Prediction of physical properties of nanofiltration membranes using experiment and theoretical models

Two commercial nanofiltration (NF) membranes, viz., Desal-HL and NF 700 MWCO were investigated experimentally using neutral and charged solutes, viz., glucose, sodium chloride and magnesium chloride. Effect of pH was studied for sodium chloride rejection and isoelectric point of the membrane was deduced. Experimental results were analyzed using Donnan steric pore and dielectric exclusion models. Dielectric exclusion arises due to the difference in dielectric constant between the bulk and the nano-pore. Born dielectric effect was used as dielectric exclusion phenomena in the present investigation. Stokes–Einstein, Born effective and Pauling radii were used for theoretical simulation, which accurately predicted different charge densities. Empirical correlations were proposed between charge density, concentration and pH for each radius. Charge density decreased drastically when dielectric exclusion term was included in the theoretical model, which showed the real physical characteristics of the membranes employed. Charge density and radius of pore was found to be an important surface parameter in predicting the separation effects in NF membranes.     

Formation and characterization of asymmetric nanofiltration membrane: Effect of shear rate and polymer concentration

In this study, we report the effects of shear rates and polymer concentrations in the formation of asymmetric nanofiltration membrane using a simple dry/wet phase inversion technique. Employing the combination of irreversible thermodynamic model, solution-diffusion model (Spiegler–Kedem equation), steric-hindrance pore (SHP) model and Teorell–Meyers (TMS) model, the transport mechanisms and membrane structural properties were determined and have been characterized for different cases of those formation parameters. The experimental and modeling showed very promising results in terms of membrane performance with interesting structural details. The optimum shear rate (critical shear rate) was found to be at about 203.20 s⁻¹ and the best polymer concentration toward the formation of high performance nanofiltration membrane is in the range of 19.60–23.10%. The modeling results suggested that the pore radius of the membranes produced lies within the range of pore radius of 29 commercial available membranes. This study also proposed that the electrolytes transport through nanofiltration membrane was dominated by a convection factor which accounted approximately 30% more than a diffusion factor. This study also indicated that shear rate and polymer concentration were found to affect the membrane performance and structural properties by providing, to a certain extent, an oriented membrane skin layer which in turn exhibiting an improvement in membrane separation ability.     

解读纳滤:一种具有纳米尺度效应的分子分离操作

纳滤膜是20世纪80年代末期发展起来的一种广泛用于液体分离的新型分离膜。早期研究中,先后提出的基于筛分效应的细孔模型,基于静电效应的电荷模型,以及同时考虑上述两种效应的静电位阻模型和道南位阻模型等为人们更好地理解纳滤膜分离机理和指导纳滤膜过程应用发挥了十分重要的作用。然而由于这些具有“疏松型反渗透膜”特点的纳滤膜没有相应的膜性能预测评价软件,使得针对具体应用过程的纳滤膜的大规模标准化应用受到了一定的制约。为此,结合上述模型,根据一些特定实验拟合确定混合盐体系同号离子间的竞争作用和异号离子间的调节作用,提出了一个适于混合盐体系的纳滤膜分离性能评价模型,促进了纳滤膜技术在水处理过程的大规模推广。最近,根据纳滤膜对离子选择性分离性能及其伴随的动电性质的细致而深入的实验研究,发现仅考虑筛分效应和静电效应并不能完全合理地解释纳滤膜的分离性能,且在动电性质的解析上也存在一定缺陷,进而对纳滤膜纳米级孔径引起的特殊效应和溶液体系中复杂相互作用引起的荷电性质变化有了更为深刻的认识和理解,提出并定量分析了离子透过纳滤膜时存在的介电排斥效应。     

Polyamide Nanofiltration Membrane from Surfactant-assembly Regulated Interfacial Polymerization of 2-Methylpiperazine for Divalent Cations Removal

