsdam where a conservation of nitrates together with chlorides, magne- to find for them another location with 3 Laue, S., Salt Weathering of Porous Struc- the epitaph was to be completed as a case sium and sodium are the main extracted moderate relative humidity around 50% tures Related to Climate Changes, Resto- study. The option of working on the epi- salt ions.12 or 60% and with low variation in relative ration of Buildings and Monuments, 11, taph under stable and relatively dry en- After the salt reduction several con- humidity. But if they have to remain in No. 6, (2005), 381-390. vironmental conditions in the workshop servation measures were implemented, the ground floor of the steeple, the pre- was a decisive factor for removing the ob- among them dry cleaning with a paint- served epitaphs should stand on plat- 4 Bionda, D., Modelling indoor climate ject from its original location where the brush, structural consolidation of the forms as suggested so that no new salt and salt behaviour in historical buil- preservation would have been more com- stone with silicic acid ester and fixing ion can penetrate into the epitaphs. dings: A case study, PhD thesis, Diss. Nr. plicated due to the humid environment. of the paint layers by using the aqueous Additionally, sustainable measures for 16567 ETH Zürich, (2006). We chose the epitaph of Leonhardt consolidant JunFunori®. Additionally, the whole church should be implemen- Kämpfe (Fig. 7) dated to the 17th century cracks were filled and repair mortars ted, among them optimizing the water 5 Price, C (Ed.), An expert chemical mo- and located at the southern wall of the were applied with silica sol as a binder. drain and removing salt ion sources like del for determining the environmen- ground floor of the steeple. The original Fig. 7 shows the epitaph of Leonhardt cement repair mortars. tal conditions needed to prevent salt polychromy of the epitaph must have Kämpfe before and after conservation. However, in Werben, as in many other damage in porous materials. European been outstanding9, but the polychromy The investigation and conservation places, there is a lack of funding to rea- Commission Research Report No 11, now shows much damage and some measures were completed during a peri- lize the best possible solutions for the (Protection and Conservation of Euro- parts are completely lost.2 od of about one year and a half before cultural heritage. Thus, step by step in- pean Cultural Heritage). London: Arch- The epitaph consists of four sandstone the epitaph was returned to the ground terventions in the surroundings and the etype Publications, (2000). fragments which are piled up on top of floor of the steeple. There, the epitaph monument itself can often improve the each other. The stone fragments are also was mounted in the same area, but in condition of monuments resulting in less 6 Weyer, A., Picazo, P.R., Pop, D., Cassar, J.A., in a degraded condition showing nume- future no rising damp or capillary sucti- degradation in the future. Özköse, A., Vallet, J.M., Srša, I., EwaGlos rous cracks, delamination and partly gra- on from the back of the walls is possible: European Illustrated Glossary of Con- nular disintegration.2 to the ground, the epitaph was sealed up servation terms for Wall painting and To minimize the risk of the transport, with lead, and to the back wall, a ventila- Acknowledgements Architectural Surfaces. Michael Imhof, the surface of the front of the epitaph ted space of 10 cm for air circulation was Petersberg, (2015). was temporary protected by cyclodo- created. Thus, no salt ion transport into We would like to thank for support decane – a temporary measure of fixing the epitaph by capillary uptake will take and fruitful discussions: Gottfried Hauff, 7 ICOMOS – ISCS, Illustrated glossary on the surface.10 place. Peter Kozub, Katja Schmeikal, Catalina stone deterioration patterns.- Monu- The application of poultices is a com- Schulz, and many other students who ment and Sites XV, Michael Imhof, Pe- mon measure of salt reduction.11 The- worked with us in Werben. We appreciate tersberg, (2010). refore, after arriving in the workshop, Summary the support of the parish of St. Johannis a moist cellulose poultice was applied in Werben. 8 Horie, C.V., Materials for Conservation on the back of the epitaph to avoid salt The investigations in the ground floor - Organic Consolidants, adhesives and crystallization at the surface of the sto- of the steeple and the conservation of coatings.- 2. edition, Butterworths, Lon- ne which was expected due to the ch- one epitaph as a case study could indicate don, (2010). ange to more arid climate conditions in how to preserve the epitaphs of St. Johan-

282 283 SWBSS 2017 | 20-22 September S. Laue et al. 4th International Conference on Salt Weathering of Buildings and Stone Sculptures

