Large amounts of oily wastewater, which can be defined as produced water, are generated in oilfields. Ultrafiltration (UF) serves as an effective and economical method to purify produced water. Unfortunately, membrane fouling during produced water treatment is severe. In this paper, the effects of the ionic strength (1, 20, and 100 mM) as well as different surfactants on the membrane fouling are investigated. Four surfactants, including SDS (anionic), APG (non-ionic), CTAB (cationic) and DDAPS (zwitterionic), were selected for this study. The Derjaguin-Landau-Verwey-Overbeek (DLVO) and extended DLVO (XDLVO) models were used to quantify interactions between the membrane-oil droplet and deposited oil layer-oil droplet surfaces and to compare these interactions with the fouling experiments. The (X)DLVO interaction energies of the membrane-oil droplet exhibited a strong agreement with the fouling tendencies at 1 mM salinity. The SiC-deposited (B20) membrane showed less reversible and irreversible membrane fouling than the Al₂O₃ (B0) membrane when filtering negatively charged O/W emulsions stabilized with SDS, APG, or DDAPS. The DLVO model predicted a higher fouling tendency at higher salinity levels during the filtration of SDS, APG, or DDAPS-stabilized O/W emulsions and a decreased fouling tendency for CTAB-stabilized emulsion with the B20 membrane. However, at higher salinity levels, the XDLVO energy barrier was affected by both the repulsive electrostatic double layer (EL) interaction and attractive Lewis acid-base (AB) interaction. By comparing both experiments and (X)DLVO modeling, this study improves the fundamental understanding of the effect of ionic strength and surfactant types on reversible and irreversible fouling of the Al₂O₃ and SiC-coated membranes fouling by O/W emulsions.

In forward osmosis, defects on the selective active layer and changes in the porous structure of the support layer can be detrimental factors affecting the membrane performance. This study focuses on the impacts of (i) the possible presence of defects and (ii) the changes in pore structures on the water flux and membrane selectivity via computational fluid dynamics analyses. Results suggest that diffusion of the draw solute through the support layer, i.e. internal concentration polarization, can be strongly enhanced or reduced by widening or narrowing the shape of the pore, respectively, while no significant effect on water permeation can be associated with the change in draw solution cross-flow velocity or in the loss of draw solute during filtration. Interestingly, defects within the active layer may affect the water flux exponentially as function of the defect size, suggesting the presence of a threshold below which the convective passage of contaminated water flux through the defect is not affecting the membrane productivity. Moreover, the presence of defects may not be a detrimental factor for membrane operating with high nominal rejection (>90%) and low percentage of defected area (<1%).

Membrane fouling remains one of the most critical drawbacks in membrane filtration processes. Although the effect of various operating parameters—such as flow velocity, concentration, and foulant size—are well-studied, the impact of particle shape is not well understood. To bridge this gap, this study investigated the effect of polystyrene particle sphericity (sphere, peanut and pear) on external membrane fouling, along with the effect of particle charge (unmodified, carboxylated, and aminated). The results indicate that the non-spherical particles produce higher critical fluxes than the spherical particles (i.e., respectively 24% and 13% higher for peanut and pear), which is caused by the looser packing in the cake due to the varied particle orientations. Although higher crossflow velocities diminished the differences in the critical flux values among the particles of different surface charges, the differences among the particle shapes remained distinct. In dead-end filtration, non-spherical particles also produced lower flux declines. The shear-induced diffusion model predicts all five particle types well. The Derjaguin-Landau-Verwey-Overbeek (DLVO) and extended DLVO (XDLVO) models were used to quantify the interaction energies, and the latter agreed with the relative critical flux trends of all of the PS particles. As for the flux decline trends, both the DLVO and XDLVO results are in good agreement.

To exploit the benefits of membrane-based separation for the pharmaceutical and chemical industries, the understanding of membrane fouling in organic solvents is crucial. Specifically for the separation of biocatalysts in the manufacture of pharmaceuticals, this study investigated membrane fouling by bovine serum albumin (BSA) in 10% v/v isopropanol (IPA), 10% v/v dimethyl sulfoxide (DMSO), 30% v/v IPA, and 30% v/v DMSO, benchmarked against that in water. The presence of either IPA or DMSO worsened fouling, with the latter comparatively worse. To understand the fouling mechanisms, Field Emission Scanning Electron Microscopy (FESEM) images were taken to assess external fouling, Evapoporometry (EP) was used to measure the pore-size distributions of the fouled membranes to examine internal fouling, a fouling model was applied to extract the fouling parameters, and the interfacial interaction energies were derived. Results indicate that the worst fouling in 30% v/v DMSO was due to both significant external fouling and internal fouling, whereas the second-worst fouling by 30% v/v IPA was caused predominantly by internal fouling. The magnitudes of the total DLVO- and XDLVO-based interaction energies were found to be poorly related to the relative flux declines. This study provides valuable insights into membrane fouling in different solvent environments.

