Pervaporation of organic liquid mixtures through poly (ether imide) segmented copolymer membranes

Pervaporation of organic liquid mixtures through poly (ether imide) segmented copolymer membranes

journal of MEMBRANE SCIENCE ELSEVIER Journal of Membrane Science 104 (1995) 181-192 Pervaporation of organic liquid mixtures through poly(ether imid...

885KB Sizes 10 Downloads 58 Views

journal of MEMBRANE SCIENCE ELSEVIER

Journal of Membrane Science 104 (1995) 181-192

Pervaporation of organic liquid mixtures through poly(ether imide) segmented copolymer membranes Nozomu Tanihara, Nobuhiro Umeo, Takashi Kawabata, Kazuhiro Tanaka, Hidetoshi Kita, Ken-ichi Okamoto * Department of Advanced Materials Science and Engineering, Faculty of Engineering, Yamaguchi University, Ube, Yamaguchi 755, Japan Received 3 August 1994; accepted in revised form 7 February 1995

Abstract Poly ( ether imide) segmented copolymers were prepared from polyether diamine and comonomer diamine with acid anhydride. They had microphase-separated structure consisting of microdomains of rubbery polyether segments and of glassy polyimide segments. Pervaporation (PV) of benzene-cyclohexane, benzene-n-hexane and acetone-cyclohexane mixtures through membranes of the copolymers were investigated. As to polyether segment, poly(ethylene oxide) (PEO) gave much better membrane performance than poly(propylene oxide) and poly(tetrahydrofuran). With an increase in PEO content, [PE], the specific permeation flux, Ql, increased significantly and the separation factor, u, decreased. Sorption and permeation occurred in microdomains of polyether segments, whereas microdomains of polyimide segments contributed to suppression of swelling of polyether microdomains and to film-forming ability. The membranes were preferentially permeable to benzene or acetone over cyclohexane or n-hexane due to preferential sorption and diffusion of benzene or acetone. The block length of the PEO segment, n, affected the membrane performance; lower Ql and higher a were obtained for shorter block length. A typical membrane with [PE] = 41 wt% and n = 9 displayed Ql = 2 kg/zm/(m2h) and a = 9 at a feed composition of 60 wt% benzene in cyclohexane and Ql = 8 kg/zm/(m2h) and a = 17 at a feed composition of 68 wt% acetone in cyclohexane at 323 K. Argon plasma treatment was effective for enhancement of the selectivity without significant reduction in Ql, especially for the acetone-cyclohexane system. Keywords: Pervaporation; Organic separations; Membrane preparation and structure; Poly(imide); Poly(ethylene oxide)

I. Introduction Pervaporation (PV) has been of much interest for the separation of azeotropic, close-boiling or aqueous organic mixtures with reduced energy consumption [ 1-18]. PV of organic liquid mixtures such as aromatics and aliphatics [3-10,12,14-18] and aromatic Cs-isomers [ 11 ] has attracted increasing attention. The permselectivity in the PV of organic liquid mixtures is * Corresponding author. 0376-7388/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved

SSDI0376-7388(95)00029-1

mainly due to the solubility selectivity, because the diffusivity selectivity is considered to be low or negative, judging from the molecular size of the penetrants [ 19]. For the high solubility selectivity, membranes should have high affinity for one component and little affinity for the other component(s). However, excessive affinity for one component causes the significant swelling of the membranes and results in loss of the selectivity and of mechanical strength. In the design of organic liquid-permselective membranes, it is important to suppress the swelling in addition to enhancing

182

N. Tanihara et aL / Journal of Membrane Science 104 (1995) 181-192

the solubility. High performance for PV of benzenecyclohexane mixtures has been reported on the membranes made of polymeric alloys [3-6] or copolymers[10], and polymers with microphase-separated [ 5,12,14 ] or crosslinking structure [ 3,6 ], and prepared by plasma graft-filling polymerization [8,9]. In this study, membranes of the poly(ether imide) segmented copolymers with microphase-separated structure were prepared, and their membrane properties for PV of benzene-cyclohexane, benzene-n-hexane and acetone-cyclohexane mixtures are investigated.

0 R: - - N ~ N - -

0

o

o

×:

BP

0

0

--N'~N-o

PM

PE01 --C3Hs-"(OCH2CH2-)nnOC3H6"-PE02 PE03 PE04

-C3H~OCH2~H~nOC3Hg--

o

n=9 n=23 n=52 n:201

CH3 --C 3H-~OCH2C.2gH2CH2.-7~3C3Hg--

PPO n=33 2. Experimental 2.1. Materials and membrane preparation Bis (3-aminopropyl)-poly (ethylene oxide) with an average block length, n, of poly(ethyleneoxide) (PEO) of 9 (PEO1), 23 (PEO2), 52 (PEO3) and 201 (PEO4), bis(3-amino-propyl) -poly(propylene oxide) with an n for poly(propylene oxide) (PPO) of 33, and bis(3-aminopropyl)-poly(tetrahydrofuran) with an n for poly(tetrahydrofuran) (PTHF) of 30 were used as diamines having polyether moieties. 4,4'Oxydianiline (ODA), 3,5-diaminobenzoic acid (DABA), 1,3-phenylenediamine (PD), and 3,3'-diaminodiphenyl sulfone (DDS) were used as comonomer diamines. 3,3',4,4'-Biphenyltetracarboxylic dianhydride (BP), pyromellitic dianhydride (PM), and 3,3',4,4'-benzophenonetetracarboxylic dianhydride (BT) were used as acid anhydrides. Polyether-containing copoly (amic acids) were prepared at a concentration of 10 wt% solids by slow addition of a stoichiometric amount of the acid anhydride to a mechanically stirred solution of the polyether diamine and the comonomer diamine(s) in N,N-dimethylacetamide under nitrogen at room temperature [20]. The solutions were stirred for ca. 12 h. Membranes of the copolyamic acids of 30-170/zm in thickness were prepared by casting the solutions into Teflon Petri dishes and drying at 353 K for ca. 10 h. The membranes were thermally imidized at 443 K for 20 h in vacuo. Structures of the poly(ether imide) segmented copolymers are shown in Fig. 1. The copolymers prepared in this study are listed in Table 1. The figures in parenthesis on abbreviation of the copolymers refer to

