Ester-Cross-linkable Composite Hollow Fiber Membranes for CO2

School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, 311 Ferst Drive NW, Atlanta, Georgia 30332, United States. Ind. Eng...
22 downloads 9 Views 3MB Size
Download PDF
Article pubs.acs.org/IECR

Ester-Cross-linkable Composite Hollow Fiber Membranes for CO2 Removal from Natural Gas Canghai Ma and William J. Koros* School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, 311 Ferst Drive NW, Atlanta, Georgia 30332, United States ABSTRACT: High-performance ester-cross-linkable composite hollow fibers were developed through simultaneous spinning of 4,4′-(hexafluoroisopropylidene) diphthalic anhydride (6FDA) based polyimide and Torlon (registered trademark of Solvay Polymers) solutions. The core layer polyamide-imide polymer, Torlon, demonstrates excellent adhesion with cross-linkable 6FDA-based polyimide during spinning and ester-cross-linking. Torlon also shows superior thermal stability after cross-linking while maintaining the open substructure of the Torlon core layer. The resultant cross-linked composite hollow fibers show a CO2 permeance of 40 GPU with a CO2/CH4 selectivity of 39 in testing at 100 psi with 50/50 CO2/CH4 feed, 35 °C. The cross-linked composite hollow fibers also show high separation performance in the presence of high-level toluene contaminants up to 1000 ppm, suggesting that cross-linked composite hollow fibers are viable under realistic operation conditions. The high natural gas separation performance and significantly reduced cost of hollow fiber formation provides a significant advancement in the state of the art for natural gas separations.

1. INTRODUCTION Dual-layer hollow fiber spinning technology is commonly used to produce defect-free composite hollow fibers through socalled dry-jet/wet-quench process by simultaneous extrusion of two polymer solutions without adding a postspinning coating step.1−4 This efficient process provides economical alternative defect-free hollow fibers that eliminate instability in “caulked” hollow fibers under aggressive feed conditions.5 Composite hollow fiber membranes, therefore, combine the advantages of low cost polymers as the supporting nonselective core layer and high performance polymer as the selective sheath layer.6,7 Ideally, the core layer provides the mechanical strength to withstand high transmembrane pressure difference and has negligible transport resistance for gas separations, while the sheath layer serves as the selective layer, which allows a high separation productivity and efficiency.8 The significantly reduced cost of membrane formation with high separation performance makes dual-layer hollow fiber spinning especially attractive for large scale gas separations that require large membrane areas.9 Development of composite hollow fiber membranes dates back to Henne et al, who disclosed dual-layer composite hollow fiber membrane for hemodialysis.10,11 The first set of composite hollow fiber membranes used for gas separations was disclosed by Du Pont in 1992.8 As for natural gas separations, Jiang et al. fabricated Matrimid/polyethersulfone (PES) dual-layer hollow fibers with a CO2/CH4 selectivity around 40 tested with 120− 270 psi 40/60 CO2/CH4 at 22 °C; however, the maximum achieved CO2 permeance was only up to 11 GPU;1 Li et al distributed PES-zeolite into dual-layer PES/BTDA-TDI/MDI copolyimide (P84) hollow fiber membranes to enhance CO2/ CH4 selectivity;12 however, further heat treatment and additional coating resulted in a CO2 permeance lower than 0.164 GPU and a CO2/CH4 selectivity below 33.4 tested with 190 psig of 50/50 CO2/CH4 at 24 °C. Besides the lower separation productivity, delamination of sheath/core layers can © 2013 American Chemical Society

significantly undermine the mechanical strength under high feed pressures, which has been discussed in fluoropolyimide/ polyethersulfone (PES) dual-layer hollow fibers.8 Lower CO2 permeance (membrane separation productivity) indicates that a high-performance polymer material as the selective sheath layer, as well as a robust core layer polymer, is needed to achieve high permeate flux and separation efficacy under aggressive feed conditions. Liu et al. applied chemical cross-linking modification on polyimide/poly(ether sulfone) dual-layer hollow fibers but the chemical cross-linked hollow fibers tended to plasticize under a CO2 feed pressure ∼50 psi.13 Researches demonstrated that 6FDA-based cross-linked polyimide hollow fibers showed a CO2 permeance over 50 GPU and a CO2/CH4 selectivity above 40 tested with 200 psi of 50/50 CO2/CH4 at 35 °C.14−17 Despite the good separation performance, the high cost of 6FDA-based cross-linkable polyimide increases cost. A duallayer hollow fiber spinning technique can be utilized to reduce the amount of expensive polyimide required, while maintaining high separation performance. Despite the attractive aspects of such advanced membranes, it is challenging to integrate a lowcost polymer core layer with the sheath layer to develop crosslinkable composite hollow fibers. A key challenge to overcome in such a membrane is the potential delamination of sheath/ core layers and collapse of the core layer polymer during aggressive heat treatment to cross-link the high performance sheath layer.6,8 Therefore, our group has focused on integrating a low-cost supporting core layer to the expensive cross-linkable sheath layer to achieve defect-free cross-linked composite Special Issue: Enrico Drioli Festschrift Received: Revised: Accepted: Published: 10495

December 19, 2012 February 19, 2013 February 21, 2013 June 3, 2013 dx.doi.org/10.1021/ie303531r | Ind. Eng. Chem. Res. 2013, 52, 10495−10505

Industrial & Engineering Chemistry Research

Article

hollow fibers without sheath/core layer delamination. Specifically, this paper will address solutions to overcome challenges to develop composite hollow fibers with the desirable open porous morphology, compatibility of sheath/core layers, and core layer thermal stability during the aggressive cross-linking. The high-performance cross-linked composite hollow fibers will be characterized under aggressive feed conditions, including high partial CO2 pressures and high-level hydrocarbon contaminants.