Fouling of ultrafiltration (UF) and microfiltration (MF) membranes by proteins is a major challenge in the bioprocessing and dairy industries, as well as in surface and wastewater treatment applications. This review attempts at presenting a comprehensive state-of-the-art understanding on protein fouling of membranes. Effects of operating conditions, along with properties of proteins and membranes, are discussed. Various tools and techniques used to characterize and monitor fouling are described. Different mitigation techniques and cleaning methods used are also presented. Two main factors have been identified as playing important roles in governing protein fouling, namely, ratio of protein size to membrane pore size and interfacial interactions (i.e., protein-protein and protein-membrane). Some directions for future research are suggested: (1) explore a wider range of proteins and their mixtures with respect to their fouling tendencies; and (2) create a comprehensive dataset that can be used to develop machine-learning models to enhance both predictive capabilities and mechanistic understanding.

Nanoparticle-laden sessile droplet drying has a wide impact on applications. However, the complexity affected by the droplet evaporation dynamics and particle self-assembly behavior leads to challenges in the accurate prediction of the drying patterns. We initiate a data-driven machine learning algorithm by using a single data collection point via a top-view camera to predict the transient drying patterns of aluminum oxide (Al2O3) nanoparticle-laden sessile droplets with three cases according to particle sizes of 5 and 40 nm and Al2O3 concentrations of 0.1 and 0.2 wt %. Dynamic mode decomposition is used as the data-driven learning model to recognize each nanoparticle-laden droplet as an individual system and then apply the transfer learning procedure. Along 270 s of droplet drying experiments, the training period of the first 100 s is selected, and then the rest of the 170 s is predicted with less than a 10% error between the predicted and the actual droplet images. The developed data-driven approach has also achieved the acceptable prediction for the droplet diameter with less than 0.13% error and a coffee-ring thickness over a range of 2.0 to 6.7 μm. Moreover, the proposed machine learning algorithm can recognize the volume of the droplet liquid and the transition of the drying regime from one to another according to the predicted contact line and the droplet height.

Membrane technology is of significant importance in water treatment applications, and also gaining momentum in other separations due to advantages such as environmentally friendly operation, less complex and lower-cost operating conditions compared to alternative options. To provide for sustainable and efficient membrane-based applications, the selection of appropriate membranes is crucial. Such a selection is based on membrane characterization, which offers critical information on parameters such as porosity, average pore size and pore size distribution (PSD). The two main classes of characterization methods are direct and indirect, with the latter having a theoretical basis, being more affordable, and also generally being able to characterize larger membrane areas compared to the direct techniques. This study reviews the indirect membrane characterization methods, the key theoretical backgrounds of which are the Young-Laplace equation, Kelvin equation, Gibbs-Thomson equation, and spectroscopy-based equations. The mathematical details are first presented, followed by the measurement details and experimental requirements, and finally the studies on membrane characterization via indirect methods. The advantages and limitations of each method are also discussed. For a complete understanding of the membrane, indirect methods may need to be complemented with direct ones and also with appropriate retention experiments of the feeds of interest.

Membrane fouling caused by natural organic matter (NOM) in water is a pressing problem. To address this, heated aluminum oxide particles (HAOPs) have been used as dynamic membranes pre-deposited onto the primary membrane to effectively remove NOM and thereby significantly diminish the fouling potential. An in-depth understanding of the mechanisms underlying the superior performance of HAOPs remains amiss, which motivated this study. Molecular dynamics (MD) simulations were conducted to systematically compare the performance of HAOPs, which have been reported to be particularly effective for high molecular weight (HMW) NOM, with the conventional powdered activated carbon (PAC) adsorbent. Six NOM constituents, three of which have HMW and three have low molecular weight (LMW), were studied. Results indicate that the mechanisms underlying the effective removal of HMW NOM by HAOPs include: (1) higher foulant-HAOPs interaction energy; (2) greater hydration of the HMW NOM, which thereby increases the affinity to the more hydrophilic HAOPs; (3) diminished mobility of the foulant once adsorbed, which deters desorption; and (4) higher peak intensities in the radial distribution functions for multiple functional groups on the HMW NOM foulants. These results are expected to be valuable towards the better design of such materials for mitigating membrane fouling.

