Preparation and characterization of an oxygen permselective polydimethylsiloxane hollow fibre membrane

Abstract

The simple preparation of a polydimethylsiloxane (PDMS) hollow fibre membrane through a condensation reaction between the hydroxyl-end groups and hydride groups of polysiloxane reactants over a porous support is reported herein. Two types of polysulfone hollow fibres with different pore asymmetry were used as the supports. A uniform PDMS membrane top-layer with PDMS intrusion inside the porous support exhibited excellent gas permeability and O₂/N₂ selectivity and showed more promise for air separation than the other membrane coating the denser support. The advantage of this membrane type is its preparation from a dilute solution system in contrast to other membrane preparation methods which involve a highly viscous solution containing odorous volatile organics.

---

Introduction

Air separation by a membrane to produce nitrogen is a competitive separation technology because of its advantages over the conventional separation systems of cryogenic distillation and adsorption processes, which include reliability, small carbon footprint and modular units.¹ The first commercial air separation plant based on a proprietary hollow fibre membrane was installed by Monsanto and later by Air Products. Now, several other companies, including DuPont, Dow, Ube, etc., have produced their own proprietary commercial air separation membrane systems. This separation process is crucial and challenging as the kinetic diameter of oxygen (3.46 Å) is similar to that of nitrogen (3.64 Å). Many membrane materials such as polysulfone (PSf), polyethersulfone (PES), polyimide, polyacrylonitrile, ethyl cellulose, polycarbonate, polyvinylacetate, polyether-block-amide, poly(4-methyl-1-pentene), poly(vinyltrimethylsilane), poly(phenylene oxide) as well as mixed matrix membranes have been investigated to prepare improved air separation membranes.^2,3 A hollow fibre membrane configuration is considered to be more suitable than other flat-sheet configured modules in terms of productivity flux, compactness, ease of fabrication and maintenance. Chung et al.⁴ studied a 6-FDA–durene polyimide hollow fibre membrane which exhibited O₂ permeance in the range of 430–1512 gas permeation units (GPU) with an O₂/N₂ selectivity of 1.1–1.2. Upon coating this 6-FDA–durene polyimide membrane with silicone rubber, the membrane O₂/N₂ selectivity increased to 3.2 with O₂ permeance of 64 GPU. On the other hand, Clausi et al.⁵ observed an O₂/N₂ selectivity of 6.5–6.6 with O₂ permeance in the range of 0.7–18.1 GPU for different Matrimid® 5218 polyimide hollow fibres prepared by varying the spinning dope composition containing volatile tetrahydrofuran. The differences in the gas permeance and selectivity of such polyimide hollow fibres could arise from the differences in their membrane structures. Even though the polyimide hollow fibre exhibited a good membrane performance for air separation, it has the disadvantages of being high cost and brittle. Cost effective membranes include either PES–polyimide blend hollow fibres obtained by spinning from a spinning dope composition containing the desired amounts of PES and polyimide⁶ or a dual layer hollow fibre membrane comprising of an expensive polymer as the top layer and cost effective PSf or PES as a support membrane.^7–9

PSf is a glassy polymer, economically cheap with good mechanical strength and is soluble in a wide variety of solvents for solution casting/spinning in the form of membranes. Asymmetric PSf hollow fibres comprising a dense top layer and porous sub-layer spun from a highly viscous PSf solution (62 000 cP) showed an O₂/N₂ selectivity of 5.8 with an O₂ permeance flux of 8.8 GPU.¹⁰ The asymmetric PSf hollow fibres were further modified with gradient pores by spinning them from dope solutions containing different volatile organic additives such as formylpiperidine, formamide, glycerol, acetic acid, propionic acid and butyric acid. However, the disadvantage of such PSf hollow fibres was the difficulty in its spinning process as the polymer dope was highly viscous and difficult to degas. In cases where the dope contained volatile organics, it was more problematic because of its odorous nature besides its high viscosity. In addition, such PSf fibres commonly contained surface pore defects which required post-modification by immersing the hollow fibres in a dilute polydimethylsiloxane (PDMS) solution to exhibit selectivity towards air separation.^11–17

In the present work, a PDMS hollow fibre membrane was prepared using a porous PSf hollow fibre as support. The PSf was conveniently spun from a dilute polymer solution at room temperature. The membrane structure was varied in terms of the top PDMS membrane layer thickness and intrusion depth of PDMS inside the PSf support. The prepared membranes were thoroughly characterized and explored for their potential in air separation.

Experimental

Materials and methods

Hydroxyl terminated poly(dimethylsiloxane) (HPDMS), viscosity 18 000–22 000 cSt; poly(methylhydro)siloxane (PMHS) viscosity 12–45 cSt and the catalyst dibutyltindilaurate from Sigma-Aldrich Chemical Company, PSf Udel P-3500 from Solvay Advanced Polymers, U.S.A and poly(vinylpyrrolidone) (PVP) from SRL Chemicals, India were purchased. The other chemicals potassium acetate (KAc), n-heptane, N,N-dimethylformamide (DMF), N,N-dimethylacetamide (DMAc) and tetrahydrofuran (THF) were purchased from SRL Chemicals, India and they were of analytical reagent grade. All the purchased chemicals were used as received without any further purification. The nitrogen (N₂ 99.99 vol%) and oxygen (O₂ 99.99 vol%) gas cylinders were obtained from the Parshav Chemicals India.