Removal of metal ions from water can not only alleviate the scaling problem of domestic and industrial water, but also solve the water safety problem caused by heavy metal ion pollution. Here, we fabricate a positively charged nanofiltration membrane via surfactant-assembly regulated interfacial polymerization(SARIP) of 2-methylpiperazine(MPIP) and trimesoyl chloride(TMC). Due to the existence of methyl substituent, MPIP has lower reactive activity than piperazine(PIP) but stronger affinity to hexane, resulting in a nanofiltration(NF) membrane with an opposite surface charge and a loose polyamide active layer. Interestingly, with the help of sodium dodecyl sulfate(SDS) assembly at the water/hexane, the reactivity between MPIP and TMC was obviously increased and caused in turn the formation of a positively charged polyamide active layer with a smaller pore size, as well as with a narrower pore size distribution. The resulting membrane shows a highly efficient removal of divalent cations from water, of which the rejections of MgCl₂, CoCl₂ and NiCl₂ are higher than 98.8%, 98.0% and 98.0%, respectively, which are better than those of most of other positively charged NF membranes reported in literatures.     

Computer program for simulation of mass transport in nanofiltration membranes

A computer program, NanoFiltran, was developed to simulate the mass transport of multi-ionic aqueous solutions in charged nanofiltration (NF) membranes, based on the Donnan steric partitioning pore and dielectric exclusion (DSPM&DE) model, with incorporation of the non-ideality of electrolyte solutions and concentration polarization effects in the membrane/feed-solution interface. With this computer program, the extended Nernst–Planck (ENP) equations are discretized inside the membrane, using the finite-difference scheme. The discretized ENP equations together with the other model equations are linearized in order to obtain a system of equations that are solved simultaneously. The linearized system of equations is based on an initial guess for the electrical potential and ions concentrations profiles, which are updated iteratively. A robust method of under-relaxation of the electrical potential and ions concentrations ensures that the convergence is achieved even for NF systems that exhibit a very stiff numerical behaviour.     

Removal of organic contaminants by RO and NF membranes

Rejection characteristics of organic and inorganic compounds were examined for six reverse osmosis (RO) membranes and two nanofiltration (NF) membranes that are commercially available. A batch stirred-cell was employed to determine the membrane flux and the solute rejection for solutions at various concentrations and different pH conditions. The results show that for ionic solutes the degree of separation is influenced mainly by electrostatic exclusion, while for organic solutes the removal depends mainly upon the solute radius and molecular structure. In order to provide a better understanding of rejection mechanisms for the RO and NF membranes, the ratio of solute radius (r(i,s)) to effective membrane pore radius (r(p)) was employed to compare rejections. An empirical relation for the dependence of the rejection of organic compounds on the ratio r(i,s)/r(p) is presented. The rejection for organic compounds is over 75% when r(i,s)/r(p) is greater than 0.8. In addition, the rejection of organic compounds is examined using the extended Nernst-Planck equation coupled with a steric hindrance model. The transport of organic solutes is controlled mainly by diffusion for the compounds that have a high r(i,s)/r(p) ratio, while convection is dominant for compounds that have a small r(i,s)/r(p) ratio.     

Nanofiltration membranes synthesized from hyperbranched polyethyleneimine

Four nanofiltration membranes, two negatively and two positively charged, were fabricated by interfacial polymerization. Three different amines, ethylenediamine (EDA), diethylenetriamine (DETA), and hyperbranched polyethyleneimine (PEI) were selected to react with two acyl chlorides, trimesoyl chloride (TMC) and terephthaloyl chloride (TPC). The two membranes containing hyperbranched PEI, PEI/TPC and PEI/TMC, are positively charged at the operational pH. But the other two membranes, EDA/TMC and DETA/TMC, are negatively charged. It is found that the two PEI membranes own special rejection characters during nanofiltration. The PEI/TPC membrane has a similar pore size to the EDA/TMC membrane but owns simultaneously the higher salt rejection and permeation flux. The PEI/TMC has a pore size as large as 1.5 nm and still has a higher NaCl rejection than the EDA/TMC membrane of which the pore size as small as 0.43 nm. We consider that the special rejection characters are derived from the special structure of PEI. The hyperbranched structure allows some of the charged amine groups drifting inside the pores and interacting with the ions in the pathway. The drifting amines increase salt rejection but have little effect on water permeation. It implies that a high flux and high rejection membrane for desalting can be obtained by attaching freely rotating charged groups.     

