One of the most obvious manifestations of the stress exerted by crystals growing in confined spaces is the damage done by salt in a remarkable range of environments, from seashores to cities to deserts , attacking the modern infrastructure as well as ancient monuments. Despite the ubiquity and seriousness of the problem, the details of the mechanism by which a growing crystal generates stress and cracking are incompletely understood. As a result, remedies to retard deterioration are based on empirical approaches and have proven, in some instances, to accelerate damage. In contrast, strategies grounded in a better fundamental understanding of the salt/water/pore wall interface, show promise of mitigating, or even eliminating, damage. The goal of the present work is to improve our understanding of the mechanisms by which salts cause damage to porous materials, and to develop a procedure for treating stone to protect against such damage. The work specifically involves use of a polymeric film to reduce the interfacial energy between salt and minerals surfaces, an idea originally suggested by the PI , which has shown considerable promise in preliminary experiments . The immediate goal of the work is to evaluate the durability of this type of treatment, to test it against a variety of relevant salts, and to optimize the application of the treatment so that field tests can be initiated (in a subsequent study).
Observations by earlier workers, discussed in ref. 4, indicate that when a crystal grows within the pores of a stone there is a liquid film between the salt crystal and the pore wall [5,6,7,8], which implies that a repulsive (disjoining) force exists between the surfaces . In the case of ice crystals, it can be shown  that the van der Waals interaction is repulsive, owing to the unusual circumstance that liquid water is denser than solid ice; this means that the Hamaker constant of the water lies between that of the ice and the mineral, resulting in repulsive forces . However, in the case of salts, the crystal should be attracted to the wall by van der Waals forces, so the repulsion must have some other origin. In a DOE-sponsored project, we are performing molecular dynamics simulation of the salt/solution/mineral interfacial region to identify the nature of this repulsion. Results to date indicate that the dominant forces are electrostatic effects related to the distribution of dissolved ions and the orientation of the water dipoles , which is consistent with the conclusions of other experimental and theoretical studies [12, 13]. As long as the liquid film exists between the crystal and the mineral surface, the crystal is in contact with a supersaturated solution, so it attempts to grow and exerts pressure on the pore wall. By eliminating this liquid film, and replacing it with a dry, low-energy interface, the adsorbed polymeric layer would eliminate this stress.
Salts in stone become concentrated owing to evaporation of the water or a change in temperature, and crystals may grow if supersaturation occurs. If a crystal nucleates in a supersaturated solution, its growth can be stopped by applying a pressure, p, to its surface. The magnitude of the required pressure is [9,14,15]
where Rg is the gas constant, T is the absolute temperature, Vm is the molar volume of the crystal, and K and K0 are the solubility products of the supersaturated and saturated solutions, respectively. It is the pore walls that must exert the pressure to prevent the growth of a crystal within a porous body. As explained in detail in ref. 4, tens of megapascals may be required to stop the growth in the presence of a modest supersaturation. Only part of the pressure in Eq. (1) is supplied by the pore wall; the rest is related to the surface energy and curvature of the crystal. When conditions of thermodynamic equilibrium prevail, the pressure on the wall is greatest in small pores in the presence of a high supersaturation [4,16]. In fact, it is unusual for stones to have pores small enough (< 100 nm) to generate high stresses, under equilibrium conditions. On the other hand, nonequilibrium conditions result during evaporation of pore liquid, when the solution retreats into the gap between an existing salt crystal and the pore wall, as shown in Figure 1. In that case, very high stresses can be generated by large crystals, because the supersaturation can only be reduced by growing toward the wall. This may be the primary mechanism by which salt does damage in the field [17, 18].
Thus, in large or small pores, it is the repulsive pressure between the salt and the pore wall that sets the upper bound on the stress that can be exerted. In either case, the damage can be avoided by eliminating the repulsion, so that the salt crystal comes into contact with the pore wall and stops growing. For example, if limestone is saturated with an aqueous solution of polyacrylic acid (neutralized with KOH), the polymer binds to the pore wall, because the deprotonated carboxyl groups bind to the calcium ions of the calcium carbonate crystal. Binding is expected to occur at a few sites along the polymer chain, leaving loops of the polymer chain free to interact with any salt crystals that approach the pore wall. As the salt crystal binds to the polymer, it stops growing and the development of pressure is avoided. The favorable interaction of sodium sulfate with calcite surfaces coated with PAA was demonstrated in ref. 3. The effectiveness of the treatment is demonstrated in Figure 2. The sample on the left is bare limestone, and it was destroyed within about one week of exposure to capillary rise of a 12 wt% solution of Na2SO4. The amount of PAA applied to the stone increases from left to right, and the two samples on the right (with coatings of PAA estimated to be 8-10 nm thick) are clearly much less damaged. In the present work, we studied whether the treatment is durable (that is, whether it desorbs), and what the optimal amount is for a given limestone.
A more quantitative measure of the effectiveness of the treatment is the warping test, in which a plate of stone is impregnated with sodium sulfate (by immersing in a solution and then drying), then glued to a plate of glass. When water wicks into the stone, it dissolves the anhydrate (thenardite), which creates a solution that is supersaturated with respect to the decahydrate (Na2SO4·10H2O, mirabilite) [19, 20]. The growth of mirabilite generates a high crystallization pressure that makes the stone expand; since it is attached to a non-expanding plate of glass, the bilayer warps, and the deflection is a measure of the pressure exerted by the salt in the pores. The results of such a test are illustrated in Figure 3, where the stress in the stone has been calculated from the measured deflection. If the limestone plate is untreated, the deflections are large and the stone is damaged, but if the stone was pretreated with PAA, the deflections are small and no damage is visible.
In the preliminary warping tests, the stress in the treated stones reached about half of the tensile strength, and subsequent measurements of the elastic modulus of the stone showed a decrease of about 10%. Therefore, the performance of the treatment, while quite impressive, could certainly be improved. Moreover, the design of the test can be improved by reducing the amount of salt, so that no damage occurs; in that case, the stress from crystallization can be accurately found from an elastic analysis of the deflection of the sample. Extensive evaluation of this test has shown that it permits quantitative evaluation of the crystallization pressure from the salt ; however, it was also demonstrated that the distribution of the salt in the stone is nonuniform, so the mathematical analysis is complicated. For that reason, we have abandoned that test in favor of dilatometry.
We have recently published reviews, partially supported by the present grant, describing the state of the art in salt crystallization theory and the experimental methods that are being used to improve our understanding of the underlying physics [22, 23].