PTHF n=30

Y:

COOH DABA

Pi)

Fig. 1. Structures of poly(ether imide) segmented copolymers.

feed wt% of the polyether diamine to total diamines. For the copolyimides containing ODA and DABA moieties, the feed weight ratio of ODA to DABA was 4. Some membranes of the poly(ether imide) segmented copolymers were treated with Ar plasma at a power of 50 W under a pressure of 13.3 Pa for 30 min. The treated side was contacted with the feed solution in PV experiments. Differential scanning calorimetry (DSC) was measured with Seiko Instruments Inc. DSC-5200 at a heating and cooling rate of 10 K/min. Glass transition temperature, Tg, and melting temperature, Tin, were determined from the initial point of the signal on the first heating run. Transmission electron micrographs were taken with a Jeol JEM-2010 electron microscope at an accelerating voltage of 200 kV at UBE Scientific Analysis Laboratory. After annealing at 343 K for 0.5 h, the sample films were stained with R u O 4 vapor at 295 K for 24 h.

2.2. PV and sorption experiments PV experiments were carried out by a literature method [ 10] at temperatures ranging from 298 to 343 K. The effective membrane area was 22.1 cm 2. Down-

N. Tanihara et al. / Journal of Membrane Science 104 (1995) 181-192

183

Table 1 Physical properties of poly (ether imide) segmented copolymers a CopolymeIa'

~ (dl/g)

Density (g/cm 3)

BP-ODA/DABA/PEOl(55) BP-ODA/DABA/PEO1 (60) BP-ODA/DABA/PEO1 (70) BP-ODA/DABA/PEO1 (75) PM-ODA/DABA/PEO1(75) PM-ODA/DABA/PEO1 (80) PM-DDS/PEO1 (85) BT-ODA/DABA/PEO1(75)

1.14 0.90 0.81 0.59 0.26 0.50 0.33 0.35 1.15 1.42 2.18 1.22 0.68 1.09 1.02 0.22 0.63 0.94 1.18 0.59 0.67 0.39

1.344 1.335 1.325 1.317 1.319 1.297 1.300 1.323 1.339 1.324 1.318 1.284 1.272 1.294 1.331 1.284 1.274 1.274 1.261 1.189 1.147 1.180

BP-ODA/DABA/PE02(50) BP-ODA/DABA/PEO2(60) BP-ODA/DABA/PEO2(70) BP-ODA/DABA/PEO2(75) BP-ODA/DABA/PEO2(80) BP-ODA/PEO2(75) BP-PD/PEO2(75) BP-ODA/DABA/PEO3(70) BP-ODA/DABA/PEO3 (75) BP-ODA/DABA/PEO4(75) PM-ODA/PEO4(80) BP-ODA/DABA/PPO(75) BP-ODA/DABA/PTHF(80) BT-ODA/DABA/PTHF(75) BP-ODA PEOI diamine (n = 9) PEO2 diamine (n = 23 ) PEO3 diamine (n = 52 ) PEO4 diamine (n = 201 ) PPO diamine (n = 33) PTHF diamine (n = 30)

Tg~ (K)

Tg2 (K)

Tm (K)

[PE] (wt%)

(225)

542

293

(225) 237 236 238

523 512 ND ND

300 303 306 301

24.3 30.6 37.5 41.2 46.9 51.1 56.9 39.4 27.0 34.6 43.3 48.1 53.4 48.8 41.2 46.1 51.9 54.7 66.6 51.2 57.6 50.0

236 231

227 231 225 229 223 222 214 216 211 211 216 190 186 194 204 207 211 194 190

523 517 511 521 514 520 519 527 521 528 ND 518 550 510 543

284 298 303 301 305 303 309 303 304 298 305 297 313 272 268

SB

Sc

SA

37

4

20

15 27 42 45 65 40

2 2 3 3 3 5

42 52

4 12

87 91

21 15

25

31

255 301 316 331 265 311

art, Inherent viscosity at 0.5 wt% and 298 K of the copoly(amic acid) in DMAc solution. Tat and Tg2 are glass transition temperatures. Tm is melting temperature. [PE] is polyether content. SB, So and SA are sorption amounts at 298 K for pure benzene, cyclohexane, and acetone, respectively, in g solvent/( 100 g dry polymer). bThe figure in parenthesis refers to feed ratio (wt%) of the polyether diamine to total diamines. stream pressure was maintained b e l o w 130 Pa. Steady state permeation was attained in less than 1 h, except for m e m b r a n e s with low content o f P E O , after the m e m b r a n e s w e r e contacted with the feed solutions. C o m p o s i t i o n analysis was p e r f o r m e d on a gas chromatograph e q u i p p e d with 3-m c o l u m n s packed with D O P 30% U n i p o r t R 6 0 / 8 0 . For sorption experiments, m e m b r a n e samples .(total weight of 1 g and thickness o f about 1 5 0 / ~ m ) dried previously were dipped into the liquids. The samples were taken out at appropriate intervals, weighed, and then dipped again into the liquids. The experiments were continued until the w e i g h t o f the samples b e c a m e constant. For sorption in a binary solution, the sorbed

liquid was r e c o v e r e d in a liquid nitrogen trap by desorbing the equilibrated sample in a v a c u u m line, and analyzed by gas chromatography.

