PIM-1 as an organic filler to enhance the gas separation performance of Ultem polyetherimide

Journal of Membrane Science 453 (2014) 614–623

Contents lists available at ScienceDirect

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

PIM-1 as an organic filler to enhance the gas separation performance of Ultem polyetherimide Lin Hao a,b, Pei Li a, Tai-Shung Chung a,n a b

Department of Chemical and Biomolecular Engineering, National University of Singapore, Engineering Drive 4, Singapore 117576, Singapore NUS Graduate School for Integrative Sciences and Engineering, National University of Singapore, 28 Medical Drive, Singapore 117456, Singapore

art ic l e i nf o

a b s t r a c t

Article history: Received 12 October 2013 Received in revised form 18 November 2013 Accepted 24 November 2013 Available online 12 December 2013

An assortment of Ultem/PIM-1 polymer blends was prepared and their transport properties to a series of gases were studied. Good dispersion between the PIM-1 and Ultem phases was found when the PIM-1 loading was low (o 20 wt%) or high ( 490 wt%). A slight shift of Tg was observed when the PIM-1 loading increased from 0 wt% to 50 wt%, suggesting likely partially miscibility. The molecular-level interactions were further confirmed by the FTIR and XRD data, where shifts of peaks were detected at several compositions. Gas transport properties of pure gases including He, N2, O2, CH4, CO2 for all polymer blends and mixed gases including CO2/CH4 (50/50) and CO2/N2 (50/50) gas pairs for Ultem/PIM-1 (90:10) and Ultem/PIM-1 (80:20) blends were explored. Considerable increments in gas permeability were observed by adding only 5 or 10 wt% PIM-1 without much compromising gas pair selectivity, i.e., the CO2 permeability increased impressively over 47% and 167%, respectively, compared with the pristine Ultem. When comparing the gas permeation properties with the predictions from semi-logarithm and Maxwell equations, they follow nicely with the semi-logarithm addition when the PIM-1 loadings are low ( o20 wt%), indicting relatively homogenous blends at these compositions, while the transport properties match closely with the Maxwell prediction at high PIM-1 loadings ( 490 wt%) due to the good dispersion of Ultem inside PIM-1. This study opens up the potential of employing PIM-1 as an organic filler to improve the permeability of low permeable materials for other industrial membrane applications. & 2013 Elsevier B.V. All rights reserved.

Keywords: Polymer blend PIM-1 Organic filler Model prediction

1. Introduction Membrane technology has been rapidly expanded to chemical and petrochemical industries for natural gas sweetening, oxygen enrichment, hydrogen purification, and separation of organic compounds attributed to its relatively small ecological footprint, easy operation and low energy consumption [1,2]. Conventional glassy polymers like polyimides, polycarbonates, polysulfone, cellulose acetate and rubbery polymers such as poly(dimethylsiloxane) occupy most of the membrane market in gas separation [3]. However, industrial applications of polymeric membrane are still limited due to their mediocre separation performance. Thus, various methods have been proposed such as thermal rearrangement, crosslinking, polymer blend, etc. for fabricating polymers with both high permeability and selectivity [4–6]. Polymer blending is a relatively straightforward and effective approach to develop industrial materials that have the advantages and attractive features of the respective constituents [7–9]. More recently, Budd and McKeown reported a kind of soluble polymer of intrinsic microporosity (PIM-1, structure shown in

n

Corresponding author. Tel.: þ 65 65166645; fax: þ 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.2013.11.045

Fig. 1) which consists of rigid ladder-like structure, large and accessible surface area [10,11], high fractional free volume (FFV), good thermal and chemical stability [12,13]. PIM-1 has a very high CO2 permeability and moderately good selectivity for CO2/CH4 and CO2/N2 gas pairs [14,15]. Its gas sorption and permeation properties have been studied in detail by several groups [16–18] and its sorption behavior follows the dual-mode sorption model [16,19]. It was found that the effective diffusion coefficients of PIM-1, which were calculated from the experimentally measured permeability and solubility coefficients using the relation D ¼P/S, are significantly higher than the diffusion coefficients of most glassy polymers such as polycarbonate, polysulfone, polyimide, etc. [13,16–20]. Due to the high gas permeability of PIM-1, it had been suggested as an organic filler to improve the permeability of low permeable polymers [21]. Mixed matrix membranes (MMMs) consisting of inorganic particles usually face challenges of poor interfaces, chain rigidification and partial pore blocking, thus the separation performance is often lower than the theoretical prediction [22–24]. Since PIM-1 is an organic polymer, blending it with conventional polymers might minimize the problems of low compatibility and improve the gas separation performance. Yong et al. were the pioneers in studying the miscibility and gas separation properties of PIM-1/Matrimid blends [25]. They

L. Hao et al. / Journal of Membrane Science 453 (2014) 614–623

found that PIM-1 and Matrimid are partially miscible. The CO2 permeability of Matrimid was increased by 25% and 77% by just adding 5 and 10 wt% PIM-1, respectively, with slight sacrifices of the CO2/CH4 selectivity. The PIM-1/Matrimid hollow fiber membranes were also demonstrated [26]. Even though Matrimid has relatively high gas permeability, it is an expensive polyimide material and has a low plasticization pressure against CO2 [27]. These concerns drive us to explore other blend pairs such as the PIM-1 and polyetherimide (PEI) Ultems 1010 (noted as Ultem thereafter) system. As illustrated in Fig. 1, Ultem is a thermoplastic resin with excellent thermal stability and has been widely used for electrical insulation and medical tools [28]. It possesses hydrolytic stability, high modulus, good UV radiation and ease of processability. Compared to Matrimid, Ultem has better chemical resistance to common solvents which makes it suitable for applications under harsh environments [29–34]. The pure gas permeability of Ultem to CO2 and O2 are 1.33 and 0.41, respectively; while the ideal selectivity of CO2/CH4, CO2/N2 and O2/N2 gas pairs are 37, 25 and 7.5, respectively [29–31]. Many Ultem polymer blends have been reported including those with polyetheramide, thermosetting polyimide and liquid crystal polymers [32–34]. In this study, we aim to investigate the miscibility and gas separation properties of the Ultem/PIM-1 blend system and examine its potential as a gas membrane material by taking advantage of PIM-1's high gas permeability and Ultem's good gas selectivity for CO2/CH4 and CO2/N2 separation. This study may provide useful insights for the exploration of new industrial materials based on PIM-1 for gas separation.

