Gas transport properties of surface fluorinated poly (vinyltrimethylsilane) films and composite membranes

Gas transport properties of surface fluorinated poly (vinyltrimethylsilane) films and composite membranes

journal of MEMBRANE SCIENCE ELSEVIER Journal of Membrane Science 90 (1994) 55-68 Gas transport properties of surface fluorinated poly (vinyltrimethy...

1MB Sizes 0 Downloads 13 Views

journal of MEMBRANE SCIENCE ELSEVIER

Journal of Membrane Science 90 (1994) 55-68

Gas transport properties of surface fluorinated poly (vinyltrimethylsilane ) films and composite membranes J.D. Le Roux, V.V. Teplyakov’,

D.R. Paul*

Department of Chemical Engineering, Centerfor Polymer Research, The Universityof Texas, Austin, TX 78712. USA (Received August 27, 1993; accepted in revised form December 22, 1993)

Abstract The surface fluorination of both thick isotropic films and composite membranes formed from poly (vinyltrimethylsilane) (PVTMS) is described. For films ranging in thickness from 23 to 200pm, fluorination treatment reduced the permeability coefficients for the gases N1, 02, CH4, COz, Kr and Xe while the permeability of He was not affected. The ideal selectivities for He, CO1, Kr and Xe relative to Nz or CH4 were increased, but the ratio for the Oz/Nz pair remained unchanged. Repeated fluorination at increasingly harsh fluorination conditions progressively lowered the permeability and increased the selectivities still further. Effective diffusivity coefficients for the entire treated film were found to be more responsive to fluorination than the solubility coefficients. For composite membranes, comprising a thin ( N 0.5 pm) selective layer of PVTMS, two sets of fluorination conditions were considered: variable treatment time at a constant fluorine feed concentration of 0.02% and different feed concentrations at a constant treatment time of 2 min. In the absence of fluorination damage, the selectivity generally increased as a function of treatment time while the permeance (P/I) was reduced. Fluorination damage to some membranes caused a decrease in both selectivity and permeance, but these defects could be caulked by applying a surface layer of poly (dimethylsiloxane). Key words:Gas separation; Composite membranes; preparation and structure

Surface modification;

1. Introduction

Controlled direct surface fluorination has been employed to modify the gas transport properties of polymer membranes based on polyethylene [ 11, poly(Cmethyl-1-pentene) (PMP) [2,3], poly ( 1-trimethylsilyl- 1-propyne ) ( PTMSP ) [ 4 1, *Corresponding author. ‘Current address: A.V. Topchiev Institute of Petrochemical Synthesis, Russian Academy of Sciences, Leninski prospect, 29, 1179 12 Moscow, Russian Federation.

Fluorination;

Poly (vinyltrimethylsilane);

Membrane

polysulfone (PSF) [ $61 and poly(2,6-dimethyl- 1,Cphenylene oxide ) (PPO) [ 7 1. The selective polymers were in the form of thick films [ 1,2,4,5] and thin skins as integrally-skinned asymmetric membranes [ 5,6] or composite structures supported on flat sheets [ 5,7] or hollow fibers [ 41. Generally, relatively mild fluorination conditions are required to optimize the benefits of the treatment and simultaneously minimize damage to the treated surface caused by the aggressive nature of gas phase fluorination. At optimum conditions, fluorination can

0376-7388/94/$07.00 0 1994 Elsevier Science B.V. All rights reserved SSDI0376-7388(94)00003-H

56

J.D. L.e Roux et al. /Journal of Membrane Science 90 (1994) 55-68

result in a small increase in the ideal selectivities of O2 over N2, while the selectivities of He and H2 relative to N2 or CH, may be increased by factors of 2 to 10. In all cases a gain in selectivity is accompanied by an often substantial decrease in the pressure-normalized gas flux or permeante (P/1) of the membrane. In some cases the decrease in productivity can be substantial. The effect of surface fluorination is to create a chemically modified surface region with different transport properties than the underlying unmodified part of the film or skin. Generally, longer fluorination times increase the thickness of the modified layer [2,3,8]. In thick films, the modified layer constitutes a much smaller fraction of the total polymer layer than in thinskinned composite membranes; consequently, the effect of fluorination is expected to be more pronounced for the latter membranes. Analysis of the fluorinated and unfluorinated regions on the transport properties, using a series resistance approach, shows that the depth of the fluorinated region can be optimized to yield a favorable trade-off between selectivity and permeability for the entire membrane [ 2 1. In thin-skinned membranes, fluorination may cause defects which penetrate the skin and drastically impair

the transport properties of the treated membrane. Favorable properties can be restored to some extent by coating the surface of the membrane with a layer of highly permeable yet unselective polymer, such as poly (dimethylsiloxane ) (PDMS) [6,7,9]. Polymers with silicon in their main chain or in their pendant groups have attracted much interest for use in gas separation membranes [ lo]. This dates from an early interest in PDMS [ 111 and continues currently to polymers such as poly ( vinyltrimethylsilane ) ( PVTMS ) and poly ( 1-trimethylsilyl- 1-propyne ) (PTMSP). The gas transport properties for thick films of these silicon containing polymers are compared to PSF and PPO in Table 1. The material PVTMS has relatively high permeabilities and reasonable selectivities for a number of gas pairs and is the basis for industrial-scale production of asymmetric gas separation membranes in Russia [ 10,12,13 1. It is, therefore, of interest to investigate the potential benefits which may be obtained by the surface fluorination of membranes with separating layers or skins formed from PVTMS. Results for the surface fluorination of PVTMS in the form of thick solution cast films and thin-

Table 1 Transport properties for dense films of poly(vinyltrimethylsilane)

(PVTMS) and other relevant polymers

Polymer

T” (“C)

Permeability

25 35

Ref.