3. Results and discussion 3.1. Characterization results o f poly(ether imide) segmented copolymers Characterization results of the p o l y ( e t h e r i m i d e ) s e g m e n t e d c o p o l y m e r s are listed in Table 1. T h e copolymers had two glass transition temperatures and one melting temperature. F r o m the c o m p a r i s o n o f Tg and

184

N. Tanihara et al. / Journal of Membrane Science 104 (1995) 181-192

ments and of hydrophobic and glassy poly(imide) segments. The micropfiase-separated structure was also confirmed by transmission electron microscope observation. In Fig. 2, the dark areas refer to microdomains of PEO segments stained with RuO4,which were rather small and in the order of 10 nm. With an increase in block length, n, of PEO, Tgl decreased for the PEObased copolymers, whereas it increased for the PEO diamines. As for PEO4, there was no difference in Tg~ between the copolymer and the diamine. Tgl values were higher for the PEOl-based copolymers than for the PEO 1-diamine and the PEO4-based copolymer by 43 and 25 K, respectively. This suggests that the segmental motion of PEO in the PEOl-based copolymers is reduced by the glassy polyimide segments. Polyether content, [PE], refers to the weight percentage of the hydrophilic diamine moieties, namely C3H6-(polyether-block)n-C3H6- chains, which was calculated from feed amount of the monomers. The volume percentage values of the hydrophilic diamine moieties, which were estimated using the densities of the corresponding polyimides, were larger than the wt% values by 3-4%; for example, [PE] =48.8 wt% or 52.6 vol% for BP-ODA/PEO2(75). 3.2. P V o f benzene-cyclohexane and benzene-nhexane

The collision diameter, de, and the solubility parameter, 8, of penetrant molecules are listed in Table 2. The 8 values of PEO, PPO and PTHF segments were calculated by the group contribution method of van Krevelen [21]. Sorption amounts of pure benzene, cyclohexane and acetone, SB, Sc and SA, in the copolymer membranes at 298 K are listed in Table 1. Fig. 2. Transmissionelectronmicrophorographof a thin filmof BPPD/PEO4(80). Tm between the copolymers and the corresponding poly(ether diamines), the lower glass transition temperature, Tgl, and Tm were attributed to polyether segments of the hydrophilic diamine moieties and the higher glass transition temperature, Tg2, was attributed to the polyimide segments of the hydrophobic diamine moieties [ 20]. This indicates that the copolymers have a microphase-separated structure consisting of microdomains of hydrophilic and rubbery poly(ether) seg-

Table 2 Collision diameterde and solubilityparameter8 of penetrant molecules and polyethersegments Penetrant or polyether

de (nm)

8 [ (J/cm3) 1/2]

Benzene Acetone Cyclohexane n-Hexane PEO PPO PTHF

0.526 0.505 0.606 0.595

18.5 19.9 16.7 14.9 19.2 17.7 17.6

N. Tanihara et al. / Journal of Membrane Science 104 (1995) 181-192 Table 3 PV properties of poly(ether imide) segmented copolymer membranes in benzene--cyclohexane and benzene--n-hexane systems at XB = 60 wt% and 323 K a Membranes

[ PE ]

BP-ODA/DABA/PEOI (55) BP-ODA/DABA/PEO1 (60) BP-ODA/DABA/PEO1 (70) BP-ODA/DABA/PEO1 (75) PM-ODA/DABA/PEO1 (75) (plasma treated) PM-ODA/DABA/PEO1 (80) PM-DDS/PEO1 (85)

24.3 30.6 37.5 41.2 46.9

BT-ODA/DABA/PEOI (75) (plasma treated) BP-ODA/DABA/PEO2(50) BP-ODA/DABA/PEO2(60)

39.4

BP-ODA/DABA/PE02(70) BP-ODA/DABA/PEO2(75) (plasma treated) BP-ODA/DABA/PEO2(80) BP-ODA/PEO2(75) BP-PD/PEO2(75) BP-ODA/DABA/PEO3(70) BP-ODA/DABA/PEO3 (75) (plasma treated) BP-ODA/DABA/PEO4(75) PM-ODA/PEO4(80) BP-ODA/DABA/PPO(75) BP-ODA/DABA/PTHF(80) BT-ODA/DABA/PTHF(75)

51.1 56.9

27.0 34.6 43.3 48.1

53.4 48.8 41.2 46.1 51.9 54.7 66.6 51.2 57.6 50.0

1 38 45 40 90 80 80 80 70 90 84 64 53 72 94 60 68 160 40 34 120 90 100 80 130 80 22

QI

a

<0.003 0.11 0.90 2.1 3.5 2.9 9.5 8.9 9.0* 1.7 1.2 0.090 1.1 1.1. 5.5 13 9.3 9.4* 31 19 2.5 18 27 17 50 83 230 290 96

49 12 9.1 7.8 10 7.0 6.7 9.1. 9.9 I1 15 9.1 10.3. 6.9 6.5 7.8 8.3* 5.5 5.5 7.7 5.4 5.2 8.2 4.5 4.6 2.1 1.9 2.2

185

The effects of [PE] on SB and Sc for a series of PEO2-based copolymer membranes are also shown in Fig. 3. For BP-ODA/DABA copolyimide membranes, Ql

1O0 r-~

i

i

10



1

0.~"

i

i

0"" 0

Z~/

~-

70 >"

/

//'O

~

60 "

~A///

50 -~ I

/ , , , , ~ ~

30 "~.~-~

~

20

O. 01

20

.z~

SC 30

40

>

50

10 o~ 60

70

[PE] [wt~] Fig. 3. Effects of [PE] on QI, SB and Sc. QI is for PEOl-based (closed circles) and PEO2-, PEO3- and PEO4-based (open circles) copolymer membranes at Xa = 60 wt% and 323 K. Sa and Sc are sorption amounts of benzene and cyclohexane in BP-ODA/DABA/ PEO2 membranes at 298 K. 50 ~

15

t

1

I

I

1"

a[pE] is polyether content in wt%, l is membrane thickness, in/~m, Ql is specific permeation flux, in kg p,m/(m 2 h), and ot is separation factor. The values of Ql and ot marked with " are for the benzene-n-hexane system and the other values are for the benzene-cyclohexane system.