615

2.2. Synthesis of polymer of intrinsic microporosity (PIM-1) PIM-1 was synthesized using the method developed by McKeown and Budd [11,35]. Purified TFTPN, TTSBI and K2CO3 (used as received) were mixed based on a stoichiometric ratio of 1:1:2 in anhydrous NMP. The solution was heated at 60 1C with vigorous stir under N2 atmosphere for 18 h. PIM-1 was precipitated in MeOH. The obtained product was washed with HCl and then deionized water. At last, PIM-1 was dried at 120 1C under vacuum for 1 day with a yield of about 90%. The molecular weight of PIM-1 was measured through gel permeation chromatography (GPC) using polystyrene standards as the calibration polymer and THF as the solvent. The lab-made PIM1 has a number average molecular weight (Mn), weight average molecular weight (Mw) and polydispersity (PDI¼ Mw/Mn) of 77,925, 170,891, and 2.19, respectively. 2.3. Preparation of the Ultem/PIM-1 films

2. Experiments

Polymer films were prepared by solution casting. Ultem pellets were dried overnight at 120 1C under vacuum to remove moisture and then dissolved in chloroform. PIM-1 with predetermined amounts were added into individual solutions to prepare Ultem/ PIM-1 blends with PIM-1 concentrations of 5, 10, 20, 30, 50, 70, 80, 90 and 95 wt%. Each solution was stirred overnight, filtered by 0.45 mm PTFE membrane filters to remove any undissolved particles and then cast on a petri dish covered by a lid to allow slow evaporation of the solvent. The obtained polymer films were further dried in a vacuum oven at 120 1C for 1 day. Films with an average thickness of 45 75 mm measured by a Digimatic indicator (Mitutoyo Mexicana) on ten various points were collected for further studies.

2.1. Materials

2.4. Characterizations

2,3,5,6-Tetrafluoroterephthalonitrile (TFTPN, 99%) and 5,50 ,6,60 tetrahydroxy-3,3,30 ,30 -tetramethyl-1,10 -spirobisinadane (TTSBI, 97%) were ordered from Matrix Scientific and Alfa Aesar, respectively. Ultem was purchased from the former GE Plastics. Chemicals such as hydrochloric acid (HCl, 37.5%), hexane (99.9%), tetrahydrofuran (THF, 99.99%) and chloroform (99.98%) were acquired from Fisher Scientific. Anhydrous potassium carbonate (K2CO3, 499.5%) was obtained from Sigma-Aldrich and used as-received. N-methyl-2pyrrolidone (NMP, 499.5%) and methanol (MeOH, 499.9%) were bought from Merck.

Surface images of the films were obtained by both atomic force microscopy (AFM, Bruker Dimention ICON), and Olympus BX50 Polarized Light Microscope (PLM). Images were further analyzed with NanoScope Analysis V1.4 and Image Pro Plus 3.0 software, respectively. The X-ray diffractor (Bruker, D8 series, General Area Detector Diffraction System (GADDS)) with a Cu Kα X-ray source (wavelength: 1.54 Å) was applied to characterize the selected polymer blends. Fourier transform infrared spectroscopy (FTIR) was employed to detect the chemical structural changes of the polymers with the

Fig. 1. The chemical structures of (a) Ultems 1010 and (b) PIM-1.

616

L. Hao et al. / Journal of Membrane Science 453 (2014) 614–623

fixed angle attenuated total reflectance (ATR) mode using a Bio-Rad FTS-3500 with a scan range from 400 to 4000 cm  1. Glass transition temperatures (Tg) of the polymer blends were obtained by using differential scanning calorimetry (DSC) (Mettler Toledo DSC 822e, Columbus, OH) with a heating rate of 10 1C/min. Samples with weights of 7–8 mg were prepared. The Tg values were taken from the second heating run. The densities of all membranes were acquired by a Mettler Toledo analytical balance ML204 and a density kit ML-DNY-43 (Zurich, Switzerland). The densities were estimated according to the Archimedean principle as shown in Eq. (1) using the values of film weights measured in both air and liquid.

ρM ¼

M air ðρ  ρair Þ þ ρair M air M liq liq

Table 1 The Vw used for the calculation of FFV in Ultem. Group

Vw (cm3/mol) 13.67 43.3

5.5

ð1Þ

43.3

where ρM, ρair and ρliq represent the densities of polymers, air and applied liquid (i.e., hexane), respectively; Mair and Mliq refer to the weights of the polymers measured in air and liquid. All measurements were conducted at 23.1 1C (lab temperature). For a postulated homogeneous Ultem/PIM-1 blend, the density is calculated as follows:

3.3

ρ12 ¼ ω1 ρ1 þ ω2 ρ2

ð2Þ

where ρ12 , ρ1 and ρ2 represent the densities of polymer blends, and individual components 1, and 2; ω1 and ω2 represent the volume fractions of each component. The fractional free volume (FFV) of polymer blends was estimated based on the following correlation: FFV ¼

V V0 V

ð3Þ

where V is the specific volume (1/ρ) and V0 is the impermeable volume occupied by the polymer and is 1.3 times that of its van der Waals volume: V0 ¼ 1.3Vw [36]. Vw can be obtained via Bondi's group contribution method [37]. Table 1 lists the individual Vw of Ultem groups for the calculation of its Vw while the Vw of PIM-1 can be adopted from Thomas et al. [14] and Yong et al. [25]. As a result, the van der Waals volumes of Ultem/PIM-1 blends could be estimated using Eq. (4). V w ¼ n1 V w1 þ n2 V w2

ð4Þ

where n1, n2 and Vw1, Vw2 refer to the mole fractions and van der Waals volumes of PIM-1 and Ultem, respectively. For a postulated homogenous Ultem/PIM-1 blend, its FFV can be estimated from Eq. (5) [38]. FFV12 ¼ ω1 FFV1 þ ω2 FFV2