Membrane characterization and solute diffusion in porous composite nanocellulose membranes for hemodialysis

The membrane and solute diffusion properties of Cladophora cellulose and polypyrrole (PPy) functionalized Cladophora cellulose were analyzed to investigate the feasibility of using electroactive membranes in hemodialysis. The membranes were characterized with scanning electron microscopy, ζ-potentiometry, He-pycnometry, N₂ gas adsorption, and Hg porosimetry. The diffusion properties across the studied membranes for three model uremic toxins, i.e. creatinine, vitamin B12 and bovine serum albumin, were also analyzed. The characterization work revealed that the studied membranes present an open structure of weakly negatively charged nanofibers with an average pore size of 21 and 53 nm for pristine cellulose and PPy-Cladophora cellulose, respectively. The results showed that the diffusion of uremic toxins across the PPy-Cladophora cellulose membrane was faster than through pure cellulose membrane, which was related to the higher porosity and larger average pore size of the former. Since it was found that the average pore size of the membranes was larger than the hydrodynamic radius of the studied model solutes, it was concluded that these types of membranes are favorable to expand the Mw spectrum of uremic toxins to also include conditions associated with accumulation of large pathologic proteins during hemodialysis. The large average pore size of the composite membrane could also be exploited to ensure high-fluxes of solutes through the membrane while simultaneously extracting ions by an externally applied electric current.     

Preparation,characterization, and performance of a novel hollow fiber nanofiltration membrane

Nanofiltration (NF) grade hollow fiber membrane was prepared by incorporation of zinc chloride into polysulfone–polyethylene glycol (molecular weight 200) blend. A 1.0 wt% zinc chloride in the blend reduced the molecular weight cut off (MWCO) of hollow fibers from 44 kDa (average pore size 64A⁰) to a nanofiltration range of MWCO 870 Da (average pore size 7.69 A^°). MWCO decreased further to 330 Da (average pore size 4.78 A^°) on addition of 2.5 wt% zinc chloride. types of NF hollow fiber were spun, corresponding to zinc chloride concentration of 1.0, 1.5, 2.0, and 2.5 wt%. Ternary phase diagram qualitatively explained the denser morphology for various concentrations of zinc chloride. This was supported by scanning electron micrographs of cross‐section and top surface of hollow fibers. NF membranes possessed negative surface charge at extreme pH conditions. Rejection of 1000 mg/l sodium chloride solution was in between 38 to 45% at pH 11, and for divalent sodium sulfate, it was in the range of 55 to 62%. Rejection of dye congo red was found to be 100%. NF membranes showed reasonable antifouling characteristics having flux recovery ratio of more than 90% and a flux decline ratio of less than 10%. Copyright © 2015 John Wiley & Sons, Ltd.     

Studies on the electrochemical and permeation characteristics of asymmetric charged porous membranes

Asymmetric charged porous membranes were prepared by imbedding 10% (W/W) ion-exchange resin in cellulose acetate binder. Membrane potential and conductance measurements have been carried out in sodium chloride solutions at different concentrations to investigate the relationship between concentration of fixed charges and electrochemical properties of developed nonselective cation- and anion-exchange membranes. Counterion transport number and permselectivity of these membranes were found to vary due to the presence of ion-exchange resin. The hydrodynamic and electroosmotic permeability of sodium chloride solutions has been studied in order to compute equivalent pore radius. For cation- and anion-exchange membranes good agreement was observed between pore radius values estimated from hydrodynamic and electroosmotic permeability coefficient separately, while for nonselective membranes no correlation was found. Membrane conductance data, along with values of concentration of fixed charges, were used for the estimation of the tortuosity factor, salt permeability coefficient, and frictional coefficient between solute and membrane matrix employing an interpretation by nonequilibrium thermodynamic principles based on frictional forces. Moreover, surface morphological studies of these membranes also have been carried out and the membranes were found to be reasonably homogeneous.     

