N2 separation

Journal of Membrane Science 443 (2013) 156–169

Contents lists available at SciVerse ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

High performance PIM-1/Matrimid hollow fiber membranes for CO2/CH4, O2/N2 and CO2/N2 separation Wai Fen Yong a, Fu Yun Li a, You Chang Xiao b, Tai Shung Chung a,n, Yen Wah Tong a a b

Department of Chemical & Biomolecular Engineering, National University of Singapore, 4 Engineering Drive 4, Singapore 117576, Singapore Suzhou Faith & Hope Membrane Technology Co. Ltd., Suzhou Industrial Park, Jiangsu Province 215123, PR China

art ic l e i nf o

a b s t r a c t

Article history: Received 13 March 2013 Received in revised form 10 April 2013 Accepted 12 April 2013 Available online 2 May 2013

Polymers of intrinsic microporosity (PIM-1) have received worldwide attention but most PIM-1 researches have been conducted on dense flat membranes. For the first time, we have fabricated PIM1/Matrimid membranes in a useful form of hollow fibers with synergistic separation performance. The newly developed hollow fibers comprising 5–15 wt% of highly permeable PIM-1 not only possess much higher gas-pair selectivity than PIM-1 but also have much greater permeance than pure Matrimid fibers. Data from positron annihilation lifetime spectroscopy (PALS), field emission scanning electron microscopy (FESEM) and apparent dense layer thickness indicate that the blend membranes have an ultrathin dense layer thickness of less than 70 nm. PIM-1 and Matrimid are partially miscible. The effect of partial miscibility on dense selective layer was studied. Defect-free hollow fibers with gas pair selectivity more than 90% of the intrinsic value can be spun directly from dopes containing 5 wt% PIM-1 with proper spinning conditions, while post annealing and additional silicone rubber coating are needed for membranes containing 10 and 15 wt% PIM-1, respectively. Comparing to Matrimid, the CO2 permeance of as-spun fibers containing 5 and 10 wt% PIM-1 increases 78% and 146%, respectively (e.g., from original 86.3 GPU (1 GPU ¼ 1  10−6 cm3 (STP)/cm2 s cmHg ¼7.5005  10−12 m s−1 Pa−1) to 153.4 GPU and 212.4 GPU) without compromising CO2/CH4 selectivity. The CO2 permeance of the fiber containing 15 wt% PIM-1 improves to 243.2 GPU with a CO2/CH4 selectivity of 34.3 after silicon rubber coating. Under mixed gas tests of 50/50 CO2/CH4, this fiber shows a CO2 permeance of 188.9 GPU and a CO2/CH4 selectivity of 28.8. The same fiber also has an impressive O2 permeance of 3.5 folds higher than the pristine Matrimid (e.g., from original 16.9 GPU to 59.9 GPU) with an O2/N2 selectivity of 6.1. The newly developed membranes may have great potential to be used for natural gas purification, air separation and CO2 capture. & 2013 Elsevier B.V. All rights reserved.

Keywords: Gas separation Polymer blends Hollow fibers PIM-1 Matrimid

1. Introduction A recent data reveals that the CO2 concentration has reached 394 ppm in the atmosphere [1] which is far beyond the upper safety limit of 350 ppm [2]. With current energy demand, experts predicted that the CO2 concentration in the atmosphere would further increase by 50% to reach 570 ppm in the year of 2100 [2]. In 2011, the U.S. Department of Energy announced their investment of $41 million over a 3-year period for the development of postcombustion technologies for CO2 from coal-fired power plants [3]. Amine absorption and cryogenic separations are the conventional technologies for CO2 capture [4]. However, the high energy and solvent consumptions as well as environmental concerns are the main bottlenecks of these technologies. In order to treat the huge

n

Corresponding author. Tel.: +65 6516 6645; fax: +65 67791936. E-mail address: [email protected] (T.S. Chung).

0376-7388/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.memsci.2013.04.037

volume of natural gas and flue gas in reducing CO2 release directly to atmosphere, a high performance and environmentally friendly technology for CO2 capture is desperately required. Membrane technology has been demonstrated to possess properties of energy savings and environmental friendliness with small footprints for CO2 capture in the integrated gasification combined cycle (IGCC) coal-fire power plant [5]. Additionally, due to the ease of processing and cost competitiveness, polymeric membranes have been widely used in both research and development and industrial applications for gas separation [4–6]. Polymeric hollow fiber membranes are favorable for industrial gas separation owing to their high surface over volume ratio, good flexibility, self-mechanical support and ease of handling during large scale module fabrication [7–11]. Typically, hollow fiber membrane has an asymmetric structure that comes with a selective layer overlaying on top of a porous support. In the past decades, commercially available polyimide, Matrimid has been widely studied as the material for both flat sheet and

W.F. Yong et al. / Journal of Membrane Science 443 (2013) 156–169

hollow fiber membrane because of its high thermal stability and good processibility [12–16]. According to the literature, the pristine Matrimid exhibits prominent gas-pair selectivity of about 30–40 for CO2/CH4 and 6–6.6 for O2/N2 [12–16]. However, it possesses relatively low gas permeability. Clausi and Koros [14] were the first to develop defect-free Matrimid hollow fiber membranes with an ultra-thin selective layer of around 100 nm. They obtained a high O2 permeance of 33.7 GPU and O2/N2 selectivity of 6.0 when the fibers were spun with an air-gap of 2.5 cm. When the air-gap was increased to 18.5 cm, fibers had an enhanced O2/N2 selectivity of 6.8 but a relatively lower O2 permeance of 11.2 GPU. Jiang et al. [15] studied the Matrimid/ polyethersulfone (PES) dual-layer hollow fiber membranes by using Matrimid as the outer selective layer and PES as the inner supportive layer. The use of dual-layer co-extrusion is one of the strategies to reduce the manufacturing cost since the costly Matrimid is only used in the outer layer. They developed the dual-layer fibers with an O2/N2 selectivity of 6.26 in pure gas tests and a CO2/CH4 selectivity of 40 in a binary 40/60 mol% CO2/CH4 mixture. Besides, Dong et al. [16] carried out a detailed study to examine the effects of spinning conditions on Matrimid hollow fibers. Their fibers had a CO2 permeance of about 11 and a CO2/CH4 selectivity of 47–67. Recently, David et al. [17] also spun Matrimid hollow fibers for hydrogen separation from multicomponent gas mixtures. However, most of the aforementioned hollow fiber membranes have relatively low gas permeance. The permeance must be further improved for disruptive industrial applications. In addition to synthesizing totally new materials [6,18–27], modifications of existing materials [6,28–34] have been extensively employed to design new membranes with enhanced performance. Among various modification approaches, polymer blend is one of the most cost- and time-effective routes as it combines the advantages of different materials into a new compound with unique and synergetic properties that are often difficult to be obtained by other synthesis means [35,36]. As a consequence of superior thermal stability and separation performance, polyimide blended hollow fiber membranes have been broadly studied [37– 40]. Matrimid blended with polybenzimidazole (PBI) [37], PES [38,39] and (BTDA-TDI/MDI) P84 [39] as hollow fiber membranes for gas separation have been explored by different research groups. Hosseini et al. [37] developed dual-layer hollow fiber membranes consisting of Matrimid/PBI as an outer selective layer and polysulfone (PSF) as the inner supportive layer. The addition of PBI significantly enhanced the membranes to be H2 selective with a H2 permeance of 29.26 GPU and a H2/CO2 selectivity of 11.11. Kapantaidakis and Koops [38] showed that the incorporation of Matrimid into PES hollow fiber membranes increased permeance significantly if comparing with the pristine PES. The CO2 permeance was about 31–60 GPU with a CO2/N2 selectivity of 35–40 after a silicon rubber coating. In the work of Visser et al. [39], the introduction of P84 in Matrimid resulted in higher O2/N2 and CO2/CH4 selectivity. In addition, P84/Matrimid fibers showed strong resistance to plasticization for CO2 feed concentrations below 80 vol%. On the other hand, the addition of PES into Matrimid increased plasticization resistance but resulted in a lower gas-pair selectivity comparing to the pristine Matrimid. Other than Matrimid-based blends, PBI and polyetherimide (PEI) hollow fiber membranes had been reported [40]. The incorporation of 0–30 wt% PBI into PEI formed miscible blends due to strong hydrogen bonding between them. However, the gas permeance decreased significantly with an increase in PBI concentration. Nevertheless, the addition of PBI successfully tuned the pristine PEI properties towards high H2/N2 selectivity. In one of our previous studies, we had explored the blend compatibility between Matrimid and polymers of intrinsic microporosity (PIMs) and studied their synergistic gas permeability [34].