69.4

ð5Þ

where FFV1, FFV2, and FFV12 refer to the fractional free volumes of the components 1, 2 and the polymer blend, respectively. 2.5. Permeation measurements Both pure and mixed gas permeabilities were measured using a constant volume permeation cell [39,40]. Membrane samples were vacuumed inside the permeation cell overnight before tests for degassing. Pure gas permeation tests were carried out in the order of He, N2, O2, CH4, CO2 at 35 1C under 3.5 ×105 N/m2 (3.5 bar). Upon pressurization of the upstream, the increase in downstream pressure was recorded for the estimation of permeability using Eq. (6).   273  1010 V dL dp P¼ ð6Þ ATðP 1  76=14:7Þ dt 760 where P is the permeability in Barrer (1 Barrer¼1  1010 cm3(STP) cm/(cm2 s cmHg)¼3.348x10-19 kmol m/(m2 s Pa)), Vd is the volume of the downstream reservoir (cm3), L stands for the membrane thickness (cm), A is the effective membrane area (cm2), T is the operating temperature (K) and P1 indicates the upstream feed

pressure (psia). Mixed gas permeability was measured for CO2/N2 (50/50 mol%) and CO2/CH4 (50/50 mol%). The mixed gas permeability was obtained at 35 1C with a partial pressure of 3.5 bar for each gas component. Permeability of each component can be estimated by Eq. (7):   ygas V d L 273  1010 dp P¼ ð7Þ 760 ATð76=14:7Þðxgas P 1 Þ dt where x and y symbolize the molar fractions of each gas on the feed and the permeate sides, respectively. The composition of the mixed gas in the permeate side was detected by gas chromatography (GC).

3. Results and discussion 3.1. Characterizations AFM measurements offer a direct observation of membrane surface morphology. Fig. 2 presents the AFM images of all membrane surfaces at the dimension of 1  1 mm2 as a function of PIM-1 content. Table 1 shows the surface roughness in terms of the arithmetic mean value, Ra, or the root mean square value, Rq. Both Ra and Rq, increase as the PIM-1 concentration increases to 50 wt%, and then decrease as the PIM-1 concentration further increases. Thus, the surfaces can be considered relatively smooth when the PIM-1 concentration is below 20 or above 80 wt% where both Ra and Rq are less than 10 nm. Consistent with the AFM data, Fig. 3 displays the optical morphology observed by a PLM as a function of PIM-1 content. Clear phase separation can be observed when the PIM-1 concentration is between 30% and 70% (i.e., between Fig. 3(e) and (g)). Reasonably good miscibility seems to take place when the PIM-1 concentration is r 10 wt% (Fig. 3(a) and (b)) or Z95 wt% (Fig. 3(k)), while partial miscibility is observed for samples containing 20%, 80% and 90% PIM-1 (Fig. 3(c), (i) and (k)). These AFM and optical results are slightly different from Liu et al.'s finding

L. Hao et al. / Journal of Membrane Science 453 (2014) 614–623

617

Fig. 2. The AFM surface images of polymer blends with weight ratios of Ultem to PIM-1 at (a) 100:0, (b) 95:5, (c) 90:10, (d) 80:20, (e) 70:30, (f) 50:50, (g) 30:70, (h) 20:80, (i) 10:90, (j) 5:95 and (k) 0:100.

a

b

c

d

e

f

g

h

i

j

k

100 µm

Fig. 3. The PLM surface images of polymer blends with weight ratios of Ultem to PIM-1 at (a) 100:0, (b) 95:5, (c) 90:10, (d) 80:20, (e) 70:30, (f) 50:50, (g) 30:70, (h) 20:80, (i) 10:90, (j) 5: 95 and (k) 0:100.

where phase separation was not observed for Ultem/PIM-1 blends with a PIM-1 concentration up to 30 wt% [21]. Fig. 4 shows the transparency of Ultem/PIM-1 blend films. High transparency is obtained by naked eyes when the PIM-1 loading is about 5 wt%

(Fig. 4(a)) or 80 wt% (Fig. 4(f)). In accordance with AFM and PLM observations, this suggests good distributions of the minor component within the major component in these concentration ranges.

L. Hao et al. / Journal of Membrane Science 453 (2014) 614–623

Matrimid/PIM-1 case [25], the peak shift is more apparent for the polymer blend with more than 90 wt% PIM-1 in Ultem, indicating a more homogeneous blending occurring at these compositions that induces reasonable interactions between the two polymers. Fig. 8 shows the XRD spectra of the pristine polymers and some partially miscible Ultem/PIM-1 blends. Similar to the reported value of 5.24 Å [49], Ultem possesses one broad peak at around 171 that corresponds to a d-space value of 5.21 Å, while PIM-1 has apparent peaks at 181 and 13.61 that correspond to d-space values of 4.9 Å and 6.5 Å, as reported in the literatures [50]. The d-space of 4.9 Å represents efficiently packed chains, while the d-space of 6.5 Å arises from micropores between chains [50]. Clear XRD shifts can be observed in these blends. Ultem/PIM-1 (10:90) has a lower d-space value than PIM-1, while Ultem/PIM-1 (90:10) has a larger d-space value than Ultem. Consistent with Tg and FTIR shifts, the d-space shifts indicate interactions between Ultem and PIM-1 at a molecular level for some compositions.