Preparation of the PSf hollow fibre porous supports from different spinning dope compositions

The desired amount of PSf granules (18%, w/w) after drying at 80 °C for 24 h was dissolved in DMF (80%, w/w) under continuous stirring at 80 °C for 10 h. PVP (2%, w/w) was added and dissolved in the above solution to form a transparent dope mixture comprising PSf, PVP and DMF which was suitable for the hollow fibre spinning process. The prepared dope solution was allowed to cool by standing at room temperature overnight to obtain a bubble free transparent spinning dope solution. Another spinning dope composition was prepared by dissolving PSf granules 22%, w/w in a mixed solvent 78%, w/w of DMAc and THF in which the w/w ratio of DMAc to THF was 1 : 1. The differences between the two spinning dope solutions were (i) the use of water soluble PVP as the additive in the PSf–PVP–DMF dope; (ii) in the other case of spinning dope composition, the PSf concentration was higher, no PVP (pore former) was added and a mixed solvent of high boiling point (165 °C) DMAc and low boiling point (66 °C) THF was used. In the first case, water soluble PVP may leach out in the coagulating water bath of the spinning process which could generate a more porous hollow fibre structure, whereas, in the other case a less porous hollow fibre structure could be obtained due to a combination of a higher PSf concentration and the fast evaporation of THF during the spinning process.

The above transparent spinning dope was extruded through a double orifice spinneret under a nitrogen environment to obtain asymmetric PSf hollow fibres using a hollow fibre spinning machine reported earlier.¹⁸ Hereinafter, the hollow fibre spun from the PSf–PVP–DMF dope is designated PSf-P while the hollow fibre spun from PSf–DMAc–THF dope is designated PSf-D. The spinning conditions are given in Table 1. The inner coagulant flowed through the inner nozzle of the spinneret acting as a bore former by causing phase inversion in the inner part of the extruded hollow fibre, while the phase inversion in the exterior part of the extruded hollow fibre occurred in the coagulating water bath. In the case of the PSf-P fibre spinning, the inner coagulant (bore fluid) was 20% KAc in water to reduce the water activity in the lumen of the PSf-P fibre as compared to pure water for PSf-D. Another difference in the spinning conditions was the slow winding of the PSf-D fibres because of the decreased rate of diffusional exchange between the solvent and non-solvent during the phase inversion process as compared to that of the PSf-P fibres.

Table 1 Spinning parameters of the PSf hollow fibres studied

Spinning parameters	PSf-P	PSf-D
Dope composition (%, w/w)	18% PSf + 2% PVP + 80% DMF	22% PSf + 39% DMAc + 39% THF
Spinneret: O.D./I.D.	1000 μm/650 μm	1000 μm/650 μm
Spinning pressure	68.94 kPa	68.94 kPa
Extrusion rate	8.9 g min^−1	8.9 g min^−1
Inner coagulant	20% KAc in water	Pure water
Bore fluid flow rate	20 ml min^−1	25 ml min^−1
External coagulant	Pure water	Pure water
Wind-up rate	30 m min^−1	5 m min^−1
Coagulant temp.	25 °C	25 °C
Air gap or gap distance	5 cm	5 cm

---

The prepared hollow fibres were washed thoroughly with deionised water and kept in methanol for two days to remove the trapped solvent and PVP completely from the hollow fibres and air-dried at room temperature for one day prior to their use in the membrane modules.

Preparation of the hollow fibre PSf composite membranes

The above PSF-P and PSF-D porous hollow fibres were modified by coating the PDMS layer to obtain a composite membrane structure using the coating procedure described below. A PDMS coating solution was prepared by dissolving HPDMS pre-polymer liquid, PMHS cross-linker, dibutyltindilaurate (DBTL) catalyst in n-heptane. The compositions of the coating solution are detailed in Table 2.

Table 2 Compositions of the PDMS coating solutions

Sample	HPDMS (g)	PMHS (g)	DBTL (g)	Heptane (g)
P1	4.41	0.44	0.15	95.00
P2	8.83	0.88	0.29	90.00
P3	17.65	1.77	0.58	80.00
P4	22.06	2.21	0.73	75.00

---

The coating solution was absorbed onto the PSf hollow fibre support by immersing the support in the coating solution until an equilibrium absorption at 25 °C followed by keeping the fibre in a vertical position for draining off the excess solution from the outer surface of fibre. The hollow fibre absorbed with the coating solution was transferred into a heating chamber and kept for a desired period wherein the polycondensation reaction occurred within the porous hollow fibre. The solvent n-heptane was gradually removed from the hollow fibre support resulting in the formation of a thin PDMS layer coating on the support. The polycondensation reaction forming the PDMS layer is shown in Scheme 1, wherein the hydroxyl end groups of HPDMS reacted with the hydride groups of PMHS in the presence of the DBTL catalyst liberating hydrogen. The membranes were finally cured at 80 °C to obtain the finished product.