9 Wollensen, E. & Haetge, E., Werben, Jo- Salt-induced flaking of wall paintings at the hanniskirche und Austattung. In: Der Kreis Osterburg. Die Kunstdenkmale der Mogao Grottoes, China Provinz Sachsen Band 4, August Hopfer Verlag Burg b.M., Osterburg, (1938). Lori Wong1*, Su Bomin2, Wang Xiaowei2, Amarilli Rava3 and Neville Agnew1 1 The Getty Conservation Institute, USA 10 Rowe, S., Rozeik, C., The uses of cyclo- 2 Dunhuang Academy, Mogao Grottoes, Dunhuang, Gansu, China 3 dodecane in conservation.- Reviews in The Courtauld Institute of Art, UK * [email protected] Conservation, Volume 9, IIC, London, (2008), 17-31. Abstract cohol (PVA) and polyvinyl acetate (PVAc) created a film-like barrier that reduced 11 Heritage, A., Heritage A. & Zezza, Des- The Mogao Grottoes, a World Heritage permeability and trapped salts below alination of historic buildings, stone Site in northwest China, is known for the painted surface. This led to a build and wall paintings.- Archetype Publica- its surviving 492 painted Buddhist cave up of salts that when exposed to periods tions Ltd, London, (2013). temples. Commissioned over a thousand of high humidity caused disruption and year period, from the fourth to the four- powdering of the plaster from cycles of 12 Hübner, B. & Schulz, C., Documentation teenth centuries, the caves were hewn deliquescence and crystallization; the of the conservation measures on the into a 1.6 km long cliff face and the wall consolidated upper layer, then separated epitaph of Leonhardt Kämpfe.- unpu- paintings executed on earthen plasters. and lifted, in a new form of flaking, refer- blished report, University of Applied Situated in a remote and arid desert red to as exfoliation. Sciences Potsdam, (2016). landscape, these painted caves have The study also aimed to improve me- endured throughout the centuries but thods of condition monitoring to better many have also suffered from salt-rela- assess when change due to salt activity ted deterioration. occurs and to implement findings from Repeat cycles of treatment for flaking the Visitor Carrying Capacity Study for on salt-damaged wall paintings have cau- the site. This included identifying caves sed worsening of conditions resulting in at risk of salt-related deterioration and significant loss of painted plaster. A re- closing them to visitation during periods search project to study this intractable of high humidity. problem and to develop and implement improved treatment methods was under- Keywords: Mogao Grottoes, Dunhuang, taken as part of a collaboration between wall painting, flaking, salt deterioration, the Getty Conservation Institute (GCI) previous treatment and the Dunhuang Academy (DA) under China’s State Administration for Cultural Heritage (SACH). 1. Introduction This case study looks at the activation mechanisms and deterioration processes Salt-induced flaking of wall paintings of salt-induced flaking and highlights is a condition phenomenon afflicting ca- the development and implementation ves at the Mogao Grottoes, a World Heri- of remedial and preventive measures tage Site in northwest China, known for to prevent further loss from occurring. its 492 surviving painted Buddhist cave Topics investigated include the material temples dating from the fourth to the fo- composition of the paintings and plaster, urteenth centuries (Figure 1). Situated in previous treatments, salt identification a remote and arid desert landscape, the- and distribution, environmental condi- se caves have endured throughout the tions and the impact of increased humi- centuries but their wall paintings have dity. Results show that past treatment of been subject to a particular form of loss flaking wall painting with polyvinyl al- related to both soluble salt deterioration

Treatment of porous, inorganic subst- high humidity, exterior damp air enters rates with synthetic polymers like poly- rapidly increasing humidity and resul- vinyl alcohol (PVA) and polyvinyl acetate ting in damaging salt activity.6, 7 This as- (PVAc) can decrease pore size with a con- pect of the collaborative project was also sequent reduction of vapor and gas per- part of the implementation of findings meability.2, 3 Upon exposure to high hu- from the Visitor Carrying Capacity Study midity the salts present undergo a phase for the Mogao Grottoes that developed change, from solid to liquid, and are free visitor management solutions to close to move through the stratigraphy. The caves at risk of salt-related deterioration less permeable surface, where the syn- – including exfoliation – during periods thetic organic materials were applied, of high humidity.8 Environmental condi- can act as a barrier trapping the ions in tions such as rain and flood events are solution; as the moisture evaporates, the occurring with increasing frequency cau- solution becomes saturated, causing the sing heightened concern over the long- salts to precipitate.4, 5 Cycles of delique- term preservation of the site and its wall scence and crystallization then causes paintings. gradual decohesion of the lower plaster layer; the upper layers are then separated from the powdering plaster below and 2. Wall painting technique are free to lift. When caves are visited, doors are ope- Commissioned over a thousand year ned allowing outdoor humid air to int- period, the caves were hewn into a 1.6 km Figure 1: Cave 85, one of the 492 painted Buddhist cave temples at the Mogao Grottoes, was selected as the test cave for this project. Its wall paintings which date from the ninth century suffer from salt-induced flaking. rude into cave interiors. In instances of long cliff face of conglomerate rock. Once