αAB =

(1)

The upstream sorption coefficient, S, in glassy polymer membranes can be described well by the so-called dual-mode model, shown in eq 2.18 C H′ ibi CA = SA = kDi + pA 1 + bA pA + bBpB

(2)

In eq 2, kDi is the Henry’s law constant, C′Hi is the Langmuir capacity constant, bi is the Langmuir affinity constant, and pi is the local effective partial pressure of component i, which represents the local chemical potential for component i. 2.2. Characterization of Membrane Performance. To characterize the separation performance of a hollow fiber membrane, two key factors, termed as permeance and selectivity, are considered in this study.18 The permeance, Pi/ l, represents the separation productivity of a hollow fiber membrane and is defined as the flux of penetrant i normalized by the partial pressure or fugacity difference across the membrane, as shown in eq 3. Pi n = i l Δpi

(3)

In eq 3, Pi represents the permeability of penetrant i; l describes the effective membrane thickness; ni represents the flux of penetrant i through the membrane; Δp refers the partial pressure or fugacity difference of each penetrant across the membrane. The common unit of permeance is the GPU, which is defined as eq 4. ⎛ cc(STP) ⎞ ⎟ GPU = 10−6⎜ 2 ⎝ cm ·s·cmHg ⎠

(5)

2.3. Cross-linking of Polymeric Hollow Fibers. As a key challenge for conventional polymeric membranes, the plasticization of polymeric membranes is often observed when an elevated feeding CO2 partial pressure increases the permeance but reduces selectivity significantly. To develop a robust membrane with solid separation performance, the CO2 induced plasticization must be suppressed to achieve a high permeance without loss of selectivity. Past studies demonstrated that a highly effective approach, ester-cross-linking, can improve the CO2/CH4 selectivity and CO2 plasticization resistance of polymer by reducing the degree of swelling and segmental chain mobility in the polymer.5,14−17,19−24 Defect-free estercross-linkable monolithic hollow fiber membranes have been developed with a significantly improved CO2 permeance (productivity) and plasticization resistance. Notwithstanding the intrinsic high separation performance and plasticization resistance, the high cost of cross-linkable polyimide used offers an opportunity if the costly high performance polymer use can be minimized. Dual-layer hollow fiber spinning reduces the usage of expensive cross-linkable sheath polymer by up to 90%;9 however, cross-linking of composite hollow fibers can cause a delamination of sheath/core layers. Moreover, aggressive heat treatment during cross-linking tends to cause a collapse of core layer polymer and to reduce the permeance significantly. To address the aforementioned challenges, defect-free and delamination-free cross-linked composite hollow fibers are pursued in this paper. 2.4. Formation of Composite Hollow Fiber Membranes. Composite hollow fiber membranes can be produced through a so-called dry-jet/wet-quench hollow fiber spinning process.2,25−27 Two homogeneous polymer solutions, called sheath dope and core dope, are simultaneously extruded with bore fluid through an annular die, called a dual-layer spinneret, into an aqueous quench bath. During the dry-jet step, the evaporation of volatile components in the sheath dope will increase the local polymer concentration of the outermost layer of nascent fibers and result in the formation of a skin layer in the sheath layer.27 When the nascent fiber enters the aqueous quench bath, solvents diffuse from fibers into the quench bath while water from the quench bath diffuses into the fibers, which cause phase separation to occur. Open porous substructures in both core layer and sheath layer can be formed during this phase separation process. A simple subsequent standard process to prepare hollow fiber modules is described in refs 5 and 28.

2. BACKGROUND AND THEORY 2.1. Membrane Separation Mechanism. The so-called sorption−diffusion model applies to polymeric gas separation membranes. In this model, the gas permeants first sorb in the upstream of a membrane and then diffuse through the membrane under a partial pressure/fugacity difference. The differences in the amount of gas sorbed in the membrane and the permeant diffusion rate through the membrane cause the gas mixture to be separated. In this case, the permeability of a polymer membrane, P, can be described by the product of the diffusion coefficient, D, and sorption coefficients, S, as shown in eq 1.18 P = D·S

PA /l PB/l

3. EXPERIMENTS 3.1. Materials. The cross-linkable sheath layer polyimide used in this study is called PDMC, which represents propanediol monoesterified cross-linkable polyimide. The molecular weight and polydispersity index (PDI) for this PDMC polymer are 74 000 (Mw) and 2.8, respectively. The PDMC is produced by monoesterification of a polyimide copolymer, 6FDADAM:DABA (3:2), with 1,3-propanediol. The 6FDA-DAM:DABA (3:2) polymer is synthesized from 4,4′-(hexafluoroisopropylidene) diphthalic anhydride (6FDA), 2,4,6-trimethyl-1,3diaminobenzene (DAM) and 3,5-diaminobenzoic acid (DABA) as described in refs 15 and 16. The selection of cross-linking agent, 1,3-propanediol, is primarily based on previous research.16,21 N-Methyl-2-pyrrolidione (NMP), tetrahydrofuran

(4)

The selectivity, αAB, measures the membrane separation efficacy for a gas pair under conditions where the upstream pressure is much greater than the downstream, as it is in this study. It is defined by the ratio of the fast gas (A) permeance to the slow gas (B) permeance, as shown in eq 5. 10496

dx.doi.org/10.1021/ie303531r | Ind. Eng. Chem. Res. 2013, 52, 10495−10505

Industrial & Engineering Chemistry Research

Article

Figure 1. Polymerization and monoesterification reaction to synthesize PDMC polyimide.