Scheme 1 Polycondensation reaction between the HPDMS pre-polymer liquid and the PMHS cross-linker.

Characterization of the hollow fibre support and membrane

The scanning electron microscopy (SEM) images of the hollow fibre supports and membranes were examined on a JEOL JEM 2100 equipped with energy dispersive X-ray (EDX) spectroscopy at an accelerating voltage of 15 to 20 kV using dried, fractured (for transverse section) and gold sputtered samples. The fractured sample was prepared by immersing the hollow fibre in liquid nitrogen and breaking it at the liquid nitrogen temperature. The water-contact angle measurements of the hollow fibre outer surfaces were carried out on a DSA100 Krűss GmbH instrument. The contact angles were measured 5 times on different areas of each of the hollow fibre surfaces to obtain average values. Attenuated total reflectance infra-red (ATR-IR) spectroscopy studies for the membrane surfaces were recorded with a PerkinElmer Spectrum GX (with a resolution of ±4 cm⁻¹, incident angle 45°). The PDMS coating layer thickness of the composite membrane was calculated by the gravimetric method using eqn (1) which is given below.where t is the membrane thickness; W₁ and A₁ are the weight and area of the coated membrane; W₂ and A₂ are the weight and area of the uncoated membrane support; d is the density of the PDMS material. 10 gravimetric measurements for each membrane were carried out to obtain an average value. A sample area of 2.6 cm² was used for the analysis. Prior to measurement, samples were kept in an oven at 80 °C overnight and the weight of each sample was measured precisely.

Gas permeation measurements

Hollow fibre membrane modules of 10–20 cm in length and 2–4 cm in diameter containing 10–50 fibres were made from PSf hollow fibre supports (PSF-P and PSF-D) and PDMS coated PSf hollow fibres using commercially available epoxy resin as the potting material. N₂ and O₂ gas permeation measurements for the hollow fibre modules were carried out at 0.5–2 bar pressure. A schematic of the experimental setup for the gas permeation study is shown in Fig. 1.

Fig. 1 Schematic presentation of the experimental set-up for gas permeation study. (1) Needle valve, (2) pressure gauge, (3) three way valve, (4) three way valve with needle, (5) bubble flow meter.

The feed gas was provided from the compressed gas cylinders, which was then passed through moisture and hydrocarbon traps. The applied pressure at the feed side was controlled by pressure regulators and mass flow controllers. The downstream pressure at the permeate side was atmospheric pressure.

The temperature was maintained at 25 °C. Flow rate measurements for the feed and permeate stream were made either by soap-film bubble flow meters or gas chromatograph (GC) TRACE 1110 Thermo Fisher Scientific equipped with Porapaq Q GC column. At least 5 single gas permeation measurements were performed for each membrane at each operating pressure and the error bars were calculated. The stage cut which is the ratio of permeate flow rate to feed flow rate was kept low so that no concentration polarization effect occurred. The gas permeance was expressed in GPU by using the following equations.

GPU = 10⁻⁶ (cm³ cm⁻² s⁻¹ cmHg⁻¹) (3)

where P/L is the permeance, Q represents the flux or volume, A is the area of the membrane, T is time and ΔP is the applied pressure. The ideal O₂ to N₂ selectivity of the membrane was calculated by taking the ratio of the O₂ and N₂ gas permeance.

Result and discussion

SEM studies and pore size measurements of the PSf hollow fibre supports spun from different PSF–PVP–DMF and PSF–DMAc–THF spinning dopes

SEM images depicted in Fig. 2 reveal the microstructure morphology of PSF-P 1.3 mm hollow fibres spun from PSF–PVP–DMF dope and PSF-D 1.7 mm hollow fibre spun from PSF–DMAc–THF dope. The smaller diameter of the PSF-P was due to a higher stretching of the hollow fibre during the spinning process (wind-up rate 30 m min⁻¹ of PSF-P as compared with 5 m min⁻¹ of PSF-D).

Fig. 2 SEM images of the PSf-P and PSf-D hollow fibre supports; full cross sections of the hollow fibres (a and d); magnified cross sectional images near the lumen surface (b and e) and near the outer surface (c and f).

The PSF-P exhibited the asymmetric porous structure of a thinner skin with large pore finger-like macrovoids or channels in the interior, while the PSF-D exhibited a less porous structure as shown more clearly in the cross-sectional SEM images near the inner (Fig. 2b and e) and outer (Fig. 2c and f) surfaces of the hollow fibres. The higher porosity of the PSF-P hollow fibres may have resulted from (i) the lower concentration of the PSf solution in the spinning dope and (ii) the PVP additive which has leached out from the hollow fibre into the coagulating bath generating more pores.