157

PIMs were chosen because not only they are a class of novel polymers but also possess extremely high free volume and large surface area due to the kinked ladder-type backbone structure [34,41–47]. Our results showed that the incorporation of 5 and 10 wt% of PIM-1 into Matrimid induces the permeability increments of 25% and 77%, respectively, from the original 9.6 to 12 and 17 Barrer (1 Barrer ¼1  10−10 cm3 (STP) cm/cm2 s cmHg) without much compromises in CO2/CH4 selectivity. Clearly, blending Matrimid and PIM-1 is a new strategy to molecularly design Matrimid based hollow fiber membranes with an enhanced flux and to simultaneously retain the high selectivity of the original Matrimid. Among all the PIMs synthesized, PIM-1 is one of the most recognized types [30,34,44–47]. Briefly, PIM-1 has a superior CO2 permeability of around 3000–8000 Barrer and a CO2/CH4 selectivity of 9–16 depending on synthesis and post-treatment conditions [30,34,44–47]. Moreover, the O2 permeability of PIM-1 is about 700–1700 Barrer with an O2/N2 selectivity of 2.2–4.3 [30,34,44– 47]. To the best of our knowledge, no prior work has been reported on PIM-1 related hollow fiber membranes in the literature possibly due to its unimpressive gas pair selectivity. Modifications of PIM-1 with a high selectivity have received worldwide attention [41–43]. Therefore, the purposes of this work are (1) to molecularly construct novel PIM-1/Matrimid blend hollow fiber membranes with synergistic separation performance by combining characteristics of high gas pair selectivity from Matrimid and high permeability from PIM-1, (2) to fundamentally understand the science and engineering for the formation of PIM-1/Matrimid hollow fiber membranes, and (3) to fabricate high-performance defect-free PIM-1/Matrimid blend hollow fiber membranes with an ultrathin dense-selective layer. The effects of spinning parameters such as dope formulation, bore fluid composition, take-up speed, PIM-1 concentration, and post-treatment conditions on gas separation performance will be systematically investigated. The miscibility and glass transition temperature (Tg) of blends will be also investigated by a polarized light microscope (PLM) and differential scanning calorimetry (DSC). It is envisioned that the fundamentals and knowledge gained throughout this study may create greater opportunities for further enhancements in polymeric hollow fiber membranes for gas separation applications.

2. Experimental 2.1. Materials The monomers 5,5′,6,6′-tetrahydroxy-3,3,3′,3′-tetramethyl-1,1′spirobisindane (TTSBI, 97%) and 2,3,5,6-tetrafluoroterephthalonitrile (TFTPN, 99%) were purchased from Alfa Aesar and Matrix Scientific, respectively. TTSBI was purified by re-crystallization in methanol, while TFTPN was sublimated under vacuum prior to use. Matrimid powder was supplied by Huntsman Advanced Materials. It was dried overnight at 120 1C under vacuum before use. N-methyl-2-pyrrolidone (NMP, 499.5%) from Merck was further purified via vacuum distillation before use. Anhydrous potassium carbonate (K2CO3, 499.5%) from Sigma Aldrich and methanol (MEOH, ≥99.9%) from Merck were used without further purification. Dichloromethane (DCM, 99.99%), hydrochloric acid (HCl, 37.5%), tetrahydrofuran (THF, 99.99%) and n-hexane (≥99.9%) from Fisher Scientific were used as received. 2.2. Synthesis and characterizations of PIM-1 The synthesis of PIM-1 was carried out by the traditional polycondensation reaction between TTSBI and TFTPN under

158

W.F. Yong et al. / Journal of Membrane Science 443 (2013) 156–169

the catalyst of anhydrous K2CO3 in an anhydrous NMP solution. The detailed procedure for the synthesis of PIM-1 could be found elsewhere [34,46]. The product yield was approximately 85% with the number-average molar mass (Mn) of 73,900, the weightaverage molar mass (Mw) of 45,700 and the polydispersity (PDI¼ Mn/Mw) of 1.62.

2.3. Characterizations of dense PIM-1/Matrimid membranes Fig. 1 depicts the chemical structures of PIM-1 and Matrimid, while Table 1 tabulates the intrinsic gas transport properties of dense PIM-1/Matrimid membranes casted from DCM under the method described elsewhere [34]. The dense PIM-1/Matrimid (15:85) membrane was deliberately prepared and tested while the other data were directly taken from our previous publication [34]. The surface morphology of PIM-1/Matrimid membranes was investigated using an Olympus BX50 polarized light microscope (PLM). The PLM micrographs were further analyzed by Image Pro Plus 3.0 software. The miscibility of dense PIM-1/Matrimid membranes was characterized by using differential scanning calorimetry (DSC, DSC822e, Mettler Toledo). The samples were tested with two consecutive scans at a heating rate of 10 1C min−1 from 40 to 450 1C. The first cycle of ramping and cooling was to eliminate any thermal history of the samples. The glass transition temperature, Tg of each sample was determined based on the mid-point transition temperature of the second heating curve.

O

O

2.4. Spinning dope formulation and preparation THF was chosen as the solvent for PIM-1 because it has a higher boiling point of 66 1C and a slower solvent evaporation rate comparing with other PIM-1 solvents such as chloroform (CHCl3) and DCM. To facilitate the spinning process, a binary solvent consisting of THF/NMP (50/50 wt%) was employed in this study. To have a better understanding of the blend dope rheology and to minimize surface defects of spun fibers for better gas separation, the critical concentrations of PIM-1/Matrimid spinning dope solutions were pre-determined before dope preparation [48]. Different polymer concentrations ranging from 6 to 22 wt% polymer in solvents for PIM-1/Matrimid (5:95), PIM-1/Matrimid (10:90) and PIM-1/Matrimid (15:85) were prepared. The viscosities of three different blend ratios of polymer solutions were measured by an ARES Rheometric Scientific Rheometer at 25 1C. Fig. 2 depicts their shear viscosity in THF/NMP solutions as a function of polymer concentration at a shear rate of 10 s−1. The critical concentration is referred to the intersection point by extrapolating the two linear regions of the viscosity curve. Based on the critical concentrations, the polymer concentrations in the PIM-1/Matrimid (5:95), PIM-1/Matrimid (10:90) and PIM-1/Matrimid (15:85) dopes were chosen as 11 wt%, 15 wt% and 17 wt% using THF/NMP (50:50) as the solvent, respectively. The following summarizes the dope preparation for spinning. PIM-1 and Matrimid were dissolved separately in THF and NMP. The solutions were stirred continuously at 25 1C for 24 h to ensure complete dissolution. On the next day, the PIM-1/THF solution was added to the Matrimid/NMP solution. The mixture was then stirred continuously for 24 h until it became a homogeneous dope solution. After that, the dope solution was degassed for 24 h before pouring into the ISCO syringe pump and then to continue degasing overnight in the ISCO syringe pump prior to spinning.

O CH3

N

N

O

O

H 3C

O

H3C

CH3

n

CH3 CN O

O

H 3C

CH3

O CN

n

Fig. 2. The critical concentration of PIM-1/Matrimid/THF/NMP dope solutions at 25 1C.

Fig. 1. Chemical structures of (a) Matrimid and (b) PIM-1.

Table 1 Intrinsic gas transport properties of PIM-1/Matrimid dense films [34]. Polymer blends

Pure gas test (35 1C, 3.5 atm) Matrimid PIM-1/Matrimid (5:95) PIM-1/Matrimid (10:90) PIM-1/Matrimid (15:85) a b

Permeability (Barrera,b)

Ideal selectivity

O2

CO2

O2/N2

CO2/N2

CO2/CH4

2.1 70.1 2.6 70.2 3.4 70.0 4.4 70.3

9.6 7 0.7 127 0.7 177 0.6 217 0.8

6.4 7 0.6 6.6 7 0.1 6.17 0.4 5.9 7 0.5

307 2.5 297 2.7 307 0.9 287 0.7

367 0.4 357 0.5 347 1.6 327 1.3

1 Barrer¼ 1  10−10 cm3 (STP)cm/cm2 s cmHg. Pure gas permeation tests performed at 35 1C and 3.5 atm.

W.F. Yong et al. / Journal of Membrane Science 443 (2013) 156–169

Table 2 Spinning conditions for Matrimid hollow fiber membranes. Hollow fiber ID

M

Dope composition Bore fluid composition Dope flow rate (ml/min) Bore fluid flow rate (ml/min) Air gap (cm) Take up rate (m/min)

Matrimid/NMP/MeOH (26.2/58.8/15 wt%)a 95/5 NMP/Water 2 0.6 2.5 12.9

a

Spinneret temperature is 50 1C.