and paint flaking, known as exfoliation (Figure 2). Repeat cycles of treatment in past decades have played a role in causing and exacerbating the deterioration phenomena, resulting in significant loss of the cave wall paintings. A joint research project to study this problem was undertaken as part of a collaboration between the Getty Conservation Institu- te (GCI) and the Dunhuang Academy (DA) under China’s State Administration for Cultural Heritage (SACH) with the aim of Figure 2: Salt-induced flaking or exfoliation of the wall developing and implementing improved paintings in Cave 85 treatment methods as well as measures to prevent this condition from recurring. Improvements in documentation and ous treatment materials and hygroscopic condition monitoring methods were also salts were identified, and environmental undertaken to better assess when chan- parameters and visitation history moni- ges to the wall paintings are occurring. tored. To provide a complete case study, In order to understand the activation Cave 85 was selected as a test cave for mechanisms and deterioration proces- this project. This ninth century cave has ses of exfoliation, the composition of the been extensively studied and informati- 1 paintings and plaster was studied, previ- on was readily available . Figure 3: Stratigraphy of the painted plaster in Cave 85.

286 287 L. Wong et al. Salt-induced flaking of wall paintings at the Mogao Grottoes, China

excavated, the roughly carved rock walls With the founding of the Dunhuang common language for the naming, recor- 4.2. Paint flaking and exfoliation and ceilings were plastered and smoo- National Art Research Institute in 1944, ding and communicating of information thed with a coarse leveling plaster of doors were installed on caves and begin- regarding this condition. The terminolo- Paint flaking occurs when the paint, or variable thickness (5-30 mm) and finis- ning in the 1950s full-scale stabilization gy also reflects the specific technique of paint and ground layer, separate from the hed with a fine plaster (4-5 mm) (Figure work on the grottoes commenced. Flood execution of the wall paintings in Cave layer below and then lifts. This deteriora- 3). The composition of the plasters were control measures were put in place in the 85 and focuses on the particular condition phenomenon is sometimes referred found to be a close match to the riverbed 1960s which prevented occurrences of tion phenomena associated with salt-re- to as “pure flaking” to distinguish it from silt from the Daquan River that runs th- flooding until June 2011 when a major lated flaking. Conditions were mapped, “exfoliation” or salt-induced flaking. Ex- rough the site with additional sand and flood caused substantial damage to the macrophotography undertaken and a de- foliation occurs deeper within the stra- plant fibers mixed in to prevent shrinka- site infrastructure. A subsequent flood tailed description of each produced. tigraphy of the painted plaster than pure ge and cracking. The plasters are bound occurred in 2012. flaking, typically within the fine plaster with clay and contain 19% clay-sized frac- Past documentation records show a layer, and is defined as lifting of the fine tion including illite, chlorite and the mi- history of salt-related deterioration exhi- 4.1. Salt deterioration plaster and/or ground and paint layers. xed-layer mineral, illite/smectite.1 bited as both surface paint loss and de- Exfoliation was found to only occur in A ground layer, composed of calcium tachment and collapse of painted plaster. Conditions related to salt deterioration caves that have been previously treated carbonate, talc, and mica and bound In the 1960s, widespread paint flaking included plaster disruption which exhi- for flaking with PVA and PVAc and have with possibly a plant gum or mucilage, was observed in a number of caves and bits itself as powdering plaster caused by active salt deterioration. Exfoliation can was applied as a thin wash to the fine water-soluble synthetic polymers, polyvi- repeat cycles of salt activity (Figure 4); be broken down into three stages of de- plaster to prepare the surface for pain- nyl acetate (PVAc) and polyvinyl alcohol salt efflorescence seen as salt crystals on velopment: ting. Onto the ground, delicate line dra- (PVA), were used to treat the paintings, the painted surface; and, punctate loss, wings were executed in red and black both to relay individual flakes and as a tiny, rounded losses, less than 1 mm in 1) Decohesion of the plaster due to salt ink. The decorative scheme was then surface consolidant. Today, other adhesi- diameter, in the paint, ground and/or activity causing volume expansion filled in with a rich palette of inorganic ves such as gelatin are also being used to fine plaster layer. Salt efflorescence and that pushes out the layers above lea- pigments and organic colorants applied treat occurrences of flaking. punctate loss are related conditions: in- ding to bulging and tenting of the singly or in combination, as thin washes In the 1980s, the site was first opened dividual salt crystals push through to the paint, ground and some fine plaster or thick layers, and finished with colored to visitors. surface creating tiny losses in the paint layers. In some cases the protruding glazes. Because of the variation in paint layer. Over time, this gradually leads to area can have associated cracking but application, the overall thickness varied complete loss of the paint layer in areas otherwise the surface is still closed (Fi- considerably (5-150 µm). Analysis found 4. Condition recording where salt problems are concentrated. gure 5a). bone glue to be the most extensively Paint flaking has been the most preva- used paint medium but samples also sho- 2) The protruding layers of paint, ground lent problem at Mogao. During an assess- wed varying amounts of polysaccharides and some fine plaster breaks open ment of 112 caves at the site, seventy-one possibly indicating the presence of plant along cracks and begins to lift further 1 were found to exhibit this condition ma- gums, honey, and mucilage . (Figure 5b). king up 63% of the caves assessed.9 The type of flaking varied considerably from 3) The breakage and opening of prot- 3. Physical history of site and localized to widespread and ranging in ruding areas progresses further to a conservation history severity; sixteen caves exhibited serious and extensive flaking. Salt-related dete- point where loss of original material Images from the early twentieth cen- rioration was found in thirty-four caves occurs (Figure 5c). tury show the site abandoned and the making up 30% of caves assessed. Twen- fronts of the cliff eroded leaving the ty-eight of these are ground level caves Photography alone cannot always ac- caves open and their wall paintings ex- with a history of flooding.9 curately describe the extent of lifting, as posed to the elements. Periodic flooding In the past, types of deterioration have the way conditions are perceived by the also occurred in the site’s history leading tended to be looked at individually rather 1 cm viewer are affected by the angle of inci- to complete loss of painting up to one than jointly and as a result salt-related dence of the light on the object. Other meter high in ground level caves and as- flaking has only recently been identified documentation techniques were emplo- sociated moisture and salt-related deteri- as a unique phenomenon. A terminolo- Figure 4: Powdering and decohesion of the plaster and yed to improve characterization of condi- punctate losses, visible in the red paint layer, are both oration in the lower zone of painting. gy was established to provide a clear and forms of salt deterioration. tions such as Reflectance Transformation

