Monument Future

References

ASTM C1515 – 2014 Standard Guide for Cleaning of Exterior Dimension Stone, Vertical and Horizontal Surfaces, New or Existing.

ASTM E1857−97 (Reapproved 2013) Standard Guide forSelection of Cleaning Techniques for Masonry, Concrete, and Stucco Surfaces.

Avagliano R., Guidetto V., Martini G., Randone E., Riccardi M. P. 2013. Artwork analisis by optical surface profilometry. Conference 8790: Optics for Arts, Architecture, and Archaeology IV 15–16 May 2013 Proc. of SPIE Vol. 8790, pp. 1–4.

BS8221-1:2012. Code of practice for cleaning and surface repair of buildings – Part 1: Cleaning of natural stone, brick, terracotta and concrete

Carvalhão M., Dionísio A. 2015. Evaluation of mechanical soft-abrasive blasting and chemical cleaning methods on alkyd-paint graffiti made on calcareous stones. J. Cult. Herit., Volume 16, pp 579–590.

Doehne, E. F., Price, C. A. 2011. Stone conservation: An Overview of Current Research; The Getty Conservation Institute: Los Angeles, CA, USA.

Demoulin, T., Girardet, F., Wangler, T. P.; Scherer, G. W., Flatt, R. J. 2016. On-site monitoring for better selection of stone repairs: A case study. Herit. Sci. 4, 38.

Gulotta, D., Toniolo, L. 2019. Conservation of the Built Heritage: Pilot Site, Approach to Design a Sustainable Process, J. Heritage, 2, 797–812; doi:10.3390/heritage2010052.

ICOMOS. 2008. Illustrated Glossary on Stone Deterioration Patterns; International Council on Monuments and Sites: Paris, France.

Mecchi A. M., Poli T., Realini M., Sansonetti A. 2008. A proposal for common approach in chosing tests for the protocol evaluation of cleaning methods. In proceedings 11th international congress on deterioration and conservation of stone, volume 1, pp 425–432.

Perez-Monserrat, E. M., Varas, M. J., Fort, R.; de Buergo, M. A. 2011. Assessment of different methods for cleaning the limestone façades of the former Workers Hospital of Madrid, Spain. Stud. Conserv, 56, 298–313.

Pozo-Antonio J. S., Ramil A., Rivas T., López A. J., Fiorucci M. P. 2016. Effectiveness of chemical, mechanical and laser cleaning methods of sulphated black crusts developed on granite, Construction and Building Materials 112 682–69.

Pozo-Antonio J. S., Rivas T., López A. J., Fiorucci M. P., Ramil A., 2016, Effectiveness of granite cleaning procedures in cultural heritage: A Review. Sci. Total Environ., Volume 57, pp 1017–1028.

Revez, M. J.; Delgado Rodrigues, J. 2016. Incompatibility risk assessment procedure for the cleaning of built heritage. J. Cult. Herit. 18, 219–228.

Schroettner H., Schmied M., Schere S. 2006. Comparison of 3D Surface Reconstruction Data from Certified Depth Standards Obtained by SEM and an Infinite Focus Measurement Machine (IFM). Microchim Acta, Volume 155, Issue 1–2, pp 279–284.

Turk, J., Mauko Pranjić, A., Hursthouse, A., Turner, R., Hughes, J. J. 2018. Decision support criteria and the development of a decision support tool for the selection of conservation materials for the built cultural heritage. J. Cult. Herit.

UNI EN 14157:2017. Natural stone test methods – Determination of the abrasion resistance.

275

FREEZE-THAW WEATHERING: DIGGING DEEPER IN THE TEMPERATURE AND LENGTH EVOLUTION OF NATURAL STONES

Maxim Deprez¹, Geert De Schutter², Veerle Cnudde^1,3, Tim De Kock^1,4

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 PProGRess, Department of Geology, Ghent University, Krijgslaan 281 S8, 9000 Ghent, Belgium

2 Magnel Laboratory for Concrete Research, Department of Structural Engineering and Building Materials, Ghent University, Technologiepark 904, 9052 Zwijnaarde, Belgium

3 Environmental Hydrogeology group, Department of Earth Sciences, Utrecht University, Princetonlaan 8a, 3584 CB Utrecht, The Netherlands

4 Antwerp Cultural Heritage Sciences (ARCHES), University of Antwerp, Blindestraat 9, 2000 Antwerp, Belgium

Introduction

Freeze-thaw (FT) weathering is an important degradation process for porous building materials. The FT susceptibility of a building material is often assessed by subjecting it to repetitive FT cycles in a climatic chamber and investigating the consequent structural changes (e. g. Martinez-Martinez et al., 2013; Nicholson and Nicholson, 2000). These investigations provide a relative freeze-thaw susceptibility, but very limited information on the actual pore-scale mechanisms causing damage. To understand the impact of FT cycles on a certain material, it is necessary to investigate proxies of such pore-scale processes. One way to do so, is by simultaneous measurement of crystallization proxies, such as e. g. temperature and strain (Prick, 1995; Ruedrich et al., 2011). The temperature indicates the onset of crystallization, as the latent heat of ice crystallization sets the temperature of the freezing system at an equilibrium, i. e. 0 °C for macroscopic ice. The length of a sample is mostly determined by its thermal expansion coefficient. However, when crystallization occurs, positive and negative stress can be generated in the pore space, which will be reflected in the strain pattern. Once stress is large enough to overcome the strength of the sample, residual strain, in other words damage, can be noticed from these strain measurements.

Such observations have previously shown that every material has a specific critical saturation degree above which the material will exhibit residual strain (Fagerlund, 1977; Prick, 1997). Since the damaging agent is actually the ice, the crystal content is also of major importance (Coussy and Monteiro, 2008). The pore size in which ice is able to form, largely depends on the temperature (Scherer, 1999). Therefore, the quantity of ice formed is related to the initial amount of water, the pore characteristics and the temperature.

The goal of this research is to estimate the in situ ice content of two natural building stones with very different pore characteristics during freezing. Therefore, high resolution temperature and length change data was acquired during FT cycles in two different experimental setups with constant degrees of water saturation. The analysis and eventual estimation of the ice content is based on the occurrence time of specific events.

276Materials and Methods
Bentheim sandstone & Savonnières limestone

The German Bentheim sandstone (B) is a well-sorted quartz arenite from the Lower Cretaceous (Valingian) and quarried near Bad Bentheim at the Dutch-German border. The stone contains over 90 % quartz and other minerals found are feldspars, some iron-oxides and rarely some clay minerals. On average, the sandstone has a porosity between 0.20 and 0.25 and the sizes of the pore throaths are narrowly distributed around 10 to 50 µm (Peksa et al., 2015). It has been used as a building stone in Germany, the Netherlands, Belgium and Denmark (Dubelaar and Nijland, 2015).

The French Savonnières oolithic limestone (S) is found in the Upper Jurassic (Tithonian) deposits of the eastern Paris Basin. It mainly consists of ooids, shell fragments and pellets, bound together by a sparite cement. Many of the nuclei of the ooids are dissolved and denominated as vacuolar ooids, which leads to a high secondary porosity. The total porosity is between 0.3 and 0.4 and the multimodal pore-size distribution can be subdivided into four groups with characteristic radii of 87, 12, 0.7 and 0.1 µm (Roels et al., 2001). This stone is commenly used as building and replacement stone in France, Belgium, Germany and the Netherlands (Fronteau et al., 2010; Graue et al., 2011; De Kock et al., 2013; Quist et al., 2013).

Temperature measurements

Of each stone, two cubes of 50 × 50 × 50 mm were prepared for a FT weathering experiment with a fixed degree of water saturation. After determining the porosity according to EN 1936 (2006), a hole of 2 mm diameter and 25 mm deep was drilled in each cube to monitor the inner temperature. Of each kind of stone, one stone was saturated 20 % and one 60 % by drying until appropriate weight was reached. Afterwards, the samples were wrapped in aluminum foil to ensure a uniform temperature distribution around the sample and with plastic tape to prevent evaporation. Two K-type thermocouples (Picolog) were attached to each stone through a small hole in the seal, with one at the top surface and one inside the sample. The holes were then filled with silicone. 20 FT cycli (EN 12371: 2010) were subsequently performed in a climatic chamber (Weiss WKL 34/40). The temperature of the stones and the chamber were registred every second.