The leaching of the water soluble pore former PVP during the phase inversion process also contributes to the formation of the porous structure.¹⁹ Furthermore, the cross-sectional PSF-P hollow fibre microstructure near the inner surface contained only a few larger macrovoids as compared to numerous large pore finger-like channels in the interior near the outer surface of the PSF-P. This was due to differences in the diffusional exchange rate between the solvent (DMF) and the non-solvent (water) in the lumen and shell sides of the PSF-P hollow fibre during the phase inversion process because of 20% KAc in water as the inner coagulant, wherein water activity in the lumen side was lower than that of pure water used as an external coagulant in the shell side. The pores are created during the phase separation because of non-solvent intrusion, surface rupture and instability in the local surface.²⁰ The use of KAc in water as the inner coagulant to control diffusional exchange between the polymer solvent (DMF) and non-solvent (water) during the phase inversion process of the hollow fibre formation has been reported earlier.²¹

On the other hand, the less porous nature of the PSF-D might have resulted from the higher PSf concentration and absence of the pore former PVP. In addition, the mixed DMAc–THF solvent in the dope influenced the solvent evaporation and polymer coagulation during the phase inversion process. The use of the mixed solvent controlling the solvent evaporation and polymer coagulation rates has been observed in some other cases.¹⁰ The faster loss of the more volatile solvent during the spinning process causes coalescence of polymer aggregates and rearrangement of polymer chains resulting in the denser (less porous) structure.

The pore sizes of the PSF-P and PSF-D hollow fibres were determined from N₂ permeation data using eqn (4) to (6) given below.^22,23

or,

J_i = K₀ + P₀p̄ (5)

where R is the gas constant (8.314 J mol⁻¹ K⁻¹); T is the absolute temperature (K); M is the molecular weight of the gas (kg mol⁻¹); r is the mean pore radius (m); L_P is the effective pore length (m); μ is the viscosity of the gas (Pa s); p̄ is the mean pressure (Pa); and J_i is the gas permeance (mol m⁻² s⁻¹ Pa⁻¹). By plotting the graph of J_i vs. p̄, the intercept (K₀) and the slope (P₀) can be calculated. r can be obtained by substituting the K₀ and the P₀ into eqn (6).

Plots of N₂ permeance against applied pressure for the PSF-P and PSF-D hollow fibres are shown in Fig. 3. The K₀ and P₀ values obtained from the plots were 3.77 × 10⁻⁷ and 2.73 × 10⁻¹³ for PSF-P and 8.42 × 10⁻⁹ and 1.53 × 10⁻¹⁵ for PSF-D. The pore radius calculated for the PSF-P was 29.4 nm and 7.2 nm for the PSF-D. Depending on the pore size and mean free path of the gas molecules, the mechanism of gas transport across the porous layer is explained. The gas transport across such hollow fibres of pore radii 7.2 and 29.4 nm may occur through Knudsen diffusion since the gas transport in the pores of 2–50 nm follow the Knudsen diffusion.²⁴

Fig. 3 Nitrogen permeances with applied pressure for PSf-P and PSf-D hollow fibre membranes.

PDMS–PSf hollow fibre composite membrane

The support substrates, PSF-P and PSF-D hollow fibres were dip coated with PDMS in order to fill the defects with PDMS and to form an active PDMS top layer on the substrate. PDMS is widely and frequently used as a coating material for gas and vapour separation due to its excellent properties such as flexibility, elasticity, good film forming capacity, compactness, large free volume, and high permeability-cum-selectivity to gases and vapours. PDMS solutions of different concentrations (P1, P2, P3 and P4 as given in Table 2) were used for the coating application. Fig. 4 shows cross-sectional SEM images of the PDMS–PSf composite hollow fibre of different membrane (top PDMS layer) thicknesses. Depending upon the polymer concentration in the coating solution, the thickness of the PDMS layer was varied. The top PDMS layers of approximately 0.5, 3, 5 and 10 μm thicknesses were visible in the cross-sectional SEM images of the composite membranes coated with P1, P2, P3 and P4 coating solutions, respectively.

Fig. 4 Cross-sectional SEM images of the PDMS–PSf composite hollow fibres (a–d) of different top PDMS layer thicknesses.

These PDMS layer thickness values were also estimated using eqn (1) from the gravimetric method. The thickness values observed by SEM and gravimetric analysis for both the PSF-P and PSF-D hollow fibre substrates are plotted in Fig. 5.

Fig. 5 The thickness values observed by the SEM and gravimetric analysis for the PSf-P and PSF-D hollow fibre.

The results show that the PDMS coating layer thickness values calculated from the gravimetric analysis were about twice the values of the top PDMS layer thicknesses observed directly by SEM in all the samples. This implied that a significant amount of PDMS was intruded inside the hollow fibre. In order to examine this, SEM-EDX elemental Si and S mapping was performed on the composite membrane samples in which PDMS and PSf may be represented by Si and S, respectively. Cross-sectional SEM-EDX images of the composite membranes with the thinnest (P1) and the thickest (P4) PDMS coatings are shown in Fig. 6.

Fig. 6 SEM-EDX images of the composite membranes. Thin PDMS coating: (i) colours showing the Si and S elemental mapping and (ii) the white colour (Si) represents the PDMS coating being mostly on the surface of the support. Thick PDMS coating: (iii) colours showing the Si and S elemental mapping and (iv) the white colour (Si) representing the PDMS coating on the surface as well as PDMS intrusion inside the support.

