Effect of filler incorporation route on the properties of polysulfone–silver nanocomposite membranes of different porosities

Abstract

Flat sheet porous polysulfone–silver nanocomposite membranes were synthesized by the wet phase inversion process. The effects of casting mixture composition and nanoparticle incorporation route on the morphological and separation properties of prepared membranes were studied by comparing nanocomposites of different preparations with silver-free controls. Silver nanoparticles were either synthesized ex situ and then added to the casting solution as an organosol or produced in the casting solution via in situ reduction of ionic silver by the polymer solvent. Nanocomposite membranes of three types differing in skin porosity and macrovoid structure were prepared. The structure and properties of nanocomposites were interpreted in terms of the coupling between the processes of nanoparticle formation and gelling of the polymer-rich phase during phase inversion. Larger nanoparticles preferentially located in the skin layer were observed in composites prepared via the ex situ method while in situ reduction of silver led to formation of smaller nanoparticles homogeneously distributed along the membrane cross-section. In some cases, incorporation of nanoscale silver formed ex situ resulted in macrovoid widening and an order of magnitude decrease in hydraulic resistance accompanied by only a moderate decrease in rejection. The accessibility of the silver nanoparticles embedded in the membrane was quantitatively assessed by the degree of the growth inhibition of a membrane biofilm due to the ionic silver released by the nanocomposites and was found to depend on the method of silver incorporation.

Introduction

Porous polymer nanocomposite membranes can be prepared using a number of nanoparticle incorporation strategies. These strategies can be grouped into four categories based on whether particles and membranes are pre-formed or synthesized in situ during the formation of the nanocomposite. Methods that involve attachment or in situ synthesis of nanoparticles on the pore surface of existing membranes [1], [2], [3], [4], [5], [6], [7] have the advantages of improved reproducibility of nanocomposite preparation, the possibility to reuse the membrane matrix or to regenerate nanoparticles, and better accessibility of immobilized nanoparticles to reactants in the permeate flow. In contrast, incorporation of nanoparticles during membrane formation [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21] is appealing because it should allow the design of nanocomposite structures with improved mechanical and separation properties along with embedded nanoparticle-based functionalities. The structure of the resulting membranes is generally a function of the physical and chemical properties of the polymer matrix and nanoparticles as well as the method of nanoparticle incorporation.

A number of published studies describe the formation of organic–inorganic porous (10–200 kDa) composite membranes by adding particles (Al₂O₃ [18], [19], ZrO₂ [11], [14], TiO₂ [15], [20], SiO₂ [8], [9], [12], [22], Ag⁰ [13], [21], [23]) to the solution of the membrane-forming polymer (cellulose acetate (CA) [13], [18], [23], polysulfone (PSf) [8], [9], [14], [17], [20], polyvinylidene fluoride (PVDF) [11], [12], [19], polyacrylonitrile (PAN) [21], polyamide-imide (PAI) [24], poly(vinylidenefluoride-co-hexafluoropropylene) [22], or poly(phthalazinone ether sulfone ketone) (PPESK) [15]). These studies indicate that skin layer properties (porosity, surface porosity, thickness) as well as the macrovoid morphology of the support layer can be affected by nanoparticle incorporation. Both of these membrane regions are important; the skin layer determines the permeability, rejection, and selectivity of membranes, while the structure of the support layer affects its compaction behavior [25], [26]. Although nanoparticle-induced changes in membrane structure are specific to the particular filler/matrix combination, certain common trends can be identified. Increasing nanoparticle loading led to: (1) increased skin layer thicknesses (SiO₂/PSf [8], [9], TiO₂/PSf [20], TiO₂/PPESK [15]), (2) higher surface porosity of the skin (ZrO₂/PSf [14], TiO₂/PSf [20], TiO₂/PPESK [15], SiO₂/PSf [9]), (3) suppressed macrovoid formation (SiO₂/PSf [8], [9], TiO₂/PSf [20], Al₂O₃/CA [18], TiO₂/PPESK [15]), and (4) higher permeability of the membrane (ZrO₂/PSf [14], SiO₂/PVDF [12], ZrO₂/PVDF [11], Al₂O₃/PVDF [19], Ag/CA [13], SiO₂/PSf [9], TiO₂/PSf [20]. In a few cases, decreased permeability (Ag/PAN [21]) or a maximum permeability at (TiO₂/PPESK [15]) intermediate nanoparticle loadings was also reported. The rejection either peaked at intermediate loadings (TiO₂/PSf [20]), or decreased after a threshold in filler loading was exceeded (SiO₂/PSf [9], SiO₂/PVDF [12], ZrO₂/PVDF [11], TiO₂/PPESK [15]), or remained unchanged (ZrO₂/PSf [14], Al₂O₃/PVDF [19]). In a recent study, nanotube fillers were used to prepare a nanocomposite membrane when polysulfone was embedded with (acid-treated) multiwall carbon nanotubes [27]. The average porosity of the composite membranes peaked at a nanotube loading of 1.5%, which also corresponded to the highest permeate flux. Interestingly, the only pronounced change in membrane macroscopic morphology was an increase in surface roughness.

Examples of simultaneous formation of the polymer matrix and the nanoparticles include chemical reduction of metal ions by a component of the membrane casting solution [13], [21], [28]. Meyer and Bhattacharyya [28] reported on the formation of Fe/Ni nanoparticles in porous cellulose acetate membranes by in situ reduction wherein the reducing agent (NaBH₄) was a component of the non-solvent bath. Yu et al. [13], [21] prepared silver-loaded cellulose acetate and polyacrylonitrile hollow fibers by the reduction of Ag⁺ by N,N-dimethylformamide, a polymer solvent, with the objective of producing biofouling-resistant ultrafiltration membranes.

The results of the studies published to-date strongly imply that the microstructural changes induced by nanoparticle incorporation can potentially be controlled to prepare membranes with desirable properties. To enable the rational design of porous nanocomposite membranes, the role of essential process variables such as nanofiller size (relative to the characteristic size of matrix pores), the affinity of the filler material for the components of the casting mixture, and the effect of the matrix pores functioning as reactors for forming filler nanoparticles need to be understood.

The overall goal of this study was to understand how the preparation conditions (nanoparticle incorporation route, porosity of the host matrix, etc.) influence the structure of the prepared nanocomposite membranes. To this end, we attempted to map nanocomposite membrane properties to preparation conditions for a series of membrane morphologies and two nanoparticle incorporation routes. Three types of membranes differing in skin porosity and macrovoid structure were prepared, and silver nanoparticles were either synthesized ex situ and then added to the casting solution as an organosol or produced in the casting solution via in situ reduction of ionic silver during the phase inversion process. Polysulfone was chosen as the matrix material due to its practical importance and wide use in the preparation of ultrafiltration membranes, while nanoscale silver was chosen for two reasons. First, there are established protocols for the preparation and characterization of silver sols in various dispersion media, including the organic solvents used in membrane preparation. Second, the presence and availability of silver can be detected by the extent of its antibacterial effect; this provides a convenient framework for the evaluation of the accessibility of silver in polymer matrices and serves as additional motivation in view of the potential application of silver-filled nanocomposites as biofouling-resistant materials.