In the benzene--cyclohexane system, the specific permeation flux, Ql, and the separation factor of benzene over cyclohexane, CtB/c, at a feed composition, XB (wt% of benzene), of 60 wt% and 323 K are listed in Table 3. The total permeation flux, Q, was inversely proportional to membrane thickness, l, ranging from 30 to 160/zm, and ctB/c was hardly dependent on it. In this study, Ql was used for comparison of the permeability of the membranes with different" thickness.' The effects of [PE] on Ql and aa/c for series of PEOl-based and the other PEO-based copolymer membranes are shown in Figs. 3 and 4, respectively.

10

o

O, 0""0

0

I

20

30

I

40 [PE]

I

50 [wt%]

I

60

70

Fig. 4. Effects of [ PE] on aa/c for PEOl-based (closed circles) and PEO2-, PEO3- and PEO4-based (open circles) copolymer membranes at Xa = 60 wt% and 323 K.

186

N. Tanihara et al. / Journal of Membrane Science 104 (1995) 181-192

Table 4 Sorption and PV properties of BP-ODA/DABA/PEO2(75) and BP-ODA/DABA copolyimide membranes for a benzene-cyclohexane mixture of Xa = 6 0 wt% at 323 K a Membranes

Ca

Cc

Ca/Cc (-)

qal

qcl

qal/qcl (-)

Da

De

DaID c (-)

BP-ODA/DABA BP-ODA/DABA/PEO2(75) PEO domain poly(imide) domain

0.2 2.0 1.90 0.10

0.001 0.40 0.40 0.0005

180 5.1 4.8 170

<0.02 4.9

0.46

11

24 26

I1 12

2.1 2.2

aC, ql, and D are in 103 m o l / ( m 3 swollen polymer), 10 -8 mol m / ( m 2 s), and 10 -t2 m2/s, respectively. The data for PEO and polyimide domains are the estimated values (see text).

was below the measurement limit (0.003 k g / z m / ( m 2 h) ) at XB = 60 wt% and 323 K, and SB and Sc were 2 and 0 g/( 100 g dry polymer), respectively, at 298 K. With an increase in [PE], SB increased drastically, whereas Sc remained at a much lower level. This indicates a significant increase in swelling of PEO segment microdomains by sorbed benzene with [PE], judging from the fact that microdomains of glassy polyimide segments hardly sorb benzene and cyclohexane (see Table 4). With an increase in [PE], Ql increased and aB/c decreased drastically in the range of [PE] less than 40 wt%. For example, with an increase in [PE] from 25 to 40 wt% for a series of PEO 1- based copolymers, Ql increased by a factor of 103 o r more and aB/ c decreased by a factor of larger than 5. The variation in Ql and aB/c with [PE] became smaller in the range of [PE] more than 40 wt%, but it was still large. With an increase in [PE] from 40 to 55 wt%, for series of PEOl-based and the other PEO-based copolymers, Ql increased by factors of 7 and 10, respectively, and aB/ c decreased by about 35%. In the range of [PE] more than 55 wt%, aB/c tended to level off. The results mentioned above indicate that sorption and permeation of the penetrants occur in PEO segment microdomains and that the fraction of PEO segment microdomains forming the continuous phase increases significantly with an increase in [PE] from 25 to 50 wt%, resulting in a significant increase in Ql. Polyimide segment microdomains contribute both to suppression of the swelling of PEO segment microdomains and to filmforming ability. With an increase in the fraction of continuous phase of PEO segments, or with a decrease in that of polyimide segments, the swelling of PEO segment domains by sorbed benzene increases, resulting in a significant reduction in aB/c.

The Ql and aB/c depended on the block length of PEO, but little on the kind of acid anhydride and hydrophobic diamine, when the comparison was carded out among the copolymers with a similar value of [PE], as can be seen from Figs. 3 and 4. The PEOl-based copolymers displayed lower Ql and higher aB/c, in comparison with the other PEO-based ones. For the PEO 1-based copolymers, the swelling of PEO segment domains due to sorbed organic liquid might be more effectively suppressed by the polyimide segment domains because of the very short block length of PEO (n=9). The PPO- and PTHF-based copolymers displayed larger Ql and smaller aB/c in comparison with the PEObased ones. They had larger SB and Sc and lower ratios of SB to Sc, because of less hydrophilic polyether segments. The 3's of PPO and PTHF segments are much closer to that of cyclohexane, as compared with PEO segment, resulting in much larger Sc of them. The PPOand PTHF-based copolymer membranes were more swelled by the feed solution than the PEO-based ones. Membranes of hydrophobic and rubbery polymers such as polyethylene [22] and polypropylene [23] have been reported to have large permeation rate (QI> I0 kg/zm/(m2h) ) with very low selectivity (aB/C < 2) for PV of benzene-cyclohexane mixtures. Poly(dimethysiloxane imide) segmented copolymers with microphase separated structure, which were prepared using bis (aminopropyl) poly ( dimethylsiloxane ) instead of poly(ether diamine) [24], had also low selectivity (aB/c < 3.5 at XB = 60 wt% and 323 K) [25]. It is interesting that the copolymers with hydrophillic PEO segment had similarly large permeation rate and much higher aB/c, compared with the hydrophobic and rubbery polymers.

