Salt crystallization and deliquescence triggered by humidity cycles in nanopores

TL;DR

This study investigates how salt solutions inside nanoporous matrices respond to humidity cycles, revealing that crystallization and deliquescence occur at much lower RH than bulk and dominate the hysteresis of sorption isotherms. Using mesoporous silica (poSi) and anodic aluminum oxide (AAO) with pore sizes spanning <4 nm to ~20 nm, and in-pore NaCl, they measure water sorption with white light interferometry and deduce pore filling via the optical path length $\mathcal{L} = 2 n H_p$ and the Kelvin radius $r_K = - 2 \sigma_w / \Delta P_c$; the vapor-liquid equilibrium follows $-\ln \mathcal{H} = r_ extell/r_K + \alpha_0 \phi S(m)$. They show deliquescence and crystallization RH are nearly independent of the initial salt content $m_i$, and argue this is set by the in-pore concentration $m$ through osmotic pressure and confinement, with $\Delta P_c = -2 \sigma_w / r_K$ and $\Pi(m) = \nu \phi k_B T \rho_w m$. A modified confined CNT model predicts that crystallization requires a supersaturation $S$ such that $r^* \le r_p$, but importantly crystallization data align with kinetic nucleation limits (no steric hindrance), while deliquescence corresponds to an unstable three-phase equilibrium; the required bulk supersaturation is found to be $S_\infty = 2.2 \pm 0.2$. Overall, the results establish that confinement alters the balance of phase transitions, reversing the classic sorption isotherm branches and enabling large metastable salt solutions in nanopores, with implications for salt weathering, membranes, and humidity-responsive materials.

Abstract

We study the response of materials with nanoscale pores containing sodium chloride solutions, to cycles of relative humidity (RH). Compared to pure fluids, we show that these sorption isotherms display much wider hysteresis, with a shape determined by salt crystallization and deliquescence rather than capillary condensation and Kelvin evaporation. Both deliquescence and crystallization are significantly shifted compared to the bulk and occur at unusually low RH. We systematically analyze the effect of pore size and salt amount, and rationalize our findings using confined thermodynamics, osmotic effects and classical nucleation theory.

Figures (5)

• Figure 1: (a) Cycles of relative humidity ($\mathcal{H} = p / p_\mathrm{sat}$) trigger crystallization and deliquescence of salt solutions upon $\mathcal{H}$ decrease or increase, respectively. (b) We study these phase transitions in mesoporous samples made of silica (poSi) or alumina (AAO); the micrographs are scanning electron microscope observations. (c) With white light interferometry (WLI), we extract from periodic modulations in reflectance spectra the optical path length ($\mathcal{L} = 2 n H_\mathrm{p}$), which varies as a function of water content in the pores.
• Figure 2: Water sorption isotherms measured with WLI on an AAO sample with (black, $m_\mathrm{i} = \qty{3.2}{\mol\per\kg}$) and without (grey, $m_\mathrm{i} = 0$) salt in the pores. The direction of the symbols indicate increasing ($\triangle$) or decreasing ($\triangledown$) $\mathcal{H}$. Sketches represent the inferred status of the pore space with the same color coding as in Fig. \ref{['fig:Intro']}, and the arrows point to the crystallization and deliquescence transitions.
• Figure 3: Effect of initial concentration ($m_\mathrm{i}$) on crystallization and deliquescence RH for two samples with different pore sizes. Uncertainties are on the order of symbol sizes or smaller. The dashed blue line corresponds to bulk crystal-solution equilibrium for NaCl ($\mathcal{H}_0 = 0.753$). Continuous brown and blue lines correspond to data average for crystallization and deliquescence respectively, surrounded by shaded areas which represent the scatter of the data.
• Figure 4: Crystallization and deliquescence RH as a function of pore size. Triangle symbols represent experimental data. Uncertainties in $\mathcal{H}$ correspond to the observed scatter across experiments at different $m_\mathrm{i}$ (shaded areas in Fig. \ref{['fig:ConcentrationEffect']}). Uncertainties in $r_\mathrm{K}$ correspond to the width of the pore size distribution as estimated from sorption isotherms Note1. The dashed blue line shows bulk deliquescence. The dotted black line represents the confined equilibrium of pure water as predicted by the Kelvin-Laplace equation. The light gray area corresponds to our theoretical prediction for sterically limited nucleation; the extension of this zone originates from the uncertainty in $\sigma_\mathrm{c\ell}$. The dark gray area corresponds to kinetically limited nucleation with a bulk supersaturation, $S_\infty = 2.2 \pm 0.2$. The inset displays the same data with transformed axes reflecting the relation predicted by Eq. (\ref{['eq:LiquidVaporEquilibrium']}).
• Figure 5: (a) In-pore growth of a spherical nucleus, (b) Corresponding Gibbs free energy profile from CNT (Eq. \ref{['eq:CNT_G_dimensionless']}). (c) Situation where $r=r_\mathrm{p}=r^\ast$. (d) Three-phase equilibrium described in Ref. Talreja-Muthreja2022.