The top PDMS layer coated over the PSf substrate was clearly observed in both cases as indicated in green colour for the P1 coating and red colour for the P4 coating [Fig. 6(i) and (iii)] which were in accordance with the SEM results as mentioned above. Significant intrusion of PDMS inside the substrate as indicated by the white colour to the depth of about 60 μm from surface was observed for the P4 coating as compared to the P1 coating, which is to be expected from the highest polymer concentration of the P4 solution and the lowest polymer concentration of P1 solution [Fig. 6(ii) and (iv)]. Furthermore, the PDMS intrusion in the different substrates using the same coating solution may be different. Therefore, the PSF-P and PSF-D hollow fibres coated with the P3 solution were studied systematically by cross-sectional SEM-EDX analysis from the substrate surface to the depth of 60 μm in order to qualitatively identify the PDMS intruded fraction in the different hollow fibre substrate. The cross-sectional SEM images of PSF-P and PSF-D coated with the P3 solution along with the Si (red) and S (green) elemental colour mapping and analysis are shown in Fig. 7 and 8. EDX analysis for each composite sample was done at 4 different locations, (i) one top PDMS layer, (ii) a penetration depth of 20 μm from the surface, (iii) 20–40 μm penetration depth and (iv) 40–60 μm penetration depth. As shown in Fig. 7 and 8, the amount of Si corresponding to PDMS was found to have decreased with penetration depth in both the samples. However, the PDMS intrusion gradually decreased in the case of the PSF-P, whereas the PDMS intrusion was concentrated in the 20 μm penetration depth in the case of the PSF-D substrate. This may be due to the differences in the pore size of the PSF-P and PSF-D.

Fig. 7 The cross-sectional SEM images of PSf-P coated with the P3 solution along with the Si (red) and S (green) elemental colour mapping and analysis. EDX analysis of the sample at 4 different locations, sample 1: one top PDMS layer, sample 2: penetration depth of 20 μm from the surface, sample 3: 20–40 μm penetration depth and sample 4: 40–60 μm penetration depth; Si and S elemental composition of the samples analysed.

Fig. 8 The cross-sectional SEM images of PSf-D coated with the P3 solution along with the Si (red) and S (green) elemental colour mapping and analysis. EDX analysis of the sample at 4 different locations, sample 1: one top PDMS layer, sample 2: penetration depth of 20 μm from the surface, sample 3: 20–40 μm penetration depth and sample 4: 40–60 μm penetration depth; Si and S elemental composition of the samples analysed.

Deeper intrusion of PDMS inside the PSF-P as compared to PSF-D may have occurred due to the larger pore size of PSF-P. The PDMS coating over the PSF-P and PSF-D hollow fibres was also characterised by ATR-IR spectroscopy in which the penetration depth of the beam into the sample surfaces was about 0.4–0.9 μm in the IR region 800–1700 cm⁻¹. The ATR-IR spectra of the hollow fibre composite membranes are shown in Fig. 9. The spectra of PSf-P and PDMS coated PSf-P are shown in Fig. 9(A) while the spectra of PSf-D and PDMS coated PSf-D are shown in Fig. 9(B). The PSf substrate shows strong IR bands at 1150 cm⁻¹ (C–SO₂–C symmetric stretching), 1245 cm⁻¹ (C–O–C stretching), 1295 cm⁻¹ (S=O stretching), 1323 cm⁻¹ (C–SO₂–C asymmetric stretching), 1488 cm⁻¹ (CH₃–C–CH₃ stretching), and 1502 cm⁻¹ (C=C aromatic ring stretching). The characteristic IR strong bands of PDMS are 800 cm⁻¹ (Si–O–Si symmetric stretching), 1020–1090 cm⁻¹ (Si–O–Si asymmetric stretching) and 1260 cm⁻¹ (Si–CH₃ of dimethylsiloxane units). The IR bands at 1488–1502 cm⁻¹ of the PSf substrate are positioned away from the IR bands of PDMS and these IR bands can be used to distinguish PSf from the PSF–PDMS composites. As shown in the ATR-IR spectra, the distinctive IR bands at 1488–1502 cm⁻¹ belonging to PSf were not visible for all the PDMS coated PSf hollow fibres, indicating that the IR penetration depth was not beyond the PDMS coated layer. The IR penetration depth in the IR region 1488–1502 cm⁻¹ was about 0.45 μm. This implies that the top PDMS layer thickness of all the coated membrane was at least larger than the IR penetration depth. The ATR-IR results of the samples agreed with the PDMS thickness observed by SEM-EDX mapping and gravimetric analysis. ATR-IR also revealed some differences between the PDMS coated PSF-P and PSF-D samples in the O–H absorption band ranges. O–H stretching vibration of the absorbed moisture in the sample was generally observed as a broad absorption band at about 3230 cm⁻¹. If the sample contains surface silanol (Si–OH) groups, various O–H absorption bands at higher wavenumber depending on free OH groups, hydrogen bonded OH pairs, and perturbed OH bands can be present.^25–27 Such OH bands at 3600–3850 cm⁻¹ (stretching) and 1500–1700 cm⁻¹ (bending) were present in PDMS coated PSF-D samples but absent in PDMS coated PSF-P samples. This suggests that the top PDMS layer over the PSF-D substrate contained silanol groups which could have arisen due to incomplete cross-linking of the initial HPDMS pre-polymer. Therefore, the PDMS cross-linking reaction of the membrane preparation process occurred differently on the PSF-P and PSF-D substrate. The hydrophobic nature of the PDMS coated PSf hollow fibres was also observed by contact angle (water) measurements. Higher contact angle (water) values indicate greater hydrophobicity. The contact angle values for PDMS coated PSf hollow fibres are shown in Fig. 10. The PSf substrate exhibited a contact angle of about 80°. The highest contact angle of about 115° was observed for the thickest PDMS layer coated samples indicating that hydrophobicity of the membrane increased with the PDMS coating layer thickness.

