Monument Future

Traditional production of gypsum mortar

For the restoration work locally available material is used. The raw material for the plaster production comes from a quarry some 17 km WNW off the Takht-e Soleyman. Two slightly different qualities of gypsum with respect to the contents of non sulfate accessory minerals are mined. The gypsum rock contains dolomite (ankerite), calcite, quartz, feldspar, clay minerals and celestite as accessory minerals which can sum up to 8–10 mass%.

The firing of the raw material is done in the traditional way as it was done many centuries ago (Soleymani, Pirak 2012, Sobott 2018). A shaft furnace built of bricks with a diametre of 1.90 m was sunk 2.90 m deep in the ground. The quarried gypsum lumps are piled up in the furnace in such a way that something like a corbeled vault is formed. Large pieces of gypsum rock are on the inside of the construction facing the firing chamber and small pieces are used to fill the space between the furnace wall and the rock pile. The apex of this artful cone-shaped construction surmounts the upper end of the furnace (Figure 4). A mix of combustible material, mostly wood, is piled up inside the vault.

Once the fire is lit the uncontrolled firing process lasts about 8 hours. Due to the construction of the furnace the temperature distribution in the gypsum filling is extremely variable. Thermocouples installed at different positions in the gypsum pile showed that the temperature difference between the central part and the margin may be as great as 800 °C so that the gypsum lumps are exposed to temperatures ranging from 200 and 1,000 °C (Jäger 2017, Jafarpanah 2017). The cooling period after the extinction of the fire lasts about 24 hours. Then the fired material is removed from the furnace and crushed with large hammers by hand at which high fired gypsum lumps disintegrate easily into powder while low fired lumps break into smaller pieces. The crushed material is sieved and filled into plastic bags. For use at the construction site the material was sieved to discard grains larger than 2 mm. The phase composition of the fired product is variable and depends mainly on the grain size and composition of the raw material, the firing temperature and time, and the resulting partial water vapour pressure on the surface of the lumps. For the assessment of the firing results, 116gypsum pieces adjacent to thermocouples were sampled and studied in situ by colouring tests and in the laboratory by polarized light microscopy of thin sections, X-ray diffraction and thermoanalysis (DTA, DTG). In order to improve the data interpretation of the field samples, experiments with raw material from the gypsum quarry were carried out under controlled temperature and time conditions in the laboratory and the resulting samples analyzed in the same manner.

Phase composition of the calcined raw material

Gypsum lumps from the furnace exhibited a temperature dependent zonation with respect to the occurrence of phases and phase morphologies (Lenz, Sobott 2008) . Low fired gypsum lumps typically consist of three zones. A thin surface layer of rehydrated anhydrite III (dehydrated hemihydrate), is followed by a layer of fibrous hemihydrate, and the central part is made up of a mixture of hemihydrate and dihydrate (unconverted gypsum). Thermoanalytic measurements prove anhydrite III to be stable up to 370 °C. It forms fibrous crystals, partly pseudomorph after gypsum. With increasing firing temperature hemihydrate and anhydrite III are converted to anhydrite II (natural anhydrite). Anhydrite II formed in the temperature range between 370 and 800 °C shows elongated fibrous crystals and at temperatures above 800 °C the crystals tend to be short prismatic (Figure 5).

Figure 5: left: short prismatic anhydrite II crystals in samples fired at 1,100 °C for 5 hours; right: prismatic anhydrite crystals in samples fired at 800 °C for 5 hours; crossed polarized light.

In a laboratory sample which was fired at 600 °C for 5 hours radial fibrous anhydrite crystals were observed. Obviously the shape of anhydrite II crystals and aggregates in fired samples depends largely on how the heat treatment was executed.

In thin sections of gypsum samples which were exposed to temperatures above 600 °C isotropic crystals with a hexagon outline were observed.

They were identified by EDXRF measurements as periclase MgO (Figure 6).

Müller et al. (2009) described the formation of periclase by a contact metamorphic reaction in dolomite marble at P = 1 kbar and T > 605 °C

CaMg(CO₃)₂ → MgO + CaCO₃ + CO₂ dolomite periclase calcite

Since idiomorphic to hypidiomorphic dolomite crystals up to 400µm in size were observed in thin sections of the raw material from the gypsum quarry near the Takht-e-Soleyman, it is reasonable to assume that the dolomite disintegrated and was converted to periclase according to the above mentioned chemical reaction. Therefore the appearance of periclase in samples can be regarded as an intrinsic temperature indicator if the pressure dependence of the reaction is known.