2.5. Hollow fiber spinning, solvent exchange and post-treatment The single-layer hollow fiber membranes were fabricated via a dry-jet wet-spinning process using a single-layer spinneret with OD and ID of 1.2 mm and 0.8 mm, respectively. The spinning process was conducted at ambient temperature. After the nascent fibers extruded from the spinneret, they passed through the airgap region and entered into the coagulant bath filled up with tap water at room temperature and finally wound up on a take-up drum. The detailed experimental procedures and set-up could be found in one of our early publications [49]. All the spinning conditions for three dopes were kept constant except for bore fluid composition and take-up velocity. A higher dope flow rate was used for PIM-1/Matrimid (15:85) to obtain a similar fiber wall thickness as the critical concentration for this dope is slightly lower compared with the others. Pure Matrimid was spun based on the dope formulation reported earlier by Jiang et al. [15] and the detailed spinning conditions are presented in Table 2. Table 3 lists the spinning conditions for the PIM-1/Matrimid dopes. After spinning, the as-spun fibers were immersed in tap water for 3 days. Subsequently, the hollow fibers were subjected to three consecutive 30-min solvent exchange by circulating in MEOH and then in n-hexane with the same procedure to allow the removal of residual solvent. Lastly, the fibers were dried in the air at ambient temperature for 24 h before further characterization, post-modifications and permeation tests. Two post-treatments procedures were studied to compare the effectiveness of sealing the outer surface defects on the hollow fibers. The first method was subjected the fibers to heat treatment. Briefly, the as-spun fibers were preheated from room temperature to 75 1C and held for 3 h, then to cool down naturally [50–52]. The other procedure was to coat the outer surface of the fibers by a 3 wt% of silicone rubber in a n-hexane solution for 3 min. The fibers were then dried in air for 2 days prior to performance tests. Silicone rubber coating has been employed widely in sealing the defects on the dense-selective layer of the fibers. The intrinsic O2/N2 selectivity of silicon rubber is about 2.1. If a membrane possesses an ultrathin dense layer, the silicon rubber coating material may affect the separation performance of hollow fibers [53].

2.6. Morphology of hollow fibers The cross-sectional, inner and outer surface morphologies of hollow fiber membranes were analyzed using field emission scanning electron microscopy (FESEM JEOL JSM-6700LV). The hollow fiber samples for the cross-sectional characterization were immersed and fractured in liquid nitrogen. All the specimens adhered on the stub using a double-side conductive carbon tape and dried under vacuum overnight. Before tests, all samples were coated with a platinum layer using a sputtering coater (JEOL LFC-1300).

159

2.7. Characterization of the dense-selective layer thickness The structural changes of hollow fiber membranes were monitored by one of the advanced techniques in positron annihilation lifetime spectroscopy (PALS); namely doppler broadening energy spectra (DBES). A slow positron beam accelerated with a variable mono-energy varies from 0 to 30 keV was used to generate DBES spectra. The mean implantation depth of a sample could be evaluated by the following equation [54]:   40 1:6 ZðEþ Þ ¼ Eþ ð1Þ ρ where Z is the mean implantation depth expressed in nm, ρ is the density of polymer material in g/cm3 and E+ is the incident positron energy in keV. One of the characteristic parameters of DBES spectra, R parameter data was then fitted by using the VEPFIT program with a three-layer model [55]. The R parameter is defined as the ratio of the total counts from the valley region with an energy width between 364.2 and 496.2 keV (from 3γ annihilation) to those from the 511 keV peak region with a width between 504.35 and 517.65 (from 2γ annihilation). In addition, the R parameter provides information about large holes (nm to mm) where o-Ps undergoes 3γ annihilation [56,57]. 2.8. Module preparation and gas permeation measurements The single-layer PIM-1/Matrimid hollow fiber membranes were tested using both pure gas and binary gases. 3–4 modules were prepared for each experimental condition. In each module, it consisted of 5–10 fibers with an effective length of around 15 cm per fiber. In the pure gas tests, O2, N2, CH4 and CO2 were tested at 1 atm at room temperature. The detailed module preparation and pure gas permeation test procedures have been reported by Chung et al. [48]. The gas permeance was calculated as follows: P 273:15  106 Q 273:15  106 Q ¼ ¼   L A ΔP nπDl ΔP T T

ð2Þ

where P/L is the gas permeance of hollow fiber membranes in GPU (1 GPU ¼ 1  10−6 cm3 (STP)/cm2 s cmHg), T is the operating temperature (K), Q is the pure gas flux (cm3/s), A is the total effective area of fibers (cm2), ΔP is the trans-membrane pressure (cmHg), n is the total number of fibers in one testing module, D is the outer diameter of fibers (cm) and l is the effective length of fibers (cm). The ideal selectivity of gas A over gas B was calculated according to the following equation: αA=B ¼

ðP=LÞA ðP=LÞB

ð3Þ

where (P/L)A and (P/L)B refer to the pure gas permeance of gases A and B, respectively. Mixed gas tests were conducted using a modified variablepressure constant volume gas permeation cell and a binary gas of CO2/CH4 (50/50 mol%). The detailed design of mixed gas permeation cell has been described elsewhere [58]. The tests were carried out at room temperature and 2 atm. The mixed gas permeance (P/L) was expressed by following equations:     yCO2 dp P 273  1010 V ð4Þ   ¼ L CO2 AT  P 0  76=14:7 xCO2 dt 760     ð1−yCO2 Þ dp P 273  1010 V ð5Þ   ¼ L CH4 AT  P o  76=14:7 ð1−xCO2 Þ dt 760

where P=L CO and P=L CH are the CO2 and CH4 gas permeance, 2 4 respectively, Po denotes the upstream feed gas pressure (psia), x refers to the mole fraction in the feed gas and y is the mole fraction

160

W.F. Yong et al. / Journal of Membrane Science 443 (2013) 156–169

Table 3 Spinning conditions for PIM-1/Matrimid hollow fiber membranes. Hollow fiber ID

1A

Polymer composition Dope composition Bore fluid composition Dope flow rate (ml/min) Bore fluid flow rate (ml/min) Take up rate (m/min)

PIM-1/Matrimid (5:95) Polymer/NMP/THF (17/41.5/41.5 wt%) 95/5 NMP/Water 2.5 2.5 6.3

10.7

15.1

Hollow fiber ID

2A

2B

2C

Polymer composition Dope composition Bore fluid composition Dope flow rate (ml/min) Bore fluid flow rate (ml/min) Air gap (cm) Take up rate (m/min)

PIM/Matrimid (10:90) Polymer/NMP/THF (15/42.5/42.5 wt%) 95/5 NMP/Water 2.5 1 2.5 6.3

15.1

6.3

Hollow fiber ID

3A

3B

3C

Polymer composition Dope composition Bore fluid composition Dope flow rate (ml/min) Bore fluid flow rate (ml/min) Air gap (cm) Take up rate (m/min)

PIM-1/Matrimid (15:85) Polymer/NMP/THF (11/44.5/44.5 wt%) 95/5 NMP/Water 80/20 NMP/Water

1B

1C

2D

80/20 NMP/Water

15.1

50/50 NMP/Water

2.8 1 2.5 6.3

100μm Fig. 3. PLM images of the dense membranes at room temperature: (a) Matrimid; (b) PIM-1/Matrimid (5:95); (c) PIM-1/Matrimid (10:90); and (d) PIM-1/Matrimid (15:85).

in the permeate. The others symbols remain the same meanings as described in Eq. (2). The selectivity of the mixed gas was determined by the mole ratio of the permeate (y) and feed (x) of the two gases (CO2 and CH4) as follows: αðCO2 =CH4 Þ ¼

yCO2 =yCH4 xCO2 =xCH4

ð6Þ

Eq. (6) becomes Eq. (3) if the permeate pressure is negligible compared to the feed pressure.

3. Results and discussion 3.1. Miscibility studies of PIM-1/Matrimid flat sheet dense membranes Miscibility and homogeneity of the blend components play important roles to determine the physicochemical properties and applications of a polymer blend. Among various methods in determining miscibility, applied optical inspection to visualize the blend morphology is one of the simplest ways [35,59]. As revealed by a PLM in Fig. 3, the optical morphology of polymer blends comprising 5–15 wt% PIM-1 in Matrimid exhibits partial miscibility between each other. The partial miscibility was further validated by measuring their Tg. Fig. 4 shows their DSC results of the second heating. The pristine Matrimid exhibits a characteristic Tg at about 332 1C

Fig. 4. DSC results of PIM-1/Matrimid dense membranes.

[34,59]. It is generally understandable that the Tg of the pristine PIM-1 could not easily be detected in the range of 50–450 1C because of the low rotational freedom of main chains and the high stiffness of PIM-1 backbone [43]. Nevertheless, it is expected that the Tg of Pristine PIM-1 is more than 350 1C [43]. Consistent with our hypothesis, the Tg of the blend membranes shifts to a higher temperature when the PIM-1 concentration increases, as shown in Fig. 4. Both optical observation and DSC results are in good agreement with our previous study by using dynamical

W.F. Yong et al. / Journal of Membrane Science 443 (2013) 156–169

161

1 2

3

1) Inner cross-section

3) Outer cross-section

2) Middle bulk

Fig. 5. Cross-sectional morphology of PIM-1/Matrimid (5:95) fibers spun at condition 1A.