288 289 L. Wong et al. Salt-induced flaking of wall paintings at the Mogao Grottoes, China

Imaging (RTI) which allows you to view ent: in Cave 85, areas with the most seve- through natural ventilation which allows a an area under different lighting angles. re damage had salt content up to 2-3.9% cave air to freely exchange with the out- (expressed in weight percent of NaCl per side when cave doors are opened during 100 gm of plaster) in contrast to less than periods of visitation. Moisture produced 5. Analytical investigation 1% salt content in areas without visible by visitors does not significantly contri- damage. Stratigraphic soluble ion cont- bute to moisture levels in the cave.10 The Sampling of exfoliating layers from ent for chloride and sodium was highest impact of the cave door being opened du- Cave 85 was undertaken in areas where at the second increment (2-5 mm) indica- ring high exterior humidity is dramatic, deterioration was concentrated. Analyti- ting enrichment of NaCl within the fine causing a sharp and sudden spike in RH cal methods employed included ESEM- plaster and at the interface between the (Figure 6). For the most part when a cave EDX on samples mounted in cross-secti- fine and coarse plaster layers. is not visited and the cave entrance door on, FTIR to identify the presence of PVAc, remains closed, a fairly stable climate is XRD to identify salt species and ion ana- maintained.1 However, when high hu- 1 cm lysis to quantify soluble ions. 6. Environmental monitoring midity persists for extended periods, the RH inside the cave increases slowly Environmental monitoring data was even when the cave door is not opened. 5.1. Treatment materials collected from three locations within The louvered panels and gaps around the b Cave 85, measuring temperature and re- door allow humid air to enter. The presence of some treatment ma- lative humidity. The moisture levels in terials on the paintings could be seen the cave are predominantly influenced with UV fluorescence in situ but in general FTIR was undertaken on unmoun- ted samples using an acetone extraction. FTIR can give semi-quantitative information on the concentration of PVAc present on samples, with standardized weight, based on the strength of signal. Out of sixteen samples, twelve indicated the presence of PVAc, two had borderline 1 cm PVAc present and two showed no positive results for PVAc. Attempts to quantify PVAc content more precisely were unsuccessful. c

5.2. Soluble salts

The main salts present were identified by X-ray diffraction (XRD) as sodium chloride (NaCl) and sodium sulfate (Na- 1 2SO4). Core sampling at different depths into the clay plaster was undertaken in three caves to provide topographic and stratigraphic distribution.1 Samples were analyzed with ion chromatography and results indicated that salts were not even- 1 cm ly distributed between caves and within individual caves. Deterioration, however, Figure 6: The humidity in Cave 85 rises even with the cave door closed after a prolonged period of rainfall on 6/16. The sharp peak in humidity that occurs between 6/19 and 6/20 is when the door is opened. In 2011, the RH exceeded 75% Figure 5a-c: The three stages of progression of exfoliation. was found to be correlated with salt cont- several times throughout the summer months.

