An experimental investigation into the salt-weathering susceptibility of building limestones

Within natural materials, dimension stones in monumental structures and artefacts are studied correlating mineralogic-petrographic and petrophysical characteristics to durability. The added cultural value encourage to study these materials to understand weathering phenomena and the response of each material to diverse environmental parameters, to preserve them (Török and Přikryl 2010).

Salt crystallization has a significant effect on the weathering of exposed ornamental and building stones (Amoroso and Fassina 1983; Siegesmund and Snethlage 2014). The significance of soluble salts as weathering agents derives from their ubiquitous occurrence in a wide range of environments (e.g., urban, desert, coastal, and rural; Goudie and Viles 1997; Smith et al. 2005), and their activity under wide ranges of relative humidity and temperature (Puehringer 1983; Sun and Zhang 2019). Therefore, proper conservation of stone artefacts relies on understanding the parameters involved in salt weathering, in particular the role of salt crystallization (Çelik and Aygün 2019).

Crystallization is the process common to all salt species inducing weathering in the host stone. The most common model considers a saline solution circulating inside the pore network, where salt precipitation begins when water starts to evaporate or temperature raises (Cardell et al. 2008; Castellazzi et al. 2013; Grementieri et al. 2017). This process occurs where the dynamic balance between evaporation rate and solution supply rate slows down or ceases. The evaporation rate depends on rock properties (permeability and pore size) and on environmental conditions (relative humidity, RH; temperature, T; Cardell et al. 2003; Grementieri et al. 2017). The supply rate depends on the properties of the solution (flux, surface tension, contact angle, viscosity) and of the rock core (permeability, pore size, and path length; Scherer 2004). If the evaporation rate is lower than the solution supply rate, then crystallization occurs at the material surface, causing the formation of efflorescence. If the evaporation rate is higher than the solution supply rate, then salts precipitate within the pore network, causing sub-efflorescence and mechanical damage.

Previous studies have focused primarily on crystallization pressure (Flatt 2002; Scherer 2004; Steiger 2005), thermal expansion (Ruedrich and Siegesmund 2007), hydraulic pressure (McMahon et al. 1992), wetting–drying pairs (Sun and Zhang 2019) and chemical weathering caused by the dissolution of minerals with weaker ionic strength than that of the saline solution (Cardell et al. 2003). These processes may all be valid and possibly co-occur, however, the most significant is the crystallization pressure. Crystal formation is favoured in larger pores, and once space is filled with a crystal, the residual solution in the smaller neighbouring capillary is affected by a curvature change due to a change decrease ? in saturation, and maximum pressure is exerted on pore walls (Steiger 2005). All the available models underline the importance of pore size and shape as a sink for salt contamination (Everett 1961; Fitzner and Snethlage 1982; Nicholson 2001; Flatt 2002; Benavente et al. 2004a, b; Angeli et al. 2008; Yu and Oguchi 2010).

To monitor the weathering induced by saline solutions, several petrophysical characteristics could be taken into account, including (i) pore size distribution (Ordóñez et al. 1997; Prikryl et al. 2003; Yu and Oguchi 2010; Akin and Özsan 2011; Di Benedetto et al. 2015), (ii) effective porosity (Yavuz and Topal 2007), (iii) water absorption by capillary uptake and total immersion (Dessandier 2000; Benavente et al. 2006), (iv) ultrasound velocity (Nicholson 2001, 2002; Sousa et al. 2005; Alonso et al. 2008; Ceryan et al. 2008), and (v) weight loss (Zedef et al. 2007).

Sulphates, chlorides, carbonates, and nitrates, combined with different cations (Na⁺, K⁺, Mg²⁺, Ca²⁺, and NH₄⁺), are the most common trigger for deterioration in dimension stones (Siegesmund and Snethlage 2014). Also deterioration caused by alums, nitrites, and oxalates have been reported (Arnold and Kueng 1985; Arnold and Zehnder 1990). Identification of the dissolved anions is crucial to interpret the origin of contamination: carbonates and sulphates can originate from rock-forming minerals, circulating water, dissolution of mortars, or pollution. Chlorides are present in coastal environments and organic matter and, rarely, as a result of the salt-storage-function of a building. Nitrates are typical of organic-matter-rich water and phosphates of water containing fertilizers or bone matter. The presence of oxalates indicates the reaction of minerals with organic matter, usually with artificial products for stone conservation treatment. Unlike anions, the identification of cations is not fundamentally important to the interpretation of the origin of weathering phenomena.

The susceptibility of diverse lithotypes to salt weathering has been studied through different, more or less standardized, ageing procedures (Lubelli et al. 2018). Usually, to reduce the number of variables (e.g., concentration of solutes, chemical interactions, environmental variables; Angeli et al. 2010), contamination by a single saline solution is considered. However, water in the pore network contains more than one species of solute, that consequently interact (Cardell et al. 2008). Some studies have considered several solute-species mixtures, but no reliable model was derived because of the numerous concurring variables (Arnold 1981; Robinson and Williams 2000).

