4 DISCUSSION

Mineralogical, petrographic, and chemical analyses confirmed the presence of large amounts of sepiolite and palygorskite concentrated along the bedding planes of the Egyptian limestone studied. The presence of salts (NaCl and NaNO3) was also confirmed. Immersion in distilled water, as well as wetting/drying cycles using distilled water and ethylene glycol, demonstrated that this limestone suffered severe damage when exposed to these liquids. No damage was observed where efflorescent growth took place. Relative humidity cycling also produced spalling and loosening of scales. From the results of the thermomechanical analysis, the damage appears to be due to the expansion of the stone perpendicular to the bedding planes where the clay minerals are concentrated. After some wetting/drying cycles, the expansion was shown to be partly irreversible. In fact, after immersion in water, the stone samples suffered the same damage that had been observed in the Naga el-Deir stela (delamination and massive loss of material), damage that was induced by relative humidity cycling in an environmentally controlled chamber. These experimental results lead us to consider that the presence of clays is a major contributor to the damage observed in the limestone. The damage mechanism involves the fixation of water molecules to the clay faces and results in eventual swelling.

Two kinds of clay swelling processes have been described (Marshall 1949; van Olphen 1977; Madsen and Muller-Vonmoos 1989): (1) crystalline swelling (short-range particle interaction, according to van Olphen [1977]), which is due to hydration of the exchangeable cations of the clay; and (2) osmotic swelling (long-range particle interaction, according to van Olphen [1977]), which is due to the large difference in the ion concentration close to the clay surface and the pore water. This last type of swelling mostly depends on ionic concentration, type of exchangeable ion (counter-ion), pH of the pore water, and type of clay.

Crystalline swelling of clays is a common phenomenon in expandable clays such as smectite and vermiculite (Norrish 1954). Nevertheless, although sepiolite and paligorskite are normally considered to be nonexpandable, they have been reported to undergo minor crystalline swelling in the presence of polar liquids such as ethylene glycol (Jones and Galan 1988). However, the extent of this swelling is small, and the swelling does not explain the large expansion observed in the limestones studied. Swelling processes have been described in nonexpandable clays, such as illite (McEwan and Wilson 1980) and kaolinite (Morris and Shepperd 1982) and are reported to be due solely to electrostatic forces (osmotic swelling). Thus, it can be hypothesized that, in this case, osmotic swelling of sepiolite and palygorskite is ultimately responsible for the expansion of and subsequent damage to the Naga el-Deir Egyptian limestone when in contact with water. The negatively charged surface of the clay can adsorb polar liquids (water) as well as various ions present. In this case, the presence of NaCl and NaNO3 within the stone offers a supply of sodium ions to the solution formed when water (either as vapor or as liquid phase) enters the pore system of the limestone. The sodium ions, in turn, become hydrated and produce an initial swelling of single fibrous clay crystals (crystalline swelling).

Then, osmotic swelling (interparticle swelling) due to electrostatic repulsion forces between nearby sepiolite and palygorskite particles, which are close-packed in very well-defined layers, as shown by SEM analyses, can create sufficient pressure to damage the layered structure of the stone. Repeated wetting/drying cycles can create irreversible deformation of the stone structure and, eventually, the loss of stone flakes and scales. In fact, this damage is the same type observed in the samples and in the stones that are used for sculptural purposes.

In this process, relative humidity cycling can lead to water condensation and swelling of the clays, promoting the damage observed in the microclimatic chamber as well as in the museum samples. It should be noted that the presence of salts within the pore system of the stone should promote condensation of water at relative humidity values lower than 100%. In the case of NaCl and NaNO3, deliquescence takes place when RH is above 75% (Arnold 1981), thus enhancing the problem due to clay expansion even when the RH changes are not very high.

Additionally, the presence of salt mixtures (nitrates and chlorides) tends to reduce the equilibrium relative humidity (RHeq) of the salts (Price and Brimblecombe 1994). Under these conditions, liquid water can be present within the pore system of the stone at relative humidity values below 75%. Hence, salts in this context are not directly responsible for the large damage suffered by this stone (no salt crystallization damage was detected); nevertheless, they contribute indirectly to the damage by reducing the RHeq (Arnold 1981; Price and Brimblecombe 1994) and by providing ions (sodium) that promote the osmotic swelling of the clays.

These results explain many observations indicating that damage of clay-rich stones is enhanced in the presence of salts (Fookes and Poole 1981; McGreevy and Smith 1984). In conclusion, all the data presented confirm that the swelling of the clays, enhanced by the presence of salts, is responsible for the observed damage in this type of Egyptian limestone.

To minimize these effects, Charola et al. (1983) and Bradley and Middleton (1988) reported that the Egyptian limestones were treated with surface coatings such as wax, but these treatments were found to be ineffective in most cases. One possible approach to stabilizing the stone could involve an unconventional method, such as reducing the swelling capacity of the clay by replacing sodium ions with calcium and/or magnesium ions (less hydrated) through cation exchange methods. Many studies indicate that clays containing exchangeable calcium ions do not swell as much as sodium clays (Norrish 1954; Chatterji et al. 1979; van Olphen 1987; Sridharan and Satyamurty 1996). Due to the high ionic exchange capacity of the clays, it should be possible to replace sodium ions (supplied by the salts) with divalent cations and reduce clay swelling (Marshall 1949; Norrish 1954; McEwan and Wilson 1980; van Olphen 1987). In fact, lime washes have been used extensively to stabilize expansive clay-rich soils (Grim 1962; Basma and Tuncer 1991), since calcium from the Ca(OH)2 can replace other highly hydrated cations (i.e., Na or K) in the clay structure (or adsorbed on the clay surface), thus reducing both the crystalline and the osmotic swelling capacity of the clays.

Another possible unconventional conservation method for reducing the swelling capacity of the clays could involve treatment with surfactants. Surfactants have been reported to influence both the rheological and the swelling properties of clays (van Olphen 1977; Permien and Lagaly 1995) because the adsorbed surfactant affects the ion distribution on the clay surface. Surfactants also have been reported to reduce the swelling capacity of clays by forming hydrophobic coatings on the clay surfaces (Theng 1974). The practical use of surface-active compounds as antiswelling agents for clay-rich ornamental stones has been reported by Snethlage et al. (1995) and Wendler et al. (1996). Using a surfactant, Wendler et al. (1996) were able to reduce the swelling capacity of the clay-rich tuff used in the carved statues (Moai) on Easter Island.

Nevertheless, much further work needs to be done in both areas before conservation treatment of ornamental stone is attempted. A much less invasive conservation measure would be to impose rigorous environmental controls (temperature and relative humidity) on the limestone storage areas to minimize cyclic changes and reduce clay swelling.