Determination of the pore radius of regenerated cellulose membranes by a dyeing technique

Membrane structure strongly affects the transport of solutes through dialysis membranes. This suggests that knowledge of membrane structure and its effects on permeability is required in order to improve the membranes. Solute transport in membrane pores is limited by steric hindrance at the pore entrances, frictional resistance of the pore walls, and the tortuosity of the pores. Differences in dyeing properties are found among the various tubular dialysis membranes made of regenerated cellulose (RC) that are commercially available. The objective of the present study is to determine intramembrane diffusivity for dyes, and from this the pore radius of RC membranes based on pore model calculations. Values of the pore radius of RC membranes obtained from intramembrane diffusivity data are in disagreement with our previously reported values obtained from solute and pure water permeability data. This indicates that RC membranes are of asymmetrical structure and slightly tight near the outside surfaces.     

用流动电位法表征纳滤膜的结构及性能

利用测量流动电位的方法考察了纳滤膜的表面电学性能对纳滤膜的截留性能的影响.首先,采用不同功能层材料制备了复合纳滤(NF)膜,考察功能层的交联时间、单体结构等对表面电性能的影响,研究纳滤膜对不同无机盐的选择截留性能与表面电性能的关系.通过流动电位法测定纳滤膜的表面电学参数,如流动电位(ΔE)、zeta电位(ζ)和表面电荷密度(σd).实验表明,这些电学参数的变化与功能层交联时间和纳滤膜截留率的变化一致,在交联时间为45 s时,3种电学参数的绝对值均最大,而纳滤膜对无机盐的截留率也最大.复合纳滤膜zeta电位的绝对值(|ζ|)按照Na2SO4>MgSO4>MgCl2变化,同截留率的变化相同.带侧基单体交联后得到的纳滤膜的表面电性能参数的绝对值小于不带侧基单体的.因此,流动电位法可用于研究复合纳滤膜的截留机理和功能层结构.     

Experimental analysis, modeling, and theoretical design of McMaster pore-filled nanofiltration membranes

A side-by-side comparison of the performance of McMaster pore-filled (MacPF) and commercial nanofiltration (NF) membranes is presented here. The single-salt and multi-component performance of these membranes is studied using experimental data and using a mathematical model. The pseudo two-dimensional model is based on the extended Nernst–Planck equation, a modified Poisson–Boltzmann equation, and hydrodynamic calculations. The model includes four structural properties of the membrane: pore radius, pure water permeability, surface charge density and the ratio of effective membrane thickness to water content. The analysis demonstrates that the rejection and transport mechanisms are the same in the commercial and MacPF membranes with different contributions from each type of mechanism (convection, diffusion and electromigration). Solute rejection in NF membranes is determined primarily by a combination of steric and electrostatic effects. The selectivity of MacPF membranes is primarily determined by electrostatic effects with a significantly smaller contribution of steric effects compared to commercial membranes. Hence, these membranes have the ability to reject ions while remaining highly permeable to low molecular weight organics. Additionally, a new theoretical membrane design approach is presented. This design procedure potentially offers the optimization of NF membrane performance by tailoring the membrane structure and operating variables to the specific process, simultaneously. The procedure is validated at the laboratory scale.     