Based on the literature review carried out by the RILEM technical committee 271 ASC (Lubelli et al. 2018), the continuous partial immersion test method (Benavente et al. 2001) was used to assess the behaviour of a selection of saline solutions. Saline solutions (Na₂SO₄, MgSO₄, Na₂CO₃, NaCl, and KNO₃) were applied to several well-characterized lithotypes; then, diffusion and physical behaviour (hygroscopicity and potential hydrate formation) were analysed and compared.

The study was carried out on a selection of rocks with homogeneous mineralogical composition but differing in texture and structure. The analysed lithotypes are all limestones employed as building or ornamental stones, and are characterized by different structural features due to their petrogenesis (Scrivano et al. 2018). Limestones are widespread as ornamental stones, but their mineralogical composition (mainly CaCO₃) makes them liable to weathering processes, being calcite highly soluble in acidic waters (Schiro et al. 2012).

Five different saline solutions were designed to identify whether salt precipitation induces chemical interactions within the analysed stones and to evaluate its importance with respect to mechanical stress induced by crystallization. Na₂SO₄, MgSO₄, Na₂CO₃, NaCl, and KNO₃ saline solutions were selected based on the mineral phases observed in efflorescence in real case studies (Arnold and Zehnder 1991; Siegesmund and Snethlage 2014; Siedel 2018). Moreover, Na₂SO₄ is normally used in determining stone durability to salt decay for its presence in acid rains and its highly destructive impact on stone structure (UNI EN 12, 370: 2001; UNI EN 13, 919: 2002; McMahon et al. 1992; Tsui et al. 2003; Ruiz-Agudo et al. 2007; Angeli et al. 2008, 2010; Diaz Gonçalves and Brito 2014). Ruiz-Agudo et al. 2007; Akin and Özsan 2011 report also the use of MgSO₄ in testing stone resistance. In this study, this saline solution was used because it forms minerals with different hydration state (epsomite: MgSO₄^.7H₂O; hexahydrate: MgSO₄^.6H₂O; pentahydrate: MgSO₄^.5H₂O; and starkeyite: MgSO₄^.4H₂O). The same behaviour characterizes also Na₂SO₄ (mirabilite: Na₂SO₄^.10H₂O; thenardite: Na₂SO₄) and Na₂CO₃ (natron: Na₂CO₃^.10H₂O; thermonatrite: Na₂CO₃^.H₂O) solutions, evidencing the influence of environmental conditions in the salt weathering process. Low solubility characterizes nitrates, that easily occur as efflorescence (Siegesmund and Snethlage 2014), and KNO₃ was selected to represent the chemical activity of nitrate compounds. Finally, also a NaCl solution was used, as in material testing procedures (UNI EN 14, 147:2003), to recreate sea spray effect on stone artefacts.

The objective of the study was to investigate the weathering mechanisms from different saline solutions, and to evaluate the influence of mineralogical and textural features on the degree of salt weathering, i.e. the control of petrophysical features on the durability of the analysed ornamental and dimension stone. Moreover, as each of the tested saline solutions represents specific environmental agents, the response of the lithotypes helps the selection of the dimension stones for diverse uses. For example, a pronounced susceptibility of a lithotype to NaCl advises against its use in a coastal environment.

Moreover, the study aims at deciphering the interaction between the solutes and the intrinsic textural and structural weaknesses of the chosen lithotypes. The identification of thisbehavioural pattern could help establishing protocols useful to define conservation strategies. For example, recognizing the features influencing susceptibility to a certain type of solution can be crucial in choosing the appropriate repair stone (% of open pores) or type of mortar to avoid the establishment of new fluid circulation equilibrium or chemical interaction (Lopez-Arce et al. 2009).

Four sedimentary carbonatic lithotypes differing in fabric, commonly used as ornamental stones (Scrivano et al. 2018), were selected as test materials (Table 1). To evaluate accurately correlations between pore network characteristics, and rock weathering susceptibility to each type of saline solution, the selection was based on homogeneous mineralogy and chemical composition, and on different pore network characteristics. The selected stones are the following:

BA and RV are mainly made up of compact micrite nodules or clasts cemented by veins of sparry calcite (BA), and micrite + phyllosilicate (RV). The grain size and textures control porosities and properties of the two lithotypes (Scrivano et al. 2018). Also, FGV and CGV have different grain size, resulting in different pore size distributions (Scrivano et al. 2019).

Starting from the mercury-intrusion-porosimetry (MIP) pore range distribution data already discussed in Scrivano et al. 2018 and summarized in Table 1, the Salt Susceptibility Index (SSI; Yu and Oguchi 2010) was calculated. This parameter links the pore size distribution of the lithotype with its liability to salt weathering. For SSI values lower than 4, the lithotype is supposed to have salt weathering resistance, while for values higher than 4 the material should be subject to salt decay.

