Chuyang Y. Tang

Colloidal interactions and fouling of NF and RO membranes: A review

Colloids are fine particles whose characteristic size falls within the rough size range of 1-1000 nm. In pressure-driven membrane systems, these fine particles have a strong tendency to foul the membranes, causing a significant loss in water permeability and often a deteriorated product water quality. There have been a large number of systematic studies on colloidal fouling of reverse osmosis (RO) and nanofiltration (NF) membranes in the last three decades, and the understanding of colloidal fouling has been significantly advanced. The current paper reviews the mechanisms and factors controlling colloidal fouling of both RO and NF membranes. Major colloidal foulants (including both rigid inorganic colloids and organic macromolecules) and their properties are summarized. The deposition of such colloidal particles on an RO or NF membrane forms a cake layer, which can adversely affect the membrane flux due to 1) the cake layer hydraulic resistance and/or 2) the cake-enhanced osmotic pressure. The effects of feedwater compositions, membrane properties, and hydrodynamic conditions are discussed in detail for inorganic colloids, natural organic matter, polysaccharides, and proteins. In general, these effects can be readily explained by considering the mass transfer near the membrane surface and the colloid-membrane (or colloid-colloid) interaction. The critical flux and limiting flux concepts, originally developed for colloidal fouling of porous membranes, are also applicable to RO and NF membranes. For small colloids (diameter≪100 nm), the limiting flux can result from two different mechanisms: 1) the diffusion-solubility (gel formation) controlled mechanism and 2) the surface interaction controlled mechanism. The former mechanism probably dominates for concentrated solutions, while the latter mechanism may be more important for dilute solutions. Future research needs on RO and NF colloidal fouling are also identified in the current paper.

Degradation of polyamide nanofiltration and reverse osmosis membranes by hypochlorite.

The degradation of polyamide (PA) nanofiltration and reverse osmosis membranes by chlorine needs to be understood in order to develop chlorine-resistant membranes. Coated and uncoated fully aromatic (FA) and piperazine (PIP) semi-aromatic PA membranes were treated with hypochlorite solution and analyzed by X-ray photoelectron spectroscopy (XPS) and Fourier transform infrared (FTIR). XPS results showed that in chlorine treated FA PA membranes the ratio of bound chlorine to surface nitrogen was 1:1 whereas it was only 1:6 in the case of PIP PA membranes. Surface oxygen of uncoated FA and PIP membranes increased with increasing hypochlorite concentration whereas it decreased for coated FA membranes. High resolution XPS data support that chlorination increased the number of carboxylic groups on the PA surface, which appear to form by hydrolysis of the amide bonds (C(O)-N). FTIR data indicated the disappearance of the amide II band (1541 cm(-1)) and aromatic amide peak (1609 cm(-1)) in both coated and uncoated chlorinated FA membranes, consistent with the N-chlorination suggested by the XPS results. Furthermore, the surface charge of chlorinated membranes at low pH (<6) became negative, consistent with amide-nitrogen chlorination. Chlorination appeared to both increase and decrease membrane hydrophobicity depending on chlorination exposure conditions, which implied that N-chlorination and hydrolysis may be competing processes. The effects of property changes on the membrane performance were also observed for NF90, BW30, and NF270 membranes.

Relating reverse and forward solute diffusion to membrane fouling in osmotically driven membrane processes

Osmotically driven membrane processes, such as forward osmosis (FO) and pressure retarded osmosis (PRO), are attracting increasing interest in research and applications in environment and energy related fields. In this study, we systematically investigated the alginate fouling on an osmotic membrane during FO operation using four types of draw solutions (NaCl, MgCl(2), CaCl(2) and Ca(NO(3))(2)) to elucidate the relationships between reverse (from draw solution to feed solution) and forward (from feed solution to draw solution) solute diffusion, and membrane fouling. At the same water flux level (achieved by adjusting the draw solution concentration), the greatest reverse solute diffusion rate was observed for NaCl draw solution, followed by Ca(NO(3))(2) draw solution, and then CaCl(2) draw solution and MgCl(2) draw solution, the order of which was consistent with that of their solute permeability coefficients. Moreover, the reverse solute diffusion of draw solute (especially divalent cation) can change the feed solution chemistry and thus enhance membrane fouling by alginate, the extent of which is related to the rate of the reverse draw solute diffusion and its ability to interact with the foulant. The extent of fouling for the four types of draw solution followed an order of Ca(NO(3))(2) > CaCl(2) >> MgCl(2) > NaCl. On the other hand, the rate of forward diffusion of feed solute (e.g., Na(+)) was in turn promoted under severe membrane fouling in active layer facing draw solution orientation, which may be attributed to the fouling enhanced concentration polarization (pore clogging enhanced ICP and cake enhanced concentration polarization). The enhanced concentration polarization can lead to additional water flux reduction and is an important mechanism governing the water flux behavior during FO membrane fouling. Findings have significant implications for the draw solution selection and membrane fouling control in osmotically driven membrane processes.