290 291 L. Wong et al. Salt-induced flaking of wall paintings at the Mogao Grottoes, China

7. Diagnostic investigation volumetric expansion and the bulging of cative of the lifting of layers that occurs with reduced water content and non- areas that is characteristic of stage 1 of with exfoliation. Time lapse images were aqueous systems. Water was used mixed Caves at the site with a history of tre- exfoliation. If conditions persist then sta- put together into a short video to record with ethanol in varying percentages: atment with PVA and PVAc were identi- ge 2 occurs, where the protrusions burst the movement. However, given the com- 100% water, 50% water: 50% ethanol, 30% fied and examined. Of these, a few were open; and, then stage 3, when the area plexity of the stratigraphy of the samples water: 70% ethanol, and 100% ethanol. also found to exhibit salt deterioration continues to open and lift to the point including paint, ground, fine plaster plus Prewetting with ethanol or ethanol and and exfoliation. However, there was no when losses occur. soluble salts and PVAc – each of which water (1:1) was also undertaken which corresponding information regarding has a different moisture uptake rate – it helped to soften the layers and acted to environmental data and visitation his- was not possible to pinpoint the exact consolidate the powdering loose material tory that could be used to correlate with 7.3. Humidity chamber reasons for the lifting.11 behind the flakes prior to relaying. deterioration. Cave 85 was the only cave Application methods also aimed to where complete information existed. In order to understand the impact of control the quantity of liquid used and fluctuating humidity on the surface and 7.4. Modeling salt behavior to keep the treated area as contained as subsurface layers treated with PVA and possible: small brushes were used and 7.1. Linking visitation and environmen- PVAc, samples taken from exfoliated A salt mixture collected from Cave 85 the liquid applied through two layers of tal data areas were placed in an environmental was tested using Dynamic Vapor Sorpti- paper tissue. The flakes were gently pres- chamber to assess their response to ch- on (DVS) between 60% and 75% RH with sed back. A Preservation Pencil, an ultra- Visitation information for Cave 85 anges in RH (possible moisture uptake increments of 0.5% RH.1 The resulting sonic humidifier, was also tested in order shows that prior to exfoliation being ob- and contraction) (Figure 7). The RH in the isotherm showed mass increase starting to further reduce water content being in- served, the cave was opened on a limited chamber ranged from approximately 38- at 67% RH, as the relative humidity introduced into the wall paintings but was basis to visitors. During summer months, 86%. A selection of samples were subjec- creased, the slope increased exponen- found to be too slow to be practical. several periods of high RH – some of ted to RH fluctuation and a few showed tially. Salt samples were also examined which reached over 80% – were recorded. movement beginning around 60%, indi- with ECOS-RUNSALT, a program that uses The repeat fluctuations of RH led to wi- a thermodynamic model to predict the 9. Condition monitoring despread plaster disruption and exfolia- behavior of salt mixtures under chan- tion in the parts of the cave where salts ging climatic conditions 1. However, as Photographic condition monitoring of are concentrated. salt mixtures can vary depending on the areas treated were established. Context cave and location and conditions of the and macro images were taken from diffe- sampling; and, as different salt species rent angles and under different lighting 7.2. Role of PVA and PVAc can form depending on the RH, main- conditions both before and after treat- taining stable RH and T within a cave ment. Microscopy areas were also estab- Prior to the use of PVA and PVAc on the was determined to be the best means of lished to monitor the emergence of salts paintings, salts would have been able to preventing future damaging salt activity. that may not be visible without magni- move through the porous plaster before fication. Monitoring aims to be under- encountering the less permeable barrier taken using the same equipment setup constituted by the paint layer. This cau- 8. Remedial treatment including the photographer, camera and sed crystallization within or underneath lighting system. the ground and paint layers, pushing off Given that exfoliation is occurring wit- the paint layer and leading to punctate hin the fine earthen plaster and in order loss. Treatment with PVA and PVAc, pe- to avoid adding a film-forming material 10. Conclusions netrated into the paint, ground and fine in the stratigraphy, treatment to relay plaster layers, affecting the permeability flakes was undertaken using only water In all treated areas, exfoliation has not and porosity of the plaster, creating a less (Figure 8). Water, can act to restore clay recurred and the decohesive areas of plas- permeable layer. When humidity levels binding properties and proved to be an ter have remained stable. However, in an fluctuated, salts present in the plaster, effective means of relaying areas of ex- otherwise arid climate, rain and periods unable to push through to the surface, foliation. However, as water can further of high humidity are occurring with in- underwent cycles of crystallization and Figure 7: A sample from Cave 85 placed in an environ- activate salts, causing damage, and there creasing frequency. Environmental data deliquescence resulting in plaster dis- mental chamber contracted at 46% RH (left) and relaxed are water sensitive elements in the pain- shows that humidity inside the caves in- at 86% RH (right) in response to changing humidity. ruption in the fine plaster. This caused Movement in the sample was observed around 60% RH. ting, treatment trials were undertaken creases even with doors closed. Though

292 293 L. Wong et al. Salt-induced flaking of wall paintings at the Mogao Grottoes, China

and NMR relaxometry: a more than 15- 9 Wong, L., Graves, K. and Wang, X., As- year study, Magnetic Resonance in Che- sessment of Cave Condition and Visita- mistry 53 (1), (2015), 34–47. tion Potential: Undertaken as part of the Visitor Carrying Capacity Study at 3 Carretti, E. and Dei, L., ‘Physiochemical the Mogao Grottoes, China, unpublis- Characterization of Acrylic Polymeric hed report, Los Angeles: Getty Conser- Resins Coating Porous Materials of Ar- vation Institute, 2012. tistic Interest’, Progress in Organic Coa- tings 49(3), (2004), 282–89. 10 Maekawa, S., Zhan, Y., Wang, B., Fu, W., and Xue, P., Environmental monitoring 4 Zhang, H., Qiang L., Ting L. and Bingjian at the Mogao Grottoes, Conservation of Z., ‘The preservation damage of hydro- Ancient Sites on the Silk Road: Procee- phobic polymer coating materials in dings of an International Conference conservation of stone relics’, Progress in on the Conservation of Grotto Sites, N. Organic Coatings 76(7–8), (2013), 1127– Agnew, ed. Los Angeles: Getty Conser- 34. vation Institute, 1997, 301–313.

5 Arnold, A., & Zehnder, K., ‘Monitoring 11 He, X., Xu, M., Zhang, H., Zhang, B., and wall paintings affected by soluble salts’, Su, B., ‘An exploratory study of the The Conservation of wall paintings: pro- deterioration mechanism of ancient ceedings of a symposium organized by wall paintings based on thermal and the Courtauld Institute of Art and the moisture expansion property analysis’,

Figure 8: Relaying of exfoliated areas was undertaken with water and water: ethanol mixes applied with brushes Getty Conservation Institute, London, Journal of Archaeological Science, 42, through paper tissue. July 13-16, 1987, S. Cather ed., Marina (2014), 194–200. del Rey, CA: Getty Conservation Institu- te, 1991, 103–35. i This project builds upon a long lasting the aluminum door acts to buffer against Fan Zaixuan, Li Yanfei, Qiao Hai, Sun partnership between the Getty Conser- the outside conditions, recommenda- Shengli and Zhang Guobin; from the GCI: 6 Agnew, N., Maekawa, S. and Wei, S., Cau- vation Institute and the Dunhuang Aca- tions were made to improve the seal by Martha Demas, Shin Maekawa and Julie ses and mechanisms of deterioration demy at the Mogao Grottoes. This inclu- closing gaps around the door and co- Chang (2014 GCI graduate intern); and, and damage in Cave 85, Conservation of des the conservation of wall paintings vering louvered panels with non-woven GCI consultants, Po-Ming Lin and Peter Ancient Sites on the Silk Road: Procee- in Cave 85 as well as master planning permeable fabric to slow air intrusion. Barker. dings of the Second International Con- for the site and visitor management [1, This should be done in conjunction with ference on the Conservation of Grotto 8]. For more information, see: http:// continued condition and environmental Sites, Mogao Grottoes, Dunhuang, Peo- www.getty.edu/conservation/our_pro- monitoring in order to assess the stabi- ple‘s Republic of China, June 28-July 3, jects/field_projects/china/app_mogao. lizing effects this has on the RH of the 2004, N. Agnew ed., Los Angeles: Getty html. cave. Finally, visitation to the cave should Conservation Institute, 2010, 412–420. be managed in a way that does not en- References ii The assessment was undertaken as part danger the wall paintings with a system 7 Li, H., Wang, W., Zhan, H., Qiu, F., Guo, Q., of the Visitor Carrying Capacity Study 1 for closing the caves to visitors during Wong, L. & Agnew, N. eds., The Conser- Sun, S. and Zhang, G., ‘The effects of at- at the Mogao Grottoes by the Getty periods of high RH. vation of Cave 85 at the Mogao Grot- mospheric moisture on the mural pain- Conservation Institute and the Dun- toes, Dunhuang: A collaborative project tings of the Mogao Grottoes’, Studies in huang Academy [8, 9]. of the Getty Conservation Institute and Conservation, 62(3-4) (2017), 229–239. 11. Acknowledgements the Dunhuang Academy, Los Angeles: iii The acetone extraction process consis- Getty Conservation Institute, 2013. 8 Demas, M., Agnew, N., and Fan, J., Stra- ted of leaving the sample at room tem- The authors would like to acknowled- tegies for Sustainable Tourism at the perature for 3 hours then centrifuging 2 ge the significant contributions of the Camaiti, M., Bortolotti, V. and Fantazzi- Mogao Grottoes of Dunhuang, China, it for 10 minutes. 0.2 µl of the solvent rest of the project team: from the Dun- ni, P., ‘Stone porosity, wettability chan- Springer International Publishing, 2015. extract was then run under FTIR. huang Academy: Chai Bolong, Chen Bo, ges and other features detected by MRI