1.40

0.30 Density (Prediction) Density (Experimental) FFV (prediction)

1.30

0.25

FFV (experimental)

1.20

0.20

1.10

0.15

1.00

FFV

Fig. 5 illustrates the variations in density and FFV with PIM-1 content in these polymer blends. As the PIM-1 loading is raised, FFV increases while density decreases because PIM-1 is a high FFV polymer. However, the FFV estimated from the density measurement is higher than the value predicted from Eq. (5) when the PIM-1 loading is smaller than 20 wt% due to the formation of interface microvoids between PIM-1 and Ultem molecules. Interestingly, the predicted FFV value is higher than the estimated FFV value when the PIM-1 concentration is higher than 30 wt%. As reported by other groups [41,42], PIM-1 has some microvoids with sizes around 10–15 Å. It is likely that the lower FFV values are resulted from the filling of these large microvoids by Ultem polymer chains (Table 2). Fig. 6 shows that Tg evolution as a function of PIM-1 loading. Tg shifts from 215 1C to 221 1C once PIM-1 reaches 50 wt%. Clearly, there are interactions between PIM-1 and Ultem at a molecular level and the presence of PIM-1 rigidifies Ultem polymer chains and hinders their rotation. Since Tg of PIM-1 is very difficult to detect [43,44], only one Tg attributed to Ultem was measured. However, the amount of Ultem Tg increment is much lower than that predicted from Fox [45] and Kwei [46] equations for fully miscible blends, this suggests that the most interacted Ultem and PIM-1 blends are only partially miscible. ATR and FTIR were both employed to investigate chemical structure changes of the blends. As shown in Fig. 7, the peaks from 600 cm  1 to 2000 cm  1 were obtained by ATR while the peaks from 2000 cm  1 to 2800 cm  1 were measured under the FTIR transmission mode since the signal obtained from ATR was weak over this range. With an increase in PIM-1 loading, the peak at near 2240 cm  1 that corresponds to C≡N [42] becomes visible and stronger, while the peaks corresponding to carbonyl C ¼ O bending near 720 cm  1 [47], stretching of imide C–N at 1361 cm  1 [5], symmetric and asymmetric stretching of C ¼ O at 1712 cm  1 and 1778 cm  1 [5,48] decrease. Consistent with the Tg shift, the C¼ O stretching at 1712 cm  1 also shifts towards a higher frequency when the PIM-1 concentration increases. Similar to the previous

Density (g/cm3)

618

0.10 0.0

0.2

0.4

0.6

0.8

1.0

PIM-1 Volume Fraction Fig. 5. The density and FFV of polymer blends as a function of PIM-1 volume fraction.

Fig. 4. Photos of membranes with weight ratios of Ultem to PIM-1 at (a) 95: 5, (b) 90:10, (c) 80:20, (d) 70:30, (e) 30:70 and (f) 20:80.

L. Hao et al. / Journal of Membrane Science 453 (2014) 614–623

3.2. Gas transport properties of the Ultem/PIM-1 polymer blends 3.2.1. Gas permeability Table 3 summarizes the pure gas permeability of He, N2, O2, CH4 and CO2 for Ultem/PIM-1 blends at different PIM-1 loadings. The permeability of the pristine PIM-1 and Ultem are comparable with the literature data [16,29]. The permeability of all gases

Table 2 The surface roughness of the polymer blend. Ultem:PIM

Surface roughness

100:0 95:5 90:10 80:20 70:30 50:50 30:70 20:80 10:90 5:95 0:100

Rq

Ra

0.90 2.26 5.26 8.04 11.5 21.4 15.3 9.23 4.48 2.15 0.78

0.65 1.82 4.14 6.77 9.45 17.1 11.8 6.69 3.23 1.58 0.61

619

increases with increasing PIM-1 loading while the selectivity of O2/N2, CO2/CH4 and CO2/N2 initially drops slightly at low PIM-1 loadings but later approaches to the respective selectivity of the pristine PIM-1. When the PIM-1 loadings are low, the permeability of the blends improves impressively. At a loading of 10 wt% PIM-1, the permeability of most gases upsurges more than 150%. When the PIM-1 loading increases to 20 wt%, the permeability of almost all gases reaches around 2 times higher than those of the pristine Ultem without significant sacrifices of the selectivity. The tremendous increment of permeability is attributed to (1) the large free volume of PIM-1 as signified from the d-spacing and (2) the microvoids formed in the interfaces between PIM-1 and Ultem since the estimated FFV values of the polymer blends are greater than the predicted values at low PIM-1 loadings. Notably, the reductions in selectivity of CO2/CH4 and CO2/N2 are less than 8% and that of O2/N2 is 14% when comparing 80/20 Ultem/PIM-1 to the pristine Ultem. However, the permeability further increases while the selectivity simultaneously drops with a raise in PIM-1 content from 30 wt% to 80 wt%. Eventually, the gas transport properties of the blend system approach to those of the pristine PIM-1 when the PIM-1 concentration exceeds 90 wt%. Fig. 9 plots the separation performance against the Robeson upper bound 2008 as a function of PIM-1 loading [51]. The separation performance of the blends surpasses the 2008 Robeson upper bound for the O2/N2 separation when the PIM-1 loading reaches 70 wt%.

Pristine Ultem Ultem/PIM-1 (90: 10)

Pristine PIM-1 Ultem/PIM-1 (10: 90)

Intensity (a.u.)

Ultem/PIM-1 (80: 20) Ultem/PIM-1 (70: 30) Ultem/PIM-1 (50: 50) Ultem/PIM-1 (30: 70) Ultem/PIM-1 (20: 80)

Ultem/PIM-1 (90: 10)

Ultem/PIM-1 (10: 90)

Pristine Ultem

160

180

200

220

240

260 5

Temperature (oC)

10

15

20

25

30

2 theta (degree) Fig. 6. The DSC curves of pristine Ultem and Ultem/PIM-1polymer blends. The arrow is to guide eyes.

Fig. 8. The XRD spectra of Ultem/PIM-1polymer blends.

Pristine Ultem Ultem/PIM-1 (95: 5) Ultem/PIM-1 (90: 10) Ultem/PIM-1 (80: 20) Ultem/PIM-1 (70: 30)

Ultem/PIM-1 (50: 50) Ultem/PIM-1 (30: 70) Ultem/PIM-1 (20: 80) Ultem/PIM-1 (10: 90) Ultem/PIM-1 (5: 95) Pristine PIM-1 2240 (C≡N)

Wavelengths (cm-1) Fig. 7. The ATR and FTIR spectra of Ultem/PIM-1polymer blends.