02/N2 N2

PVTMS

Selectivity

(barrers ) b CH,

02

11 12.5

He

H2

coz

18 21

200

49

180 170

190 150

4.0 3.92

19,20 this work

44

PDMS

25

281

604

_

649

354

3230

2.15

18

PTMSP

25 30

6745 4970

10040 7730

17000 13000

16150

6600

33100 28000

1.49 1.56

4 14

PPO

25 25

3.8 2.96

4.16 4.46

17 this work

25 25 25

0.205 0.250 0.179

5.76 5.2 6.43

16 15 this work

PSF

15.8 13.2 1.18 1.3 1.15

*Temperature at which properties were measured. bl barrer= 1O-Lo cm3(STP) cm/cm* s cmHg.

3.71 0.39 0.158

112.5 86.9 19.3 10.8 11.54

77.8 61.5

10.65

75.5 56.5 7.06 5.70 6.07

J.D. Le Roux et al. /Journal of Membrane Science 90 (1994) 55-68

skinned composite membranes are reported here. The latter membranes were formed by spin coating a solution of PVTMS onto a microporous ceramic substrate to form a selective layer less than 1 ,um thick [ 2 11. One objective of this work was to investigate whether the fluorination of PVTMS membranes could yield improved selectivities with higher productivities than fluorinated PSF or PPO membranes. Another objective was to observe the effect of relatively harsh fluorination conditions and multiple treatment cycles on relatively thick homogeneous PVTMS films, a procedure which cannot be followed for thin-skinned membranes without the risk of surface damage and consequent loss of favorable transport properties. Thick films of known thickness also permit the determination of average permeability, diffusion and solubility coefficients over the entire film, without distinguishing between the relative contributions of the fluorinated and unfluorinated regions.

2. Experimental 2.1. Formation of composite membranes Homogeneous films were cast on a horizontal glass plate from a 7% solution of PVTMS in toluene. The PVTMS was supplied by the A.V. Topchiev Institute of Petrochemical Synthesis and had an intrinsic viscosity in cyclohexane of 1.0 dl/g (at 25°C) with a reported molecular weight of N 830,000 as determined by light scattering [ lo,22 1. A glass transition temperature of 152.5 20.7”C was determined by thermal analysis. The films were placed in a vacuum oven at 50’ C until they reached a stable weight and thermogravimetric analysis confirmed that the final film was free of solvent. The composite membranes were formed by spin coating a dilute solution of PVTMS directly onto microporous ceramic substrates. These yalumina microfilters ( AnoporeTM, from Anotec Separations) are asymmetric with an average surface pore diameter of 200 A, a bulk pore diameter of 2,000 8, and a molecular weight cutoff of 10’ daltons; the bulk porosity is 30-35%

51

and the surface pore density of the order of lO+ lo cmm2 [ 231. The spin coating solution comprised 1.O% (by weight) of PVTMS in cyclohexane. A photo-resist spin coater (Series EC 10 1D ) from Headway Research Inc. (Garland, Texas) was used to apply the PVTMS solution to the ceramic substrate as described in detail elsewhere [ 7,2 11. A spinning speed of 400-800 rpm was used. The coated substrate was air-dried for several hours and then dried under vacuum at 70’ C for at least 16 h to ensure complete removal of solvent. Prior to solvent treatment and gas permeation measurements, the circular composite membranes (4.7 cm in diameter) were masked peripherally with adhesive aluminum foil to expose a permeation area of 5.1 cm2; masking helped to minimize damage during handling. To repair minor skin defects, all membranes were treated with liquid methyl ethyl ketone by drawing a liquid-saturated sponge applicator across the exposed surface of the membrane, followed by vacuum drying at 70 ’ C.As described for other polymers [ 241, the solvent treatment usually raises the 02/N2 selectivity of the membrane to slightly above the value observed for thick films (see Table 1) .The unfluorinated membranes of this study had an average thickness of 4,550 8, with a standard deviation of 880 A. Membranes were screened by measuring their O2 and N2 fluxes and the ideal selectivity at 24” C. The O2 permeance (Table 1) was used to estimate the skin thickness and the 02/N2 selectivity was used to gauge the integrity of the skin. A membrane was considered defect-free if it had an 02/N2 selectivity equal to or greater than 4.0, i.e., that of thick dense PVTMS films prepared by solution casting (Table 1) .The final membranes had a composite structure comprising the microporous ceramic support covered by a substantially defect-free PVTMS skin with apparent thicknesses ranging from 0.45 to 0.60 pm. After measuring their gas transport properties, the membrane discs were prepared for fluorination treatment as described below. 2.2. Fluorination and surface coating Prior to fluorination the composite membranes were mounted on a solid aluminum back-

58

J.D. Le Roux et al. /Journal of Membrane Science 90 (I 994) 55-68

ing using adhesive aluminum foil, so that only the PVTMS skin surface was exposed. The masked membranes were fluorinated in a stainless steel flow-through reactor with a volume of 14.6 cm3. The feed gas consisted of commercially available 2% (by volume) F2 in He which was further diluted to the desired concentration with pure He. For this reactor, the fluorine concentration at the membrane surface approaches that in the feed within 1 min of treatment; thus, after N 1 min, the treatment time and fluorine concentration at the membrane surface can be varied independently. With this reactor configuration it is necessary to use extremely low feed concentrations of F2 to prevent overfluorination of the membrane. Background and further experimental details of the fluorination process are given elsewhere [ 2,6]. The gas transport properties of the membranes were measured before and after coating their surfaces with a layer of PDMS. As noted in Table 1, PDMS is highly permeable and has a relatively low Oz/N2 selectivity compared to PVTMS. These properties make it well suited for caulking minor skin defects [ 91. A 3% (by weight) solution of PDMS in methyl ethyl ketone was applied to the membrane surface by means of a sponge applicator and allowed to dry at ambient temperature for several hours and then overnight in a vacuum oven at 70°C to remove all traces of solvent. 2.3. Gas permeation measurements Gas permeability coefficients of dense films were measured at an upstream pressure of 1.51.7 atm and essentially zero pressure at the downstream, according to techniques described elsewhere [ 25 1. The permeability coefficient can be factored as follows P=DS

(1)

where D is the diffusion coefficient averaged across the membrane and S is the solubility coefficient at upstream conditions [ 26,27 1. The diffusion coefficient was estimated from the time lag, 8, of the transient permeation measurement [ 28 1, according to the relation

(2) where 1 is the average thickness of the membrane. The temperatures and the gases used in the measurements are shown with each data set below. The permeability coefficients for solution cast PPO and PSF thick films were measured independently at 25 ‘C and Table 1 contains these and other values from the literature. For the composite membranes, the permeation rates of several gases were measured before and after fluorination, and after coating with PDMS, using a commercially available permeation cell (Millipore Corporation, Bedford, MA). Steady state permeation rates were measured at 24°C with gas pressures in the range of 2 to 4 atm to obtain the permeance (P/I) or pressurenormalized flux, as described in greater detail elsewhere [ 6,7 1.

3. Gas transport 3.1. Dense homogeneous films Dense PVTMS films of known average thickness were prepared by solution casting to investigate the effect of surface fluorination on their gas transport properties. This permitted the determination of steady state permeability coefflcients and, from time lag measurements, the calculation of effective diffusivities and selectivity coefficients from Eqs. ( 1) and (2). The thickness of these films ranged from 23 to 200 pm, typically two orders of magnitude thicker than the separating layers of the composite membranes formed on microporous ceramic supports by spin coating. Since the thickness of the latter PVTMS layers could not be measured independently, only ideal separation factors and pemeante (P/I), as opposed to permeability, could be determined. From these the ideal separation factors can be computed. The results for dense films are presented in Tables 2,3 and 4. Table 2 shows the progressive changes in permeability of a 200 ,um film fluorinated at increasingly harsh conditions, by increasing either the treatment time or

59

J.D. Le Roux et al. /Journal ofMembrane Science 90 (1994) 55-68

Table 2 Permeability and ideal selectivity of solution cast PVTMS film fluorinated consecutively for various times and fluorine concentrations (film thickness -2OOpm) Fluorination conditions Time” 0

2 4 16 5 10

Ideal selectivity

Permeabity’

[Fib

He

CO2

O2

N2

CH,

WNz

He/N2

He/CH4

C02KH4

0

170.0 169.0 170.4 170.0 169.7 170.0

150.0

49.0 44.9 45.6 43.6 41.6 40.6

12.5 12.3 12.2 11.0 10.6 11.3

21.0 19.9 20.9 17.8 15.8 14.8

3.9 3.7 3.7 4.0 3.9 3.6

13.6 13.7 14.0 15.5 16.0 15.0

8.1 8.5 8.2 9.6 10.7 11.5

7.14 7.0 7.9 8.5 8.9

0.02 0.04 0.04 1.0 2.0

146.1 140.8 133.6 132.0

“Pluorination treatment time (min). bFluorine concentration in the feed (~01% F2 in He ) . ‘In barrers ( 1 barrer= lo- lo cm3 (STP) cm/cm2 s cmHg). Table 3 Apparent diffusivity and solubility of solution cast PVTMS film fluorinated consecutively for various times and fluorine concentrations (film thickness - 200 pm ) Fluorination conditions

Apparent diffusivity”

Time” 0

2 4 16 5 10

Apparent solubilityd

[F] b

He

CO2

O2

N2

CH,

He

0

370=

5.3

9.4 9.1 9.4 9.2 8.8 8.1

4.1 4.1 4.1 4.0 3.5 3.5

2.10 1.74 1.71 1.47 1.44 1.33

0.46’ -

0.02 0.04 0.04 1.0 2.0

5.6 5.2 4.8 4.8

co2

02

N2

C&

28.3

5.2 4.9 4.9 4.7 4.7 5.0

3.0 3.0 3.0 2.8 3.0 3.2

10.0 11.4 12.2 12.1 11.0 11.1

26.0 27.0 27.8 27.5

Wuorination treatment time (min ) . bFluorine concentration in the feed (~01% F2 in He). “Units: lo-’ cm2/s. dUnits: low3 cm3 (STP) /cm2 s cmHg. ‘Data from Teplyakov and Meares [ 19 1. Table 4 Permeability and ideal selectivity of solution cast PVTMS films fluorinated consecutively for various times and fluorine concentrations (film thickness - 25-33 pm) Fluorination conditions Time’ 0

5 10 5

Thickness

T”

(pm)

(“C)

[Flb 0

1.0 1.0 2.0

25 33 28 23

20 20 20 20

Permeabilityd

Ideal selectivity

He

Xe

Kr

CH4

He/CH,

Xe/CH,

260 270 260 260

24 18 17 21

30 17 11 12

17 9 6 7

15.3 30.0 43.3 37.1

1.4 2.0 2.8 3.0

“Fluorination treatment time (min ) . bFluorine concentration in the feed (~01% F2 in He). Temperature at which transport properties were evaluated. din barrers [ 1 barrer= 1O-r0 cm3 (STP) cm/cm2 s cmHg].

1.8 1.9 1.8 1.7

60

J.D. Le Roux et al. /Journal of Membrane Science 90 (1994) 55-68

the fluorine concentration, or both. The permeability coefficients of the five gases examined are comparable to previously reported values at 25°C (Table 1). Multiple fluorination treatments of increasing severity generally led to a decrease in the permeability for four of the five gases. The only exception was He, which has the smallest kinetic diameter and largest initial permeability. The selectivities (also shown in Table 2) were generally increased, except for 02/N2 which remained essentially unaffected by the treatment. It is noteworthy that the He/CH, selectivity was increased by a factor of 1.4 after all the treatments, while the He permeability showed no concomitant decrease. Eqs. ( 1) and (2 ) strictly apply only to films that contain no defects and whose properties are uniform throughout their thickness. After fluorination, all of these conditions will not be met. While the permeability of a graded film is a meaningful quantity, solubility and diffusivity parameters are not so simple. Nevertheless, useful trends can be found by continuing this separation of the permeability into effective or apparent “solubility” and “diffusivity” parts. Table 3 shows the apparent diffusivity and solubility of the same dense film under identical fluorination conditions as Table 2. The variation of the apparent diffusivity in response to fluorination is clearly much more pronounced than for the apparent solubility. Thus, the effect of fluorination treatment on permeability appears to be dictated more by a change in diffusivity, as might be expected. The effect of fluorination on the transport behavior of a series of gases with a large range of molecular dimensions (He, CH4, Kr and Xe) was examined using a number of thin PVTMS films fluorinated once under different conditions. Table 4 shows the permeability coefficients for these inert gases and their selectivities (relative to methane) at 20°C. As found for the thick 200 pm film (Table 2), the permeability of He remains essentially unchanged at these fluorination conditions. On the other hand, fluorination considerably reduces the permeabilities of the other gases relative to the unfluorinated poly-

mer, the gases with larger molecular dimensions being most affected. The changes in permeability are directly reflected in the selectivities which emphasize the potential utility of surface fluorination to improve the separation factor for separations involving He, as well as HZ, though its transport properties were not measured for these dense lilms.

4. Composite membranes 4.1. Thick-skinned composite membranes To explore the middle ground between the transport properties of dense films and those of thin-skinned composite membranes, a relatively thick-skinned composite membrane (7.5 pm) was formed by spin coating several layers of PVTMS onto a ceramic support. This membrane was fluorinated at 1% F2 in He for 5 min and its gas transport properties were determined at 35 ‘C in the same manner as for the dense films. The apparent thickness of the PVTMS layer was estimated from its permeance (P/I) and the permeability coefficient of dense films (Tables 1 and 2 ) since the film could not be removed from the substrate for physical thickness measurements without damaging it. The permeability and selectivities before and after fluorination are shown in Table 5. The trends in permeability are superficially similar to those shown in Tables 2 and 4 for thick films, i.e., except for He, the permeability is reduced after fluorination and the selectivities of all but the 02/N2 pair are increased. The extent of the changes after only one treatment cycle is more pronounced than for the thicker films. Since the overall film thickness is smaller, the ratio between the thickness of the fluorinated and unfluorinated regions of the film should be greater, and so too the contribution of the fluorinated region to the overall transport properties. Again it is significant that the permeability for He is unchanged while its selectivity relative to the other gases is substantially increased. The results for thin-skinned composite membranes, discussed below, show that similar considerations should apply for Hz separation.