N. Tanihara et al. / Journal of Membrane Science 104 (1995) 181-192

Sorption and PV experiments for a binary benzenecyclohexane solution (XB=60 wt%) in BP-ODA/ DABA/PEO2(75) and BP-ODA/DABA copolyimide membranes were carried out. The results are listed in Table 4. The concentration of component i in the membrane, Ci (i = B for benzene or C for cyclohexane), was calculated from the sorption amount of the component, assuming the additivity of the density. The diffusion coefficient of component i, Di, was evaluated as Di = qil/Ci, where qil is the specific permeation flux of component i. The glassy BP-ODA/DABA copolyimide had much smaller CB and Cc in comparison with BP-ODA/DABA/PEO2(75). Assuming that the sorption is the same between polyimide segment microdomains in the PEO-based copolymer and BP-ODA/DABA copolyimide, the CB and Cc values for the PEO segment microdomains were estimated using [PE] and the sorption data for these two polymers, and are also listed in Table 4. The DB and D c values for the PEO segment microdomains were estimated neglecting the contribution of the polyimide segment microdomains to the permeation. The results indicate that the sorption is much larger for the PEO segment microdomains than for the polyimide segment microdomains, and that the preferential permeation of benzene component through the PEO-based copolymer membranes is attributed to both preferential sorption and diffusion of benzene. The contribution to separation is larger for the sorption process than for the diffusion one. A little larger diffusion coefficient for benzene is due to its smaller molecular size. Effects of feed composition on the specific permeation flux, permeate composition, YB (wt% of benzene), and C~B/care shown in Fig. 5. With an increase in XB, qBl increased significantly and aB/c decreased. This is because the membrane swelling increased with an increase in XB. Therefore, it is necessary to suitably select the membranes depending on the feed composition. The PEOl-based copolymers with [PE] of 35 to 40 wt% are suitable for a feed solution rich in benzene (XB > 60wt%), because of their performance well-balanced between the permeation flux and the selectivity. The PEO2-, PEO3- or PEO4-based copolymers with [PE] larger than 45 wt% are suitable for a feed solution poor in benzene (XB < 40wt%). With increasing temperature, qB and qc increased and aB/c decreased. For BP-ODA / DABA / PEO2 (75) at XB of 60 wt%, the activation energies were 45 and

187

300

1oo

Zs

E

~" • lo u_...i

o_~

0.3

I

I

I

I

20

40

60

80

1O0

X, [wt~] IOO 8O

4-J i..__1

~40 20 0 0

I

I

{

I

20

40

60

80

O0

X. [wt~] i

12

i

(c)

i

i

\0

I0 i

8 6 4 2 0 0

I

I

I

I

20

40

60

80

1O0

X B[wt~] Fig. 5. Effects of feed composition on (a) specific permeation flux (qBl: open symbols and solid lines; qcl: closed symbols and broken lines), (b) permeation composition, and (c) separation factor for PV ofbenzene---cyclohexanemixture at 323 K for membranes of PMODA/PEO4(80) (O,O), BP-ODA/DABA/PEO2(75) ( 1 , • ) , B P - O D A / D A B A / P E O l ( 7 5 ) (I-q), and B P - O D A / D A B A / PEO1 (70) (O).

188

N. Tanihara et al. / Journal of Membrane Science 104 (1995) 181-192

Table 5 Membrane performance toward PV of benzene-cyclohexane mixtures Membranesa

l (/xm)

XB (wt%)

Temperature (K)

Ql (kg/xm/(m 2 h) )

ctB/c (-)

Ref.

BP-ODA/DABA/PEO1 (75) PM-ODA/DABA/PEO1 (75) (plasma treated) BP-ODA/DABA/PEO2(75) (plasma treated) Poly(esterurethane)-I BP-3MPD/PD(3/1 ) polyimide

90 80 80 94 61

60 60 60 60 60 60 50 50 50 50 50 50 55 55 55 51 53 50 50 50 50

323 323 323 323 323 303 323 343 323 343 346 346 351 351 351 327 353 298 298 298 323

2.1 3.5 2.9 13 9.3 1.5c 3.4 5.4 0.41 2.7 0.35 ¢ 2.9c 20.0 24.1 16.2 4.0 50.3 7.4 2.8 0.5 4.6

9.1 7.8 10 6.5 7.8 3.8 10 10 27 14 10 8.7 12 40 9.0 9.7 5.2 oo 7 14.8 8

b b b b b 12 10 10 10 10 15 16 3 3 4 4 4 5 7 8 8

BP-3MPD/PD( 1/ 1) polyimide PI- 1 asymmetric membrane PI-2 asymmetric membrane Alloy of PPN-I(6.1%P) and CA Alloy of PPN-I(10.5%P) and CA Alloy of PSP and CA Alloy of PPOBrP and CA Modified cellulose ester MA-g-HEMA Poly(),-methyl ]-glutamate) HDPE-g-MA

26 31 10 10

20 20 20 20 16 30 43 23 23

a3MPD: 2,4,6-trimethyl-l,3-phenylenediamine; PI-I: aromatic poly(imide) from diphenylsulfonetetracarboxylic dianhydride; P1-2: aromatic poly(imide) from 6FDA; PPN-l: poly (styrene diethylphosphonate); CA: cellulose acetate; PSP: copolymer of vinylidene chloride and styrene diethylphosphonate; PPOBrP: poly(bromophenylene oxide dimethylphosphonate ester); MA-g-HEMA: graft copolymer of 2-hydroxylethyl methacrylate and methylacrylate; HDPE-g-MA: high density poly(ethylene-graft-methyl-acrylate) membrane prepared by plasma graft-filling polymerization, ~l'his study. CQwalue is given because membrane thickness was unknown. Table 6 Membrane performance toward PV of aromatic-aliphatic mixtures Membranesa PM-DDS/PEOI ( 85 ) Poly (esterurethane)- 1 Poly(esterurethane) -2 Poly(etherurethane) Muitiblock polymer PI/PA copolymer

l (/zm) 70 43 43 378 7

Separation system

X (wt%)

Temperature (k)

QP

tzc (-)

Ref.