PIM-1/Matrimid (10:90)

Dope:PIM-1/Matrimid (5:95)

20 x

20 x

PIM-1/Matrimid (15:85)

20 x

Fig. 6. Dense-selective layer thickness of pristine (a) PIM-1/Matrimid (5:95); (b) PIM-1/Matrimid (10:90); and (c) PIM-1/Matrimid (15:85) hollow fiber membranes.

mechanical analyses where the characteristics of partial miscibility was found for samples containing PIM-1 up to 80 wt% [34]. 3.2. Morphology of PIM-1/Matrimid hollow fiber membranes Fig. 5 depicts the cross-sectional morphologies of PIM-1/Matrimid (5:95) hollow fibers spun at condition 1A. As illustrated in the figure, the membrane consists of two distinct layers, where the inner layer is highly porous with full of finger-like macrovoids and the outer layer is ultrathin and dense. The ultrathin denseselective layer provides the gas pair selectivity and high flux, while the porous inner layer functions as a mechanical support. Since the inner layer has a highly porous structure, it has the advantage of minimizing the sub-layer resistance for gas transport. Fig. 6 shows the cross-sectional morphologies close to the outer dense-selective layer spun from dopes with different blend ratios

of PIM-1/Matrimid. The dense-selective layer is normally formed by the rapid demixing at the outer surface when the nascent fiber exposes to water that causes rapid coalescence and aggregation of polymer chains [60]. A close look at the dense-selective layer with a high magnification of 20,000 reveals that all the fibers spun from different blend ratios have a very thin dense-selective layer of about 30–70 nm. In addition, it is interesting to note that the macrovoids tend to be eliminated when a higher PIM-1 loading is incorporated in the polymer dope as shown in Fig. 6. This is attributed to the higher viscosity of the PIM-1/Matrimid (15:85) polymer dope (when compared to PIM-1/Matrimid (5:95) and (10:90)) that effectively suppresses the formation of macrovoids. Besides, regardless of blend compositions from 5 to 15 wt% PIM-1 in Matrimid, no clear phase separation could be observed in the polymer matrix as referred to the micrographs in Fig. 6. Nevertheless, the effect of

162

W.F. Yong et al. / Journal of Membrane Science 443 (2013) 156–169

PIM-1/Matrimid (15:85)-3B

PIM-1/Matrimid (15:85)-3A

Borefluid:NMP/water (95:5)

PIM-1/Matrimid (15:85)-3C

NMP/water (80:20)

NMP/water (50:50)

40

6.0

35

5.5

30

5.0

25

4.5

20

4.0 10 15 Take-up speed (m/min)

26

170

24

160

22

150

20

140

18

130

16

120

14 5

10 15 Take-up speed (m/min)

30

CO2Permeance (GPU)

180 170

25 160 20

150 140

15 130

CO2/CH4 Selectivity

5

180

CO2/N2 Selectivity

6.5 CO2Permeance (GPU)

45

O2/N2 Selectivity

O2Permeance (GPU)

Fig. 7. A comparison of the inner surface morphology of PIM-1/Matrimid (15:85) fibers spun with different bore fluid compositions: (a) NMP/water (95:5); (b) NMP/water (80:20); and (c) NMP/water (50:50).

10

120 5

10 Take-up speed (m/min)

15

Fig. 8. Effect of take-up speed on gas separation performance of PIM-1/Matrimid (5:95) hollow fiber membranes: (a) O2 permeance and O2/N2 selectivity; (b) CO2 permeance and CO2/N2 selectivity; and (c) CO2 permeance and CO2/CH4 selectivity.

partial miscibility with a high concentration of PIM-1 is apparently magnified during gas separation tests and the details will be disclosed in the later section. 3.3. Effect of bore fluid compositions on inner surface morphology Fig. 7 reveals the inner surface morphologies of fibers spun from the PIM-1/Matrimid (15:85) dope using 95/5, 80/20 and 50/ 50 wt% NMP/water as bore fluids. As can be seen, the fiber spun with the highest solvent concentration (e.g., 95 wt% NMP) displays the most porous structure as shown in Fig. 7(a). This is directly resulted from the delayed demixing with the existence of the high solvent content. On the other hand, when a high concentration of non-solvent (e.g., water) is used in the bore fluid, it would lead to a

fast precipitation in the inner surface, thus a relatively denser skin is formed. As a result, the order of the inner surface porosity of the fibers follows a trend of 3A 43B 43C. Since a highly porous substructure is desirable to minimize gas transport resistance, a bore fluid comprising a high solvent content (e.g., 95 wt% NMP) is preferred. 3.4. Effect of take-up speed on gas separation performance The take-up speed is another crucial parameter that affects the formation of defect-free hollow fiber membranes. Generally, a higher take-up speed correlates to a greater elongational stress during spinning. Fig. 8 shows the gas permeance of PIM-1/ Matrimid (5:95) hollow fiber membranes spun from different

W.F. Yong et al. / Journal of Membrane Science 443 (2013) 156–169

163

Table 4 Gas separation performance of Matrimid and PIM-1/Matrimid hollow fiber membranes at different blend ratios before post-treatment. Permeance (P/L) (GPUa)

Hollow fibers ID

Matrimid PIM-1/Matrimid PIM-1/Matrimid PIM-1/Matrimid PIM-1/Matrimid PIM-1/Matrimid PIM-1/Matrimid PIM-1/Matrimid PIM-1/Matrimid PIM-1/Matrimid PIM-1/Matrimid a

(5:95)—1A (5:95)—1B (5:95)—1C (10:90)—2A (10:90)—2B (10:90)—2C (10:90)—2D (15:85)—3A (15:85)—3B (15:85)—3C

Ideal selectivity

O2

N2

CH4

CO2

O2/N2

CO2/CH4

CO2/N2

16.9 39.5 38.4 37.3 56.6 53.0 49.6 42.1 147.0 115.0 133.5

2.6 7.3 6.2 6.2 11.6 8.7 9.7 7.5 57.1 55.6 81.6

2.5 7.7 5.4 5.6 11.3 8.1 8.6 7.6 55.8 51.5 67.8

86.3 172.9 153.4 151.5 227.0 212.4 184.5 194.4 373.8 329.9 413.2

6.5 5.4 6.2 6.1 4.9 6.1 5.4 5.6 2.6 2.1 1.6

34.5 22.4 28.4 27.1 20.1 26.2 21.5 25.6 6.7 6.4 6.1

33.2 23.8 24.9 24.6 19.6 24.5 19.0 25.8 6.5 5.9 5.1

1 GPU¼ 1  10−6 cm3 (STP)/cm2 s cmHg ¼7.5005  10−12 m s−1 Pa−1.

Fig. 9. Comparison of O2 permeance and O2/N2 selectivity as a function of PIM-1 content in different PIM-1/Matrimid blend hollow fiber membranes.

take-up speeds. An optimum selectivity for O2/N2, CO2/N2 and CO2/CH4 separation is achieved for the fiber spun from condition 1B when the take-up speed is at 10 m/min. For all gas pairs tested, the selectivity displays a general trend with an initial increase followed by a decrease when the take-up speed increases. The initial increase in gas-pair selectivity with increasing take-up speed is likely attributed to the enhanced polymer chain packing and alignment in the extruded fiber. On the other hand, a continued increase in take-up speed would result in overstretching of polymer chains and formation of minor defects on the membrane surface, thus a decrease in gas-pair selectivity [11,61]. Moreover, with the effect of enhanced polymer chain packing and alignment, the gas permeance decreases with an increase in takeup speed. 3.5. Effect of PIM-1 concentration and bore fluid chemistry on gas separation performance Table 4 summarizes the gas separation performance of as-spun blend hollow fibers as a function of PIM-1 concentration, while Fig. 9 illustrates the effects of PIM-1 concentration on O2 permeance and O2/N2 selectivity of the as-spun fibers. Generally, the addition of 5–15 wt% PIM-1 in the polymer blends results in an increase in gas permeance and a decrease in gas-pair selectivity. For example, with a mere 5 wt% and 10 wt% PIM-1 in Matrimid, the CO2 permeance of the as-spun fibers (e.g., PIM-1/Matrimid (5:95) (condition 1B) and PIM-1/Matrimid (10:90) (condition 2B)) increase 78% and 146%, respectively (e.g., from original 86.3 GPU to 153.4 GPU and 212.4 GPU, respectively) without compromising its

CO2/CH4 selectivity. The increase in gas permeance is attributed to the increase in free volume in the polymer matrix with the addition of PIM-1 into Matrimid. Nevertheless, except fibers spun from PIM-1/Matrimid (5:95) show air separation performance as good as defect-free membranes, fibers spun from PIM-1/Matrimid (10:90) (condition 2A and 2C) and PIM-1/Matrimid (15:85) have deteriorated gas-pair selectivity comparing to the intrinsic selectivity of dense membranes. This is presumably attributed to the formation of defects at the selective layer because compatibility decreases as PIM-1 content is increased. In other words, different degrees of chain entanglement and molecular packing may take place with increasing PIM-1 loading [62]. Besides, the rich Matrimid and the rich PIM-1 domains in the blend dope may have different phase inversion rates during the membrane formation, thus create defects at the outer selective layer. The last three rows of Table 4 also compares the gas separation performance as a function of bore fluid chemistry for as-spun blend hollow fibers spun from the dope containing 15 wt% PIM-1. It is observed that the fiber spun with a higher solvent content (e.g.; NMP) in the bore fluid has a higher O2 and CO2 permeance. This phenomenon is directly resulted from the highly porous inner surface structure as a consequence of the delay demixing induced by a high solvent concentration in the bore fluid, as observed in Fig. 7. 3.6. Defect-free PIM-1/Matrimid hollow fiber membranes with an ultra-thin dense-selective layer A hollow fiber membrane could be considered as defect free when its gas-pair selectivity is more than 90% of its intrinsic value