J.D. Le Roux et al. /Journal of Membrane Science 90 (1994) 55-68

61

Table 5 Permeability and ideal selectivity of a thick composite PVTMS membrane (skin thickness _ 7.5 pm) before and after fluorination Ideal selectivity

Permeability”

Fluorination conditions

[Fib

He

CO2

O2

N2

CH4

MN2

He/N2

He/CH,

C02D-b

0

0

5

0.1

173.2 173.9

160.6 117.0

49.0 37.0

12.4 9.5

23.1 8.9

3.9 3.9

14.0 18.3

7.5 19.5

6.9 13.1

Timd

“Fluorination treatment time (min). bFluorine concentration in the feed (vol% F2 in He). “In barrers [ 1 barrer= lo-” cm3 (STP) cm/cm2 s cmHg].

II 1

2

3

4

5

6

7

a

9

10

0

11

1

2

11

3

4

6

1

11

11

5

7

a

9

10

Fluorination time [min] Fig. 1. The effect of fluorination times for a constant feed concentration of 0.02% F2 on the permeance (P/l) of PVTMS composite membranes for six gases: before any treatment (0 ), after fluorination (0 ) and after coating the fluorinated membrane surface with a layer of PDMS ( 0 ). The lines are intended to emphasize trends in the data. The units of permeance are expressed in gas permeation units [ 1 GPU = 10m6 cm3 (STP ) /cm2 s cmHg 1.

4.2. Thin-skinned composite membranes: effect of treatment time The study of the fluorination of thick PVTMS

membranes was extended to thin-skinned composite membranes to increase the proportion of the fluorinated to unfluorinated regions. Since the gas fluxes are generally higher, thin-skinned

J.D. Le Roux et al. /Journal ofMembrane Science 90 (I 994) 55-68

62 50

40 A .5 ‘5 fii 5 *

30 20 10 0

30 25 .c ._ ii $j

20 I5 10 5 0 25 20

x .=> ‘C i? z

15 10 5 0

kIIIIII”‘II4 0

1

2

3

4

5

Fluorination

6

7

8

9

10 11

time [min]

o

I

I

I

I

I

I

1

2

3

4

5

6

Fluorination

I1

7

I

8

9

I

I,

10 11

time [min]

Fig. 2. The selectivity before ( 0 ) and after (0 ) fluorination and after coating with PDMS ( 0 ) for the six gas pairs at different fluorination times and a constant feed concentration of 0.02% F2. The symbols and lines have the same meaning as in Fig. 1.

membranes also have greater practical significance. Due to the small volume of the fluorination reactor, the effect of fluorination time and fluorine feed concentration could be studied separately. Experience with the fluorination of two other polymers, PSF [ 61 and PPO [ 71, in the same reactor, indicated that low fluorine concentrations and short treatment times were required to minimize damage to the polymer which is associated with a loss of favorable properties. Figs. 1 to 3 show the effect of varying treatment time at a constant fluorine concentration in the feed of 0.02% (by volume). Solid lines have been included to illustrate broad trends in the data. As found for the fluorination of PSF and PPO membranes [ 6,7 1, the data show considerable

scatter after fluorination and no trends are indicated for the post-fluorination data (solid symbols). As shown in Fig. 1, fluorination decreases the permeance (P/Z) of the membranes for all gases studied. The extent of this decrease in permeante was greater at longer fluorination times. Helium was the least affected, but unlike the thick films discussed above, its permeance relative to the untreated value ranged from 92% (at short fluorination times) to N 65% (after 10 min ) of the untreated value. For He and H2 the application of a layer of PDMS lowered the permeance still further, while the effect for the other gases was minimal. It appears likely that the decrease in permeance as a function of fluorination time

63

J.D. Le Roux et al. /Journal ofMembrane Science 90 (1994) 55-68

5.0

.E? g

4.0

w 2"

3.0

. 0” 2.0 1.0

0.8

20

40

60

80

O2 Permeance

100

120

(P/r) [GPU]

0.7

30 0.6

‘1’1’,‘1’1,1,,‘,‘1’1’ 10 0

25 -

‘5 0.5

0.4t I I 012345678

I

I

I

Fluorination

I

I

I

I

9

I

10

I

II

-I

12

time (min]

Fig. 3. The selectivity before ( 0 ) and after (0 ) fluorination and after coating with PDMS ( Cl ) for 0*/N* and N&H4 at the same conditions as those for Figs. 1 and 2.

is due to a corresponding increase in the depth of a less permeable fluorinated skin [ 2,3,5,8]. The addition of a layer of PDMS serves as an additional resistance to permeation which results in a further decrease in the permeance. The accompanying changes in ideal selectivity for eight gas pairs are shown in Figs. 2 and 3. Mild fluorination for up to 5 min of treatment generally increases the selectivity, though it is clear that in some cases the membrane skin is damaged and the selectivity is lowered. The selectivity after 10 min is more variable and not all membranes show an improvement. After coating with PDMS, the scatter is considerably less, probably because the defects are caulked and the actual properties of the treated membrane more apparent. This var-

0

~‘~‘~‘,‘~‘~I~‘~‘~‘~‘~ 200 300

100

H, Pertneance

400

500

600

(P/l) [GPU]

Fig. 4. The relation between the selectivity and productivity for a set of similar untreated membranes fluorinated for different times, before ( 0 ) and after ( 0 ) fluorination and after coating with PDMS (Cl ). The upper graph shows the relation between O2 permeance and 02/N2 selectivity and the lower graph the H2 permeance versus H&H4 selectivity. Each fluorinated and coated membrane is identified by its fluorination time given in minutes.

iability, even on a controlled laboratory scale, has been noted for the direct fluorination of membranes made from several polymers [ 5-7 1. The damage can be attributed to the aggressive nature of direct fluorination and the fact that the membranes are often made with the thinnest possible skins. These skins may be extremely thin

140

ljy

,,,(,

I,,,,1 I * 1 !. I I I II I 0 0.02 0.04 0.06 0.08 0.10 0.12 0.14 I

0 0.02 0.04 0.06 0.08 0.10 0.12 0.14

I

Fluorine concentration 5. The effect of fluorine feed concentration lines have the same meaning as in Fig. 1.

at a fixed fluorination

in some areas or have latent defects which are exacerbated by the treatment. The behavior of those membranes which were coated with PDMS after fluorination may be better illustrated by Fig. 4, where the permeance and the Oz/Nz and HJCH4 selectivities are shown for a set of membranes before fluorination, after fluorination and after coating. In order to trace the response of each membrane, the treatment times (in minutes) are given beside the symbols. In the interest of clarity, only a single set of five membranes is shown here. Figs. 1 to 3 contain more extensive data showing the same trends in a different way. Considering first the 02/N2 gas pair in the upper plot, some membranes undergo an increase in selectivity and a decrease in permeance after fluorination, while for others (e.g., after 2 and 10 min of treatment )

I

I

I.I.,,,,,,,,,,_

0 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16

[vol%]

time of 2 min on the permeance

(P/I). The symbols and

both the permeance and selectivity decreases. The reduction in selectivity is apparently due to minor skin defects resulting from damage during fluorination. After surface coating, a well-defined trend of increasing selectivity and decreasing permeance emerges as a function of fluorination time. The selectivities of the damaged membranes appear to be restored by caulking the defects. The selectivity of the undamaged membranes decreases due to the series contribution of the layer of PDMS which has a lower selectivity than that of PVTMS; from Table 1 the OJNz selectivity for PDMS is 2.2 while for PVTMS it is 4.0. For both the damaged and undamaged membranes, the layer of PDMS increases the resistance to permeation which reduces the permeante. Similar considerations apply to the effects of fluorination and coating on the HJCH, selec-

J.D. Le Roux et al. /Journal of Membrane Science 90 (1994) 55-68

ot*‘*“““““‘*l 0

0.02

0.04

I,I.I.II

0.06

0.08

0.10

Fluorine concentration

0.12

0.14

0

[vol%]

I.I*II-

0.04

0.06

0.08

0.10

Fluorine concentration

Fig. 6. The effect of fluorine feed concentration for a fixed fluorination in Fig. 2. The symbols and lines have the same meaning as in Fig. 1.

tivity and Hz permeance, although the reduction in permeability of the smaller Hz molecule is less pronounced than for 02. As a control, one membrane (shown at a fluorination time of 0 min) was coated with PDMS without first being fluorinated. As expected, Fig. 1 shows that the added resistance of the coated layer causes a small decrease in the permeance of all gases, the lighter gases (Hz and He) being most affected. From Figs. 2 and 3 it is observed that the selectivities for O/N2 and CO2 relative to N2 and CH4 are virtually unaffected while the remaining gas pairs are reduced slightly. This can again be attributed to the added series resistance which affects the faster gases more than the slower ones. The effects observed here are simi-

0.02

65

0.12

0.14

[vol%]

time of 2 min on the selectivity of the six gas pairs shown

lar to those pointed out above for fluorinated membranes which apparently were not damaged by the treatment. 4.3. Thin-skinned composite membranes: effe ofjluorine feed concentration Another series of composite membranes was exposed to different feed concentrations up to 0.15% F2 in He for a constant treatment time of 2 min. Fig. 5 shows the permeance as a function of fluorine concentration before and after fluorination and, for some membranes, after coating with PDMS. For all six gases fluorination resulted in a decrease in permeance, while the coating treatment caused no significant reduc-

66

J.D. Le Roux et al. /Journal ofMembrane Science 90 (1994) 55-68

PDMS reduces the overall selectivity of the composite membrane, since the added layer has a lower selectivity than both the fluorinated and unfluorinated PVTMS. Optimum results are obtained for uncoated membranes fluorinated at a concentration of 0.1% for 2 min. Under these conditions two to three-fold gains in selectivity are realized for He and Hz relative to N2 or CH4 with a relatively modest loss of productivity for the fast gases (Fig. 5 ).