35

620

L. Hao et al. / Journal of Membrane Science 453 (2014) 614–623

Ultem/PIM-1 blends (90:10 and 80:20) were further tested by mixed gas pairs of CO2/N2 (50/50 mol%) and CO2/CH4 (50/50 mol%) at 35 1C with a partial pressure under 3.5 bar. The data listed in Table 4 are comparable with the pure gas data. Interestingly, for the CO2/N2 and CO2/CH4 mixed gas pairs, the membrane has slightly higher mixed gas selectivities than the pure gas ones, which could be possibly associated with the higher affinity of PIM1 towards CO2 and better sorption competition during mixed gas tests. This phenomenon was also observed by Guiver and his co-workers [52,53] that the sorption of CO2 suppresses that of N2. They observed a higher CO2/N2 selectivity of 30.2 when the concentration of CO2 was higher than 40%. 3.2.2. Prediction of permeability Since a good dispersion of PIM-1 was observed inside the Ultem matrix at low PIM-1 loadings and vice versa, the Maxwell

equation was applied to calculate the permeability for this blend system [54].   P D þ 2P C  2φD ðP C  P D Þ P ef f ¼ P C ð9Þ P D þ 2P C þ φD ðP C  P D Þ where Peff represents the effective permeability of a heterogeneous blend system, PC and PD stand for the permeability of the continuous phase and the dispersed phase, respectively, and φD represents the volume fraction of the dispersed phase. Three types of predictions were made as illustrated in Fig. 10. The upper dotted-lines in all graphs represent the prediction from the Maxwell model assuming PIM-1 as the continuous phase and Ultem as the dispersed phase. The lower dotted-lines symbolize the Maxwell prediction by considering Ultem as the continuous phase and PIM-1 as the dispersed phase. The middle dotted-lines denote the prediction from a homogenous system where permeability follows the semi-logarithmic addition [55] as expressed in Eq. (10): ln P ¼ ϕ1 ln P 1 þ ϕ2 ln P 2

where ϕ refers to the volume fraction and subscripts 1 and 2 represent the two components. The following facts can be observed by comparing the experimental and model prediction data. When the PIM-1 loadings are high (i.e., 490%), a good match in permeability is found between the experimental data and the Maxwell prediction (upper dotted-line). Clearly, well dispersed Ultem molecules in the PIM-1 matrix are formed. At low PIM-1 loadings (i.e.,o20%), the experimental data are higher than the Maxwell prediction (i.e., lower dotted-line where Ultem is the continuous phase) but very close to the semi-logarithmic addition. This implies that relatively high miscibility takes place at these compositions, thus their permeability follows the rule of the semi-logarithmic addition. When the PIM-1 loading is in the range of 30–80 wt%, its permeability approaches to the semi-logarithmic prediction but with some deviations. The deviations may result from two factors: (1) as confirmed in Fig. 5, the FFV values of the blends differ from

Table 3 The gas transport properties of Ultem/PIM-1 system. Permeability (Barrer)

8.8 0.054 10.3 0.083 14.8 0.16 22.6 0.26 29.9 0.38 75.9 1.9 231.4 21.9 435.4 65.2 794.1 143.7 888.7 155.0 947.4 168.1

O2

CH4

0.38 0.040 0.58 0.060 1.1 0.12 1.6 0.19 2.2 0.28 10.4 2.2 93.9 28.5 230.1 97.1 512.4 247.6 556.5 277.0 585.5 317.7

CO2

CO2/ N2

CO2/ CH4

O2/ N2

1.48 2.18 3.95 6.58 9.27 51.7 477.1 1259.7 2876.6 3275.9 3488.7

27.4 26.2 25.2 25.7 24.8 26.9 21.8 19.3 20.0 21.1 20.8

37.0 36.5 33.8 34.6 34.7 23.2 16.7 13.0 11.6 11.8 11.0

7.1 7.0 6.8 6.1 5.8 5.4 4.4 3.5 3.6 3.6 3.5

1000

CO2/N2 Selectivity

100:0 95:5 90:10 80:20 70:30 50:50 30:70 20:80 10:90 5:95 0:100

N2

Selectivity

1000

CO2/CH4 Selectivity

He

100

10

1

100

10

1 1

10

100

1000

10000

1

10

100

1000

10000

CO2Permeability (Barrer)

CO2Permeability (Barrer) 100

O2/N2 Selectivity

Ultem: PIM

ð10Þ

Pristine Ultem 10

Pristine PIM-1

Ultem/PIM-1 (95: 5)

Ultem/PIM-1 (5: 95)

Ultem/PIM-1 (90: 10)

Ultem/PIM-1 (10: 90)

Ultem/PIM-1 (80: 20)

Ultem/PIM-1 (20: 80)

Ultem/PIM-1 (70: 30)

Ultem/PIM-1 (30: 70)

Ultem/PIM-1 (50: 50) 1

1 0.1

1.0

10.0

100.0

1000.0

O2Permeability (Barrer) Fig. 9. Comparison between the separation performance of Ultem/PIM-1polymer blends and the 2008 Robeson upper bound.

L. Hao et al. / Journal of Membrane Science 453 (2014) 614–623

621

their theoretical predictions due to the formation of microvoids and interfaces between these two materials and (2) both the semilogarithmic addition and the Maxwell model are too simple to describe the real complicated phase morphology. As pointed by previous researchers, predicting the performance of polymer blends is challenging owing to the unidentified degree of miscibility and possible changes of phase continuity at any stage [56,57]. The polymer rich phase does not necessarily form the continuous phase. Both Hao et al. [58] and Chen et al. [59] have reported the same phenomenon that under certain circumstances the polymer lean phases can take over the continuous phase.

PIM-1/Matrimid blends [25]. At lower PIM-1 loadings, Ultem is the continuous phase, thus its selectivity has a more dominant effect on the overall selectivity. In addition, there might be charge transfer complexes (CTCs) in Ultem that tighten the polymer chains and result in higher CO2/CH4 and O2/N2 selectivity than the semi-logarithmic predictions, as pointed out by Yong et al. [25]. When the PIM-1 loading is high, PIM-1 becomes the continuous phase and the CTCs effect becomes weak, thus leading to a lower selectivity than the prediction.

