Asymmetric hollow fiber (HF) membranes have become ubiquitous in their applications in ultra- and microfiltration, gas separation, and reverse osmosis. HFs illustrate many favorable characteristics, namely the larger surface and separation area due to the geometry of the HFs, relatively high mechanical stability, and are uncomplicated to handle and use [1]. The HF membranes observed in this work were produced through the nonsolvent-induced phase separation (NIPS) process.
The NIPS process comprises the interactions between multiple (usually three) components, in which a homogenous system of a chosen polymer (P), solvent (S), and any desired additives (dope solution) undergo phase separation with a specific geometry (dictated by the type of spinneret utilized) in a nonsolvent (NS; most typically water) [2]. Due to the immiscibility of the polymer in the water, diffusion is initiated, leading to phase separation. This, in turn, leads to membrane formation [2].
While many parameters play an essential role in the fabrication of HF membranes, only the dope characteristics and their consequences on forces the fiber experiences during spinning are investigated in this work. These properties have an enormous impact on fiber morphology and porosity, and these, in turn, affect fiber performance. While performance with the appropriate testing equipment (gas separation unit/ultrafiltration unit) was investigated, it is not included in the scope of this work. The dope composition and temperature affect the dope’s viscosity, influencing the kinetics of phase inversion and the S-NS exchange. Additionally, the shear and elongational stresses the fiber experiences due to the take-up speed and the gravitational force in the airgap play a significant role in the fiber morphology and porosity. Unfortunately, it is difficult to determine the impact of each phenomenon unambiguously. Thus, the motivation of this work is two-fold: (1) to highlight the significance of the dope characteristics in the precipitation process and final fiber morphology and (2) to investigate the effects of the different forces mentioned above on the end-product independently.
For the dope solution, polyethersulfone (PES; mixed molecular weight) was purchased in pellet form, and synthetic grade solvent 1-methyl-2-pyrrolidone (99,8% NMP; 99,13 g/mol) were both purchased from VWR. PES and NMP were homogenously mixed to create the dope solution. Deionized water (Arium water purification system, Sartorius AG, Goettingen, Germany) was utilized as a borefluid. Distilled water was used for the coagulation bath. The NIPS process plant and the fibers spun using the plant were built and produced in-house (Figure 1). To characterize the HF membranes, the scanning electron microscope (SEM; Coxem E-30Plus, Korea) was utilized. HFs were broken with liquid nitrogen to create a clean cross-section and sputtered as the HFs were not conductive. Geometric porosity was also measured. The dimensions of the spun fibers were measured from the SEM images. A small piece of fiber was weighed using an analytical weighing scale (Kern, Germany), and porosity was calculated together with the fiber dimensions.
The dope characteristics (PES concentration, viscosity, and temperature) that affect solvent exchange and kinetics of membrane formation was investigated. This, in turn, affects fiber morphology and porosity. Significant delay in coagulation of the inner structure leads to thicker fiber dimensions due to the shifting of the binodal curve towards the P-S axis, affecting the kinetics and thermodynamics of the phase inversion diagram. This is precisely the observation that was witnessed in this work.
Dope viscosity, PES concentration, and fiber porosity are undeniably interconnected. Therefore, dope viscosity must be considered when discussing the effects of PES concentration on fiber dimensions, morphology, and porosity.
Table 1 depicts a rising trend in fiber dimensions (with a dip in fiber dimensions at 27 wt% PES) with a simultaneous decreasing trend in porosity with increasing PES concentrations. In addition to this effect, arguably more critical effects need to be accounted for, namely the take-up speed (elongational stresses caused due to the rate at which the fiber is winded) and the die swell phenomenon.
Take-up speeds (as seen in Table 1) were unfortunately not kept constant. During experiments, an attempt was made to set the take-up speed to that the fiber is under an appropriate amount of elongational stress (visually). The take-up speed is supposed to correspond with the free fall velocity at which the fiber emerges from the spinneret. Due to the varying viscosities, the fibers’ free fall velocity also varied. Despite the relatively low variance in the take-up speeds, the differing amounts of elongational stress affect the thickness of the fibers. This effect would describe the sudden decrease in the fiber dimension for identifier C40. B40 was taken up at a lower speed than A40 and C40 and thus is thicker than either identifier. D40 was taken up at a faster speed than the other identifiers and had thicker fiber dimensions than the others. Due to the constant dope extrusion rate (DER, and hence the dope extrusion velocity DEV), each identifier underwent a different stretch ratio (Equation 1). The varying stretch ratios illustrate the significance of the take-up speed on fiber thickness as the (inner diameter/outer diameter) ID/OD ratios of all four identifiers were higher than the spinneret ID/OD ratio. Thus, it is expected that higher stretch ratios lead to even lower fiber thickness in this work. However, this phenomenon of D40 fibers as an outlier can be attributed to the die swell effect.
Due to the varying viscosities, the fibers’ free fall velocity also varied. Despite the relatively low variance in the take-up speeds, the differing amounts of elongational stress affect the thickness of the fibers. This effect would describe the sudden decrease in the fiber dimension for identifier C40. B40 was taken up at a lower speed than A40 and C40 and thus is thicker than either identifier. D40 was taken up at a faster speed than the other identifiers and had thicker fiber dimensions than the others. Due to the constant dope extrusion rate (DER, and hence the dope extrusion velocity DEV), each identifier underwent a different stretch ratio (Equation 1). The varying stretch ratios illustrate the significance of the take-up speed on fiber thickness as the (inner diameter/outer diameter) ID/OD ratios of all four identifiers were higher than the spinneret ID/OD ratio. Thus, it is expected that higher stretch ratios lead to even lower fiber thickness in this work. However, this phenomenon of D40 fibers as an outlier can be attributed to the die swell effect.
With increasing PES concentrations in the dope solution (and thus increasing viscosities), the die swell is expected to have a more considerable impact on the fiber. Literature [3] states that die swell can be reduced by winding the fibers onto a spool, and this exact occurrence was also observed in this work.
Die swell can also be counteracted by elongational stresses the fiber experiences due to gravity between the time they are extruded and get taken onto the first spool (white bottom spool in Figure 1). This effect, however, is almost negligible and eliminated once the fiber is taken up on the winder drum [3]. Therefore, the influence of the take-up speed, elongational stresses due to gravity, and die swell make the comparison of each fiber from different dope concentrations difficult.
Additionally, as seen in Figure 2, all fibers depict long finger-like pores growing from the inner surface. The ratio of the finger pores’ length to the wall thickness seems to decrease as the fiber becomes larger as well. The density of finger-like pores from the outer surface becomes larger with increasing PES concentration. The decrease in finger-like pores and the creation of spongy structures with higher concentrations is an observation also recorded in literature.
This work aimed to investigate the effects of the physical effects the fibers experience during the spinning process. It was found that the take-up speed, die swell, and the elongational stresses due to gravity fundamentally change the morphology of the HFs. This work has shown that further optimization is necessary to investigate the effects listed in this work. Finally, additional dope characteristics like temperature and viscosity should be further examined, as they have an equally crucial impact on the fiber structure. Such investigation and examination are necessary to understand the complicated membrane formation and its concrete effects on hollow fiber membranes.
[1] Z.-P. Wang et al., “A novel polysulfate hollow fiber membrane with antifouling property for ultrafiltration application,” Journal of Membrane Science, vol. 664, p. 121088, Dec. 2022, doi: 10.1016/j.memsci.2022.121088.
[2] H. Strathmann, Introduction to Membrane Science and Technology. Wiley-VCH Verlag GmbH & Co., 2011.
[3] N. Bolong, A. Ismail, and M. Salim, “Effect of Jet Stretch in the Fabrication of Polyethersulfone Hollow Fiber Spinning for Water Separation,” Journal of Applied Membrane Science & Technology, vol. 6, Nov. 2017, doi: 10.11113/amst.v6i1.50.
