Mechanistic understanding of internal membrane fouling via real time monitoring techniques during microfiltration

• Main Author: Lay, Huang Teik
• Other Authors: Chew Jia Wei
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Chemical engineering; Engineering::Environmental engineering
• Online Access: https://hdl.handle.net/10356/163865
• Institution: Nanyang Technological University

In recent decades, membrane-based filtration has become more prominent and increasingly employed in wide-ranging fields. Unfortunately, membrane fouling continues to be one of the major impediments in commercial implementation. Although membrane fouling has been widely researched, a complete mechanistic understanding of the membrane fouling mechanism is still far from perfect, mainly due to the complexity of various factors that can affect membrane fouling. To date, most studies were dedicated only to the deposition of foulant particles on the membrane surface (i.e., external membrane fouling), in which the understanding of the deposition of foulant particles within the membrane pores (i.e., internal membrane fouling) remains amiss. To achieve the goal of revealing the fouling mechanisms within the membrane pores, three-dimensional imaging methods through optical coherence tomography (OCT) have been employed to achieve real time monitoring of layers above and below the feed-membrane interface throughout the filtration process. By analyzing the patterns of deposition of foulant particles at each respective layer, a complete image of fouling mechanism within the membrane pores can be pictured. In parallel with the real time, non-invasive characterization experimental efforts, Field Emission Scanning Electron Microscope (FESEM), Quartz crystal Microbalance with Dissipation (QCM-D), modelling equations and Derjaguin-Landau-Verwey-Overbeek (DLVO) theory have been utilized to provide information about the fouling cake layer characteristics and interfacial interaction energies. In this work, the internal membrane fouling mechanisms study was first started by investigating the feeds containing mixtures of oppositely charged latex particles. It had been found out that the mixtures of the latex particles (which were less positively charged than that of purely aminated and carboxylated latex particles) gave the worst flux declines due to the greater extent of pore blockages by mixtures of latex particles. After understanding the fouling mechanisms of ideal colloidal latex particles, the study was carried on understanding more realistic oppositely charged protein feed solutions, specifically the negatively charged protein (BSA and pepsin) and positively charged protein (lysozyme). The results showed that the protein cake layers displayed a unique dynamic movement within the polycarbonate track-etched membrane pores which was not observed by other foulants (e.g. oil emulsion, latex particles). Furthermore, it had been noticed that the realistic protein feed solutions were not only differed in surface charge but also in shape. To understand the effect of foulant particle shape on membrane fouling, the next study was focused on examining colloidal latex particles with different particular shapes (namely, peanut, pear and sphere). The sphere-shaped particles displayed the worst flux decline, while the peanut- and pear-shaped ones fouled similarly as the preferential vertical orientations favored the nonspherical particles to form a less denser fouling cake layer within the membrane pores that allowed more permeate to pass through. After understanding the effect of mixture of oppositely charged particles and particle shape on the internal fouling mechanisms, the study was brought forward to more complex levels which were to investigate the effect of coating layer and to verify the validity of different DLVO approximation methods on predicting the affinity between foulant-membrane surfaces. The results highlighted the accuracy of LSA equation compared to that of P-B equation in predicting the EL interaction energy component as the former accounts for three different scenarios (namely, constant potential, constant charge and intermediate cases between two) while the latter only assumes constant potential.