3.2.3. Prediction of selectivity Fig. 11 compares the CO2/CH4 and O2/N2 selectivity versus the predictions from the semi-logarithmic addition [55] as follows:       P PA PA ln A ¼ ϕ1 ln þ ϕ2 ln ð11Þ PB PB 1 PB 2

3.2.4. Comparison with similar systems So far, only limited polymer blends containing PIM-1 have been studied for gas separation. As listed in Table 5, they are Matrimid, Ultem and 6FDA-m-PDA [21,25]. In general, the permeability of all blends increases with the addition of small amounts of PIM-1 without much sacrifice in selectivity. When comparing our Ultem/PIM-1 system with Liu et al.'s work, their performance trends in terms of PIM-1 effects on permeability and selectivity are quite similar. However, both works have different values of permeability. As pointed out by Li et al. [16], both the solvents applied (chloroform, dichloromethane or tetrahydroforan) and the film preparation protocols have significant effects on the gas transport properties of PIM-1. As shown in Table 5, the slight deviations in their gas separation performance may be resulted from the differences in polymer molecular weights, membrane preparation, thermal history and tests conditions. Yong et al. [25] have reported that CO2 permeability increases 25% and 77% with the addition of 5 and 10 wt% PIM-1 into Matrimid, respectively. Since Ultem is a less permeable polymer, the effect of incorporating PIM-1 on the blend permeability should be much remarkable. Consistent with our hypothesis, CO2 permeability increases 47% and 167% with 5 and 10 wt% PIM-1, respectively. Clearly, PIM-1 has great potential to work as an organic filler to enhance the permeability of low permeable polymers and give them new opportunities for gas separation applications.

The experimental selectivity is initially higher than the prediction but becomes slightly lower than the prediction when the PIM-1 loading is over 80 wt%. A similar trend had been observed from

Table 4 The mixed gas CO2/N2 (50/50 mol%) and CO2/CH4 (50/50 mol%) testing results of 90:10 and 80:20 Ultem/PIM-1 blends. Permeability

Selectivity

CO2

N2

CH4

CO2/N2

CO2/CH4

Ultem/PIM-1 (90:10) Pure gas tests Mixed gas (CO2/N2) Mixed gas (CO2/CH4)

3.95 3.00 3.37

0.16 0.11 –

0.12 – 0.09

25.2 27.3 –

33.8 – 37.0

Ultem/PIM-1 (80:20) Pure gas tests Mixed gas (CO2/N2) Mixed gas (CO2/CH4)

6.58 6.45 6.23

0.26 0.21 –

0.19 – 0.17

25.7 30.7 –

34.6 – 36.6

1000.00

CH4 Permeability (Barrer)

CO2 Permeability (Barrer)

10000

1000

100

10

1000

100.00 10.00 1.00 0.10 0.01

1 0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

He Permeability (Barrer)

1000.0

O2 Permeability (Barrer)

0.4

0.6

0.8

1.0

PIM-1 Volume Fraction

PIM-1 Volume Fraction

100.0

10.0

1.0

0.1

10000

1000

100

10

1 0.0

0.2

0.4

0.6

0.8

PIM-1 Volume Fraction

1.0

0.0

0.2

0.4

0.6

0.8

1.0

PIM-1 Volume Fraction

Fig. 10. Comparison between the model prediction and the experimental data of Ultem/PIM-1 polymer blends.

622

L. Hao et al. / Journal of Membrane Science 453 (2014) 614–623

10

O2/N2 Selectivity

CO2/CH4 Selectivity

100

10

-----Prediction Ideal selectivity Mixed - gas data

-----Prediction

Δ ideal selectivity

1

1

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

PIM-1 Volume Fraction

PIM-1 Volume Fraction

Fig. 11. Comparison between the model prediction and the experimental data of Ultem/PIM-1 polymer blends.

Table 5 Comparison of gas transport properties of different PIM-1 blend systems. Permeability

Selectivity

CO2

O2

O2/N2

CO2/N2

CO2/CH4

Ultem/PIM-1 (this work) Pristine Ultem 90:10 80:20 70:30

1.48 3.95 6.58 9.27

0.38 1.1 1.6 2.2

7.1 6.8 6.1 5.8

27.4 25.2 25.7 24.8

37.0 33.8 34.6 34.7

Ultem/PIM-1[21]a 90:10 80:20 70:30

2.89 5.69 5.77

– – –

– – –

– – –

31.6 31.2 30.2

Matrimid/PIM-1[25] (Liu et al. [21]a) Pristine Matrimid 9.6 (10.0) 95: 5 12 (–) 90:10 17 (20.3) 70:30 56 (35.9)

2.1 2.6 3.4 11

6.4 6.6 6.1 5.8

30 29 30 28

36 35 34 31

6FDA-m-PDA/PIM-1 [21]a Pristine 6FDA-m-PDA 14.8 92.5:7.5 22.3

– –

– –

– –

48.4 48.7

a

respectively. The mixed gas results exhibit comparable performance with their respective pure gas tests. (6) The permeability of the blends follow nicely with the semilogarithm addition when the PIM-1 loadings are low (o20 wt %), indicting relatively homogenous blends at these compositions, while they match closely with the Maxwell prediction at high PIM-1 loadings (490 wt%) due to the good dispersion of Ultem inside PIM-1. (7) This study also opens up the potential of using PIM-1 as an organic filler to enhance the permeability of low permeable materials for industrial gas separation applications.

Acknowledgments (28.2) (–) (27.1) (24.8)

Tests were conducted at 50 1C and  100 psig.

The authors would like to thank the Singapore National Foundation (NRF) for supporting the project “New Biotechnology for Processing Metropolitan Organic Wastes into Value-Added Products” (grant number R-279-000-311-281). Special thanks go to Dr. Wang Huan, Ms. Yong Wai Fen and Mr. Wan Chunfeng for their kind help and valuable suggestions.