164

W.F. Yong et al. / Journal of Membrane Science 443 (2013) 156–169

of dense films. The fibers spun from PIM-1/Matrimid (5:95) (conditions 1B and 1C) and PIM-1/Matrimid (10:90) (conditions 2B and 2D) have shown the defect-free characteristics because their O2/N2 selectivity is higher than 90% of their intrinsic values of dense films (as listed in Table 1). On the other hand, it could also be found that the newly developed fibers exhibit relatively high gas permeance. This is likely attributed to the formation of an ultra-thin dense-selective layer on the as-spun fibers. A detailed calculation of the apparent dense layer thickness of the as-spun hollow fiber membrane could be obtained by the following equation: L¼

P O2 ðdense f ilmÞ  103 ðP=LÞO2 ðhollow f iber membraneÞ

the calculated apparent dense layer thickness. As can be seen, both selected membranes have a dense-selective layer thickness of less than 100 nm, which is considered as ultra-thin in membrane research society. As a result, a defect-free PIM-1/Matrimid hollow fiber membrane with an ultra-thin dense-selective layer has been successfully fabricated in this work. To validate the dense-selective layer thickness and to understand the morphological change of the newly developed hollow fiber membrane, the as-spun fibers with different blend compositions were examined by PALS analyses. Fig. 10 shows the evolution of R parameter as a function of incident positron energy and the penetration depth of the positron into the membrane. S-parameter was not chosen in this study because of the presence of imide functional groups in Matrimid, which would cause the strong quenching and inhibition of the positronium [63,64]. In general, the dense-selective layer thickness is determined at the first region when the R parameter decreases dramatically to the minimum value when the incident positron energy increases.

ð7Þ

where L is the apparent dense-selective layer thickness (nm), and P is the gas permeability of the dense film in Barrer (1 Barrer¼ 1  10−10 cm3 (STP) cm/cm2 s cmHg). Table 5 summarizes

Table 5 R parameter and dense-selective layer thickness of PIM-1/Matrimid hollow fiber membranes. R1a

Hollow fibers ID

−3

0.32417 3.29  10 0.33407 3.82  10−3 0.33417 3.30  10−3

PIM-1/Matrimid (5:95)—1A PIM-1/Matrimid (10:90)—2A PIM-1/Matrimid (15:85)—3A a

L1 (nm)a

Apparent dense selective layer thickness, La (nm)

25 7 4 24 7 4 36 7 7

66 60 –

Rl and Ll are the R parameter and the thickness of dense layer obtained from the VEPFIT simulator.

Mean depth (μm) 0.03

0.09

0.18

0.28

0.40

0.54

0.43 1A (Experimental) 2A (Experimental) 3A (Experimental) 1A (VEPFIT) 2A (VEPFIT) 3A (VEPFIT)

0.42

0.41

PIM-1/Matrimid (15:85)-3A

R

0.40

0.39 PIM-1/Matrimid (10:90)-2A 0.38

0.37

PIM-1/Matrimid (5:95)-1A

0.36 0

1

2

3 4 Positron Incident Energy (keV)

5

6

Fig. 10. R parameters of PIM-1/Matrimid hollow fiber membranes with different blend compositions.

Table 6 Comparison of gas separation performance of PIM-1/Matrimid (10:90) hollow fiber membranes before and after heat treatment. Hollow fibers ID

Permeance (P/L) (GPUa)

Ideal selectivity

O2

N2

CH4

CO2

O2/N2

CO2/CH4

CO2/N2

Pristine fibers PIM-1/Matrimid (10:90)—2A PIM-1/Matrimid (10:90)—2C

56.6 49.6

11.6 9.7

11.3 8.6

227.0 184.5

4.9 5.4

20.1 21.5

19.6 19.0

After heat treatment at 75 1C and 3 h PIM-1/Matrimid (10:90)—2A PIM-1/Matrimid (10:90)—2C

52.9 40.5

9.1 6.7

8.5 6.4

210.2 177.3

5.8 6.0

23.1 26.3

24.8 27.8

a

1 GPU ¼ 1  10−6 cm3 (STP)/cm2 s cmHg ¼7.5005  10−12 ms−1 Pa−1.

W.F. Yong et al. / Journal of Membrane Science 443 (2013) 156–169

165

Table 7 Gas separation performance of PIM-1/Matrimid (15:85) hollow fiber membranes before and after different post-treatment methods. Hollow fibers ID

Permeance (P/L) (GPUa)

Ideal selectivity

O2

N2

CH4

CO2

O2/N2

Pristine fibers PIM-1/Matrimid (15:85)—3A PIM-1/Matrimid (15:85)—3B PIM-1/Matrimid (15:85)—3C

147.0 115.0 133.5

57.1 55.6 81.6

55.8 51.5 67.8

373.8 329.9 413.2

2.6 2.1 1.6

6.7 6.4 6.1

6.5 5.9 5.1

After heat treatment at 75 1C and 3 h PIM-1/Matrimid (15:85)—3A PIM-1/Matrimid (15:85)—3B PIM-1/Matrimid (15:85)—3C

132.9 99.7 125.0

47.6 45.3 72.8

47.9 43.4 59.1

326.8 319.8 396.4

2.8 2.2 1.7

6.8 7.4 6.7

6.9 7.1 5.4

After silicon rubber coating for 3 min PIM-1/Matrimid (15:85)—3A PIM-1/Matrimid (15:85)—3B PIM-1/Matrimid (15:85)—3C

59.9 57.1 50.1

9.9 9.2 8.0

7.1 7.3 6.8

243.2 234.6 217.1

6.1 6.2 6.2

34.3 32.1 32.0

24.6 25.5 27.1

a

CO2/CH4

CO2/N2

1 GPU¼ 1  10−6 cm3 (STP)/cm2 s cmHg ¼7.5005  10−12 m s−1 Pa−1.

Fig. 11. Comparison of O2 permeance and O2/N2 selectivity for PIM-1/Matrimid hollow fiber membranes after different post-treatment conditions (* is the pristine fibers without post-treatment).

The second region is named as the transition layer where the R parameter increases with a further increase in incident positron energy. The transition layer indicates the fiber morphology transforming from a dense to a porous structure. Following that, the third region is where the R parameter reaches a flat region, indicating a more homogenous porous substructure. Since R parameter provides information about large holes (nm to mm) and the fiber spun from PIM-1/Matrimid (5:95) has the lower R value comparing with the other two fibers, this indicates that the fiber spun from PIM-1/Matrimid (5:95) has the densest structure and the lowest gas permeance as shown in Table 4. By applying a three-layer model, the dense layer thickness can be estimated by the VEPFIT software. As tabulated in Table 5, the simulated dense layer thickness is less than 100 nm which agrees with the observation from SEM and the calculated apparent dense-layer thickness from gas permeation data. 3.7. Effect of post-treatment on gas separation performance of as-spun PIM-1/Matrimid hollow fiber membranes Two post-treatment methods; namely, heat treatment [50–52] and silicon rubber coating [53] have been employed to cure the defects on membrane surface. The fibers are expected to be easily sealed by heat treatment if there are minor defects on the selective skin. On the other hand, if the defects cannot be mitigated by heat treatment, the fibers

will then be treated with a silicon rubber coating. The gas separation performance of PIM-1/Matrimid (10:90) hollow fiber membranes before and after heat treatment is depicted in Table 6. The selectivity of fibers 2A and 2C increases significantly close to the intrinsic selectivity of the dense film after heat treatment while there is a slight drop in gas permeance. The minor decrease in gas permeance is directly resulted from the densification of the selective layer. The gas separation performance of PIM-1/Matrimid (15:85) hollow fiber membranes before and after two consecutive post-treatments is summarized in Table 7. Firstly, the same approach of heat treatment was employed. However, the heat treatment was not effective to seal the defects and the fibers still displayed a low selectivity. Thus, a silicon rubber coating was employed on the outer surface of the fibers. As can be seen, all fibers reveal remarkable improvements in gas-pair selectivity. Compared to the pristine Matrimid, the O2 and CO2 permeance of PIM-1/Matrimid (15:85) (condition 3A) hollow fiber membranes after silicon rubber coating is 3.5 folds and 2.8 folds higher, respectively, (e.g., O2 permeance of 59.9 GPU and CO2 permeance of 243.2 GPU) with a similar O2/N2 and CO2/CH4 selectivity. Fig. 11 compares the gas separation performance of PIM-1/Matrimid single layer hollow fiber membranes after post-treatments. The results reveal an enhanced gas permeance with an increase in PIM-1 loading. Interestingly, all fibers show comparable gas-pair selectivity with the pristine Matrimid membrane. In addition, mixed gas studies using CO2/CH4 (50%/50%) as the feed were conducted for PIM-1/Matrimid