5. Conclusions

e:: 0

0.02

0.04

0.06

0.08

Fluorine concentration

0.10

0.12

0.14

0.16

[vol%]

Fig. 7. The effect of fluorine feed concentration for a fixed fluorination time of 2 min on the selectivities of the Oz/Nz and N2/CH4 gas pairs. The symbols have the same meaning as in Fig. 1.

tion. The response of the various selectivities to increasing fluorine concentration is shown in Figs. 6 and 7. It appears that fluorination treatment results in increased selectivities for all but the 02/Nz pair. The selectivities for the latter pair are not improved by coating with PDMS after fluorination; in fact, the selectivity decreases slightly in most cases. Again the membrane properties after fluorination are highly variable compared to the results after coating with PDMS, presumably because only some of the membranes are damaged to the extent that skin lesions occur. As discussed above, when fluorination does not damage the skin of the membrane, the addition of a layer of

A series of PVTMS films having a wide range of effective thicknesses were surface fluorinated and the effect of this treatment on their gas transport properties was evaluated. For dense homogeneous films ranging in thickness from 7.5 to 200 pm, fluorination treatment reduced the permeability coefficients for the gases NZ, 02, CH4, COz, Kr and Xe while the permeability of He was not affected. The ideal selectivities for He, C02, Kr and Xe relative to N2 or CH4 were increased, but the ratio for the 02/N2 pair remained unchanged. Repeated fluorination of the 200 pm film at increasingly harsh fluorination conditions progressively lowered the permeability and increased the selectivities still further, though the He permeability and OJN2 selectivity remained unchanged. Generally, the permeability of the heavier gases with larger kinetic diameters were more affected by the treatment. For the entire treated film, the apparent diffusion coefficients were found to be much more responsive to fluorination than the apparent solubility coefficients. The effect of the treatment was also more pronounced for thinner membranes, because the fluorinated region constituted a greater proportion of the overall film thickness. The investigation included the surface fluorination of composite membranes comprising a thin ( N 0.5 pm ) layer of PVTMS formed by spin coating a cyclohexane solution onto a microporous alumina substrate. Two sets of fluorination conditions were considered: variable treatment time at a constant fluorine feed concentration of 0.02% and different feed concentrations at a

J.D. L.e Roux et al. /Journal of Membrane Science 90 (I 994) 55-68

constant treatment time of 2 min. In the absence of fluorination damage, the selectivity generally increased as a function of treatment time while the permeance was reduced. Fluorination damage to some membranes caused a decrease in both selectivity and permeance. The surfaces of some membranes were coated with a layer of PDMS to caulk defects which may have formed. This aftertreatment increased the selectivity of the membranes whose selectivity had been reduced by fluorination but reduced the selectivity of those without defects by a small factor (see Fig. 4 ) . The increase in selectivity is ascribed to the caulking of minor defects, and calculatiohs using dense films transport properties confirmed that the reduction in selectivity is due to the series resistance effect of adding a layer of low selectivity PDMS on top of the fluorinated PVTMS surface. In terms of selectivity, the optimum fluorine feed concentration at 2 min of treatment was found to be 0.1%. At lower concentrations the selectivity was increased and permeance decreased as a function of concentration. At a higher concentration of 0.15% the selectivity decreased after fluorination but was raised by coating with PDMS; this behavior suggests overfluorination, resulting in damage to the skin of the membrane. On the whole, optimized fluorination treatment has the most beneficial effect on the H2/CH4 selectivity, which can be increased by a factor of two to three. Although higher selectivities were usually realized for uncoated membranes, surface coating is recommended to achieve more uniform results after fluorination. In thin-skinned composite membranes the maximum increase in selectivity is accompanied by a reduction in permeance to -70% of that of the unfluorinated membrane. In dense, thick PVTMS films no loss of permeability was observed, though the permeance of such films would be quite low due to their thickness.

6. Acknowledgments The authors thank the National Academy of Sciences (Washington, DC) for the award of a Visiting Fellowship to V.V. Teplyakov which en-

67

abled him to conduct part of this research at The University of Texas at Austin. Thanks are also due to Dr. Joel Kampa, James DeYoung and Prof. R.J. Lagow for their technical assistance with the surface fluorination, as well as to the A.V. Topchiev Institute of Petrochemical Synthesis, Moscow, for supplying the polymer. This work was supported by the Texas Advanced Technology Program under Grant No. 1607, by the Separations Research Program at the University of Texas, Austin, TX, USA, and by the CSIR, Pretoria, South Africa.