4. Conclusion

References

A series of Ultem/PIM-1 blend films were fabricated and investigated for their physiochemical and gas transport properties. The following conclusions can be made:

[1] M.-B. Hägg, A. Lindbråthen, CO2 capture from natural gas fired power plants by using membrane technology, Ind. Eng. Chem. Res. 44 (2005) 7668–7675. [2] D.M. D’Alessandro, B. Smit, J.R. Long, Carbon dioxide capture: prospects for new materials, Angew. Chem. Int. Ed. 49 (2010) 6058–6082. [3] Y. Osada, T. Nakagawa (Eds.), Membrane Science and Technology, Marcel Dekker, Inc., New York, 1992. [4] 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. [5] P.S. Tin, T.S. Chung, Y. Liu, R. Wang, S.L. Liu, K.P. Pramoda, Effects of crosslinking modification on gas separation performance of Matrimid membranes, J. Membr. Sci. 225 (2003) 77–90. [6] C. Zhao, J. Xue, F. Ran, S. Sun, Modification of polyethersulfone membranes—a review of methods, Prog. Mater. Sci. 58 (2013) 76–150. [7] N.C. Liu, W.E. Baker, Reactive polymers for blend compatibilization, Adv. Polym. Technol. 11 (1992) 249–262. [8] M. Xanthos, Interfacial agents for multiphase polymer systems: recent advances, Polym. Eng. Sci. 28 (1988) 1392–1400. [9] C. Koning, M.V. Duin, C. Pagnoulle, R. Jerome, Strategies for compatibilization of polymer blends, Prog. Polym. Sci. 23 (1998) 707–757. [10] P.M. Budd, B.S. Ghanem, S. Makhseed, N.B. McKeown, K.J. Msayib, C.E. Tattershall, Polymers of intrinsic microporosity (PIMs): robust, solution-processable, organic nanoporous materials, Chem. Commun. (2004) 230–231. [11] N.B. McKeown, P.M. Budd, K. Msayib, B. Ghanem, Microporous Polymer Material, International Patent, WO05012397, 2005. [12] N.B. McKeown, P.M. Budd, Polymers of intrinsic microporosity (PIMs): organic materials for membrane separations, heterogeneous catalysis and hydrogen storage, Chem. Soc. Rev. 35 (2006) 675–683. [13] P.M. Budd, N.B. McKeown, B.S. Ghanem, K.J. Msayib, D. Fritsch, L. Starannikova, N. Belov, O. Sanfirova, Y. Yampolskii, V. Shantarovich, Gas permeation

(1) Ultem and PIM-1 form good dispersions when the PIM-1 loading is low ( o20 wt%) or high (4 90 wt%) as observed from the morphologies measured by both PLM and AFM. (2) A slight shift of Ultem Tg from 215 1C to 221 1C was detected by DSC for the Ultem/PIM-1 (50:50) blend, suggesting the system might be partially miscible. (3) The shifts of FTIR and XRD peaks suggest that interactions between Ultem and PIM-1 are at a molecular level. (4) According to the density measurements, the free volume of Ultem/PIM-1 blends increases as the PIM-1 loading increases. Positive deviations in free volume from the linear additional rule at low PIM-1 loadings may be resulted from interface voids, while negative deviations were observed at higher PIM1 loadings possibly due to the filling of Ultem molecules into PIM-1 pores. (5) Permeability of the blends increases remarkably with the addition of a small amount of PIM-1 filler without much scarifying the selectivity, e.g. the permeability of CO2 increases 47% and 167% when the PIM-1 loadings are 5 wt% and 10 wt%,

L. Hao et al. / Journal of Membrane Science 453 (2014) 614–623

[14]

[15]

[16] [17]

[18]

[19] [20]

[21] [22] [23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31]

[32]

[33] [34] [35]

parameters and other physiochemical properties of a polymer of intrinsic microporosity: polybenzoidioxane PIM-1, J. Membr. Sci. 325 (2008) 851–860. S. Thomas, I. Pinnau, N. Du, M.D. Guiver, Pure- and mixed-gas permeation properties of a microporous spirobisindane-based ladder polymer (PIM-1), J. Membr. Sci. 333 (2009) 125–131. P.M. Budd, K.J. Msayib, C.E. Tattershall, B.S. Ghanem, K.J. Reynolds, N.B. McKeown, D. Fritsch, Gas separation membranes from polymers of intrinsic microporosity, J. Membr. Sci. 251 (2005) 263–269. P. Li, T.S. Chung, D.R. Paul, Gas sorption and permeation in PIM-1, J. Membr. Sci. 432 (2013) 50–57. S.V. Adymkanov, Y.P. Yampol’skii, A.M. Polyakov, P.M. Budd, K.J. Reynolds, N.B. Mckeown, K.J. Msayib, Pervaporation of alcohols through highly permeable PIM-1 polymer films, Polym. Sci. Ser. A 50 (2008) 444–450. O. Vopicka, K. Friess, V. Hynek, P. Sysel, M. Zgazar, M. Sipek, K. Pilnacek, M. Lanc, J.C. Jansen, C.R. Mason, P.M. Budd, Equilibrium and transient sorption of vapours and gases in the polymer of intrinsic microporosity PIM-1, J. Membr. Sci. 434 (2013) 148–160. P. Li, T.S. Chung, D.R. Paul, Temperature dependence of gas sorption and permeation in PIM-1, J. Membr. Sci. 450 (2014) 380–388. O. Hölck, M. Böhning, M. Heuchel, M.R. Siegert, D. Hofmann, Gas sorption isotherms in swelling glassy polymers-detailed atomictic simulations, J. Membr. Sci. 428 (2013) 523–532. C. Liu, S.T. Wilson, Mixed Matrix Membranes Incorporating Microporous Polymers as Fillers, US Patent 7,410,525, 2008. R. Mahajan, W.J. Koros, Factors controlling successful formation of mixedmatrix gas separation materials, Ind. Eng. Chem. Res. 39 (2000) 2692–2696. Y. Li, T.S. Chung, C. Cao, S. Kulprathipanja, The effects of polymer chain rigidification, zeolite pore size and pore blockage on polyethersulfone (PES)– zeolite A mixed matrix membranes, J. Membr. Sci. 260 (2005) 45–55. 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. 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, F.Y. Li, Y.C. Xiao, T.S. Chung, Y.W. Tong, High performance PIM-1/ matrimid hollow fiber membranes for CO2/CH4, O2/N2 and CO2/N2 separation, J. Membr. Sci. 443 (2013) 156–169. A. Bos, I.G.M. Pünt, H. Strathmann, M. Wessling, Suppression of gas separation membrane plasticization by homogeneous polymer blending, AIChE J. 47 (2001) 1088–1093. S. Carroccio, C. Puglisi, G. Montaudo, Thermal degradation mechanisms of polyetherimide investigated by direct pyrolysis mass spectrometry, Macromol. Chem. Phys. 200 (1999) 2345–2355. Husain, Mixed matrix dual layer hollow fiber membranes for natural gas separation, Georgia Institute of Technology, Atlanta, GA, 2006. (Ph.D. dissertation). J. Xia, S. Liu, P.K. Pallathadka, M.L. Chng, T.S. Chung, Structural determination of Extem XH 1015 and its gas permeability comparison with polysulfone and Ultem via molecular simulation, Ind. Eng. Chem. Res. 49 (2010) 12014–12021. Y. Dai, J.R. Johnson, O. Karvan, D.S. Sholl, W.J. Koros, Ultems/ZIF-8 mixed matrix hollow fiber membranes for CO2/N2 separations, J. Membr. Sci. 401– 402 (2012) 76–82. A. Etxeberria, S. Guezala, J.J. Iruin, J.G. De La Campa, J. De Abajo, Blends of poly (ether imide) and an aromatic poly(ether amide): phase behavior and CO2 transport properties, J. Appl. Polym. Sci. 68 (1998) 2141–2149. V. Choudhary, E.M. Pearce, Studies on the thermal and curing behavior of polyimide blends, J. Appl. Polym. Sci. 51 (1994) 1823–1834. J.A. Rock, Polyetherimide–Liquid Crystal Polymer Blends, US patent 4,871,817, 1989. P.M. Budd, E.S. Elabas, B.S. Ghanem, S. Makhseed, N.B. McKeown, K.J. Msayib, C.E. Tattershall, D. Wong, Solution-processed, organophilic membrane derived from a polymer of intrinsic microporosity, Adv. Mater. 16 (2004) 456–459.