Combined temperature and length measurements

Six cores with a height of 38 mm and 20 mm diameter were drilled for each stone. Their porosity was determined using EN 1936 (2006). Also in these cylinders, a hole of 2 mm diameter and 10 mm deep was drilled to attach a thermocouple to the centre. For each kind of stone, all six cylinders obtained a different saturation degree going from 0 to 100 % saturated in steps of 20 %. The cylinders were sealed with aluminum foil and tape and one thermocouple was inserted into the sample, while another was attached to the top surface. The cylinders were placed in an invar holder while LVDT sensors (Geotron) were inserted through a hole in the holder and placed onto the top surface of the cylinder, which was partially strapped from its seal. Due to the small size of the samples and to limit the gathered data quantity, a shortened version of the FT cycle proposed in EN 12371 (2010) was made for this experiment with a total duration of 6 hours. To start, the temperature was set at 10 °C for half an hour. Then the temperature was decreased to –15 °C over 1.5 hours where it remained constant for half an hour. After this, the temperature was raised again to 10 °C over 1.5 hours and continued at this temperature for the next 2 hours. Three of these cycles were applied on each drilled core, with a sampling interval for temperature and length of 1 second.

Results and Discussion
Initial ice crystallization

The evolution of the sample temperature was examined to estimate the ice content evolution (Fig. 1). After a period of cooling, the temperature reaches negative values. Crystallization expresses itself by a sudden temperature rise towards 0 °C, which is caused by latent heat release. As long as sufficient 277heat is produced, the temperature remains at 0 °C. This phenomenon is also referred to as the zero curtain (Hall, 2007). After this, the temperature decreases to –10 °C and it follows the temperature of the chamber until the thawing starts. There, the temperature first remains at 0 °C until all ice has melted and then rises to 18 °C.

First, an estimation of the ice that initially forms during and immediately after nucleation, is made. Since the temperature at onset of nucleation is below zero, a certain mass of ice forms while the temperature increases to the zero curtain. Plotting the measured maximal undercooling temperature (T_u) and the duration of the subsequent zero curtain (t_zc) (Fig. 1) shows a linear correlation. From the trendline of the 20 cycles, the duration of the zero curtain when nucleation would occur at 0 °C (t₀) can be estimated from the intersection with the time axis (Fig. 2).

Figure 1: Temperature evolution inside sample B60 during one of the 20 FT cycles. The temperature of undercooling before nucleation (T_u) and the duration of the zero curtain (t_zc) are indicated.

Figure 2: T_u plotted against t_zc shows a linear relationship for each sample. The letter in the name indicates the kind of stone, while the numbers indicate the saturation degree. The intersection of the calculated trendline and the t_zc-axis equals t₀ or the time necessary to crystallize all the water if nucleation would have occurred at 0 °C.

Assuming that most of the present water is frozen at the end of the zero curtain, the mass of ice crystals formed at nucleation (m_ic) can then be estimated from the difference between this t₀ and t_zc.

m_ic = (1 − t_zc/t₀)m_w

With m_w the initial water mass. When this initial ice mass is plotted against the undercooling temperature (Fig. 3), it is noticed that the linear relations align per saturation degree. This means that the amount of initial formed ice could be more dependent on the saturation degree, the undercooling temperature and the cooling rate rather than the total mass of water. No trend was observed in the T_u over the different cycles.

Figure 3: From the temperature of undercooling at nucleation (T_u), it is possible to estimate the mass of ice that has formed at the time of nucleation (m_ic). Sample and saturation degree are again indicated by the letter-number combination.

The total duration of crystallization

The length and temperature were monitored simultaneously during three consecutive FT cycles. If the output is plotted over time (Fig. 4), the temperature evolves similarly as previously described. The length evolution is generally quite different for both stones, however, important is that highly saturated samples of both stones reach a maximum in expansion later than the end of the zero curtain. Therefore, crystallization is clearly still ongoing after t_zc, and the end of the expansion (t_t) could give a better estimate of the ice content evolution. This maximum expansion is only seen in 278the 80 and 100 % saturated B-samples and the 60, 80 and 100 % saturated S-samples.

Figure 4: The temperature (T) evolution in grey and length (dL) evolution in black of an 80 % saturated B-sample during a shortened FT cycle. The length is expressed as the initial sample length (38 mm) minus the sample length at a certain point during the cycle. A clear expansion phase is seen near the end of the zero curtain.

When plotting t_t against T_u, the linear relation discussed earlier is again noticed (Fig. 5). However, it is difficult to derive accurate trendlines, and hence initial ice contents, from only three measurements per sample. Providing we are able to calculate the m_ic and assuming that the crystallization rate is constant, the ratios between the time of events, such as the end of the zero curtain (t_zc) or the initiation of expansion (t_ex) and the total crystallization time (t_t) could provide insights on the crystal content at those events (Fig. 6).

Figure 5: The correlation between the ending time of the expansion phase (t_t) and the undercooling prior to nucleation (T_u) seems also linear from the three data points gathered per sample. This limited amount of data impedes an accurate estimation of the ice mass.