166

W.F. Yong et al. / Journal of Membrane Science 443 (2013) 156–169

(10:90) (condition 2B) and PIM-1/Matrimid (15:85) (condition 3A) hollow fiber membranes. Table 8 shows that the permeance and selectivity measured under mixed gas tests are slightly lower than pure gas data. The lower in mixed gas permeance of CO2 and CH4 is possibly due to the competition effects of the one gas component over the other gas component for sorption and transport pathways in the membrane. As a consequence, the lower in the permeance of the fast gas (e.g., CO2) results in a decrease in the mixed gas selectivity [64,65]. 3.8. Comparison with previous studies on gas separation performance Polyimide hollow fiber membranes have been studied by various researchers in recent years due to its thermal stability and high selectivity [14–17,38–40]. Both neat polyimide and blend polyimide hollow fibers have been developed. Table 9 shows a comparison in separation performance between some of these fibers and the PIM-1/ Matrimid hollow fiber membranes for O2/N2, CO2/CH4 and CO2/N2 separation. The newly developed PIM-1/Matrimid hollow fiber membranes show superior permeance to other polyimide blend membranes with slightly lower selectivity. The O2 and CO2 permeance of the PIM-1/Matrimid hollow fiber membranes are about 3 and 2.5 folds, respectively, higher than the other literature data. Specifically, a comparison of gas separation performance of the PIM-1/Matrimid hollow fibers for different gas pairs with other commercial materials

[66–71] has been plotted in Fig. 12. As can be seen, there is an obvious enhancement in gas separation performance for the newly developed PIM-1/Matrimid hollow fiber membranes. As discussed in earlier sections, the improvement is mainly due to the inclusion of the high free volume PIM-1 in Matrimid and the formation of defect-free hollow fiber membranes with an ultra-thin dense-selective layer. The high performance PIM-1/Matrimid hollow fiber membranes would definitely enlighten research interest in designing next generation hollow fiber membranes for air separation, pre-combustion and postcombustion CO2 capture and energy development.

4. Conclusions Novel and high performance PIM-1/Matrimid hollow fiber membranes comprising 5–15 wt% of highly permeable PIM-1 with an ultrathin dense-selective layer have been developed in this study. The newly developed membranes show synergistic characteristics of higher selectivity with respect to PIM-1 and higher permeability with respect to Matrimid. The following conclusions could be drawn from this work: 1. High flux defect-free as-spun blend membranes with gas-pair selectivity more than 90% of the intrinsic values of dense films can be fabricated by incorporating 5 wt% and 10 wt% PIM-1 into

Table 8 Binary gas separation performance of PIM-1/Matrimid (15:85) hollow fiber membranes after silicon rubber coating. Hollow fibers ID

Pure gasa

Binary gasa

Permeance (GPU)

PIM-1/Matrimid (10:90)—2B PIM-1/Matrimid (15:85)—3A

Selectivity

Permeance (GPU)

Selectivity

CH4

CO2

CO2/CH4

CH4

CO2

CO2/CH4

8.1 7.1

212.4 243.2

26.2 34.3

6.9 6.6

159.7 188.9

23.1 28.8

Table 9 A comparison of polyimide hollow fiber membranes for gas separation [14–17,37–40]. Polymer

Matrimid Matrimid Matrimid Matrimid Matrimid Matrimid PBI/Matrimid (50:50)c PES/Matrimid (20:80)c PES/Matrimid (50:50)c PES/Matrimid (80:20)c P84/Matrimid (50:50)c PBI/PEI (10:90)c PBI/PEI (20:80)c Matrimid-M PIM-1/Matrimid (5:95)—1A PIM-1/Matrimid (5:95)—1B PIM-1/Matrimid (10:90)—2Ad PIM-1/Matrimid (10:90)—2Cd PIM-1/Matrimid (15:85)—3Ac PIM-1/Matrimid (15:85)—3Bc a

Testing T/P (1C/atm)

25/7.9 25/7.9 25/6.8 20/10 20/15b 30/2.3 35/10 25/3.9 25/3.9 25/3.9 35/4 25/7.8 25/7.8 25/1 25/1 25/1 25/1 25/1 25/1 25/1

Permeance (P/L) (GPU)

Selectivity

Ref.

O2

N2

CH4

CO2

O2/N2

CO2/CH4

CO2/N2

33.7 11.2 4.6 9 – – – – – – 4.9 4.9 0.8 16.9 39.5 38.4 52.9 40.5 59.9 57.1

5.6 1.6 0.7 1.3 – 0.7 – – – – 0.6 4.5 0.4 2.6 7.3 6.2 9.1 6.7 9.9 9.2

– – – – 0.16 – 0.12 – – – – – – 2.5 7.7 5.4 8.5 6.4 7.1 7.3

– – – – 11 10.1 4.81 40 31 60 – – – 86.3 172.9 153.4 210.2 177.3 243.2 234.6

6.0 6.8 6.3 6.7 – – – – – – 7.9 1.1 2.0 6.5 5.4 6.2 5.8 6.0 6.1 6.2

– – 40a – 67 – 41.8 – – – – – – 34.5 22.4 28.4 23.1 26.3 34.3 32.1

– – – – – 14.7 – 40 35–38 39 – – – 33.2 23.8 24.9 24.8 27.8 24.6 25.5

[14] [14] [15] [16] [16] [17] [37] [38] [38] [38] [39] [40] [40] This This This This This This This

Dual-layer hollow fiber membranes with Matrimid as outer layer and PES as inner supportive layer, tested with binary gas test consists of 40/60 CO2/CH4. Tested with binary gas test consists of 20/80 CO2/CH4. Fibers after silicon rubber coating. d Fibers after heat-treatment at 75 1C and 3 h. b c

work work work work work work work

W.F. Yong et al. / Journal of Membrane Science 443 (2013) 156–169

167

Fig. 12. A comparison of (a) CO2/CH4; (b) O2/N2 and (c) CO2/N2 separation performance of PIM-1/Matrimid membranes with other commercial materials.

Matrimid. Comparing to the pristine Matrimid, the CO2 permeance of the as-spun blend fibers containing 5 and 10 wt% PIM-1 increases significantly to 78% and 146%, respectively (e. g., from original 86.3 GPU to 153.4 GPU and 212.4 GPU) without showing much compromises in CO2/CH4 selectivity. With a higher PIM-1 content of 15 wt%, the CO2 permeance of the blend fiber displays a greater improvement of 2.8 folds to 243.2 GPU with a comparable CO2/CH4 selectivity to the blend dense film after the silicon rubber coating. 2. The O2 permeance of the PIM-1/Matrimid (15:85) (condition 3A) hollow fiber membrane after silicon rubber coating increases abundantly. It has a value of 3.5 folds higher than the pristine Matrimid (e.g., from original 16.9 GPU to 59.9 GPU) with a similar O2/N2 selectivity of 6.1. This fiber is very promising as a new generation membrane for air separation. 3. Optical observation and DSC results have revealed partial miscibility between PIM-1 and Matrimid. The effect of partial miscibility on defect formation at the dense selective layer is more prominent with increasing PIM-1 loading in the hollow fibers. It is believed that the higher PIM-1 content in Matrimid, the less the compatibility between polymers is. This is a direct result of different degrees of chain–chain interactions and phase inversion rates during membrane formation. The minor defects on the dense selective skin of the PIM-1/Matrimid (10:90) (condition 2A and 2C) fiber can be effectively eliminated by heat treatment at 75 1C for 3 h, while the large defects in PIM-1/Matrimid (15:85) fibers have to be cured with a combination of heat treatment and silicon rubber coating. 4. Membrane morphology revealed that macrovoids tend to be eliminated when a higher PIM-1 loading is incorporated in the polymer dope. It is likely attributed to the higher viscosity of

the PIM-1/Matrimid (15:85) polymer dope (when compared to PIM-1/Matrimid (5:95) and (10:90)) that effectively suppresses the formation of macrovoids. 5. The fiber spun with the highest solvent concentration (e.g., 95 wt% NMP) displays the most porous inner surface morphology with a higher O2 and CO2 permeance compared to the fibers spun with 80 and 50 wt% NMP as bore fluids. The highly porous substructure with minimal transport resistance is directly resulted from the delayed demixing induced by the existence of high solvent content. 6. The increase in take-up speed improves the gas-pair selectivity to an optimum value followed by a decrease. This is mainly due to the enhanced polymer chain packing and alignment in the extruded fiber initially, while a further increase in take-up speed would result in overstretching the polymer chains and cause minor defects on the membrane surface. 7. The newly developed PIM-1/Matrimid hollow fiber membranes demonstrate an exceptional gas separation performance that surpasses other reported polyimide blend membranes for CO2/ CH4, O2/N2 and CO2/N2 separation. Future works will be focused on the physical aging behavior of the newly developed hollow fiber membranes and the sustainability of these membranes for industrial use.