(THF), ethanol, lithium nitrate (LiNO3), polyvinylpyrrolidone (PVP), methanol, hexane, and 1,3-propanediol were all purchased from Sigma-Aldrich (Milwaukee, WI). The carboxylic acid groups in DABA groups can react with ester groups to form new ester bonds and interconnect the polymer matrix; this so-called cross-linking process can stabilize the polymer against plasticization. The synthesis of 6FDADAM:DABA (3:2), monoesterification, and cross-linking reaction are shown in Figures 1 and 2.16 The selection of a core layer material is challenging due to the complexity of composite hollow fiber formation and the further cross-linking required after the spinning process, which may cause sheath/core layer delamination or the core layer collapse of composite hollow fibers. Cellulose acetate is compatible with 6FDA-DAM:DABA (3:2) and PDMC.9,16 However, subsequent cross-linking can cause serious collapse of cellulose acetate and reduce the permeance significantly due to the lower glass transition temperature (Tg) of cellulose acetate. The collapse of a core layer suggests that a polymer with a glass transition temperature higher than cellulose acetate should be pursued to provide both the desirable compatibility with PDMC and thermal stability during cross-linking. In this paper, we report for the first time that a polyamide-imide copolymer, Torlon, can be cospun with PDMC as the supporting core layer

polymer to provide the desirable morphology, thermal stability, and high separation performance. The structure of Torlon is shown in Figure 3. 3.2. Dope Development. The spinning dope consists of polymer, solvents, and nonsolvents and should be a homogeneous solution. Solvents dissolve polymers to produce polymer solutions for spinning and nonsolvent move the dope composition toward the binodal to allow rapid phase separation. N-Methylpyrrolidone (NMP) can dissolve both PDMC and Torlon and serves as the primary solvent for both the sheath dope and core dope. Besides NMP, tetrahydrofuran (THF) is used as a secondary solvent in the sheath dope to promote the formation of a skin layer in the outermost sheath layer.27 Typical nonsolvents, including water and ethanol, are chosen in this work due to their relatively low toxicity and easy processing. Polyvinylpyrrolidone (PVP) or lithium nitrate (LiNO3) was also added in dopes in this work as a “pore former”, which can help form desirable open porous substructures in composite hollow fibers. To develop a spinnable dope, the ternary phase diagram, containing the binodal curve, must be constructed to ensure that the dope composition is not only located in the one-phase region but also is close to the binodal to allow rapid phase separation. The dope compositions must also not be too close or cross the 10497

dx.doi.org/10.1021/ie303531r | Ind. Eng. Chem. Res. 2013, 52, 10495−10505

Industrial & Engineering Chemistry Research

Article

Figure 2. Ester-cross-linking mechanism to produce cross-linked PDMC polyimide.

3.3. Composite Hollow Fiber Spinning. Composite hollow fibers can be formed through a dry-jet/wet-quench hollow fiber spinning system, as shown in Figure 4. After the dope development, the sheath dope and core dope are coextruded with bore fluid through a dual-layer spinneret. As discussed in section 2.4, the skin layer is formed in the outermost layer of a fiber during the dry-jet process. The nascent composite fiber then enters the aqueous quench bath and phase separation occurs immediately. The hollow fibers are collected on a rotating drum and soaked in water bath for ∼3 days to remove residual solvents and nonsolvents in the fibers. Typically, if the wet fibers are dried directly, capillary force in the fibers induced by dehydration can collapse the pores and

Figure 3. Chemical structure of Torlon polymer used as a robust core layer material.

binodal, which may cause defects or eliminate the skin layer.28 A straightforward technique to determine the approximate binodal location is provided by the so-called “cloud point determination”, as described in refs 5 and 29.

Figure 4. Dry-jet/wet-quench dual-layer spinning to form composite hollow fiber membranes. 10498

dx.doi.org/10.1021/ie303531r | Ind. Eng. Chem. Res. 2013, 52, 10495−10505

Industrial & Engineering Chemistry Research

Article

Figure 5. Ternary phase diagram showing the binodal (black solid line) of PDMC polymer/solvent/nonsolvent system. Solid points and open circles represent one-phase dope and two-phase dope, respectively. In this diagram, solvents are NMP and THF; Nonsolvents include ethanol and LiNO3.

Figure 6. Ternary phase diagram showing the binodal (black solid line) of Torlon/solvent/nonsolvent system. Solid points and open circles represent one-phase dope and two-phase dope, respectively. In this diagram, the solvent is NMP; nonsolvents are H2O and PVP.

even damage the fiber selective layer. To control the morphology, solvent exchange is conducted to remove water from the fibers, which is particularly important for preserving the transition layers of composite hollow fibers. A typical solvent exchange consists of two steps: first soak the fibers in methanol to remove water from fibers, and then use hexane to replace the residual methanol in fibers. After solvent exchange, the fiber surface tension is significantly reduced and the fibers can be further dried under vacuum or heating without collapsing the pores of fibers or damaging the defect-free selective layer. 3.4. Cross-linking of Composite Hollow Fibers. The cross-linking reaction was conducted by annealing fibers in a preheated vacuum oven at a constant temperature for a set period of time. The cross-linking temperature is critical to

