PHYSICAL REVIEW E 88, 032404 (2013) Pore-scale dynamics of salt precipitation in drying porous media Mansoureh Norouzi Rad, 1 Nima Shokri, 1,* and Muhammad Sahimi 2 1 School of Chemical Engineering and Analytical Science, University of Manchester, Manchester M13 9PL, United Kingdom 2 Mork Family Department of Chemical Engineering and Materials Science, University of Southern California, Los Angeles, California 90089-1211, USA (Received 3 December 2012; published 9 September 2013) We study the pore-scale dynamics of salt precipitation in three-dimensional drying porous media, utilizing high resolution x-ray microtomography and scanning electron microscopy. Our results illustrate that the salt precipitation patterns in drying porous media are nonuniform, manifesting the influence of the spatial distribution of pore sizes on the dynamics of salt crystallization and formation of discrete efflorescence. Results reveal that during stage-1 evaporation from saline porous media, the salt precipitation rate initially increases which is followed by a constant precipitation rate. This non-linear behaviour is attributed to the preferential liquid vaporization and salt precipitation in finer pores located at the surface of the porous medium contributing in evaporation according to the pore sizes. We also show that, contrary to common practice, the macroscopic convection-diffusion equation cannot provide accurate predictions for the dynamics of salt precipitation, at least at the early stages, due to the microscale heterogeneity of evaporation sites at the surface that results in salt precipitation exclusively in the finer pores. DOI: 10.1103/PhysRevE.88.032404 PACS number(s): 68.08.−p, 47.56.+r, 47.53.+n, 89.75.Fb I. INTRODUCTION Understanding the physics of salt precipitation in porous media is of fundamental importance to many natural and industrial processes, such as preservation of pavement and historical monuments, mineral-fluid interactions, CO 2 seques- tration in rock, and the evaporation rates from porous media [1,2]. As water evaporates, the salt concentration in the pore space increases continuously until it exceeds the solubility limit, at which time it precipitates. The precipitation pattern modifies the morphology of the pore space and, consequently, influences flow and transport processes in it. Despite their high importance, particularly to the problem of drinking water for the world’s rapidly increasing population, direct pore-scale imaging and the study of salt precipitation and crystallization patterns in three-dimensional (3D) drying porous media are very rare, largely due to the imaging complexity of the dynamics of salt precipitation. Therefore, most of the previous studies were focused on the description of salt deposition patterns at the macroscale and were restricted mostly to two-dimensional (2D) imaging of evaporating surfaces [2,3]. We have employed x-ray microtomography to study the key effects of the pore sizes and the initial salt concentration on dynamics and patterns of salt precipitation in a drying porous medium. Detailed visualization of salt precipitation enables us to identify unambiguously the mechanism of discrete efflorescence on the evaporation surface as a result of the invasion of large pores by air and the subsequent capillary- induced liquid flow toward the fine pores on the surface, where water evaporation occurs preferentially. We establish that the pore-size heterogeneity at the surface of porous media results in heterogeneous distribution of the vaporization sites on the surface and the preferential efflorescence that occurs exclusively in the fine pores. In addition, we quantify the influence of the initial salt concentration on its deposition * E-mail address: nima.shokri@manchester.ac.uk rate in 3D drying porous media. Finally, we demonstrate that, contrary to common practice, the observed trends cannot be accurately modeled by the convection-diffusion equation (CDE), at least at the early stages. II. EXPERIMENTAL CONSIDERATIONS The drying experiments were carried out in cylindrical plastic columns, 35 mm in height and 11 mm in diameter, packed with sand grains with particle sizes ranging from 0.17 to 1.0 mm, with an average size of 0.58 mm, saturated with NaCl (reagent grade powder from EMD Chemicals, Inc., New Jersey, USA) solutions of 3.5 M (moles of NaCl/kg of water), 4 M, and 6 M. The drying of the sand columns was visualized by means of a HMXST x-ray microtomography system (at 70 kV and 140 μ ˚ A) with spatial and temporal resolutions of 0.021 mm and 30 min. The duration of each round of the experiments was nearly 24 h. Each 3D scan included 1500 horizontal cross sections. We used the standard machine learning technique of a support vector machine (SVM) [4] with a quadratic kernel to segment each cross section into solid, liquid, and air. The SVMs were trained using sequential minimal optimization [4] from MATLAB’s machine learning package. More details about the method are given elsewhere [4]. To distinguish the precipitated salt from the sand grains, each 2D cross section was compared to its corresponding image obtained at the beginning of the experiment, when there was no solid salt in the sample, with the difference indicating the precipitated salt. The resulting typical cross sections are illustrated in Figs. 1(a)–1(c). Figure 1(d) shows a typical example of the vertical growth of the precipitated salt above the sand surface. The spectrum of yellow to orange (light to medium gray) indicates the time of the scan such that the closer the color is to yellow (light gray), the later the column was scanned. Using this method, we quantify the dynamics of salt growth above the evaporation surface at various times and locations. 032404-1 1539-3755/2013/88(3)/032404(5) ©2013 American Physical Society