Figure 6: Periclase crystals in anhydrite II matrix; sample from furnace, adjacent to thermocouple that recorded a maximum temperature of 800 °C; plane polarized light.

117Reactivity of the new gypsum mortar

Variable amounts of dehydrated or partly dehydrated calcium sulfate phases in the dry mortar and different crystal habits of the phases are the reasons for a variable reactivity of the mixture. For example, the low degree of hydration of the sample fired for 5 hours at 600 °C is possibly due to the radial fibrous structure of anhydrite crystals. The degree of hydration after 3 days and 1 year was determined gravimetrically and by X-Ray diffraction (Rietveld method), respectively, for samples fired at different temperatures and for different times. The results are summarized in Table 1.

As stated above the fired material is a variable mixture of components with different hydration reaction rates (Glasenapp 1910). Unlike α- and β-hemihydrate, anhydrite III and low fired anhydrite II which hydrate very quickly, high fired anhydrite II reacts very slowly with water. This retarded reaction of anhydrite II can produce cracks in the plaster if the mortar fabric is very dense and cannot compensate the volume increase of the hydrated calcium sulfate phase. Therefore the hydration of high fired anhydrite II has to be activated by a suitable chemical such as citric acid or potassium sulfate. Experiments and analyses proved superfine calcium hydoxide Ca(OH)₂ to be a suitable activator to prevent a retarded hydration of high fired anhydrite II. The linear expansion coefficient of the high fired gypsum injection mortar measured after 280 days was reduced to less than 2 mm/m. The compressive strength of the set mortar measured after 28 days was 6.1 MPa and decreased to 2.9 MPa after storage of the samples under water. No cracking was observed in test walls which were built in 2016 and injected with this mortar. Obviously its properties are favourable for the repair and reinforcement of the historic masonry and it is now successfully applied since two years.

Table 1: Degree of rehydration of fired gypsum as a function of firing temperature and time.

Static safeguarding of the west iwan

In a first step of the strengthening of the west iwan the masonry was grouted with a gypsum suspension developed by University of Technology Dresden (TUD) and Jäger Consulting Engineers Ltd. The worm pump SP-20 from Desoi was absolutely necessary for the grouting of large caverns, cavities and crack systems in the north wall of the west iwan. The water-to-gypsum ratio of the aggregate-free suspension was 0.63. The chemical Retardan 200 P by SIKA Company was added as a retarder. Tests with ViscoCrete by SIKA Company showed that the addition of a plasticizer to the suspension was not necessary to guarantee sufficient flowability. The gypsum suspension could easily be mixed and processed on-site. Observation of the injection process confirmed the very good flow behaviour of the suspension which ensured the closing of the cracks and cavities in the masonry. The grout remained flowable long enough and the quantity of the water taken from the surroundings were within limits. The masonry showed heavy vertical cracks and numerous holes and cavities in the interior, which had to be grouted from the bottom to the top of the wall with the developed gypsum suspension. First holes were drilled at a spacing of 30 to 50 cm along the course of the crack to grout the masonry. The holes had a diameter of 24 mm and depths of 30 to 80 cm. After completion of drilling, the entire masonry including the cracks was prepared for the subsequent rising grouting. For this purpose the cracks were cleaned of dust and loose objects and the parts of the masonry facade near the masonry facade were secured with loam mortar. This prevented contamination of the masonry during the grouting work and the preceding filling of the cracks. Then grouting was started from bottom to top (Figure 7).

118

Figure 7: Gypsum mortar injection into the cracked north wall of the west iwan. Right side: Bottom to top injection in sealed crack.

It was frequently the case that the masonry could be grouted up to 1 m of height from one hole. This demonstrated an effective internal transport of material, with continuous grouting of the entire masonry being guaranteed. At each level, grout escaped at the various holes drilled round the buttress. In total, 7 tons of gypsum suspension were grouted to the east part of the north wall of the west iwan.

References

Bräunel, M. (2016). Takht-e Soleyman – vertiefende Bestandsaufnahme der Ruinenteile des westlichen Iwans in Vorbereitung der notwendigen statisch-konstruktiven Sicherung. Diplomarbeit an der Technischen Universität Dresden, Lehrstuhl für Tragwerksplanung, Fakultät Architektur, 2016, unpublished.