294 295 SWBSS 2017 | 20-22 September L. Wong et al. 4th International Conference on Salt Weathering of Buildings and Stone Sculptures

iv The source of the salts in the Cave 85 Development of a network-based climate monitoring wall painting may have originated in the conglomerate substrate, in the ori- system for climate assessment and regulation ginal materials of the painted plaster or was brought in with flood events. For Christian Leonhardt*, Sabine Leonhardt and Julika Heller complete information on the salt in- Werkstatt für Kunst und Denkmalpflege, Kiel, Gerrmany vestigation in Cave 85 see Chapter 16, * [email protected] Salt-induced Deterioration of Wong & Agnew 2013 [1]. Abstract le online, which is more convenient and v Copyright © 2002-2005 by Davide Bionda. advantageous than conventional data In the cloister (the so-called Schwahl) log-ger records for long-term climate vi Manufactured by Preservation Equip- of the St. Petri Cathedral in the German measurement and evaluation, which can ment Ltd. town of Schleswig, the current situation only be accessed in situ. It allows direct of the historical wall paintings was inves- response to critical climatic values by tigated in the context of a research pro- connected alarm, control and regulation ject funded by the German Federal Foun- systems. This way, measurements can be dation for the Environment (German: taken promptly thus preventing damage “Deutscher Bund für Umwelt” – DBU). and high resto-ration costs. Previously, The project is focused on the investiga- climatic data had been collected in the tion of the intense salt contami-nation course of the research project since 2007, and its connections with the climatic si- the comprehensive evaluation of which tuation. served as a basis for the setting of the pa- In this context, a network-based cli- ram-eters for the new system. mate monitoring system was installed Measurement data are transferred di- and tested in 2016. The data is accessib- rectly to an open-source based server in-

Figure 1: The Schwahl in Schleswig, exterior walls with the chronological depictions of Christ‘s life cycle

296 297 C. Leonhardt et al. Development of a Network-based Climate Monitoring System for Climate Assessment and Regulation