develop delamination-free composite hollow fibers without collapsing the core layer. A cross-linking temperature higher than the glass transition temperature of the core layer polymer will tend to collapse the core layer and reduce the permeance significantly. On the contrary, an excessively low cross-linking temperature may not be able to stabilize hollow fibers against CO2 plasticization due to a lower degree of cross-linking. In this work, the cross-linking temperature was 200 °C and the crosslinking time was 2 h, primarily based on successful development of cross-linked monolithic hollow fibers.16,28 THF dissolution experiments were typically used to verify the cross-linking ability of membrane samples and cross-linked samples were not be dissolved by THF. Details about this method were described in ref 24. Since the Torlon is insoluble in THF, it is relatively difficult to characterize the cross-linkability of PDMC/Torlon 10499

dx.doi.org/10.1021/ie303531r | Ind. Eng. Chem. Res. 2013, 52, 10495−10505

Industrial & Engineering Chemistry Research

Article

composite hollow fibers by using THF dissolution experiment. Nevertheless, the PDMC sheath layer of composite hollow fibers is believed to be cross-linked since the 200 °C, 2 h cross-linked monolithic PDMC hollow fibers spun from the same batch PDMC as composite fibers are insoluble in THF.30 3.5. Hollow Fiber Membrane Characterization. 3.5.1. Scanning Electron Microscope (SEM). A scanning electron microscope (SEM) can be used to observe the cross-section of asymmetric hollow fibers. To prepare a fiber sample for the SEM test, the fibers are cryogenically fractured in liquid nitrogen to preserve their cross-section structures. Since the polymeric fibers are nonconductive, the cross-section of fibers must be coated with a gold layer for the SEM measurement. SEM images are particularly important to characterize the adhesion of sheath/core layers for composite hollow fibers. In fact, the cryogenic fracturing itself is a demanding test of adhesion. 3.5.2. Mixed Gas Permeation. The natural gas separation performance of hollow fibers is typically studied by using mixed gas permeation to simulate the realistic gas feed compositions. Compared to pure gas permeation, mixed gas permeation can probe the separation performance of membrane in the presence of both plasticization and competition effects.28 Both bore-fed and shell-fed systems can be used for gas permeation; however, a shell-fed system is preferred in this study since shell-fed modules can overcome the concentration polarization in the feed if mixed gas feeding is at a low stage cut. Moreover, in actual applications for high pressure feeds, shell side feed flow is preferred in the event of a fiber failure, which simply results in collapse and fiber shutdown. The stage cut is defined by the ratio of the permeate rate to feed flow rate. In mixed gas permeation, the retentate flow rate is determined by the permeate rate to meet a stage cut of less than 1% to completely avoid the concentration polarization in the feed in this fundamental study.

4.2. Composite Hollow Fiber Development. On the basis of the ternary phase diagrams and the successful singlelayer hollow spinning on PDMC, the dope was determined and consists of 30.5% polymer, 30.5% NMP, 19.46% ethanol, 13.04% THF, and 6.5% LiNO3.31 This same dope was used for the sheath layer, and the core layer was adjusted for compatibility to arrive the composition as noted in Table 1. The spinning conditions for the composite hollow fiber spinning are summarized in Table 1. Table 1. Composite Hollow Fiber Spinning Conditions dual-layer hollow spinning conditions core dope composition (Torlon/NMP/H2O/ PVP) sheath dope extrusion rate core dope extrusion rate bore fluid composition bore fluid rate spinneret temperature air gap quench bath take-up rate

16.0%/74.4%/5.6%/4.0% 30−60 mL/h 180 mL/h NMP/H2O 80/20 wt % 60 mL/h 50−70 °C 1−15 cm 20−50 °C 10−50 m/min

The spinning conditions shown in Table 1 are primarily based upon previous researches.9,32 The complex nature of polymer solutions and the sheath/core layer interface, together with aggressive cross-linking, for the composite hollow fiber spinning requires considerable trial and error to develop the desirable defect-free and delamination-free cross-linked composite hollow fibers. In this work, composite hollow fibers spun from various spinning conditions were first characterized by using SEM to probe the optimum spinning process variables, as shown in Table 1. Three main spinning factors were examined in this study: the quench bath temperature, sheath to dope flow rate ratio, and air gap residence time, as discussed below. For monolithic hollow fiber spinning, a higher quench bath temperature (say 50 °C) can accelerate the phase separation and promote the formation of open porous substructure.15 However, due to the complicated interface interaction of sheath/core layers, a higher quench bath temperature may cause delamination of sheath/core layers. Therefore, in this work, both cold water and hot water quench baths were considered and SEM images of the resultant composite hollow fibers are shown in Figure 7. Figure 7 shows that the hot quench bath temperature of 50 °C can cause a serious delamination of sheath/core layers while the cold water quench bath (20 °C) produces delamination-free composite hollow fibers. A lower temperature can reduce the diffusion rate of solvents and nonsolvents between nascent fibers and quench bath during the wet-quench process. This will increase the contact time of sheath dope and core dope before phase separation occurs. Since a longer “interdiffusion” time of polymer chains is hypothesized to lead to a better adhesion,33 a lower quench temperature is preferred to develop delamination-free composite hollow fibers. However, the quench bath temperature should be controlled carefully since an excessively low temperature may cause inadequate phase separation of nascent fibers before reaching the guide roller in the quench bath. This may produce undesirable oval hollow fibers, which is detrimental for fibers to withstand high feed transmembrane pressures. The lower quench bath temperature

4. RESULTS AND DISCUSSIONS 4.1. Dope Development for the Sheath and Core Layer. As discussed in section 3.2, the ternary phase diagrams must be constructed to prepare the spinnable sheath dope and core dope. A common and effective method to determine the binodal is through the cloud point technique.5,28 The results of cloud point experiments are summarized and plotted in ternary phase diagrams for both PDMC and Torlon, shown in Figures 5 and 6. In Figures 5 and 6, the solid points represent the one-phase samples and open circles represent the two-phase samples for the PDMC or Torlon solution system. The binodal (shown as black solid line) lies between the one-phase region and twophase region. After the construction of ternary phase diagrams, the dope composition for the hollow fiber spinning are chosen near the binodal to allow both rapid phase separation and high separation productivity. The location of the binodal is, of course, affected by the components in the dope, for instance, using H2O instead of ethanol will affect the binodal and move it to the left relative to the case for ethanol as a nonsolvent in Figure 5. The primary purpose of the ternary phase diagram was to determine the appropriate dope composition to produce a defect-free skin by moving dope composition closer to the binodal without crossing it during passage through the air gap as THF and ethanol are lost. 10500

dx.doi.org/10.1021/ie303531r | Ind. Eng. Chem. Res. 2013, 52, 10495−10505

Industrial & Engineering Chemistry Research

Article

Figure 7. SEM images showing un-cross-linked PDMC/Torlon composite hollow fiber samples spun at different quench bath temperatures: (a) cold water, ∼20 °C; (b) hot water, ∼50 °C.

Figure 8. SEM images showing un-cross-linked PDMC/Torlon composite hollow fibers spun at different sheath/core dope flow rate ratios at a quench bath temperature of 20 °C: (a) sheath:core = 1:3; (b) sheath:core = 1:6.

Figure 9. SEM images showing un-cross-linked PDMC/Torlon composite hollow fibers spun at different air gas residence time, t, at a quench bath temperature of 20 °C: (a) t = 0.04 s; (b) t = 0.1 s.

could produce composite hollow fibers with skin defects. This is possibly due to inadequate THF evaporation when reducing the sheath dope flow rate from 60 to 30 mL/h. It is believed that the evaporation of THF moves the dope composition toward to the vitrified region during the air gap period. However, the local polymer concentration in the outmost layer of a nascent fiber may be insufficient to produce a defect-free skin from the lower sheath dope flow rate. During the passage of nascent fibers in the air gap, the high solvent concentration of the core dope may be able to increase the sheath dope solvent concentration due to inward diffusion of solvent from the core dope. Since the diffusion rate is inversely proportional to thickness, the sheath dope with a lower flow rate (thus a smaller thickness) tends to be greater affected by the core dope

may also cause substructure resistance, which is formed during the phase separation process. The effect of sheath dope flow rate was also studied as it determines the amount of expensive PDMC used for dual-layer spinning. A sheath to core dope flow rate ratio of 1: 3 and 1:6 was studied in this work. The SEM images of composite hollow fibers spun from these two different sheath/core dope flow ratios are shown in Figure 8. As shown in Figure 8, the PDMC sheath layers are adhered well with the Torlon core layers at those sheath/core dope flow rate ratios (1:3 and 1:6). Moreover, open porous substructures are apparent in both of those composite hollow fibers, suggesting phase separation occurs rapidly. However, preliminary permeation showed that the lower dope flow ratio of 1:6 10501

dx.doi.org/10.1021/ie303531r | Ind. Eng. Chem. Res. 2013, 52, 10495−10505

Industrial & Engineering Chemistry Research

Article

Figure 10. Scanning electron micrographs of the cross section of an un-cross-linked PDMC/Torlon hollow fiber, showing the dense skin layer, sheath layer, core layer, and the porous substructure.

Figure 11. Scanning electron micrographs of the cross section of a cross-linked PDMC/Torlon hollow fiber, showing the dense skin layer, sheath layer, core layer, and the porous substructure.

solvent diffusion vs the case of a higher flow rate. This effect may allow some diffusive mixing of sheath and core dopes and reduce THF concentration in the sheath layer; thereby possibly offsetting evaporation of THF and promoting skin defects. On the other hand, the faster diffusion rate can also promote the phase separation and produce excessively porous fibers, which may cause skin defects of the fibers for the sheath dope with a lower flow rate. Therefore, the 1:3 dope flow ratio was preferred in this work to produce both delamination-free and defect-free composite hollow fiber membranes. The air gap residence time has been identified as a key factor in determining the skin layer formation and gas separation performance. Studies on Matrimid hollow fibers showed that a lower air gap residence time can reduce the skin layer thickness and improve the gas separation permeance.27 Therefore, the effect of air gap residence time was also explored in this work to produce the most productive composite hollow fibers. The air gap residence time, t, is determined by the ratio of air gap height to the take-up rate. SEM images of composite hollow fiber spun at different air gas residence time are shown in Figure 9. Figure 9 shows that composite hollow fibers show good adhesion of sheath/core layers at an air gap residence time from 0.04−0.1 s. A longer air gap residence time can produce hollow fibers with a relatively smaller dimension and increase the adhesion since the interdiffusion time of polymer chains is

increased;33 however, it may also cause a relatively thicker skin layer and reduce the gas permeance (separation productivity). Therefore, an air gap residence time of 0.04 s was preferred to produce thin-skinned and high-performance composite hollow fibers. SEM images of an un-cross-linked PDMC/Torlon composite hollow fiber spun from some of the best spinning states from this preliminary campaign are shown in Figure 10. Figure 10 shows that Torlon is compatible with PDMC as there is no delamination of sheath/core layer in the fibers during spinning. The open porous substructures are apparent in both the sheath and core layer. The composite hollow fibers after cross-linking were also checked by SEM, as shown in Figure 11. Figure 11 shows that the Torlon core layer demonstrates good adhesion with the sheath layer after cross-linking, suggesting the cross-linking does not cause delamination of core layer and sheath layer. Moreover, the open porous substructure of Torlon core layer did not collapse during crosslinking, showing that the Torlon core layer has a desirably strong thermal stability. The morphologies shown in the SEM images suggest that Torlon provides the desirable open porous morphology, compatibility with PDMC and thermal stability after cross-linking and is a promising polymer candidate for composite hollow fiber spinning. The natural gas separation performance of PDMC/Torlon composite hollow fibers will be 10502

dx.doi.org/10.1021/ie303531r | Ind. Eng. Chem. Res. 2013, 52, 10495−10505

Industrial & Engineering Chemistry Research

Article

CO2 plasticization. The gas separation performance was double-checked by depressurization. Figure 12 shows that the cross-linked composite hollow fibers demonstrated a slightly smaller degree of swelling after depressurization than the uncross-linked fibers, suggesting that cross-linking can stabilize hollow fibers against CO2 induced polymer swelling. Although cross-linking tends to cause a loss of permeance due to the densification of transition layer after cross-linking,16 the crosslinked composite hollow fibers still show an attractive CO2 permeance of 40 GPU at 100 psi, 35 °C. This fact notwithstanding, the neat PDMC cross-linked monolithic hollow fibers show a CO2 permeance of 117 GPU, suggesting that there is still room to improve the separation productivity of composite hollow fibers.31 On the other hand, Figure 13 shows that the cross-linking improved the separation selectivity from 28 (un-cross-linked) to 39 (cross-linked) at 100 psi feed pressure, 35 °C. The crosslinked composite hollow fibers show a CO2/CH4 selectivity of 39 at 65 psi, which is close to the intrinsic CO2/CH4 selectivity measured by dense films (αCO2/CH4 ∼ 42), suggesting that there is no apparent skin layer defect in the cross-linked composite hollow fibers. The permeation results of PDMC/Torlon composite hollow fibers suggest that Torlon is a promising material as the core layer to develop high performance composite hollow fibers with a significant reduction of membrane formation cost. 4.4. Hydrocarbon Contaminants Effects. As shown in section 4.3, the cross-linked composite hollow fibers show attractively high separation performance in a model natural gas feed consisting of only CO2 and CH4. Despite this attractive capability, typical raw natural gas feeds often contain a certain amount of hydrocarbon impurities. To explore this issue, highperformance cross-linked composite hollow fibers were tested in the presence of high levels of contaminants to probe their performance under even challenging feed streams. In this study, the cross-linked composite hollow fibers were further characterized with a feed having a toluene content from 30− 1000 ppm in 50/50 CO2/CH4 mixture at 35 °C. The permeation results are summarized in Figures 14 and 15.

further probed by using mixed gas permeation under aggressive feed conditions. 4.3. Natural Gas Separation Performance. The natural gas separation performance of PDMC/Torlon composite hollow fibers were evaluated by using the 50/50 CO2/CH4 mixed gas at a total feed pressure up to 300 psi. Figures 12 and 13 show the mixed gas permeation results of both un-crosslinked and 200 °C, 2 h cross-linked PDMC/Torlon composite hollow fibers.

Figure 12. CO2 permeance of un-cross-linked and cross-linked PDMC/Torlon composite hollow fiber membranes at elevated feed pressures in this work. Permeances calculated by using fugacity driving forces. Test conditions: 50/50 CO2/CH4, at 35 °C. Permeance uncertainty: ±10%.

Figure 13. CO2/CH4 selectivity of un-cross-linked and cross-linked PDMC/Torlon composite hollow fiber membranes at elevated feed pressures in this work. Selectivities calculated by using fugacity driving forces. Test conditions: 50/50 CO2/CH4, at 35 °C. Selectivity uncertainty: ±6%.

The mixed gas permeation data in Figure 12 shows that the un-cross-linked PDMC/Torlon composite hollow fibers show a slight CO2 permeance upswing at 200 psi of feed pressure, indicative of possible plasticization at a partial CO2 pressure of 100 psi. The plasticization of un-cross-linked monolithic hollow fibers was also found in ref 16. However, no apparent CO2 permeance upswing was found in the cross-linked PDMC/ Torlon composite hollow fibers, an indication of absence of

Figure 14. CO2 permeance of cross-linked PDMC/Torlon composite hollow fiber membranes at different toluene levels from this work. Toluene activity at 300 psi, 1000 ppm was calculated to be 0.33. Permeances calculated by using fugacity. Test conditions: 50/50 CO2/ CH4, at 35 °C. 10503

dx.doi.org/10.1021/ie303531r | Ind. Eng. Chem. Res. 2013, 52, 10495−10505

Industrial & Engineering Chemistry Research

Article

importantly, Torlon shows a strong thermal stability during the aggressive cross-linking without the collapse of the core layer, which is critical for preserving the high permeance and selectivity of composite hollow fibers. Mixed gas permeation shows that the cross-linked PDMC/Torlon composite hollow fibers have a CO2 permeance of 40 GPU with a CO2/CH4 selectivity of 39 in testing at 100 psi with 50/50 CO2/CH4 feed, 35 °C. Besides the permeation with clean CO2/CH4 mixture, the cross-linked PDMC/Torlon composite hollow fibers also demonstrate decent separation performance in the presence of a toluene contaminant level up to 1000 ppm, suggesting crosslinked PDMC/Torlon composite hollow fibers are viable under realistic operation conditions. The advanced dual-layer spinning technique described in this paper can not only significantly reduce the cost of materials used for the hollow fiber formation but also achieve the high separation performance of sheath layer polymer. The significantly improved natural gas separation performance and reduced cost of hollow fiber membrane formation provides a significant advancement in the state of the art for natural gas separations, which is essentially beneficial for commercialization of expensive high performance materials by integrating low-cost supporting polymers with expensive sheath layer polymers.

Figure 15. CO2/CH4 selectivity of cross-linked PDMC/Torlon composite hollow fiber membranes at different toluene levels from this work. Selectivities calculated by using fugacity. Test conditions: 50/50 CO2/CH4, at 35 °C.

Figure 14 shows that the increasing of toluene levels reduces the CO2 permeance, which is believed to be due to the antiplasticization induced by the toluene contaminant. Antiplasticization occurs when an antiplasticizer reduces the fractional free volume and increase the stiffness by lowering the glass transition temperature (Tg) of the polymer.18,34 Toluene induced antiplasticization can reduce fractional free volume (FFV) and thereby hinder the diffusion of each penetrant and reduce the permeance accordingly. Despite the permeance loss induced by the antiplasticization, cross-linked PDMC/Torlon composite hollow fibers maintained a CO2 permeance over 15 GPU even in the presence of 1000 ppm toluene, which is about 50% higher than what reported for the cross-linked monolithic hollow fibers in the literature (CO2 permeance ∼ 10 GPU).17 However, the CO2 permeance of composite cross-linked hollow fibers under 1000 ppm toluene is relatively lower than the monolithic fiber samples (CO2 permeance ∼ 50 GPU) spun from another 120 000 Mw batch PDMC, which result from optimized spinning dopes and spinning processes.31 Details about the toluene effect on monolithic hollow fibers can be found in ref 30. On the other hand, as shown in Figure 15, the presence of toluene somehow reduces the CO2/CH4 selectivity only moderately, since both penetrants are affected, and the composite hollow fibers showed attractively high CO2/CH4 selectivity above 30 at all testing toluene levels. The high separation performance in the presence of high-level toluene contaminants demonstrates that cross-linked PDMC/Torlon composite hollow fibers are promising for natural gas separations under extremely challenging feed conditions. With moderate pretreatment to achieve a 30 ppm contaminant level, the antiplasticization effect is greatly moderated. Future studies will seek to reduce the selective layer in the sheath while maintaining defect-free nature of the membrane.

■

AUTHOR INFORMATION

Corresponding Author

*Tel.: +1 404 385 2845. Fax: +1 404 385 2683. E-mail address: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors acknowledge the financial supports by Chevron Energy Technology Company, the Roberto C. Goizueta Chair in Chemical Engineering, and the Georgia Research Alliance.

■

REFERENCES

(1) Jiang, L. Y.; Chung, T. S.; Li, D. F.; Cao, C.; Kulprathipanja, A. Fabrication of Matrimid/polyethersulfone dual-layer hollow fiber membranes for gas separation. J. Membr. Sci. 2004, 240 (1−2), 91− 103. (2) Pinnau, I.; Wind, J.; Peinemann, K. V. Ultrathin Multicomponent Poly(Ether Sulfone) Membranes for Gas Separation Made by Dry Wet Phase Inversion. Ind. Eng. Chem. Res. 1990, 29 (10), 2028−2032. (3) Pesek, S. C.; Koros, W. J. Aqueous Quenched Asymmetric Polysulfone Membranes Prepared by Dry Wet Phase-Separation. J. Membr. Sci. 1993, 81 (1−2), 71−83. (4) Husain, S.; Koros, W. J. Mixed matrix hollow fiber membranes made with modified HSSZ-13 zeolite in polyetherimide polymer matrix for gas separation. J. Membr. Sci. 2007, 288 (1−2), 195−207. (5) Wallace, D. W. Crosslinked Hollow Fiber Membranes for Natural Gas Purification and Their Manufacture from Novel Polymers. Ph.D. Dissertation, The University of Texas at Austin, Austin, TX, 2004. (6) Pereira, C. C.; Nobrega, R.; Peinemann, K. V.; Borges, C. P. Hollow fiber membranes obtained by simultaneous spinning of two polymer solutions: a morphological study. J. Membr. Sci. 2003, 226 (1−2), 35−50. (7) Strathmann, H. Membrane separation processes: Current relevance and future opportunities. AIChE J. 2001, 47 (5), 1077− 1087. (8) Li, D. F.; Chung, T. S.; Wang, R.; Liu, Y. Fabrication of fluoropolyimide/polyethersulfone (PES) dual-layer asymmetric hollow fiber membranes for gas separation. J. Membr. Sci. 2002, 198 (2), 211− 223.

5. CONCLUSIONS High-performance cross-linked PDMC/Torlon composite hollow fibers were successfully developed without skin layer defects or delamination of sheath/core layers. Torlon demonstrates excellent compatibility with cross-linkable PDMC polyimide sheath layer during spinning. More 10504