Fig. 9 ATR-IR spectra of the (A) PDMS coated PSf-P and (B) PSf-D composite hollow fibres.

Fig. 10 Contact angle (water) values of PDMS coated PSf-P and PSf-D composite hollow fibres.

Performance evaluation of the hollow fibre composite membranes for application in air separation

The gas permeability of the composite hollow fibre membranes and supports was measured in terms of single-gas N₂ and O₂ permeance. N₂ and O₂ permeance were found to be the same, implying no preferential gas selectivity for the hollow fibre support existed, which can be expected due to the relatively very large pore size of the supports (29 nm for PSF-P and 7 nm for PSF-D) as compared to the small size of gas molecules in which the gas transport across such porous support occurs through Knudsen diffusion. The PSF-P support exhibited a higher gas permeance of 1300 GPU as compared to 27 GPU of the PSF-D due to the differences in their pore sizes. The composite membranes formed upon PDMS coating of the PSF-P and PSF-D supports exhibited different gas permeation rates of N₂ and O₂. The change in N₂ and O₂ gas permeation rates at various pressures ranging from 60 to 160 kPa for all the membranes are shown in Fig. 11. Gas permeation rates for the composite membranes were found to decrease with an increase in the PDMS coating thickness. The gas transport across the composite membrane is explained below. The composite membrane structure comprising a dense PDMS layer and porous PSf support is illustrated schematically in Fig. 12. The r pores and micropore channels of the PSf support may be filled with the PDMS from the surface to some extent in the interior as well as top PDMS continuous layer over the PSf surface depending on the amount of the PDMS coating (Fig. 12(b) and (e)), as such can be inferred from the SEM-EDX cross-sectional microstructure morphology analysis of the composite membranes as discussed above. Before the PDMS coating, the initial PSf hollow fibre support can be depicted as a porous layer comprising tortuous pore channels (∼0.01–0.1 μm), finger or teardrop shaped macrovoids (∼10–100 μm) underneath and micropore channels (<0.002 μm) within the polymer chain networks as shown in Fig. 12(a). The barrier resistance R of the membrane surface area A and length l (Fig. 12(b)) is related to the intrinsic permeability P of the membrane by the following expression.

Fig. 11 N₂ and O₂ gas permeation rates at pressures ranging from 60 to 160 kPa for the membranes.

Fig. 12 Schematic cross-sectional view of the PDMS coated PSf-P hollow fibre composite membrane; cross-sectional views of the PSf-P support before and after PDMS coating (a–c); PSf-D hollow fibre before and after the coating (d and e); the resistance model depiction (f).

The total resistance R_t can be calculated using the resistance model approach proposed by Henis and Tripodi given below.²⁸

where R₁ is the resistance from the PDMS top layer of thickness l₁ and the effective membrane surface area of A₁ which allows gas permeation through its free volume. R₂ is the resistance from the filled PDMS in the pores of the PSf hollow fibre support with corresponding membrane surface area A₂; whereas R₃ is the resistance from the PDMS coated PSf chain networks of the membrane surface area A₃. Thus, R₂ and R₃ are the combined resistance from the intruded PDMS of length l₂ inside the PSf support. The resistance from the remaining porous hollow fibre support is represented by R₄ (Fig. 12(f)).

The gas transport in the PDMS top layer of the resistance R₁ may occur through the sorption diffusion model in which the gas molecules get solubilized in the membrane material and then diffusion of the gas occurs across the membrane. The overall permeability of such a dense membrane is the combination of a thermodynamic sorption process and kinetic diffusion process.

P = SD (9)

where P is the permeability, S is the sorptivity and D is the diffusivity.