Acknowledgments The authors would like to thank the Singapore National Research Foundation (NRF) for their support through the Competitive Research Program for the project entitled, “New

168

W.F. Yong et al. / Journal of Membrane Science 443 (2013) 156–169

Biotechnology for Processing Metropolitan Organic Wastes into Value-Added Products” (Grant no. R-279-000-311-281). The authors express their appreciation to Mr. Y.K. Ong, Mr. J. Zuo, Miss H. Wang and Miss N.L. Le for their valuable suggestions to this work.

Nomenclature CHCl3 DBES DCM DSC FESEM HCl ID IGCC OD K2CO3 MEOH NMP P84 PALS PES PBI PDI PEI PES PIM PLM PSF TFTPN THF TTSBI

chloroform Doppler broadening energy spectra dichloromethane differential scanning calorimetry field emission scanning electron microscopy hydrochloric acid inner diameter integrated gasification combined cycle outer diameter potassium carbonate methanol N-methyl-2-pyrrolidone BTDA-TDI/MDI positron annihilation lifetime spectroscopy polyethersulfone polybenzimidazole polydispersity polyetherimide polyethersulfone polymers of intrinsic microporosity polarized light microscope polysulfone 2,3,5,6-tetrafluoroterephthalonitrile tetrahydrofuran 5,5′,6,6′-tetrahydroxy-3,3,3′,3′-tetramethyl-1,1′spirobisindane

List of symbols A l ρ P/L ΔP Tg E+ Z x y Mn T D Q N Po Mw

effective area of the testing fibers (cm2) effective length of the testing fiber (cm) density of polymer material (g/cm3) gas permeance of hollow fiber membrane (GPU) gas pressure difference cross the membrane (cmHg) glass transition temperature (1C) incident positron energy (keV) mean implantation depth (nm) mole fraction in the feed gas mole fraction in the permeate number-average molar mass operating temperature (K) outer diameter of the testing fiber (cm) pure gas flux (cm3/s) total number of fibers in one testing module upstream feed gas pressure (psia) weight-average molar mass

References [1] IPCC Fourth Assessment Report: Climate Change (AR4), 2007, 〈www.ipcc.ch/ publications_and_data/publications_and_data_reports.html〉. [2] Mauna Loa Observatory: Scripps Institution of Oceanography, Atmospheric CO2 on 5 December 2012, website: 〈http://co2now.org/〉.

[3] Fossil Energy Techline, U.S. Department of Energy, August 2011. [4] D.R. Paul, Y.P. Yampol’skii, Polymeric Gas Separation Membranes, CRC Press, Boca Raton, FL, 1994. [5] T.C. Merkel, H. Lin, X. Wei, R. Baker, Power plant post-combustion carbon dioxide capture: an opportunity for membranes, J. Membr. Sci. 359 (2010) 126–139. [6] Y.C. Xiao, B.T. Low, S.S. Hosseini, T.S. Chung, D.R. Paul, The strategies of molecular architecture and modification of polyimide-based membranes for CO2 removal from natural gas—a review, Prog. Polym. Sci. 34 (2009) 561–580. [7] H. Strathmann, K. Kock, P. Amar, R.W. Baker, The formation mechanism of asymmetric membranes, Desalination 16 (1975) 179–203. [8] J.A. van’t Hof, A.J. Reuvers, R.M. Boom, H.H.M. Rolevink, C.A. Smolders, Preparation of asymmetric gas separation membranes with high selectivity by a dual bath coagulation method, J. Membr. Sci. 70 (1992) 17–30. [9] T.S. Chung, A review of microporous composite polymeric membrane technology for air-separation, Polym. Compos. 4 (1996) 269–283. [10] S.P. Nunes, K.V. Peinemann, Membrane Technology in the Chemical Industry, Wiley-VCH Verlag GmbH, Weinheim, 2001. [11] N. Peng, N. Widjojo, P. Sukitpaneenit, M.M. Teoh, G.G. Lipscomb, T.S. Chung, J. Y. Lai, Evolution of polymeric hollow fibers as sustainable technologies: past, present, and future, Prog. Polym. Sci. 37 (2012) 1401–1424. [12] A. Bos, I.G.M. Punt, M. Wessling, H. Strathmann, Suppression of CO2-plasticization by semiinterpenetrating polymer network formation, J. Polym. Sci. B: Polym. Phys. 36 (1998) 1547–1556. [13] Y. Huang, D.R. Paul, Physical aging of thin glassy polymer films monitored by gas permeability, Polymer 45 (2004) 8377–8393. [14] D.T. Clausi, W.J. Koros, Formation of defect-free polyimide hollow fiber membranes for gas separations, J. Membr. Sci. 167 (2000) 79–89. [15] L.Y. Jiang, T.S. Chung, D.F. Li, C. Cao, S. Kulprathipanja, Fabrication of Matrimid/ polyethersulfone dual-layer hollow fiber membranes for gas separation, J. Membr. Sci. 240 (2004) 91–103. [16] G. Dong, H. Li, V. Chen, Factors affect defect-free Matrimids hollow fiber gas separation performance in natural gas purification, J. Membr. Sci. 353 (2010) 17–27. [17] O.C. David, D. Gorri, K. Nijmeijer, I. Ortiz, A. Urtiaga, Hydrogen separation from multicomponent gas mixtures containing CO, N2 and CO2 using Matrimids asymmetric hollow fiber membranes, J. Membr. Sci. 419–420 (2012) 49–56. [18] H.B. Park, C.H. Jung, Y.M. Lee, A.J. Hill, S.J. Pas, S.T. Mudie, E. Van Wagner, B. D. Freeman, D.J. Cookson, Polymers with cavities tuned for fast selective transport of small molecules and ions, Science 318 (2007) 254–258. [19] H. Lin, E. Van Wagner, B.D. Freeman, L.G. Toy, R.P. Gupta, Plasticizationenhanced H2 purification using polymeric membranes, Science 311 (2006) 639–642. [20] H. Reardon, J.M. Hanlon, R.W. Hughes, A. Godula-Jopek, T.K. Mandal, D. H. Gregory, Emerging concepts in solid-state hydrogen storage: the role of nanomaterials design, Energy Environ. Sci. 5 (2012) 5951–5979. [21] P.M. Budd, E.S. Elabas, B.S. Ghanem, S. Makhseed, N.B. McKeown, K.J. Msayib, C.E. Tattershall, D. Wang, Solution-processed, organophilic membrane derived from a polymer of intrinsic microporosity, Adv. Mater. 16 (2004) 456–459. [22] D.J. Liaw, K.L. Wang, Y.C. Huang, K.R. Lee, J.Y. Lai, C.S. Ha, Advanced polyimide materials: syntheses, physical properties and applications, Prog. Polym. Sci. 37 (2012) 907–974. [23] D. Grainger, M.B. Hägg, The recovery by carbon molecular sieve membranes of hydrogen transmitted in natural gas networks, Int. J. Hydrogen Energy 33 (2008) 2379–2388. [24] C.H. Lau, S.L. Liu, D.R. Paul, J.Z. Xia, Y.C. Jean, H.M. Chen, L. Shao, T.S. Chung, Silica nanohybrid membranes with high CO2 affinity for green hydrogen purification, Adv. Energy Mater. 1 (2011) 634–642. [25] M. Carta, R. Malpass-Evans, M. Croad, Y. Rogan, J.C. Jansen, P. Bernardo, F. Bazzarelli, N.B. McKeown, An efficient polymer molecular sieve for membrane gas separations, Science 339 (2013) 303–307. [26] Y.C. Xiao, T.S. Chung, Grafting thermally labile molecules on cross-linkable polyimide to design membrane materials for natural gas purification and CO2 capture, Energy Environ. Sci. 4 (2011) 201–208. [27] H.A. Patel, C.T. Yavuz, Noninvasive functionalization of polymers of intrinsic microporosity for enhanced CO2 capture, Chem. Commun. 48 (2012) 9989–9991. [28] T.X. Yang, Y.C. Xiao, T.S. Chung, Poly-/metal-benzimidazole nano-composite membranes for hydrogen purification, Energy Environ. Sci. 4 (2011) 4171–4180. [29] S. Kulprathipanja, R.W. Neuzil, N.N. Li, Separation of fluids by means of mixed matrix membranes, US patent 4740219, 1988. [30] N. Du, M.M. Dal-Cin, G.P. Robertson, M.D. Guiver, Decarboxylation-induced cross-linking of polymers of intrinsic, microporosity (PIMs) for membrane gas separation, Macromolecules 45 (2012) 5134–5139. [31] T.S. Chung, L.Y. Jiang, Y. Li, S. Kulprathipanja, Mixed matrix membranes (MMMs) comprising organic polymers with dispersed inorganic fillers for gas separation, Prog. Polym. Sci. 32 (2007) 483–507. [32] M.R. Coleman, W.J. Koros, Conditioning of fluorine-containing polyimides. 2. Effect of conditioning protocol at 8% volume dilation on gas-transport properties, Macromolecules 32 (1999) 3106–3113. [33] M.L. Chua, Y.C. Xiao, T.S. Chung, Effects of thermally labile saccharide units on the gas separation performance of highly permeable polyimide membranes, J. Membr. Sci. 415–416 (2012) 375–382. [34] W.F. Yong, F.Y. Li, Y.C. Xiao, P. Li, K.P. Pramoda, Y.W. Tong, T.S. Chung, Molecular engineering of PIM-1/Matrimid blend membranes for gas separation, J. Membr. Sci. 407–408 (2012) 47–57.