dx.doi.org/10.1021/ie303531r | Ind. Eng. Chem. Res. 2013, 52, 10495−10505

Industrial & Engineering Chemistry Research

Article

(9) Liu, J. Development of Next Generation Mixed Matrix Hollow Fiber Membranes for Butane Isomer Separation. Ph.D. Dissertation. Georgia Institute of Technology, Atlanta, GA, 2010. (10) Li, D. F.; Chung, T. S.; Rong, W. Morphological aspects and structure control of dual-layer asymmetric hollow fiber membranes formed by a simultaneous co-extrusion approach. J. Membr. Sci. 2004, 243 (1−2), 155−175. (11) Henne, W.; Dunweg, G.; Schmitz, W.; Pohle, R.; Lawitzki, F. Method of producing dialyzing membrane. U.S. Patent 4,164,437, 1979. (12) Li, Y.; Chung, T. S.; Huang, Z.; Kulprathipanja, S. Dual-layer polyethersulfone (PES)/BTDA-TDI/MDI co-polyimide (P84) hollow fiber membranes with a submicron PES-zeolite beta mixed matrix dense-selective layer for gas separation. J. Membr. Sci. 2006, 277 (1− 2), 28−37. (13) Liu, Y.; Chung, T. S.; Wang, R.; Li, D. F.; Chng, M. L. Chemical cross-linking modification of polyimide/poly(ether sulfone) dual-layer hollow-fiber membranes for gas separation. Ind. Eng. Chem. Res. 2003, 42 (6), 1190−1195. (14) Omole, I. C.; Adams, R. T.; Miller, S. J.; Koros, W. J. Effects of CO2 on a High Performance Hollow-Fiber Membrane for Natural Gas Purification. Ind. Eng. Chem. Res. 2010, 49 (10), 4887−4896. (15) Omole, I. C.; Miller, S. J.; Koros, W. J. Increased molecular weight of a cross-linkable polyimide for spinning plasticization resistant hollow fiber membranes. Macromolecules 2008, 41 (17), 6367−6375. (16) Omole, I. C. Crosslinked Polyimide Hollow Fiber Membranes for Aggressive Natural Gas Feed Streams. Ph.D. Dissertation, Georgia Institute of Technology, Atlanta, GA, 2008. (17) Omole, I. C.; Bhandari, D. A.; Miller, S. J.; Koros, W. J. Toluene impurity effects on CO2 separation using a hollow fiber membrane for natural gas. J. Membr. Sci. 2011, 369 (1−2), 490−498. (18) Koros, W. J.; Fleming, G. K. Membrane-Based Gas Separation. J. Membr. Sci. 1993, 83 (1), 1−80. (19) Wind, J. D.; Staudt-Bickel, C.; Paul, D. R.; Koros, W. J. The effects of crosslinking chemistry on CO2 plasticization of polyimide gas separation membranes. Ind. Eng. Chem. Res. 2002, 41 (24), 6139− 6148. (20) Wind, J. D.; Staudt-Bickel, C.; Paul, D. R.; Koros, W. J. Solidstate covalent cross-linking of polyimide membranes for carbon dioxide plasticization reduction. Macromolecules 2003, 36 (6), 1882− 1888. (21) Hillock, A. M. W.; Koros, W. J. Cross-linkable polyimide membrane for natural gas purification and carbon dioxide plasticization reduction. Macromolecules 2007, 40 (3), 583−587. (22) Hillock, A. M. W.; Miller, S. J.; Koros, W. J. Crosslinked mixed matrix membranes for the purification of natural gas: Effects of sieve surface modification. J. Membr. Sci. 2008, 314 (1−2), 193−199. (23) Wallace, D. W.; Staudt-Bickel, C.; Koros, W. J. Efficient development of effective hollow fiber membranes for gas separations from novel polymers. J. Membr. Sci. 2006, 278 (1−2), 92−104. (24) Wallace, D. W.; Williams, J.; Staudt-Bickel, C.; Koros, W. J. Characterization of crosslinked hollow fiber membranes. Polymer 2006, 47 (4), 1207−1216. (25) Carruthers, S. B. Integral-skin formation in hollow fiber membranes for gas separations. Ph.D. Dissertation, The University of Texas at Austin, Austin, TX, 2001. (26) Clausi, D. T. Formation and characterization of asymmetric polyimide hollow fiber membranes for gas separations. Ph.D. Dissertation, The University of Texas at Austin, Austin, TX, 1998. (27) Clausi, D. T.; Koros, W. J. Formation of defect-free polyimide hollow fiber membranes for gas separations. J. Membr. Sci. 2000, 167 (1), 79−89. (28) Ma, C. Optimization of Asymmetric Hollow Fiber Membranes for Natural Gas Separation. M.S. Thesis, Georgia Institute of Technology, Atlanta, GA, 2011. (29) Boom, R. M.; Vandenboomgaard, T.; Vandenberg, J. W. A.; Smolders, C. A. Linearized Cloudpoint Curve Correlation for TernarySystems Consisting of One Polymer, One Solvent and One Nonsolvent. Polymer 1993, 34 (11), 2348−2356.

(30) Ma, C. Highly Productive Ester Crosslinkable Composite Hollow Fiber Membranes for Aggressive Natural Gas Separations. Ph.D. dissertation, Georgia Institute of Technology, Atlanta, GA, 2012. (31) Ma, C.; Koros, W. J. High-performance Ester-crosslinked Hollow Fiber Membranes for Natural Gas Separations. J. Membr. Sci. 2013, 428, 251−259. (32) Husain, S. Mixed Matrix Dual Layer Hollow Fiber Membranes for Natural Gas Separation. Ph.D. Dissertation, Georgia Institute of Technology, Atlanta, GA, 2006. (33) He, T.; Mulder, M. H. V.; Strathmann, H.; Wessling, M. Preparation of composite hollow fiber membranes: co-extrusion of hydrophilic coatings onto porous hydrophobic support structures. J. Membr. Sci. 2002, 207 (2), 143−156. (34) Madden, W. C. The Performance of Hollow Fiber Gas Separation Membranes in the Presence of an Aggressive Feed Stream. Ph.D. Dissertation, Georgia Institute of Technology, Atlanta, GA, 2005.

10505

dx.doi.org/10.1021/ie303531r | Ind. Eng. Chem. Res. 2013, 52, 10495−10505