The PDMS layer is a rubbery material of dense polymer chain networks with empty voids (free fraction volume) within the chain aggregates wherein the overall membrane selectivity is obtained from the difference of the penetrant permeability on the basis of its sorption and diffusion coefficient.^29,30

According to the aforementioned eqn (8), the total resistance of the PDMS–PSf composite membranes may vary because of their structural variation depending upon the hollow fibre supports PSF-P and PSF-D of different pore asymmetry. The intrusion of the PDMS layer in the porous support may vary according to the pore asymmetry of the support. For the porous support of the thin skin containing small pores and large macrovoids in the interior, the PDMS intrusion is mostly in the skin region and to some extent, in the polymer structure surrounding the macrovoids. Plugging of the macrovoids with PDMS was not observed; possibly because of their large 10–100 μm channels or voids. Therefore, the PSf support of the thinner skin along with large macrovoids can generate a PDMS–PSf composite membrane of lower barrier resistance allowing for a higher flux. Further, if all the pores of the support are not properly coated by the PDMS, there is a possibility of defects that might affect adversely the gas separation selectivity but result in a higher flux. On the other hand, a thicker PDMS coating reduces the chances of defects but might decrease the productivity flux. The PDMS–PSF-P composite membrane had a PSF-P support of a thinner skin and a more porous structure containing large macrovoids, as a result, the membrane permeability is higher as compared to that of the PDMS–PSF-D composite membrane. A cross-sectional view of the PSF-D support and PDMS–PSF-D composite membrane are depicted in Fig. 12(d) and (e), respectively. The plots of the oxygen permeance (GPU) and O₂/N₂ selectivity for the series of PDMS coated PSF-P and PSF-D hollow fibres are shown in Fig. 13. The PDMS coated PSF-P sample series exhibited an O₂ permeance range of 6–60 GPU which was at least 3 times greater than the O₂ permeance (2–11 GPU) for the PDMS–PSF-D series. In both the series, the O₂/N₂ selectivity was increased for the composite membrane with an increase in the PDMS coating layer up to an optimal level beyond which the selectivity remained constant. A maximum O₂/N₂ selectivity of about 6 was observed for the PDMS–PSF-P membrane while it was only 4 for the PDMS–PSF-P membrane. According to the resistance model approach given in eqn (8), the transport property of the membrane is related to the total resistance given by combination of resistances from the top PDMS layer, the filled PDMS in the pores of the support, the PDMS coated PSf chain and porous PSf support. Therefore, the differences in the resistance parameters between the two types of membranes may result in differences in their O₂/N₂ separation properties.

Fig. 13 Plots of the oxygen permeance (GPU) and O₂/N₂ selectivity for the series of PDMS coated PSf-P (A) and PSf-D (B) hollow fibres.

Conclusions

A new preparation method for a hollow fibre air separation membrane comprising a top PDMS membrane layer and a porous PSf hollow fibre support is presented in this report. The PSf hollow fibres were conveniently spun from a dilute solution at ambient conditions. The membrane structure was varied by changes in the PDMS membrane layer thickness and the depth of PDMS intrusion inside the porous support of different pore asymmetry. An optimum PDMS membrane layer thickness over the porous PSf support was required to obtain the maximum O₂/N₂ selectivity. Pore asymmetry of the support is an important parameter to obtain the membrane for high permeability and selectivity. The support pore asymmetry comprising a thinner skin with numerous small pores with higher porosity could result in obtaining a membrane with a high performance both in permeability and selectivity. On the other hand, the support of a thicker skin with lower porosity could generate a non-uniform coating less-selective membrane with a higher resistance to gas flow. The prepared membrane exhibited O₂ flux from 6–60 GPU with a high O₂/N₂ selectivity 2–6 at 25 °C which means it is a promising material for air separation and has a performance comparable to that of a commercial membrane.¹⁶

Acknowledgements

Financial assistance as research grants from the Council of Scientific & Industrial Research (CSIR network project CSC0104) Government of India as well as the instrumentation facility provided by Analytical Discipline & Centralized Instrument Facility, CSIR-CSMCRI, Bhavnagar are gratefully acknowledged. CSIR-CSMCRI Registration No. 076/2016.