Benzene-n-hexane Toluene-cyclohexane Toluene-cyclohexane Toluene-cyclohexane Toluene-n-octane Toluene-n-octane

60 60 60 60 r f

323 303 303 303 363 423

9.0 1.1e 16.7 11.6 10 4.2

9.1 4.8 4.5 3.7 8.3 8.5

d 12 14 14 17 18

"Membranes contain the following soft segments: poly (hexylene adipate) by 56 wt% for poly(esterurethane)-1; poly(butylene adipate) by 85 wt% for poly(esterurethane); poly(butylhexylether) by 85 wt% for poly(etherurethane); poly(ethylene adiphate) for multiblock polymer; poly ( ethylene adipate ) by 80 wt% for PI / PA ( polyimide / poly ( adipate ) ) copolymer. bUnit, k g / z m / ( m 2 h). cSeparation factor of aromatic component over aliphatic one. dThis study. eQ value. rFeed composition: I0 wt% toluene, 40 wt% p-xylene, 20 wt% iso-octane and 30 wt% n-octane.

N. Tanihara et al. / Journal of Membrane Science 104 (1995) 181-192 Table 7 Membrane performance toward PV of acetone-cyclohexane system at XA= 68 wt% (azeotropic composition) and 323 K a Membranes

l

Ql

OtA/C

BP-ODA/DABA/PEO1 (75) (plasma treated) BP-ODA/DABA/PEO2(75) BP-ODA/DABA/PEO3(75) PM-ODA/DABA/PEO1 (75) (plasma treated) B P - 3 M P D / P D ( 3 / I ) copolyimide b BP-3MPD/PD( 1/ 1 ) copolyimideb BT-3MPD polyimideb

65 96 94 120 80 80 30 5 7

8.2 9.5 34 63 13 15 29 14 36

17 42 6.7 5.1 9.6 15 13 23 8.3

"Units: l, p.m; Ql, kg ~ m / ( m 2 h) ~ r o m [ 10l.

55 kJ/mol for qB and qc, respectively, and aa/c decreased from 8.5 at 298 K to 5.5 at 343 K. Effect of plasma treatment of the membranes on the PV was examined. As can be seen from Table 3, the plasma treatment increased aB/c accompanied by a small decrease in Ql. This is attributed to reduction of the membrane swelling due to the crosslinking of PEO segments only in the surface layer. PV of benzene-n-hexane mixtures was also investigated for some poly (ether imide) segmented copolymer membranes. Specific permeation flux, Ql, and separation factor of benzene over n-hexane, aB/H, at XB of 60 wt% and 323 K are listed in Table 3. As compared with the benzene--cyclohexanesystem, these membranes displayed similar Ql and a little higher aa/ n for the benzene-n-hexane system. The sorption amounts of n-hexane, SH in these copolymer membranes were below the measurement limit (SH < 1 g/ ( 100 g dry polymer) ), because of the large difference in 8 between n-hexane and the PEO segments. There-

189

fore, the higher selectivity factor in the benzene-nhexane system seems due to larger solubility selectivity. It is interesting to compare the membrane performance of the PEO-based copolymers toward PV separation of aromatic and aliphatic mixtures with that of the other polymers reported in the literature. The membrane performance for PV of benzene-cyclohexane mixtures, being at a relatively high level, is listed in Table 5. Table 6 lists the membrane performance of block-copolymer membranes with microphase separated structure for PV of other aromatic-aliphatic mixtures. In Tables 5 and 6, the poly(esterurethane)-1, which is composed of 56 wt% soft segments of poly (hexylene adipate) and 46 wt% hard segments ofpoly (urethane), has a little smaller Q and a little larger a for toluenecyclohexane mixtures than for benzene-cyclohexane mixtures [ 12]. Therefore, it is not meaningless to compare membrane performance of one polymer for PV of benzene-aliphatic mixtures with that of another polymer for PV of toluene-aliphatic mixtures. The comparison shows that the PEO-based copolymers has much higher membrane performance for PV of aromatic-aliphatic mixtures as compared with the poly (esterurethane) - 1 and -2 and poly (etherurethane) [12,14]. The latter polymers have longer alkylene groups such as butylene and hexylene in soft segments but the solubility parameters of their soft segments are as large as that of PEO segment; for example, 8 = 19.5 (J/cm 3) 1/2 for poly(butylene adipate). Patent literatures have reported on membrane performance of the multiblock copolymer [17] and PI/PA copolymer [ 18], which have soft segment of poly(ethylene adipate) (8=22.2 ( J / c m 3 ) l / 2 ) , for PV of quartary mixtures. Neglecting the large difference in the feed

Table 8 Sorption and PV properties of BP-ODA/DABA/PEO2 (75) and BP-ODA/DABA copolyimide membranes for acetone-cyclohexane mixture OfXA=68 wt% at 323 K" Membranes

CA

Cc

CA/Cc

qAl

qcl

q^l/qcl

DA

Dc

DA/Dc

BP-ODA/DABA BP-ODA/DABA/PEO2(75) PEO domain polyimide domain

2.3 2.9 1.9 1.0

0.0099 0.21 0.21 0.0043

240 14 9.0 230

<0.02 15

0.74

20

51 79

36 35

1.4 2.2

ac, ql, and D are in 103 m o i / ( m 3 swollen polymer), 10 - s mol m / ( m 2 s), and 10-t2 m2/s, respectively. The data for polyether and polyimide domains are estimated values (see text).

190

N. Taniharaet al./ Journal of Membrane Science 104 (1995) 181-192

~.

lO0 f F (a)

.

.