7. References [ I] C.L. Kiplinger, D.F. Persico, R.J. Lagow and D.R. Paul, Gas transport in partially fluorinated low-density polyethylene, J. Appl. Polym. Sci., 31 (1986) 2617-2626. [ 21 J.M. Mohr, D.R. Paul, T. Mlsna and R.J. Lagow, Surface fluorination of composite membranes. Part I. Transport properties., J. Membrane Sci., 55 ( 199 1) 13 l148. [ 3lJ.M. Mohr, D.R. Paul, Y. Taru, T. Mlsna and R.J. Lagow, Surface fluorination of composite membranes. Part II. Characterization of the fluorinated layer, J. Membranesci., 55 (1991) 149-171. [4]M. Langsam, M. Anand and E.J. Karwacki, Substituted propyne polymers. I. Chemical surface modification of poly [ 1-( trimethylsilyl)propyne] for gas separation membranes, Gas Sep. Purif. 2 ( 1988) 162-l 70. [ 51J.M. Mohr, D.R. Paul, I. Pinnau and W.J. Koros, Surface fluorination of polysulfone asymmetric membranes and films, J. Membrane Sci., 56 ( 1991) 77-98. [6]J.D. Le Roux, D.R. Paul, J. Kampa and R.J. Lagow, Modification of asymmetric polysulfone membranes by mild surface fluorination. Part I. Transport properties, J. Membrane Sci., in press. [7]J.D. Le Roux, D.R. Paul, J. Kampa and R.J. Lagow, Surface fluorination of poly(phenylene oxide) composite membranes. Part I. Transport properties, J. Membrane Sci., 90 (1994) 21-35. [ 81A.P. Kharitonov, Y.L. Moskvin and G.A. Kolpakov, The direct fluorination of polyethylene terephthalate films, Sov. J. Chem. Phys., 4 (1987) 877-885. [9]J.M.S. Henis and M.K. Tripodi, Composite hollow iiber membranes for gas separation: the resistance model approach, J. Membrane Sci., 8 ( 198 1) 233-246. [ 1OlN.A. Plate, S.G. Durgaryan, V.S. Khotimskii, V.V. Teplyakov and Yu.P. Yampol’skii, Novel poly(silicon oletins) for gas separations, J. Membrane Sci., 52 (1990) 289-304.

68

J.D. Le Roux et al. /Journal of Membrane Science 90 (1994) 55-68

[ II] K. Kammermeyer, Silicone rubber as a selective barrier: gas and vapor transfer, Ind. Eng. Chem., 49 (1957) 1685-1686. [ 1215. Bouchillon, A. Fabre and A. Fame, Anisotropic organosilicon polymer membrane, US Pat. 3,754,375 ( 1973), assigned to R.-P.S.A. [ 13]N. Plate and Y.P. Yampol’skii, Relationship between structure and transport properties for high free volume polymeric materials, in: D.R. Paul and Y.P. Yampol’skii (Eds. ), Polymeric Gas Separation Membranes, CRC Press, Boca Raton, FL, 1994. [ 14]Y. Ichiraku, S.A. Stern and T. Nakagawa, An investigation of the high gas permeability of poly( l-trimethylsilyl-1-propyne), J. Membrane Sci., 34 (1987) S-18. [ 15lH.J. Bixler and O.J. Sweeting, Barrier properties of polymer films, in O.J. Sweeting (Ed.), The Science and Technology of Polymer Films, Vol. II, Wiley, New York, 1971,pp. l-130. [ 16]Y. Nagase, A. Naruse and K. Matsui, Chemical modification of polysulphone. 2. Gas and liquid permeability of polysulphone/polydimethylsiloxane graft copolymer membranes, Polymer, 31 (1990) 121-125. [ 17 ] S. Pauly, Permeability and diffusion data, in J. Brandrup and E.H. Immergut (Eds.), Polymer Handbook, Wiley, New York, 1989, pp. VI/435-449. [ 18]H. Yasuda and K.J. Rosengren, Isobaric measurement of gas permeability of polymers, J. Appl. Polym. Sci., 14 ( 1970) 2839-2877. [ 19lV.V. Teplyakov and P. Meares, Correlation aspects of the selective gas permeabilities of polymeric materials and membranes, Gas Sep. Purif., 4 (1990) 66-74.

[ 2OlV.V. Teplyakov and S.G. Durgaryan, On the ratio of permeability parameters for constant gases and hydrocarbons in polymers, Vyskomolek. Soedin. Ser. A, 26 (1984) 1498-1505 (in Russian). [ 2 1]J.D. Le Roux and D.R. Paul, Preparation of composite membranes by a spin coating process, J. Membrane Sci., 74(1993)233-252. [22]N.S. Nametkin, V.S. Khotimskii and SC. Durgaryan, Synthesis of high molecular weight poly(vinyltrimethylsilane) and some of its properties, Dokl. Akad. Nauk USSR, 1665 (1966) 1118-l 120 (in Russian ) . [ 23 ] Anotek Separations, AnoporeTM Membrane: Technical data, New York, 199 1. [24]M.E.Rezac,J.D.LeRoux,H.Chen,D.R.PaulandW.J. Koros, Effect of mild solvent post-treatments on the gas transport properties of glassy polymer membranes, J. Membrane Sci., 90( 3) ( 1994) in press. [25] W.J. Koros, D.R. Paul and A.A. Rocha, