Modification of track-etched membrane structure and performance via uniaxial stretching
MetadataShow full item record
The objective of this research is to demonstrate changes in performance characteristics of track-etched microfiltration (MF) membranes by uniaxial stretching, and to predict changes in the characteristics of pore entrances, such as major and minor axis, aspect ratio, and pore area, from known membrane properties and stretching conditions. Pore size characteristics (distribution of lengths of the major and minor axes and aspect ratio) impact membrane performance (flux, rejection, and permeate particle size distribution). Polyester (PET) track-etched (TE) membranes (Whatman) with uniform pores sizes ranging from 0.2 to 10 µm have been stretched at 80C < T< 170C with total strains up to 40 %. Surface pore characteristics were measured using SEM photos and digital image analysis. For the PET 1 µm membrane, an 11 % stretch increased pore area 31 %, increased the major axis 55 %, decreased the minor axis 14 %, and thereby increased the aspect ratio 78 %. Aspect ratios (ratio of major axis length to minor axis length) increased up to 100 %. Pure water flux was shown to improve upon stretching for some samples, but almost all samples exhibited improvement in flux with particle challenge. For the 1 µm membrane, resistance was shown to increase upon stretching through a Hermia analysis, and flux increased by 100 % after 1 m3 /m2 of normalized cumulative volume throughout. vii Particle rejection was shown to increase in some cases and stay the same or decrease in others. The mathematical model relating final membrane structure (deformation) to initial membrane characteristics and stretching conditions uses a constitutive model for membrane material that accounts for non-linear behavior under stress. The material properties are described mathematically by the experimentally determined Prony series, used to predict effects of stretching on pore size and shape. Additionally, uniaxial stressstrain data are used to fit a hyperelastic model, like the Marlow or Arruda-Boyce. The model proved to be useful in predicting pore size and shape based on material properties and stretching conditions, especially for samples of mid-range porosity, or approximately 6 %. The model did not, however, predict the permanent nature of the deformation successfully, allowing the sample to rebound in the simulation significantly more than is observed experimentally.