Membrane technology is increasingly becoming a promising alternative in the chemical and pharmaceutical industries, wherein organic solvents may form the continuous phase. Regarding the inevitable membrane fouling phenomenon, although the knowledge base is rich for feeds involving water, an analogous understanding for feeds involving organic solvents is limited. Accordingly, in this study, we systematically investigated the fouling behaviors of a model colloidal foulant (namely, silica) dispersed in water and five organic solvents (namely, methanol, ethanol, acetone, toluene and hexane) during ultrafiltration. The flux decline trends were clearly different. The XDLVO model and a fouling model were employed to extract mechanistic insights. Firstly, zeta potential alone was a poor indicator of the fouling extent. Secondly, solvents with high polarity (i.e., methanol, ethanol) had repulsive foulant-foulant and foulant-membrane interfacial interactions, which were beneficial in mitigating membrane fouling, leading to lesser flux decline and lower cake resistance. Thirdly, solvents with no or low polarity (i.e., n-hexane, toluene and acetone) had attractive interfacial interactions, which worsened membrane fouling. However, attractive foulant-foulant interaction was beneficial in augmenting shear-induced diffusion, which mitigated fouling. Fourthly, the fouling parameters extracted from the fouling model generally were lesser and greater respectively for the high-polarity and lower-polarity solvents, which agree with the interfacial interaction values and flux decline trends. The insights emanating from this study on membrane fouling in organic solvents are expected to be valuable in the design and operation of such emerging membrane-filtration systems.

The internal fouling of membranes is typically presumed and inferred, but the direct characterization of representative samples is challenging. This study targeted to assess internal fouling using the Optical Coherence Tomography (OCT) and Evapoporometry (EP) techniques for real-time monitoring during filtration and off-line measurement of the pore size distributions (PSDs) of the fouled membranes, respectively. The results were validated by atomic force microscopy (AFM) measurements and also the three-mechanism fouling model. The foulant used was the well-studied bovine serum albumin (BSA), while the membranes were three commercially available polymeric microfiltration membranes with similar nominal pore sizes and porosities. Although the OCT affirmed the progressive worsening of internal fouling in real-time, the quantitative comparison of the extents of internal fouling among the three membranes was not possible. As for EP, it was able to quantitatively compare the pore size distributions and average pore diameters to ascertain the different extents of internal fouling. The phenomenological model available was effective in quantitatively comparing the extents of pore-constriction among the three membranes, while AFM tied the worst internal fouling to the most attractive BSA-membrane affinity. More in-depth understanding of internal fouling is warranted to not only recommend better membranes but also facilitate the advancement of models.

To comparatively assess the Evapoporometry (EP) technique vis-à-vis the Liquid-Liquid Displacement Porosimetry (LLDP) technique, the pore size distributions, mean pore diameters (d_avg) and porosities of five polymeric (namely, nylon, PES, PTFE, PET and PVDF) and one inorganic (namely, alumina) UF/tight MF membranes were quantified by both techniques. For all the membranes, the pore size ranges were generally narrower and the pore size distributions had distinctive peaks for the LLDP technique. For the nylon, PES and PTFE membranes, the d_avg values obtained from the two techniques agreed well. However, for the PET and PVDF membranes, the differences were twofold due to the higher pressure needed for the LLDP tests. Specifically, for PET, the d_avg value obtained via EP was half that via LLDP, because the higher pressure compacted the lower mechanical strength polymer, leading to pore closure. On the other hand, for PVDF, due to the rubber nature, the higher pressure caused the pores to be stretched, leading to larger pores. As for the alumina membrane, because of the more ideal cylindrical pores, the d⁴-weighting of the LLDP measurement gave a greater d_avg value than that of the d²-weighting of the EP measurement. Also, porosity measurements were erroneous for LLDP if the active layer cannot be precisely quantified. With respect to MWCO, while EP does not explicitly quantify this, the LLDP generally over-estimated the values, because of the errors associated with the measurement of the first (largest) pores at the lowest pressures.

Evapoporometry (EP) determines the pore-size distribution (PSD) based on the Kelvin equation that relates the evaporation rate of a volatile wetting liquid to the pore diameter. EP offers several advantages relative to other PSD characterization methods. EP is adapted here to characterize only the continuous pores rather than all the pores. This adaptation involves sequentially injecting a volatile wetting liquid under the membrane followed by a denser, non-volatile, non-wetting liquid to seal the pores as the overlying volatile liquid evaporates. EP characterization of only the continuous pores shifts the PSD based on all the pores towards smaller pores as indicated by a decrease in the average pore diameter from 15.2 nm to 13.2 nm for a 500 kDa PVDF membrane and from 12.8 nm to 11.7 nm for a 300 kDa PES membrane. The difference between the total pore mass for the two EP characterization protocols was 46.2% and 36.3% for the PVDF and PES, respectively. This provides an assessment of the interconnectivity of the pores, which was quite high as would be expected for these two solvent-cast membranes. By injecting an underlying, non-volatile, non-wetting liquid, EP can be adapted to use non-wetting volatile liquids of interest in some applications.