Both expanding B-samples reach their maximal expansion at an inner temperature (T_min) of –4 °C (Fig. 6), while the maximal expansion of the three S-samples occurs at temperatures that vary between –5.6 and –4.5 °C. The relation between the temperature (T) and the pore radius (r_p) in which crystals can form is given by (Scherer, 1999):

r_p – δ = 2γCL/((T – Tm)∆Sfv)

with δ the size of a small water film between the crystal and the pore wall (0.9 nm), γ_CL the specific energy of the crystal/liquid interface (0.04 J/m²), T_m the equilibrium melting temperature (0 °C) and ∆S_fv the molar entropy of fusion (1.2 J.cm^-3.K^-1). Filling in T_min results in pores of 18 nm radius for B and between 15 and 12 nm radius for S. These pore sizes correlate to the smallest pores found in pore-size distributions of both stones (Roels et al., 2001; Zhang et al., 2018).

Figure 6: Starting at the time of nucleation (t = 0) the temperature (T) is drawn in full lines and the length change (dL) in dashed lines. The time at which expansion starts is t_ex and expansion ends is t_t. The T_min corresponding to t_t, can be read from the simultaneous T-curve and for both 80 and 100 % saturated B-samples, this is –4 °C. This is the case for all three cycles.

Water/ice content during 20 FT cycles

The total crystallization time (t_t) for the 20 FT cycles can be based on T_min, which is –4 °C and –5 °C for B and S respectively. This is plotted against the minimal undercooling and a corrected t₀ (t_0c) can be derived. Using m_ic=(1–t_t/t_0c)m_w then again leads to a new estimation of the ice content formed directly after nucleation (Fig. 7).

Figure 7: By using t_t instead of t_zc, a corrected estimation of m_ic can be made. The resulting correlation displayed here is however slightly less linear than when the t_zc was used. This is probably due to the assumption that the water is only fully crystallized at a fixed T_min. This assumption could be false, with all the water already crystallized at higher or lower temperatures.

The ice that crystallizes during the zero curtain all forms in the pores larger than 0.1 µm (macro). Ice will only grow into smaller pores (micro) when the temperature decreases following the r^p_-T relation 279previously described. For a 100 % saturated sample, the volume fraction of a certain pore radius also represents the volume fraction of water (S_L) inside those pores, which enables us to produce a S_L-T relation (Fig. 8) (Coussy and Monteiro, 2008). For B, the temperature decreases rapidly if 90 % of the water is crystallized. Hence, 10 % of the water resides in the micropores. Similarly for S at 100 % saturation, the temperature decreases abruptly when 85 % of the water is crystallized, while 15 % of the pores has a radius smaller than 0.1 µm.

Figure 8: The liquid water content (S_L) and the temperature (T) curve for 100 % saturated S (full line) and B (dashed line) derived from the temperature – pore size relation. The arrows indicate the liquid saturations at which the temperature decreases rapidly.

The mass of ice (m_zc) formed at time t_zc can be estimated by: m_zc = m_ic+(m_w–m_ic)(t_zc/t_t). During t_t, the ice forms in the macropores while after, the temperature decreases and ice forms in the micropores. Hence, the remaining mass of water (m_{w mi}) should reside in the micropores and thus: m_zc = m_w–m_{w mi}. The investigated samples are only 20 and 60 % saturated and, in equilibrium conditions, the capillary forces cause the micropores to be mostly filled with water prior to freezing. Therefore, it is expected that m_{w mi}/m_{w 100}, where m_{w 100} is the mass of water when the sample is 100 % saturated, approximates the percentages of micropores derived from the S_L-T relation. For the 20 and 60 % saturated S-samples, the average m_{w mi}/m_{w 100} equals 6.6 % and 8.6 % respectively. These values for the 20 % and 60 % saturated B-samples are respectively 5.4 % and 8.1 %. The discrepancies with the theoretically derived values for S (15 %) and B (10 %) can be explained firstly by cryosuction of water in the micropores towards ice crystals forming in the macropores. Stone S has a multimodal pore-size distribution and tapered micropores, which both stimulate the cryosuction process. Therefore, the obtained mass of water in the micropores is well below 15 %. Secondly, during nucleation, water that was residing micropores could have been used to build up m_ic since T_u is negative.

Although these results are very promising, the approach relies on the assumption that the crystallization rate is constant, which might be untrue. However, the linear relation between both t_zc and t_t and T_u, suggests a constant rate in most parts of the freezing stage. The approach presented here should nevertheless be compared to results from an established technique and should be tested on other porous materials.