All the studied lithotypes were employed as building materials in a different part of Italy and with different functions. In particular, FGV and CGV were adopted for stairs, walls and other architectonic uses, as well as for sculpture. Under the different condition of use, weathering patterns are developed; those linked to saline solutions are mainly registered for outdoor staircases and wall basements that can suffer even severe material loss. RV and BA are primarily employed as cladding or paving in sheltered locations, however, salt induced weathering can easily become the first origin of differential erosion.

The precipitation of salt crystals occurred inside and outside the pore network, depending on the physical features of the pore network or on the solutions. Precipitation is associated inside the stone with modification of petrophysical parameters, and outside with evidence of salt crusts at the macroscale. Following salt precipitation, also deterioration by cracks, efflorescence and sub-efflorescence, chemical reactions, grain loss, and rounding developed. Cracks develop in the RV samples mainly along the micrite + phyllosilicate veins. This texture, characterized by higher porosity than the nodules and by highly reactive surface, due to the fine calcite diameter, is the weakest element of the lithotype.

Grain loss and rounding characterize mainly the two bioclastic VS limestones. Efflorescence, sub-efflorescence and chemical reactions affect at different extent all the lithotypes. The chosen solution are generally weakly acidic, fact that explains the patterns of secondary precipitation in every sample. Conversely, Na₂CO₃ having pH around 10, have a basic behaviour, more effective in dissolving mineral phases.

Under SEM, several distinct efflorescence morphologies for the different saline species were revealed. Herein, some of the saline solutions, under alternating hydration–dehydration, and variable temperature–humidity rate during the testing procedure, allowed crystals to precipitate and subsequently modify. Furthermore, the diverse patterns of crystallization depend on the environmental conditions: euhedral crystals or compact crusts represent precipitation from solution in a damp environment, whereas skeletal forms and cracks appear to be the most diffuse crystallization habits developed during evaporation (Rodriguez-Navarro et al. 2000; Charola et al. 2007, 2017; Lopez-Arce et al. 2009; Lubelli et al. 2018). The development of overlapping series of different morphologies can be interpreted as signs of a stepwise crystallization influenced by the environmental conditions.

The most porous lithotypes, FGV and CGV, presented the highest SSI values and, in fact, were the most liable to material loss and modification of their hydric properties. In particular, the stones showed high weathering rates when exposed to Na₂SO₄, Na₂CO₃, and MgSO₄ solutions. The modification of the micro-pore network, i.e. increase in capillary rise during the first minutes of soaking, and hygroscopic sorption variations, took place during exposure to all saline solution, and was confirmed by relative mass variation (Table 4). Moreover, as the complete capillary absorption dataset includes a decrease of the total content of water absorbed by the specimens, the widening of the larger pores, can be inferred as related with mechanical stresses and calcite dissolution. The latter is assumed after the observation of secondary calcite re-precipitation for all the saline solutions. Moreover, the formation of tabular crystals of calcium sulphate, probably gypsum, on the specimens treated with sulphate solutions confirms that chemical reactions occurred at the rock surface.

Despite its low SSI parameter, RV was affected by Na₂SO₄ and Na₂CO₃ solutions. Visual inspection and SEM images showed that the most conspicuous crystallization occurred at the micrite–phyllosilicates veins, sparing the compact nodules, with consequent differential behaviour between the two zones with diverse porosity (Benavente et al. 2007). The differential durability of nodules and veins place this lithotype in the category of dual-porosity materials. Therefore, the SSI value rather reflects the hardness of the nodules, whereas the veins are more weatherable. In RV the pore network within the micrite-phyllosilicates veins was modified by pore enlargement (Scrivano et al. 2018) especially after Na₂SO₄, MgSO₄ and Na₂CO₃ salt contamination. The joints interested by the mechanical and chemical stresses induced by the saline solutions explains the small rate of decay, compared to CGV and FGV.

The compactness of BA resulted in high resistance to salt decay, as demonstrated by the slight modification of hydric properties and the virtual lack of mass loss, consistent with the values of SSI.

The main difference in the behaviour of RV and BA resides in the veins. In fact, while the recrystallized veins of BA are as compact as the breccia clasts, the micrite + phyllosilicates joints cementing the micritic nodules of RV are characterized by diffuse inter-granular + intra-granular pores resulting in a widespread reactive surface. Moreover, the different behaviour of nodules and veins in the lithotype led to differential weathering and breaking, especially in RV + Na₂CO₃.

Overall, the SSI proved efficient in predicting the behaviour of homogenous lithotypes (i.e. BA, FGV and CGV), however, this index does not consider the chemical activity of the different saline solutions, thus overrating the susceptibility of FGV and CGV to MgSO_4, NaCl and KNO₃.

Generally, the most compact fabric (BA and RV nodules) account for a greater durability to salt action, while high open porosity values (FGV and CGV) and abundance of micrite (RV) caused higher susceptibility to decay; in particular, FGV, CGV and RV micrite + phyllosilicate veins showed low durability for sulphate and carbonate contamination. Therefore, caution must be adopted when using these stones in outdoor for the presence of acid rains, and in the presence of lime-based bedding mortars and high capillary-rise rate.

The experiments on the four carbonatic sedimentary stones allow the following conclusions: