MEMBRANE MODIFICATION: Performance Enhancement via Membrane Stretching

Recent publications have shed some light on how flux and rejection in microfiltration (MF) and ultrafiltration (UF) processes can be influenced by the shape of the pores on the membrane surface.  In particular, membrane performance may be impacted by the aspect ratio of the surface pores, where aspect ratio is the ratio major axis length / minor axis length.  Our research expands upon these earlier studies by systematically varying the aspect ratio of membrane surface pores and investigating the relationship between membrane performance and pore aspect ratio.  The long-term objective of this research is to improve membrane performance by controlling the aspect ratio of pores on the surface of MF and UF membranes.  Specifically, the research:

·               establishes quantitative relationships between membrane pore size characteristics (maximum, minimum, and mean values of the major axis, minor axis, and aspect ratio as well as surface porosity) and membrane performance (flux and rejection).

·               develops and verifies a mathematical model to relate final membrane structure to initial membrane structure, membrane material mechanical properties, and membrane stretching conditions.

Strategy

The research is designed to achieve the stated objectives by accomplishing the following tasks:

1.      Experimentally stretch membranes, measuring initial and final surface pore size characteristics.

2.      Conduct particle challenge experiments on pre- and post-stretched membranes to determine the effects of pore size characteristics on membrane performance and fouling layer accumulation.

3.      Conduct permeation experiments to determine the effects of pore size characteristics on fouling by protein fouling, oily water, and NOM fouling.

4.       Predict or experimentally determine the mechanical properties of the membrane materials (e.g., Prony series constants) important in modeling the stretching.

5.      Develop a model to simulate the stretching of microporous membranes, thereby relating final pore size characteristics to initial pore size characteristics, membrane material, and stretching process parameters (rate and extent of stretching, thermal conditions).

6.      Compare the experimental results to the simulation results and refine the simulation accordingly.

Simulations have been developed for track-etched (T-E) and phase inversion / phase separation (PS) membrane stretching process.  These simulations, which were created using finite element methods in ABAQUS software, use baseline information on bulk membrane properties to predict the effects of stress on pore shape and size.  Specifically, polyethylene teraphthalate (PET) T-E membranes with pore sizes of 0.6, 2, and 10 μm have been investigated at elongations up to 40%.  The simulations take into account pore density, angles of pores, pore interactions, and material properties.  Quantitative results from these simulations match up well with experimental data as seen in Figure 1. 

Figure 1.  The major axis and minor axis from simulations for 2 micrometer PET track etch membranes shows fairly good agreement with experimental values. 

The highly complex structure of PS membranes forced a different approach to modeling their deformation.  The deformation process for PS membranes can be broken into three stages.  The first stage entails the alignment of all of the membrane pores in the direction of stretching.  The second stage of deformation exhibits a uniform increase in pore aspect ratio with increasing strain.  The third stage of the deformation process begins when the micro-structure of the membrane (the individual polymer filament) break, opening large macro-voids in the membrane.  The first stage of the deformation was analyzed with a single pore continuum model.  This model provided insight into the level of strain required to completely align all of the membranes pores in the direction of strain.  The second stage of the deformation was modeled using a geometrically arrayed highly idealized beam model.  The results from these two models are given in Figure 2.  The models showed excellent qualitative agreement with experimental data.  The simplifications required to model the complex shape of the PS membranes prevented exact quantitative agreement between the model and experimental data.  The third phase of membrane deformation could not be modeled because ABAQUS did not allow for the breaking of individual polymer filaments.

Figure 2 – Comparison of experimental data for 0.2 μm PVDF to model predictions

             Atomic force microscopy (AFM) has been employed to determine the effect of stretching on membrane surface properties of PI membranes.  AFM maps the surface of a substrate by tapping it with a microscopic cantilever.  The use of this instrument has allowed for precise characterization of membrane surface pores as shown in Figure 3.  As shown in Figure 3, stretching PS membranes results in a dramatic change in surface morphology. Standard methods of analyzing surface morphology, such as roughness and bearing analysis were unable to capture the differences between stretched and non-stretched membranes.  It is hypothesized that the change in the surface morphology due to stretching could be exploited in applications such as cross flow filtration.  It can be imagined that particles traversing the membrane surface would behave very differently if crossing the membrane in the direction of stretching versus crossing the membrane perpendicular to the direction of stretching.  Current research efforts are directed toward developing the appropriate membrane performance challenges to determine the effects of flow direction on stretched membranes.

Figure 3 – AFM surface plots of 0.2 μm PVDF both non-stretched (A) and stretched (B).

         After membranes have been stretched it is important to determine how their performance is affected by the change in pore structure.  The change in flux and flux decline was measured initially using dead end filtration with silicon dioxide particles.  Stretching has been shown to increase flux through all membranes.  Another important result of T-E membrane stretching is that membranes show a decreased flux decline during particulate flow as shown in Figure 4.  Figure 4 also shows that excellent repeatability has been achieved between experiments.   Similar results have been seen for other pore sizes of PET track-etched membranes.  For phase inversion membranes, stretching has been shown to dramatically increase pure water flux through PVDF membranes as shown in Figure 5.  Membrane fouling results for phase inversion membranes using a variety of different foulants: silicon dioxide particles, latex micro-spheres, oil-water, and natural organic matter have not shown dramatic reductions in fouling. 

Figure 4 – Effect of stretching on flux decline for 1 μm PET membrane using Min-u-sil 5 silicon dioxide particles at a concentration of 10 mg/L and a pressure of 137 kPa.

Figure 5 – Effect of stretching on pure water flux for 0.2 μm PVDF membrane at a pressure of 137 kPa.

Publications from this project:

• J.A. Morehouse, L.S. Worrel, D.L. Taylor, B.D. Freeman, and D.F. Lawler, “The Effect of Uni-axial Orientation on Macroporous Membrane Structure.”  Journal of Porous Materials, 13, 61–72 (2006).
• J.A. Morehouse, D.L. Taylor, D.R. Lloyd, D.F. Lawler, B.D. Freeman, and L.S. Worrel, “The Effect of Uni-axial Stretching on the Roughness of Microfiltration Membranes.”  J. Membr. Sci., 280, 712–719 (2006).
• J.A. Morehouse, D.R. Lloyd, B.D. Freeman, D.F. Lawler, K.M. Liechti, and E.B. Becker, “Modeling the Stretching of Microporous Membranes.”  J. Membr. Sci., 283, 430–439 (2006).
• L.S. Worrel, J.A. Morehouse, L.A. Shimko, D.R. Lloyd, D.F. Lawler, and B.D. Freeman, “Enhancement of Track-etched Membrane Performance via Stretching.” Separations and Purification Technology,  53 71–80 (2007).
• P.L. Hanks, K.J. Kaczorowski, E.B. Becker, and D.R. Lloyd, “Modeling of uni-axial stretching of track-etch membranes.”  J. Membr. Sci.  305, 196–202 (2007).