Membrane-based filtration is promising for the treatment of oily wastewater with stable micron-sized oil droplets, but is unfortunately limited by the inevitable membrane fouling phenomenon. To advance the understanding on membrane fouling by oil emulsion, this study made use of the direct observation through the membrane (DOTM) technique to experimentally visualize the evolution of fouling and determine critical flux, and also molecular dynamics simulations to unveil the interfacial interactions and behaviors underlying the different fouling behaviors. Three oil emulsion types with similar mean droplet sizes were studied, namely, one without surfactant, one stabilized by sodium dodecyl sulfate (SDS; a negatively charged surfactant) and one stabilized by dodecyltrimethylammonium bromide (DTAB; a positively charged surfactant). DOTM results indicate that the critical flux was the highest in the absence of surfactant and lowest for the DTAB-stabilized ones, while simulation results indicate that the interaction energies are clearly different among the different oil emulsion types. Both hence affirm that the presence of surfactant and different surfactant type distinctly changes the fouling tendencies. The key conclusions on the different fouling tendencies among the three oil emulsion types are summarized as follows. Firstly, the highest critical flux in the absence of any surfactant is linked to greatest oil-water interaction and least oil-membrane interaction, both of which cause the oil molecule to prefer being in the bulk aqueous feed. The radial distribution function (RDF) further substantiates this. Secondly, the higher critical flux exhibited by the SDS-stabilized oil emulsion compared to the DTAB-stabilized ones could be attributed to the negative charge of the former. Simulations affirm the comparatively greater oil-membrane repulsion through the RDF profile, lower oil-membrane interaction and higher oil-water interaction of the SDS-stabilized oil emulsion. Thirdly, despite surfactants generally have a stabilizing effect on oil emulsions, DOTM images show that coalescence was most extensive for the DTAB-stabilized oil emulsion, because of the greatest oil-membrane attraction and least oil-oil repulsion. Coupling both experiments and simulations, this study enhanced the mechanistic understanding on the effect of surfactant on membrane fouling by emulsions of the same oil.

The striping phenomenon in membrane filtration, whereby the foulants deposit as regular streaks rather than a more uniform layer, was first observed more than 30 years ago (Jonsson, 1987), but the understanding has remained limited to a few subsequent studies (Henriksen and Hassager, 1993; Larsen, 1991; Li et al., 2016; Tanudjaja et al., 2017; Tarabara et al., 2002). In view of the potential practical implications of the stripes in terms of membrane fouling and fouling mitigation, this study was targeted at an in-depth characterization of the stripes. The direct observation through the membrane (DOTM) technique was employed to observe the conditions when the stripes formed by oil emulsions stabilized by three Tween surfactants at a lower cross-flow velocity, and higher oil concentrations and permeate fluxes. The results indicate that (i) other than hydrodynamic factors (e.g., permeate drag, tangential shear), foulant-membrane and foulant-foulant interactions played a role in stripe formation or disappearance, as evident in the formation of stripes only by the oil emulsion stabilized by the Tween surfactants and the effect of pH; (ii) stripes made up of oil droplets appeared to be easier to remove than the more uniform layer of oil droplets, as evident in the shorter time taken for the former to detach; and (iii) among the three Tween surfactants, the striping characteristics investigated were largely similar, except for the time taken for the deposits to detach. These are expected to have implications for fouling control and mitigation.

Membrane distillation (MD) has been gaining increasing attention as a promising alternative to the more conventional separation processes like distillation and reverse osmosis (RO), due to the higher rejection, lower fouling propensity and ability to treat water using low-quality waste heat. Hydrophobic membranes with high mass transfer and low heat conduction are desirable in direct contact membrane distillation (DCMD), and this contradicting transport properties, along with the chemical and thermal stability required, remains a challenge to date. Herein, we report a membrane modification strategy based on the array of functionalities conferred by emerging 2D materials. Specifically, the current effort targeted at investigating the potential of MXene, a novel 2D material with both photothermal and anti-fouling functionalities, as a coating to improve DCMD performance. Feeds containing bovine serum albumin (BSA) and sodium chloride (NaCl) were filtered through uncoated and MXene-coated PVDF membranes for 21 h, with results indicating a reduction of 12% of heater energy input per unit volume distillate, along with a reduction in flux decline in the range of 56–64%. This study demonstrated that MXene is a promising 2D material for improving the practical feasibility of MD.