W.F. Yong et al. / Journal of Membrane Science 443 (2013) 156–169

[35] L.M. Robeson, Polymer Blends: A Comprehensive Review, Hanser Gardener Publications, Cincinnati, OH, 2007. [36] D.R. Paul, S. Newman, Polymer Blends, Academic Press, New York, 1978. [37] S.S. Hosseini, N. Peng, T.S. Chung, Gas separation membranes developed through integration of polymer blending and dual-layer hollow fiber spinning process for hydrogen and natural gas enrichments, J. Membr. Sci. 349 (2010) 156–166. [38] G.C. Kapantaidakis, G.H. Koops, High flux polyethersulfone–polyimide blend hollow fiber membranes for gas separation, J. Membr. Sci. 204 (2002) 153–171. [39] T. Visser, N. Masetto, M. Wessling, Materials dependence of mixed gas plasticization behavior in asymmetric membranes, J. Membr. Sci. 306 (2007) 16–28. [40] T.S. Chung, Z.L. Xu, Asymmetric hollow fiber membranes prepared from miscible polybenzimidazole and polyetherimide blends, J. Membr. Sci. 147 (1998) 35–47. [41] N.B. McKeown, P.M. Budd, Exploitation of intrinsic microporosity in polymerbased materials, Macromolecules 43 (2010) 5163–5176. [42] M.D. Guiver, Y.M. Lee, Polymer rigidity improves microporous membranes, Science 339 (2013) 284–285. [43] N. Du, H.B. Park, G.P. Robertson, M.M. Dal-Cin, T. Visser, L. Scoles, M.D. Guiver, Polymer nanosieve membranes for CO2-capture applications, Nat. Mater. 10 (2011) 372–375. [44] J. Song, N. Du, Y. Dai, G.P. Robertson, M.D. Guiver, S. Thomas, I. Pinnau, Linear high molecular weight ladder polymers by optimized polycondensation of tetrahydroxytetramethylspirobisindane and 1,4-dicyanotetrafluorobenzene, Macromolecules 41 (2008) 7411–7417. [45] P. Li, T.S. Chung, D.R. Paul, Gas sorption and permeation in PIM-1, J. Membr. Sci. 432 (2013) 50–57. [46] F.Y. Li, Y.C. Xiao, T.S. Chung, S. Kawi, High performance thermally self-crosslinked polymer of intrinsic microporosity (PIM-1) membranes for energy development, Macromolecules 45 (2012) 1427–1437. [47] F.Y. Li, Y.C. Xiao, Y.K. Ong, T.S. Chung, UV-rearranged PIM-1 polymeric membranes for advanced hydrogen purification and production, Adv, Energy Mater. 2 (2012) 1456–1466. [48] T.S. Chung, S.K. Teoh, X.D. Hu, Formation of ultrathin high-performance polyethersulfone hollow fiber membranes, J. Membr. Sci. 133 (1997) 161–175. [49] D.F. Li, T.S. Chung, R. Wang, Morphological aspects and structure control of dual-layer asymmetric hollow fiber membranes formed by a simultaneous coextrusion approach, J. Membr. Sci. 243 (2004) 155–175. [50] A.Z. Gollan, Anisotropic membranes for gas separation, US Patent 4,681,605, 1987. [51] B. Bikson, J.K. Nelson, Composite membranes and their manufacture and use, US Patent 4,826,599, 1989. [52] Y. Li, C. Cao, T.S. Chung, K.P. Pramoda, Fabrication of dual-layer polyethersulfone (PES) hollow fiber membranes with an ultrathin dense-selective layer for gas separation, J. Membr. Sci. 245 (2004) 53–60. [53] J.M.S. Henis, M.K. Tripodi, Composite hollow fiber membranes for gas separation: the resistance model approach, J. Membr. Sci. 8 (1981) 233–246. [54] Y.C. Jean, P.E. Mallon, D.M. Schrader, Principle and Application of Positron & Positronium Chemistry, World Scientific, Singapore, 2002. [55] A. Van Veen, H.H. Schut, J. de Vries, R.A. Hakvoort, M.R. IJpma, Analysis of positron profiling data by means of “VEPFIT”, in: AIP Conference Proceedings, vol. 218, AIP, New York, 1990, pp. 171–176.

169

[56] S. Zhang, K.Y. Wang, T.S. Chung, H.M. Chen, Y.C. Jean, G. Amy, Well-constructed cellulose acetate membranes for forward osmosis: minimized internal concentration polarization with an ultra-thin selective layer, J. Membr. Sci. 360 (2010) 522–535. [57] W.S. Hung, M.D. Guzman, S.H. Huang, K.R. Lee, Y.C. Jean, J.Y. Lai, Characterization and layer structure in asymmetric thin film polymeric membranes in the wet condition using the variable monoenergy slow positron beam, Macromolecules 43 (2010) 6127–6134. [58] R. Wang, S.L. Liu, T.T. Lin, T.S. Chung, Characterization of hollow fiber membranes in a permeator using binary gas mixtures, Chem. Eng. Sci. 57 (2002) 967–976. [59] Y. Wang, S.H. Goh, T.S. Chung, Miscibility study of Torlons polyamide-imide with Matrimids 5218 polyimide and polybenzimidazole, Polymer 48 (2007) 2901–2909. [60] R.E. Kesting, A.K. Fritzsche, Polymeric Gas Separation Membranes, WileyInterscience, New York, 1993. [61] N. Peng, T.S. Chung, The effects of spinneret dimension and hollow fiber dimension on gas separation performance of ultra-thin defect free Torlonss hollow fiber membranes, J. Membr. Sci. 310 (2008) 455–465. [62] M.M. Teoh, T.S. Chung, K.Y. Wang, M.D. Guiver, Exploring Torlon/P84 copolyamide-imide blended hollow fibers and their chemical cross-linking modifications for pervaporation dehydration of isopropanol, J. Membr. Sci. 61 (2008) 404–413. [63] K.I. Okamoto, K. Tanaka, M. Katsube, O. Sueoka, Y. Ito, Positronium formation in various polyimides, Radiat. Phys. Chem. 41 (1993) 497–502. [64] F.Y. Li, Y. Li, T.S. Chung, H. Chen, Y.C. Jean, S. Kawi, Development and positron annihilation spectroscopy (PAS) characterization of polyamide imide (PAI)polyethersulfone (PES) based defect-free dual-layer hollow fiber membranes with an ultrathin dense-selective layer for gas separation, J. Membr. Sci. 378 (2011) 541–550. [65] J.H. Kim, S.Y. Ha, Y.M. Lee, Gas permeation of poly(amide-6-b-ethylene oxide) copolymer, J. Membr. Sci. 190 (2001) 179–193. [66] N. Peng, T.S. Chung, J.Y. Lai, The rheology of Torlons solutions and its role in the formation of ultra-thin defect-free Torlons hollow fiber membranes for gas separation, J. Membr. Sci. 326 (2009) 608–617. [67] Y. Li, T.S. Chung, Y.C. Xiao, Superior gas separation performance of dual-layer hollow fiber membranes with an ultrathin dense-selective layer, J. Membr. Sci. 325 (2008) 23–27. [68] D.L. Wang, K. Li, W.K. Teo, Highly permeable polyethersulfone hollow fiber gas separation membranes prepared using water as non-solvent additive, J. Membr. Sci. 176 (2000) 147–158. [69] D.L. Wang, W.K. Teo, K. Li, Preparation and characterization of high-flux polysulfone hollow fibre gas separation membranes, J. Membr. Sci. 204 (2002) 247–256. [70] N. Peng, T.S. Chung, K.Y. Li, The role of additives on dope rheology and membrane formation of defect-free Torlons hollow fibers for gas separation, J. Membr. Sci. 343 (2009) 62–72. [71] N. Peng, T.S. Chung, M.L. Chng, W.Q. Aw, Evolution of ultra-thin denseselective layer from single-layer to dual-layer hollow fibers using novel Extems polyetherimide for gas separation, J. Membr. Sci. 360 (2010) 48–57.