.

composition between the patents and the present study, their membrane performance is comparable to that of the PEO-based copolymers. In Table 5, the membrane performance of the PEObased copolymers are comparable to that of the rigid and glassy polyimides without PEO segment and of the other glassy polymers, except for the polymer alloys of poly(styrene diethylphosphonate) and cellulose acetate, for which both higher permeability and higher selectivity have been reported [ 3 ].

I

.

3.3. P V o f acetone-cyclohexane 0 01

0

'£"

20

40

60 XA[wt%]

80

1O0

Ioo

80 ~,~ 60

.. 'Sa;ot,,qu,o equilibrium

>.< 40 20 0 0

500

i

i

20

\1

i

40 60 X^ [wtg~]

(c)D ~

I

I

i

80

1O0

I

S 100

10

2

A

o~

l

20

I

I

40 60 X^ [wt%]

I

80

100

Fig. 6. Effects of feed composition on (a) specific permeation flux

(qAl: open symbolsand solid lines; qcl: closed symbolsand broken lines), (b) permeationcomposition,and (c) separationfactor for PV of acetone-cyclohexanemixtureat 323 K for membranesof BPODA/DABA/PEO2(75) (O,O), BP-ODA/DABA/PEOl(75) (11, • ) and plasma-treated membranes of BP-ODA/DABA/ PEO1(75) (VI).

Specific permeation flux and separation factor of acetone over cyclohexane, OCA/C,for the PEO-based copolymer membranes at a feed composition, XA (wt% of acetone), of 68 wt% are listed in Table 7. For comparison, the data for the glassy polyimides from 2,4,6trimethyl-l,3-phenylene diamine (3MPD) are also listed in Table 7 [ 10]. The membranes were preferentially permeable to acetone over cyclohexane, aA/C was smaller for the PEO-based copolymers than for the 3MPD-based copolyimides having the similar values of Ql. The plasma treatment of the PEO-based copolymers significantly enhanced aA/C with rather a smaller increase in Ql. The performance of the plasma-treated PEO-based copolymer membranes was apparently superior to that of the 3MPD-based copolyimide membranes. The results of sorption and PV for a binary solution of XA = 68 wt% in BP-ODA/DABA/PEO2(75) and BP-ODA/DABA copolyimide membranes are listed in Table 8. The values of CA, Co DA and Dc for the PEO segment microdomains were estimated by the same way as in the benzene--cyclohexane system. It is noted that the BP-ODA/DABA copolyimide sorbed a fairly large amount of acetone but the specific permeation flux was below the measurement limit. Therefore, it is safely considered that the permeation in the PEO-based copolymer membranes occurs through the PEO segment microdomains and the polyimide segment microdomains have only a negligible contribution to the permeation. The preferential permeation of acetone component is attributed to preferential sorption and diffusion of acetone. The contribution to separation is larger for the sorption process than for the diffusion one. The larger permeation selectvitiy for the acetonecyclohexane system as compared with the benzene-

N. Tanihara et al. / Journal of Membrane Science 104 (1995) 181-192

cyclohexane system is due to the larger solubility selectivity as a result of the larger difference in ~ between acetone and cyclohexane. The effects of feed composition on the specific permeation flux, permeate composition, YA (wt% of acetone), and aA/C are shown in Fig. 6. With an increase in XA, qAl increased and C~A/cdecreased significantly, especially in the range OfXA less than 50 wt%, because of increasing membrane swelling. Therefore, the PEOl-based copolymers with [PE] of 35 to 40 wt% are more suitable for the acetone-cyclohexane system. PV of polar and nonpolar organic liquid mixtures through perfluorosulfonic acid (PFSA) composite membranes have been reported [ 13]. The membranes (H ÷ form) displayed Ql of 16 kg/xm/(m2h) and OtlPA/ Cof 8 at XIVA= 40 wt% and 328 k in isopropanol ( I P A ) cyclohexane system, and Ql of 43 kg/xm/(m2h) and aEtOH/C of 10 at XEtOH=50 wt% and 318 K in ethanol(EtOH)-cyclohexane system. The membrane performance of the PEO-based copolymer membranes mentioned above is considered to be comparable to that of PFSA composite membranes. The PEO-based copolymer membranes are potentially applicable to PV separation of other organic polar and nonpolar liquid mixtures.

4. Conclusions 1. Poly(ether imide) segmented copolymers had microphase-separated structure consisting of microdomains of rubbery polyether segments and of glassy polyimide segments. Sorption and permeation occurred in the former domains and the latter domains contributed to both film-forming ability and membrane durability. 2. In PV of benzene-cyclohexane, benzene-n-hexane and acetone-cyclohexane mixtures, they were more permeable to benzene and acetone. The selectivity was much better for PEO-based copolymers than for PPO- and PTHF-based ones because of larger solubility selectivity. 3. The permeation flux and the selectivity varied significantly with feed composition, and depended on content and block length of PEO segments ( [ PE] and n, respectively). The PEO 1 (n = 9)-based copolymers with [PE] of 35-40 wt% were suitable for feed solutions rich in benzene or acetone.

191

4. Plasma treatment of the membranes was effective because of the increase in the selectivity accompanied by only a small reduction in the flux. 5. PEO-based copolymer membranes displayed high PV performance comparable to that of other glassy polymers and good membrane durability.