frastructure, which is scalable and ready most important medieval monuments of 2. Necessity for a network-based perties of gypsum due to the interaction for future requirements. Due to the data the federal state of Schleswig-Holstein. climate-monitoring system with highly soluble salts depending on being directly available on the internet, The church, which was mentioned for the indoor and outdoor climate. Here, it is possible to provide a minute-accura- the first time in 1134, was extended on Starting with a comprehensive restora- besides long term changes in climate climate monitoring in real-time. Thus, the north side by a three-winged cloister tion and conservation measure in 2007, te conditions during a period of several the evaluation of an arbitrary number of between 1310 and 1320 by order of Bis- the condition of the stock of the Schwahl days and weeks, short term changes in measurement locations and the analysis hop Johannes II von Bockholt. The clois- had been continuously observed. Here, the indoor climate could have a relevant of local causes (events, weather changes) ter is called “Schwahl” (Danish: Svalen) an accelerated accumulation of gypsum influence on the damage process. Thus, can be put into practice. which means “semi-open passage”. The and hygroscopic salts was observed on for instance, opening the exterior doors The system was conceived and tested 24 yokes of the Schwahl are opened to- the outer walls causing intense damage for a longer period of time under certain for the requirements of the Schleswig wards the courtyard by pointed-angled to the wall paintings. Consequently, se- weather conditions could cause an unfa- Cathedral. It can also be used in archives, windows. Small slits are integrated into veral work campaigns have been carried vourable change in the interior climate depots and churches and thus constitu- the windows, so that the cloister is a per- out to address this problem since 2010. thus causing a temporarily accel-erated tes a reliable basis for climate monito- manently unheated, slightly ventilated Among other measures, these included damage progression on the wall pain- ring. room under the partial influence of the the reduction of the accumulations of tings. external climate. gypsum and salt by the application of Since this is a common problem affec- Keywords: Climate monitoring, climate The Schwahl has been used as a walk- poultices (compare figures 2 and ).3 The ting a great number of architectural mo- data, networking-based monitoring, pre- way for processions and shows polychro- campaigns were accompanied by com- numents, appropriate measures need to ventive conservation, regulation systems me paintings on the arches of the 24 prehensive analyses prior to and after be taken to deal with this issue. Conclu- yokes. The paintings represent fabulous the concerning measures. sively, this gave rise to the development creatures and are largely well pre-ser- Subsequently, the complex causes and of a modular and reusable solution for a 1. The cloister in the St. Petri cathedral ved. Moreover, the wall panels shown in conditions of the given damage proces- network-based and real time-capable cli- at Schleswig the picture below are covered by pain- ses are being in-vestigated within in the mate monitoring system. Thus, a remote tings giving a chronological depiction of framework of a research project funded observation of the conditions in situ can The St. Petri cathedral in the north Christ‘s life cycle. by the German Federal Environ-mental be realised and proper advice can be gi- German town of Schleswig is one of the Foundation (German: Deutsche Bundes- ven in real time. This involves automatic stiftung Umwelt (DBU) since 2016 (com- alarms based on guidelines derived for pare [2]). the individual object as well as a remote The primary goal of the project was to evaluation of critical situations. In doing investigate the modification of the pro- so, the authors aim to provide a tool to si-

Figure 2: Application of desalination poultices Figure 3: After the application of desalination poultices Figure 4: Principle sketch of intelligent monitoring and its integration in cultural heritage maintenance, based on [1]

298 299 Development of a Network-based Climate Monitoring System for Climate Assessment and Regulation C. Leonhardt et al.

al-time. This network-based climate mo- Finally, the read-out and evaluation References nitoring system is named IM-MOMON of the climate data was usually carried (www.immomon.net) and was created out in an annual cycle prior to the ins- 1 ReMonitoring of historic structures, by collaboration of restorers involved in tallation of the network-based climate project smoohs, p. 17, www.smoohs.eu the pro-ject and IT system architect Dirk monitoring. Accordingly, the process of Hoffmeister. data evaluation becomes cumbersome 2 Project 3216945: The historical pain- As an outcome of this cooperation, and laborious due to the huge amount tings in the silt of St. Peter‘s Cathedral proper measuring locations, measuring of data to be analysed. Furthermore, the in Schleswig: investigations on the con- intervals, holding time of the data, th- association of the data to specific events version, mobilization and recrystalliz- reshold values were determined based (e. g. concerts, exhibitions, etc) and their ation of gypsum on heavily salt-laden on the restorers‘ expert knowledge. Also, effects on the climate becomes difficult ground appropriate measures were derived to this way. This issue can now easily be de- deal with alarm messages caused by cri- alt with by regular evaluation of the data tical climate condi-tions due to e.g. open available in real-time (figures 4 and 5) on Figure 5: Read the climate data from the Schwahl, via exterior doors. The control of external the servers. Mobilephone and PC. devices (e. g. ventilation) can also be initi- ated by the system. gnificantly reduce the amount of dama- The climate data are measured with 4. Conclusion and outlooks ge caused by temporarily unfavourable set intervals by sensors distributed in climate conditions. the monument, collected in a local data The recording and evaluation of the hub and then transferred directly to a se- interior climate of historic buildings and cured internet server. In addition to the historic items or art objects in exhibition 3. Development of a network-based commonly requested data on room tem- or storage is a fundamental, preventive climate monitoring – IMMOMON perature and relative humidity, sensors measure for preservation. Even costin- for outside climate, etc. can be easily intensive interventions due to climatein- Within the scope of the DBU project it tegrated into the system. Via authorized duced damage processes can be avoided was possible to develop a system for mea- access rights clients can access their data by characterisation of the climate profile suring and moni-toring the climate in re- in real-time. and development of corresponding restoration recommendations. Here, IMMOMON can provide a significant contribution. Thus, the network-based measurement, control and warning system saves the necessity for regular control in-situ and allows for immedi-ate response to irregularities or unfavourable climatic situations. More information on the network-based climate monitoring can be found on www.immomon.net.

Figure 6: Visualization of the measuring points on site, defective devices are displayed in real time (meas-uring point 5).

300 301 all about salts: www.saltwiki.net