Correlation Between Structure and Transport Properties of Polymeric Membranes for Immunoisolation

Laboratory-made asymmetric polyurethane membranes designed for immunoisolation were investigated. Two types of EK and ES membranes were prepared in different spinning conditions. The membrane structure was characterised by the skin pore radius measurements using differential scanning calorimetry (DSC). Diffusive transport properties of membranes were determined by in vitro method for albumin and creatinine. The scanning electron microscopy (SEM) was applied to study the morphology of membranes. It has been found that the DSC technique is a useful tool for the evaluation of pore radii in the skin layer for PU membranes. Calculated pore radii were in the range from 1.95 to 2.47 nm for the EK and ES types. A correlation between the skin pore radii and the transport properties was not found in this case of investigated membranes. However, the transport properties data can serve for the estimation of selectivity of membranes. Thus, the selectivity of membranes for solutes of various molecular size was estimated from the D _m/D _w ratio of diffusion coefficients for albumin and creatinine. The SEM micrographs reveal the finger-like internal structure of capillary membranes, as well as various skin thickness. This revised version was published online in July 2006 with corrections to the Cover Date.     

The effect of gel layer thickness on the salt rejection performance of polyelectrolyte gel-filled nanofiltration membranes

The effect of gel layer thickness on salt separation of positively charged pore-filled nanofiltration membranes has been examined both theoretically and experimentally. The extended Nernst-Planck (ENP) equation coupled with the Teorell-Meyer-Sievers (TMS) model were used to calculate the pressure-driven sodium chloride rejections for membranes having gel densities in the range typically used in nanofiltration applications. It was found that salt rejection was dependent on membrane (gel-layer) thickness with salt rejections increasing rapidly with thickness up to 50–75 μm. Further increases in thickness beyond this point had a much smaller effect on salt rejection. The theoretical predictions were examined experimentally by preparing a series of membranes with cross-linked poly(3-acrylamidopropyl)-trimethylammonium chloride (PAPTAC) gels with varying densities within the pores of a thin microporous polyethylene (PE) support. The membranes were characterized by their polymer volume fractions (gel concentration), thicknesses and effective charge densities. The effect of membrane thickness was examined by using single and stacks of two membranes. The pure water fluxes and salt rejections of the membranes and membrane stacks were determined in the pressure range 50–550 kPa. The single salt rejections of the membranes which were very dependent on the thickness of the membrane or membrane stack, were fully in accord with the calculated salt rejections of the membranes.     

Solute rejection by porous thin film composite nanofiltration membranes at high feed water recoveries

The authors developed a rigorous framework to model nanofiltration (NF) membrane selectivity at high feed water recoveries and verify it experimentally. The phenomenological model and the Donnan steric partitioning pore model (DSPM) were incorporated into a differential element approach for predicting removal of a variety of solutes from single salt solutions and natural water by NF membranes up to 90% feed water recovery in the temperature range 5-41 degrees C. In this approach, the entire membrane ensemble was divided into numerous sub-elements analogous to real-world full-scale NF installations, where concentrate (or reject) from one element feeds into the next element. Fundamental membrane properties (average pore radius, surface charge density, and ratio of thickness to porosity) and the reflection coefficient and permeability coefficient were first independently obtained for each solute-membrane-temperature combination using separate low recovery experiments with negligible concentration polarization and later used as model inputs to calculate solute removal in a purely predictive fashion for 5-90% recovery. This modeling approach accurately predicted removals from single salt solutions of NaCl and MgSO(4) as well as natural organic matter, disinfection by-product precursors, and several ions from pretreated Lake Houston water in a wide range of operating conditions demonstrating its use to simulate NF permeate water quality under real-world conditions of high feed water recovery.     

