Membrane distillation (MD) is a thermal separation process, in which a porous hydrophobic membrane separates components in a homogeneous liquid mixture based on their volatility. The driving force in MD is based on the difference in vapor pressure of the components induced by a temperature gradient between the two liquid phases separated by a membrane. Mass transfer only occurs in the vapor phase through the membrane pores from a warmer feed inflow to a colder permeate discharge. Therefore, the primary prerequisite for MD operation is that the liquid phase must be avoided to penetrate through the dry pores of the membrane (pore wetting). When pore wetting takes place, it will at least impair product quality or at most will incapacitate the process. Thus, developments in MD are bounded by the wetting prevention. Wettability of a membrane depends on the membrane and liquid feed properties as well as operating conditions. ^Wetting occurs when the transmembrane hydraulic pressure surpasses the liquid entry pressure (LEP) of a membrane. LEP depends on the maximum membrane pore size and the hydrophobicity of the membrane material (Franken et al. 1987). Hydrophobicity of a membrane is dictated by the membrane surface chemistry (SC) and surface geometrical structure (SGS). SC and SGS are governed by surface free energy and surface roughness of the membrane, respectively whereas both contribute to the water contact angle (WCA) of the membrane. To control wetting incident in MD process, three approaches were examined: (i) modification of membrane surface roughness (ii) modification of membrane surface free energy (iii) prevention of adsorption of the wetting agent at the membrane surface.
The surface geometry of the membrane was modified by dip coating of hydrophobic SiO2 nanoparticles onto the surface of a polypropylene membrane to increase the membrane surface roughness. ^Increasing the membrane surface roughness caused the state transition of the surface from Wenzel model to Cassie-Baxter model. It is found that depositing nanoparticles on membrane macrostructure increased the membrane roughness from 87 nm to 126 nm (and consequently the WCA from 139 to 154). MD tests were carried out with 1.0 M sodium chloride solution in which sodium dodecyl sulfate (SDS) as the surface-active species was gradually added to the feed. The results showed that SDS in the feed solution wetted the non-coated hydrophobic membranes and reduced the salt rejection to 35%, however, the coated membranes resisted to the pore wetting with salt rejection >99% at 0.3 mM SDS concentration. Nonetheless, these membranes that resisted wetting were wetted for the SDS concentrations higher than 0.3 mM. ^In the second step, the multiscale-layer (MSL) superhydrophobic membrane with WCA>150 was prepared by casting polyvinylidene fluoride solution on hydrophilic-hydrophobic non-woven supports using dry-wet phase inversion technique. The phase inversion was performed in a water bath containing hydrophobic silanized nanoparticles. In this case, not only the chemistry and geometry of membrane surface but also the SC and roughness of the membrane pores at the macroscopic level was altered through self-assembly of organosilane on the embedded nanoparticle on the membrane surface. Direct contact membrane distillation (DCMD) experiments showed that the salt rejection >99% was achieved at SDS concentrations up to 0.4 mM. Nevertheless, MSL membranes were wetted for the SDS concentrations higher than 0.4 mM.
In the next step, besides using the superhydrophobic membrane, air bubbling in MD was implemented for the wetting prevention. ^Introducing air bubbles in the feed flow forms air dispersion adjacent to the membrane surface that enables restoring the hydrophobicity of the surface-wetted area of the membrane by displacing the liquid film containing the wetting species from the surface of the membrane. The presence of air bubbles on the surface of the superhydrophobic membrane in the DCMD setup inhibited the occurrence of wetting with salt rejection >99% at SDS concentrations up to 0.8 mM.
Finally, the influences of flow pattern and flow regime on the wetting rate of a gas bubbling membrane distillation (GBMD) process was studied by introducing air bubbles into the feed stream of a tubular membrane cell. The results showed that the performance of MD is increased by the air bubble addition, in particular also for low surface tension solutions. It was shown that the wetting occurrence was prevented within the turbulent slug flow pattern at the gas/ liquid ratio above 0.35. ^Overall, the achieved results revealed that the efficiency of the gas bubbling to control wetting in MD depends on not only the flow pattern but also the liquid and gas flow rate.