623

[36] A. Bondi, van der Waals volumes and radii, J. Phys. Chem. 68 (1964) 441–451. [37] J.Y. Park, D.R. Paul, Correlation and prediction of gas permeability in glassy polymer membrane materials via a modified free volume based group contribution method, J. Membr. Sci. 125 (1997) 23–39. [38] W.H. Lin, R.H. Vora, T.S Chung, Gas transport properties of 6FDA-Durene/1,4phenylenediamine (pPDA) copolyimides, J. Polym. Sci. Polym. Phys. 38 (2000) 2703–2713. [39] Y.C. Xiao, L. Shao, T.S. Chung, D.A. Schiraldi, Effects of thermal treatments and dendrimers chemical structures on the properties of highly surface crosslinked polyimide films, Ind. Eng. Chem. Res. 44 (2005) 3059–3067. [40] H.Z. Chen, Y.C. Xiao, T.S. Chung, Synthesis and characterization of poly (ethylene oxide) containing copolyimides for hydrogen purification, Polymer 51 (2010) 4077–4086. [41] N.B. McKeown, P.M. Budd, Exploitation of intrinsic microporosity in polymerbased materials, Macromolecules 43 (2010) 5163–5176. [42] A.G. McDermott, G.S. Larsen, P.M. Budd, C.M. Colina, J. Runt, Structural characterization of a polymer of intrinsic microporosity: X-ray scattering with interpretation enhanced by molecular dynamics simulations, Macromolecules 44 (2011) 14–16. [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] N. Du, H.B. Park, M.M. Dal-Cin, M.D. Guiver, Advances in high permeability polymeric membrane materials for CO2 separations, Energy Environ. Sci. 5 (2012) 7306–7322. [45] M. Jaffe, P. Chen, E.W. Choe, T.S. Chung, S. Makhija, High performance polymer blends, in: P. Hergenroth (Ed.), Advances in Polymer Science, vol. 117, Springer, New York, NY, 1994, p. 297. [46] T.K. Kwei, The effect of hydrogen bonding on the glass transition temperatures of polymer mixtures, J. Polym. Sci. Polym. Lett. 22 (1984) 307–313. [47] D.T. Clausi, W.J. Koros, Formation of defect-free polyimide hollow fiber membranes for gas separations, J. Membr. Sci. 167 (2000) 79–89. [48] 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. [49] R.H. Vora, S.H. Goh, T.S. Chung, Synthesis and properties of fluoropolyetherimides, Polym. Eng. Sci. 40 (2000) 1318–1329. [50] J. Weber, Q. Su, M. Antonietti, A. Thomas, Exploring polymers of intrinsic microporosity—microporous, soluble polyamide and polyimide, Macromol. Rapid. Commun. 28 (2007) 1871–1876. [51] L.M. Robeson, The upper bound revisited, J. Membr. Sci. 320 (2008) 390–400. [52] N. Du, G.P. Robertson, M.M. Dal-Cin, L. Scoles, M.D. Guiver, Polymers of intrinsic microporosity (PIMs) substituted with methyl tetrazole, Polymer 53 (2012) 4367–4372. [53] S. Thomas, I. Pinnau, N. Du, M.D. Guiver, Hydrocarbon/hydrogen mixed-gas permeation properties of PIM-1, an amorphous microporous spirobisindane polymer, J. Membr. Sci. 338 (2009) 1–4. [54] J.C. Maxwell, A Treatise on Electricity and Magnetism, Dover Publications, Inc., New York, 1954. [55] A.E. Barnabeo, W.S. Creasy, L.M. Robeson, Gas permeability characteristics of nitrile-containing block and random copolymers, J. Polym. Sci. Polym. Chem. 13 (1975) 1979–1986. [56] S.C. George, S. Thomas, Transport phenomena through polymeric system, Prog. Polym. Sci. 26 (2001) 985–1017. [57] L.M. Robeson, Polymer blends in membrane transport processes, Ind. Eng. Chem. Res. 49 (2010) 11859–11865. [58] L. Hao, P. Li, T. Yang, T.S. Chung, Room temperature ionic liquid/ZIF-8 mixedmatrix membranes for natural gas sweetening and post-combustion CO2 capture, J. Membr. Sci. 436 (2013) 221–231. [59] H.Z. Chen, P. Li, T.S. Chung., PVDF/ionic liquid polymer blends with superior separation performance for removing CO2 from hydrogen and flue gas, Int. J. Hydrog. Energy 37 (2012) 11796–11804.