5. List of symbols

Ci

concentration of component i in membrane (mol/m 3) dc collision diameter (nm) Di diffusion coefficient of component i (m2/s) l membrane thickness (/xm) n average block length of polyether segment [PE] polyether content in copolyimide (wt%) q~ permeation flux of component i ( m o l / ( m 2 s), k g / ( m 2 h) ) q~l specific permeation flux of component i (mol m / ( m 2 s), k g / x m / ( m 2 h)) Q total permeation flux (kg / (m 2 h) ) Ql total specific permeation flux ( k g / x m / ( m 2 h) ) Si sorption amount of pure component i (g solute/( 100 g dry polymer) ) Xi composition of component i in feed (wt%) Y~ composition of component i in permeate (wt%) a separation factor ( - ) 8 solubility parameter ( (J/cm 3) 1/2) 7/ inherent viscosity (dl/g) 5.1. Subscripts

A B C H

acetone benzene cyclohexane n-hexane

References

[ 1] P. Aptei,N. Challard,J. Cunyand J. Neel,Applicationof the pervaporation process to separate azeotropic mixtures, J. MembraneSci., 1 (1976) 271, and referencestherein. [2] J. Neel, P. Aptel and R. Clement, Basic aspects of pervaporation,Desalination,53 (1985) 297. [3] I. Cabasso, Organic liquid mixtures separation by permselective polymer membranes. 1. Selection and characteristicsof dense isotropicmembranesemployedin the

192

N. Tanihara et al. / Journal of Membrane Science 104 (1995) 181-192

pervaporation process, Ind. Eng. Chem. Prod. Res. Dev., 22 (1983) 313. [4] H. R. Acharya, S. A. Stem and Z. Z. Liu and I. Cabasso, Separation of liquid benzene/cyclohexane mixtures by perstraction and pervaporation, J. Membrane Sci., 37 (1988) 205, and references therein. [5 ] J. Terada, T. Hohjoh, S. Yoshimasu, M. Ikemi and I. Shinohara, Separation of benzene-cyclohexane azeotropic mixture through polymeric membranes with microphase separated structures, Polym. J., 14 (1982) 347. [6] S. Yoshimasu, H. Nomura, K. Komiya, T. Okano and I. Shinohara, Separation of an azeotropic mixture by the crosslinked polymeric membrane with trapped hydrogen chloride. Design of polymeric membrane for completely selective permeation, Nippon Kagaku Kaishi, (1980) 1785. [7] F. Suzuki and K. Onozato, Pervaporation of benzenecyclohexane mixtures by poly(y-methyl l-glutamate) membrane and synenergetic effect of their mixture on diffusion rate, J. Appl. Polym. Sci., 27 (1982) 4229. [8] T. Yamaguchi, S. Nakao and S. Kimura, Plasma-graft filling polymerization: Preparation of a new type of pervaporation membrane for organic liquid mixtures, Macromolecules, 24 (1991) 5522. [9] T. Yamaguchi, S. Nakao and S. Kimura, Solubility and pervaporation properties of the filling-polymerized membrane prepared by plasma-graft polymerization for pervaporation of organic-liquid mixtures, Ind. Eng. Chem. Res., 31 (1992) 1914. [10IN. Tanihara, K. Tanaka, H. Kita and K. Okamoto, Pervaporation separation of organic liquid mixtures through membranes of poly(imide) containing methyl-substituted phenylenediamine moieties, J. Membrane Sci., 95 (1994) 161. [ 11 ] M. Wessling, U. Wemer and S.-T. Hwang, Pervaporation of aromatic Ca-isomers, J. Membrane Sci., 57 ( 1991 ) 257. [12] L. Enneking, W. Stephan and A. Heintz, Sorption and diffusivity measurements of cyclohexane+benzene and cyclohexane + toluene mixtures in polyurethane membranes. Model calculations of the pervaporation process, Bet. Bansenges. Phys. Chem., 97 (1993) 912.

[ 13] B. K. Dutta and S. K. Sikdar, Separation of azeotropic liquid mixtures by pervaporation, AIChEJ., 37 (1991 ) 58 I. [ 14] H. Ohst, K. Hildebrand and R. Dhein, Polymers structure/ properties-correlation of polyurethane PV-membranes for aromatic/aliphatic separation, Proc. 5th Int. Conf. Pervaporation Processes Chem. Ind., Bakish Mat. Corp., Englewood, NJ, 1991, p. 7. [15] M. Nakatani, S. Matsuo and K. Nakagawa, Assymmetric polyimide membranes and pervaporation separation of organic liquid mixtures, Jpn. Kokai Tokkyo Koho, JP 03,284,335, 1991. [16] M Nakatani, S. Matsuo and K. Nakagawa, Pervaporation separation of organic mixtures using asymmetric membranes, Jpn. Kokai Tokkyo Koho, JP 03,284,336, 1991. [17] R. C. Schucker, Multi-blockpolymer comprising an ester prepolymer, made by combining epoxy with a compatible second prepolymer, the membrane made therefrom and its use for separations, US Pat. 5,221,481, 1993. [18] W. S. Winston Ho, G. Sartori and S. J. Han, Polyimide/ aliphatic polyester copolymers without pendant carboxylic acid groups (C2662), US Pat. 5,241039, 1993. [ 19] R. T. Chem, W. J. Koros, H. B. Hopfanberg and V. T. Stannet, Materials selection for membrane-based gas separations, ACS Symp. Set., 269 (1985) 25. [20] K. Okamato, N. Umeo, S. Okamyo, K. Tanaka and H. Kita, Selective permeation of carbon dioxide over nitrogen through poly(ethylene oxide)-containing polyimide membranes, Chem. Lett., (1993) 225. [21] D. W. Van Krevelan, Properties of Polymers, Elsevier, Amsterdam, 1976. [ 22] R. Y. M. Huang and V. J. C. Lin, Separation of liquid mixtures by using polymer membranes. 1. Permeation of binary organic liquid mixtures through polyethylene, J. Appl. Polym. Sci., 12 (1968) 2615. [23] M. Kucharski and J. Stelmaszek, Separation of liquid mixtures by permeation, Int. Chem. Eng., 7 (1967) 618. [24] C. A. Arnold, J. D. Summers, Y. P. Chen, R. H. Bott, D. Chen and J. E. McGrath, Structure-property behavior of soluble polyimide-polydimethylsiloxane segmented copolymers, Polymer, 30 (1989) 986. [25] T. Kawabata and K. Okamoto, unpublished data.