Burkert, T., Fuchs, C., Sobott, R. (2019). Statisch-konstruktive Sicherungsarbeiten am westlichen Iwan der UNESCO-Welterbestätte Takht-e Soleyman, Iran. in Mauerwerk Kalender 2019, Ernst und Sohn, Berlin, 295–331.

Glasenapp, M. (1910). Plaster, Overburnt Gypsum and Hydraulic Gypsum. – Cement & Engineering News, Chicago, Illinois, 47 pp.

Fucke, D., Hansen, M. (2012). Takht-e Soleyman – vorbereitende Untersuchungen und Varianten zur Sicherung der Ruinenteile des westlichen Iwans. Diplomarbeit an der Technischen Universität Dresden, Lehrstuhl für Tragwerksplanung, Fakultät Architektur, 2012, unpublished.

Huff, D. (2006). The Ilkhanid Palace at Takht-i Sulayman. Excavation Results, in: Komaroff, Linda (Ed.): Beyond the Legacy of Genghis Khan. Brill, Leiden 2006, 94–110.

Jäger, W. (2017). Burning process of gypsum in the kiln in Tahkt-e Soleyman 20.05.2016. – Technische Universität Dresden, Faculty of Architecture, Chair of Structural Design, 12 pp.

Jafarpanah, M. (2017). Burning process of gypsum in the kiln in Tahkt-e Soleyman 13.08.2017. – Takht-e Soleyman – UNESCO World Heritage Site, 13 pp.

Lenz, R., Sobott, R. (2008). Beobachtungen zu Gefügen historischer Gipsmörtel. In: Gipsmörtel im historischen Mauerwerk und an den Fassaden. – Hrsg. von M. Auras und H.-W. Zier, WTA Schriftenreihe, Heft 30, 23–34.

Lucas, H. G. (1992). Gips als historischer Außenbaustoff in der Windsheimer Bucht. Dissertation, Fakultät für Bergbau, Hüttenwesen und Geowissenschaften der RWTH Aachen.

Müller, T., Baumgartner, L. P., Foster, C. T. Bowman, J. R. (2009). Crystal Size Distribution of Periclase in Contact Metamorphic Dolomite Marbles from Southern Adamello Massif, Italy. – Journal of Petrology, Vol. 50/3, 451–465.

Naumann, R. (1977). Die Ruinen von Tacht-e-Suleiman und Zendan-e-Suleiman und Umgebung.

Dietrich Reimer Verlag, Berlin, 126 pp.

Sobott, R. (2018). Historic and modern gypsum mortar application at the Takht-e Soleyman, Iran. Report about on-site studies and results of sample analyses. unpublished, 25 pp.

Soleymani, A., Pirak, M. (2012). Nachstellung von halbgebranntem und halbzerstoßenem Gips. (Transkribiert aus dem Iranischen) Quarterly Research Review of Razavi Architecture, Vol. 1, No. 1, 61–71.

119

EFFECTS OF ZEOLITES AND SWELLABLE CLAY MINERALS ON WATER-RELATED PROPERTIES AND THERMAL DILATATION IN VOLCANIC TUFF ROCKS

Alexandra Kück¹, Christopher Pötzl¹, Rubén López-Doncel², Reiner Dohrmann³, Siegfried Siegesmund¹

IN: SIEGESMUND, S. & MIDDENDORF, B. (EDS.): MONUMENT FUTURE: DECAY AND CONSERVATION OF STONE.

– PROCEEDINGS OF THE 14TH INTERNATIONAL CONGRESS ON THE DETERIORATION AND CONSERVATION OF STONE –

VOLUME I AND VOLUME II. MITTELDEUTSCHER VERLAG 2020.

1 Geoscience Centre of the Georg August University Göttingen, Germany

2 Geological Institute of the Autonomous University of San Luis Potosí, Mexico

3 Federal Institute for Geosciences and Natural Resources (BGR), Hanover, Germany

Abstract

The weathering of natural building stones in the South of Mexico is mostly controlled by the influence of atmospheric and meteoric water, thermal stress and the input of salts from the environment. Twelve varieties of volcanic tuff rocks from South and Central Mexico were analyzed regarding their petrographical and petrophysical properties, as well as their weathering behavior. The tuffs show a broad range of properties and differential weathering behavior. Moisture properties like water uptake, water vapor diffusion and hygroscopic water sorption as well as hydric expansion and salt crystallization show great dependence on the pore space properties and the content of swellable clay minerals or zeolites. The deterioration during salt bursting tests is controlled by both salt crystallization pressure and hydric expansion. Especially zeolite-rich samples show intense water-related weathering. Their sorption values and hydric expansion are high, while capillary water uptake is comparably low. The clay and zeolite-rich tuffs furthermore suffer from intense shrinking and fracturing during drying. The very high hydric expansion superimposes the salt crystallization pressure and the effect of thermal dilatation.