Calculation of the salt separation by negatively charged gel-filled membranes

The salt separations of negatively charged gel-filled membranes composed of poly(2-acrylamido-2-methylpropanesulfonic acid) gels anchored within a polypropylene microporous substrate have been determined experimentally and modeled theoretically. The separation of these membranes were calculated by both the Teorell, Meyer and Sievers (TMS) model and the Donnan–Steric Pore (DSP) model coupled with the extended Nernst–Planck equation. For modeling, the membrane effective thickness, effective charge density, and pore radius were either directly measured or calculated from theories without the use of fitting procedures. Good agreement between the experimental measurements and the theoretical calculations of salt separation was observed. For the theoretical calculations, the TMS model is suitable for membranes with moderate gel polymer volume fractions, while the DSP model is more suitable for membranes with high gel polymer volume fractions. Moreover, with a calculated constant effective charge density, the salt separation with different salt concentrations could be accurately predicted. The separation of various other salts could also be predicted with good accuracy.     

Comparison of methods for characterizing microporous membranes for plasma separation

To design membranes suitable for therapeutic use, the relationship between membrane structure and permeability needs to be studied. In this work, the solute permeability of small tubular membranes for plasma separation was determined by using radioisotope-labeled solutes. Through analysis of data on solute and pure water permeability and on water content, by means of a tortuous pore model that we have proposed, we can obtain pore diameter, surface porosity and tortuosity. Membrane structure was also analyzed by mercury intrusion and scanning electron microscopy, and the results were compared with each other. The mercury intrusion method is unsuitable for the structural analysis of polymer membranes because of the damage and/or expansion resulting from highly elevated pressure. The tortuous pore model is recommended for the elucidation of membrane structure.     

Study on the thin-film composite nanofiltration membrane for the removal of sulfate from concentrated salt aqueous: Preparation and performance

Thin-film composite (TFC) nanofiltration (NF) membrane was prepared through the interfacial polymerization between piperazine (PIP) and trimesoyl chloride (TMC) on the polysulphone support membrane. The chemical structure of membrane surface was studied by attenuated total reflectance infrared (ATR-IR) and X-ray photoelectronic spectroscopy (XPS). Parametric studies were conducted by varying reaction time, curing temperature, curing time and additives in PIP solution for obtaining the optimum polymerization conditions. Systematic performance studies were conducted with different feed solutions, feed concentrations, feed pHs, operating temperatures and pressures. Continuous and comparative tests were also conducted to determine the performance stability and separation efficiency of the thin-film composite NF membrane prepared. High performance thin-film composite NF membrane for the selective sulfate removal from concentrated sodium chloride aqueous with the water permeability coefficient of 75 L/(m² h MPa) could be prepared under specific conditions. Experimental results on concentrated mixed solution of NaCl and Na₂SO₄ demonstrated that the NF membrane developed could be successfully used for the removal of sodium sulfate from the concentrated brine of chloralkali industry with high permeate flux, selectivity and performance stability.     

Influence of steric, electric, and dielectric effects on membrane potential

The membrane potential arising through nanofiltration membranes separating two aqueous solutions of the same electrolyte at identical hydrostatic pressures but different concentrations is investigated within the scope of the steric, electric, and dielectric exclusion model. The influence of the ion size and the so-called dielectric exclusion on the membrane potential arising through both neutral and electrically charged membranes is investigated. Dielectric phenomena have no influence on the membrane potential through neutral membranes, unlike ion size effects which increase the membrane potential value. For charged membranes, both steric and dielectric effects increase the membrane potential at a given concentration but the diffusion potential (that is the high-concentration limit of the membrane potential) is affected only by steric effects. It is therefore proposed that membrane potential measurements carried out at high salt concentrations could be used to determine the mean pore size of nanofiltration membranes. In practical cases, the membrane volume charge density and the dielectric constant inside pores depend on the physicochemical properties of both the membrane and the surrounding solutions (pH, concentration, and chemical nature of ions). It is shown that the Donnan and dielectric exclusions affect the membrane potential of charged membranes similarly; namely, a higher salt concentration is needed to screen the membrane fixed charge. The membrane volume charge density and the pore dielectric constant cannot then be determined unambiguously by means of membrane potential experiments, and additional independent measurements are in need. It is suggested to carry out rejection rate measurements (together with membrane potential measurements).