Effect of solution chemistry on the adsorption of perfluorooctane sulfonate onto mineral surfaces

Perfluorooctane sulfonate (PFOS) is an emergent contaminant of substantial environmental concerns, yet very limited information has been available on PFOS adsorption onto mineral surfaces. PFOS adsorption onto goethite and silica was investigated by batch adsorption experiments under various solution compositions. Adsorption onto silica was only marginally affected by pH, ionic strength, and calcium concentration, likely due to the dominance of non-electrostatic interactions. In contrast, PFOS uptake by goethite increased significantly at high [H+] and [Ca2+], which was likely due to enhanced electrostatic attraction between the negatively charged PFOS molecules and positively charged goethite surface. The effect of pH was less significant at high ionic strength, likely due to electrical double layer compression. PFOS uptake was reduced at higher ionic strength for a strongly positively charged goethite surface (pH 3), while it increased for a weakly charged surface (pH 7 and 9), which could be attributed to the competition between PFOS-surface electrostatic attraction and PFOS-PFOS electrostatic repulsion. A conceptual model that captures PFOS-surface and PFOS-PFOS electrostatic interactions as well as non-electrostatic interaction was also formulated to understand the effect of solution chemistry on PFOS adsorption onto goethite and silica surfaces.

Rejection of pharmaceuticals by forward osmosis membranes.

Rejection of four pharmaceutical compounds, carbamazepine, diclofenac, ibuprofen and naproxen, by forward osmosis (FO) membranes was investigated in this study. For the first time, the rejection efficiency of the pharmaceutical compounds was compared between commercial cellulose triacetate (CTA) based membranes and thin film composite (TFC) polyamide based membranes. The rejection behavior was related to membrane interfacial properties, physicochemical characteristics of the pharmaceutical molecules and feed solution pH. TFC polyamide membranes exhibited excellent overall performance, with high water flux, excellent pH stability and great rejection of all pharmaceuticals investigated (>94%). For commercial CTA based FO membranes, hydrophobic interaction between the compounds and membranes exhibited strong influence on their rejection under acidic conditions. The pharmaceuticals rejection was well correlated to their hydrophobicity (log D). Under alkaline conditions, both electrostatic repulsion and size exclusion contributed to the removal of deprotonated molecules. The pharmaceuticals rejection by CTA-HW membrane at pH 8 followed the order: diclofenac (99%)>carbamazepine (95%)>ibuprofen (93%) ≈ naproxen (93%). These results can be important for FO membrane synthesis, modification and their application in water purification.

Boric acid permeation in forward osmosis membrane processes: modeling, experiments, and implications.

Forward osmosis (FO) is attracting increasing interest for its potential applications in desalination. In FO, permeation of contaminants from feed solution into draw solution through the semipermeable membrane can take place simultaneously with water diffusion. Understanding the contaminants transport through and rejection by FO membrane has significant technical implications in the way to separate clean water from the diluted draw solution. In this study, a model was developed to predict boron flux in FO operation. A strong agreement between modeling results and experimental data indicates that the model developed in this study can accurately predict the boron transport through FO membranes. Furthermore, the model can guide the fabrication of improved FO membranes with decreased boron permeability and structural parameter to minimize boron flux. Both theoretical model and experimental results demonstrated that when membrane active layer was facing draw solution, boron flux was substantially greater compared to the other membrane orientation due to more severe internal concentration polarization. In this investigation, for the first time, rejection of contaminants was defined in FO processes. This is critical to compare the membrane performance between different membranes and experimental conditions.

Gypsum scaling in pressure retarded osmosis: Experiments, mechanisms and implications

Pressure retarded osmosis (PRO) is an osmotically-driven membrane process that can be used to harvest salinity-gradient power. The PRO performance (both water flux and power density) can be severely limited by membrane fouling. The current study, for the first time, investigates PRO scaling in a bench-scale pressurized system using calcium sulfate dihydrate (gypsum) as a model scalant. In addition to the bulk feed solution (FS) saturation index (SI bulk), gypsum scaling was found to be strongly affected by the draw solution (DS) type and concentration, the applied hydraulic pressure, and the membrane orientation. The commonly recommended active layer facing draw solution (AL-DS) orientation was highly prone to internal scaling. In this orientation, severe internal concentration polarization (ICP) of scaling precursors induced gypsum clogging in membrane support layer even when the FS was undersaturated (e.g., SI bulk = 0.8). At higher SI bulk values, external gypsum crystal deposition occurred in addition to internal scaling. More severe scaling was observed when the DS contained scaling precursors such as Ca(2+) or SO4(2-), suggesting that the reverse diffusion of these precursors into the FS can significantly enhanced gypsum scaling. Increasing applied hydraulic pressure could enhance reverse solute diffusion and thus result in more severe gypsum scaling when the DS contained scaling precursors. A conceptual model, capturing the two important PRO scaling mechanisms (ICP of scaling precursors from FS and reverse diffusion of scaling precursors from the DS), is presented to rationalize the experimental results. Our results provide significant implications for PRO scaling control.

Preparation of supported lipid membranes for aquaporin Z incorporation

There has been a recent surge of interest to mimic the performance of natural cellular membranes by incorporating water channel proteins-aquaporins (AQPs) into various ultrathin films for water filtration applications. To make biomimetic membranes one of the most crucial steps is preparing a defect-free platform for AQPs incorporation on a suitable substrate. In this study two methods were used to prepare supported lipid membranes on NF membrane surfaces under a benign pH condition of 7.8. One method was direct vesicle fusion on a hydrophilic membrane NF-270; the other was vesicle fusion facilitated by hydraulic pressure on a modified hydrophilic NF-270 membrane whose surface has been spin-coated with positively charged lipids. Experiments revealed that the supported lipid membrane without AQPs prepared by the spin coating plus vesicle fusion had a much lower defect density than that prepared by vesicle fusion alone. It appears that the surface roughness and charge are the main factors determining the quality of the supported lipid membrane. Aquaporin Z (AqpZ) proteins were successfully incorporated into 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) liposomes and its permeability was measured by the stopped-flow experimental procedure. However, after the proteoliposomes have been fused onto the modified substrate, the AqpZ function in the resultant membrane was not observed and AFM images showed distinct aggregations of unfused proteoliposomes or AqpZ proteins on the substrate surface. It is speculated that the inhibition of AqpZ function may be caused by the low lipid mobility on the NF membrane surface. Further investigations to evaluate and optimize the structure-performance relationship are required.

Effects of Chlorine Exposure Conditions on Physiochemical Properties and Performance of a Polyamide Membrane—Mechanisms and Implications

Understanding the effects of chlorine exposure on polyamide (PA) based membranes is essential in membrane lifespan improvement. In this study, NF90 nanofiltration membrane was treated with sodium hypochlorite at different concentrations, pHs and durations. The changes in membrane elemental composition and bonding chemistry obtained from XPS and ATR-FTIR revealed the impacts of two competing mechanisms: N-chlorination and chlorination-promoted hydrolysis. More chlorine was incorporated into the PA matrix at pH <7, at which HOCl is dominant, while chlorine-promoted hydrolysis was more favorable at pH >7 with abundant hydroxyl groups. The membrane surface became more hydrophobic when chlorination was dominant, which in turn caused the water permeability of chlorinated membrane to decrease. Meanwhile, membrane became more hydrophilic and less cross-linked when hydrolysis effects were governing, which made the membrane more permeable for water. Rejection of charged solutes [NaCl, As(V)] improved in most chlorinating conditions due to increased charge density. However, when hydrolysis was severe (≥ 1000 ppm, pH 7 and 9), the enhanced charge repulsion effect could not compensate for the extensive amide bond cleavage, resulting in declined rejection. The lower rejection of neutral boric acid provided strong evidence of a less cross-linked separation layer.

Mining nutrients (N, K, P) from urban source-separated urine by forward osmosis dewatering

Separating urine from domestic wastewater promotes a more sustainable municipal wastewater treatment system. This study investigated the feasibility of applying a forward osmosis (FO) dewatering process for nutrient recovery from source-separated urine under different conditions, using seawater or desalination brine as a low-cost draw solution. The filtration process with the active layer facing feed solution exhibited relatively high water fluxes up to 20 L/m(2)-h. The process also revealed relatively low rejection to neutral organic nitrogen (urea-N) in fresh urine but improved rejection of ammonium (50-80%) in hydrolyzed urine and high rejection (>90%) of phosphate, potassium in most cases. Compared to simulation based on the solution-diffusion mechanism, higher water flux and solute flux were obtained using fresh or hydrolyzed urine as the feed, which was attributed to the intensive forward nutrient permeation (i.e., of urea, ammonium, and potassium). Membrane fouling could be avoided by prior removal of the spontaneously precipitated crystals in urine. Compared to other urine treatment options, the current process was cost-effective and environmentally friendly for nutrient recovery from urban wastewater at source, yet a comprehensive life-cycle impact assessment might be needed to evaluate and optimize the overall system performance at pilot and full scale operation.