Introduction

Zeolites have an important influence on the weathering behaviour of natural building stones. Korkuna et al. (2006) found very small pore sizes for zeolite-rich samples (< 2 nm) resulting in a high specific surface area. The specific surface area, along with the pore size and connectivity of the pores is a controlling factor for water transport and retention in porous rocks and has a great influence on values like capillary water uptake, water vapor diffusion and hygroscopic water sorption (Siegesmund and Dürrast 2011).

Hydric dilatation can cause a significant volume change in tuff rocks and is therefore an important weathering factor. Reasons for hydric expansion may be the presence of swellable clay minerals or a large percentage of micropores (e. g. Gonzales and Scherer 2004). The swelling of natural building stones with a high percentage of micropores may be explained by the process of disjoining pressure, but the process is still under discussion (e.g. Ruedrich et al. 2011; Wedekind et al. 2013; Pötzl et al. 2018a; Pötzl et al. 2018b).

Zeolites adsorb and desorb water molecules reversibly (Di Tchernev 1978). They are therefore able to store heat during desorption and give it back to the environment during adsorption. Between 0 °C and 90 °C zeolites counter the expected thermal 120expansion because of shrinkage due to reversible water loss (Colella et al. 2001). Zeolites accordingly have an influence on the thermal expansion behavior of tuff rocks.

In this study zeolite and clay-rich tuff rocks from the states of Oaxaca in Southern Mexico and Queretaro in Central Mexico were investigated and compared with zeolite-free samples.

Petrography

The investigated samples are mainly felsic pyroclastic rocks. Cantera Verde Oaxaca (CV_O), Cantera Amarilla Oaxaca (CA_O) and Cantera Rosa Oaxaca (CR_O) are tuff rocks from the city of Oaxaca de Juárez. Cantera Verde Etla (CV_E), Cantera Amarilla Etla (CA_E) and Cantera Rosa Etla (CR_E) are from the city of Etla (20 km to the northwest of Oaxaca de Juárez). Mitla Gris (MG) and Mitla Rosa (MR) are from the village Mitla 45 km southeast of Oaxaca. Querétaro Blanco (QB), Querétaro Melon (QM), Querétaro Amarilla (QA) and Querétaro Naranja (QN) are from the state of Querétaro in Central Mexico. Lithoscans and thin section images of all samples are presented in figures 1 and 2.

Cantera Verde Oaxaca (CV_O) is a homogenous, dark-green vitric, rhyolitic ash tuff with massive to dense texture and weak orientation of pumice clasts (Fig. 1a). The glassy matrix is strongly altered to zeolite and clay minerals and is partly devitrified. Cantera Verde Etla (CV_E) is a homogenous, light-pistaccio green vitrophyric ash tuff of rhyolitic composition. Noticable macrosopically are white clay lenses (Fig. 1b), which disintegrate when in contact with water. CV_E has a hypocrystalline-cryptocrystalline matrix with vitrophyric texture. The crystals are often strongly weathered (Fig. 2b). The matrix is rich in zeolites (clinoptilolite and mordenite). SiO₂ is present in the form of crystobalite. Pablo-Galán (1986) describes the formation of zeolites in the Etla tuff by alkaline diagenesis of rhyolitic glass. He distinguishes between two varieties of tuff, a clinoptilolite-rich variety and a mordenite-rich variety.

Cantera Amarilla Oaxaca (CA_O) is a homogenous, yellow to white vitric ash tuff of rhyolitic composition (Fig. 1c). The rock is very soft and contains a high amount of swellable clay minerals (Tab. 1). Sometimes foliation can be observed in the ash layers. Lithoclasts of up to 0.5 cm are present in the cryptocrystalline to glassy matrix (Fig. 2c).

Figure 1: Lithoscans of the investigated tuffs. a) CV_O, b) CV_E, c) CA_O, d) CA_E, e) CR_O, f) CR_E, g) MG, h) MR, i) QB, j) QM, k) QA, l) QN.