References

1. P. Bernardo, E. Drioli and G. Golemme, Membrane gas separation: a review/state of art, Ind. Eng. Chem. Res., 2009, 48, 4638–4663.
2. G. Dong, H. Li and V. Chen, Challenges and opportunities for mixed-matrix membranes for gas separation, J. Mater. Chem. A, 2013, 1, 4610–4630.
3. R. S. Murali, T. Sankarshana and S. Sridhar, Air separation by polymer based membrane technology, Sep. Purif. Rev., 2013, 42, 130–186.
4. T. S. Chung, W. H. lin and R. H. Vora, The effect of shear rates on gas separation performance of 6-FDA–durene polyimide hollow fibres, J. Membr. Sci., 2000, 167, 55–66.
5. D. T. Clausi and W. J. Koros, Formation of defect free polyimide hollow fibre membranes for gas separations, J. Membr. Sci., 2000, 167, 79–89.
6. G. C. Kapantaidakis and G. H. Koops, High flux polyethersulfone–polyimide blend hollow fibre membranes for gas separation, J. Membr. Sci., 2002, 204, 153–171.
7. X. Ding, Y. Cao, H. Zhao, L. Wang and Q. Yuan, Fabrication of high performance Matrimid/polysulfone dual-layer hollow fibre membrane for O₂/N₂ separation, J. Membr. Sci., 2008, 323, 352–361.
8. D. Li, T. S. Chung and R. Wang, Morphological aspects and structure control of dual-layer asymmetric hollow fibre membranes formed by a simultaneous co-extrusion approach, J. Membr. Sci., 2004, 243, 155–175.
9. D. F. Li, T. S. Chung, R. Wang and Y. Liu, Fabrication of fluoropolyimide/polyethersulfone (PES) dual layer asymmetric hollow fibre membranes for gas separation, J. Membr. Sci., 2002, 198, 211–223.
10. S. C. Pesek and W. J. Koros, Aqueous quenched asymmetric polysulfone hollow fibres prepared by dry/wet phase separation, J. Membr. Sci., 1994, 88, 1–19.
11. J. M. Henis and M. K. Tripodi, Multicomponent membranes for gas separations, US Pat., no. 4,230,463, 1980.
12. A. K. Fritzsche, M. K. Murphy, C. A. Cruse, R. F. Malon and R. E. Kesting, Characterization of asymmetric hollow fibre membranes with graded-density skins, Gas Sep. Purif., 1989, 3, 106–116.
13. A. K. Fritzsche, C. A. Cruse, R. E. Kesting and M. K. Murphy, Polysulfone hollow fibre membranes spun from Lewis acid:base complex. (I) Structure determination by oxygen plasma ablation, J. Appl. Polym. Sci., 1990, 40, 19–40.
14. A. K. Fritzsche, C. A. Cruse, R. E. Kesting and M. K. Murphy, Polysulfone hollow fibre membranes spun from Lewis acid:base complex. (II) The effect of Lewis acid to base ratio on membrane structure, J. Appl. Polym. Sci., 1990, 39, 1949–1956.
15. R. E. Kesting, A. K. Fritzsche, M. K. Murphy, C. A. Cruse, A. C. Handermann, R. F. Malon and M. D. Moore, The second-generation polysulfone gas-separation membrane. (I) The use of Lewis acid:base complexes as transient templates to increase free volume, J. Appl. Polym. Sci., 1990, 40, 1557–1574.
16. R. E. Kesting, A. K. Fritzsche, C. A. Cruse and M. D. Moore, The second-generation polysulfone gas-separation membrane. (II) The relationship between sol properties, gel macrovoids and fibre selectivity, J. Appl. Polym. Sci., 1990, 40, 1575–1582.
17. D. Wang, K. Li and W. K. Teo, Effects of temperature and pressure on gas permselection properties in asymmetric membranes, J. Membr. Sci., 1995, 105, 89–101.
18. A. V. R. Reddy, P. Ray, P. S. Singh, P. Kallem, S. Maurya and J. J. Trivedi, High flux hollow fibre ultrafiltration membranes and process for the preparation thereof, US Pat., Appl. 0213875 A1, 2013.
19. M. J. Han and S. T. Nam, Thermodynamic and rheological variation in polysulfone solution by PVP and its effect in the preparation of phase inversion membrane, J. Membr. Sci., 2002, 202, 55–61.
20. N. Widjojo and T. S. Chung, Thickness and air gap dependence of macrovoid evolution in phase-inversion asymmetric hollow fibre membranes, Ind. Eng. Chem. Res., 2006, 45, 7618–7626.
21. A. F. Ismail, I. Dunkin, S. L. Gallivan and S. J. Shilton, Production of super selective polysulfone hollow fibre membranes for gas separation, Polymer, 1999, 40, 6499–6506.
22. A. C. Sun, W. Kosar, Y. Zhang and X. Feng, Vacuum membrane distillation for desalination of water using hollow fibre membranes, J. Membr. Sci., 2014, 455, 131–142.
23. D. Wang, K. Li and W. K. Teo, Preparation and characterization of polyvinylidene fluoride (PVDF) hollow fibre membranes, J. Membr. Sci., 1999, 163, 211–220.
24. R. W. Baker, Membrane Technology and Applications, John Wiley & Sons Ltd, New Jersey, 2004.
25. A. M. Kansara, V. K. Aswal and P. S. Singh, Preparation and characterization of new poly(dimethylsiloxane) membrane series via a ‘cross-linking’ reaction using monomolecular trichloro (alkyl) silane of different alkyl chain and type, RSC Adv., 2015, 5, 51608–51620.
26. B. A. Morrow and A. J. McFarlan, Surface vibrational modes of silanol groups on silica, J. Phys. Chem., 1992, 96, 1395.
27. A. M. Efimov, V. G. Pogareva and A. V. Shashkin, Water-related bands in the IR absorption spectra of silicate glasses, J. Non-Cryst. Solids, 2003, 332, 93.
28. J. M. Henis and M. K. Tripodi, Composite hollow fibre membranes for as separation: the resistance model approach, J. Membr. Sci., 1981, 8, 233–246.
29. K. Ghosal and B. D. Freeman, Gas separation using polymer membranes: an overview, Polym. Adv. Technol., 1994, 5, 673–697.
30. B. Freeman and I. Pinnau, Separation of gases using solubility-selective polymers, Trends Polym. Sci., 1997, 5, 167–173.

---