Cantera Amarilla Etla (CA_E) is a yellow to orange rhyolitic, vitric ash tuff with characteristic layering (Fig. 1d). The orange layers or rings are colored by iron oxides and make the orientation of the sample clearly visible. The grain size is small (up to 3 mm) and the glassy matrix is homogenous (Fig. 2d). The tuff contains zeolites and swellable smectite (Tab. 1).

Cantera Rosa Oaxaca (CR_O) is a homogenous, fine- to medium-grained rhyolitic ash tuff of red to pink color with a glassy matrix and small crystals < 1 cm (Fig. 1e). The rock easily loses sand-sized particles on the surface. The thin section reveals an eutaxitic texture (Fig. 2e) with high amounts of welded pumice and glass. The sample is strongly devitrified in parts and contains clay lenses where the glass is altered.

121Table 1: CEC, BET as well as clay mineral content and zeolites of the investigated tuff rocks.

Cantera Rosa Etla (CR_E) is a rhyolitic, vitric ash tuff and very similar to CR_O in terms of texture, grain size and color (Fig. 1f). The thin section shows the similarity of CR_O and CR_E (Fig. 2e–f). CR_E contains smaller grains with smaller pores and the color is slightly brighter than CR_O.

The Mitla Tuff Gris (MG) is a rhyolic to dacitic crystal tuff of gray to white color (Fig. 1g). It has very variable porosities depending on the origin within the quarry. MG contains pumice clasts and angular crystals in a glassy matrix (Fig. 2g).

Mitla Tuff Rosa (MR) is a dense, rhyolitic tuff with small pores and crystals of up to 3 mm size. It has a pink matrix, which is in some parts discolored to white (Fig. 1h). The thin section (Fig. 2h) shows an overall small proportion (10 %) of cryptocrystalline to microcrystalline matrix with seriate texture. The crystals are hypidiomorphic and poorly altered.

Querétaro Blanco (QB) is a white to gray rhyolitic lapilli tuff with a high amount of pumice and glass within the matrix. Lithic fragments of more than 1 cm size are macroscopically visible in the lithoscan (Fig. 1i). The thin section of QB shows a vitrophyric texture with very few crystals and lithoclasts embedded in a glassy matrix (Fig. 2i).

Figure 2: Thin sections of he investigated tuffs. a) CV_O, b) CV_E, c) CA_O, d) CA_E, e) CR_O, f) CR_E, g) MG, h) MR, i) QB, j) QM, k) QA, l) QN with parallel nicols except g and h with crossed nicols.

Querétaro Melon (QM) is a rhyolithic, vitric lapilli tuff of orange-melon color with lithic fragments, as well as small crystals and big clasts of pumice in a glassy matrix (Fig. 1j). QM shows an eutaxitic texture with firmly welded glassy matrix and few smaller crystals under the microscope (Fig. 2j).

Querétaro Amarillo (QA) is a dacitic, heterogenous tuff breccia with clasts of several centimeters in size. The matrix is yellowish-brown and contains clasts that are often whitish-gray. The clasts are partly pumice, mostly xenoliths and a small amount of crystals (Fig. 1k). The matrix of QA is 122strongly altered to clay minerals. The pumice are mostly unwelded and show little signs of elongation or compression (Fig. 2k).

Querétaro Naranja (QN) is a vitric lapilli tuff of trachydacitic composition with lithic fragments in the centimeter scale. Small fragments of orange pumice and strongly altered crystals in a devitrified matrix of orange color are visible in Figure 1l. The thin section shows round and angular xenocrysts and altered feldspar in the fine-grained matrix (Fig. 2l).

In sum, all samples have an acidic composition. Particularly glass-rich samples are CR_O, CR_E, QM, QA and QN. Feldspar could be detected in all samples, while quartz was not detected by XRD in CV_E, CR_E, QB and QM. CV_E and QB instead contain the high temperature modifications of SiO₂ (tridymite and cristobalite). The cation exchange capacity (CEC) and the specific surface area (BET), as well as the content of clay minerals and zeolites in all samples are presented in Table 1.

Samples with significant amounts of zeolites are CV_O, CV_E and CA_E. The specific surface area (BET) is high for these samples. The main zeolites are heulandite and clinoptilolite, minor amounts of mordenite were detected in CV_E. Clay minerals were detected in all samples. The content of swellable clay minerals indicated by a high CEC value is especially high in CA_O, MG and CV_O. Smectite and muscovite-illite make up the main content of the clay fraction. Chlorite was only detected in CV_E and kaolinite was found in MG and could be present in small amounts in CV_O, CV_E, CA_O and QA.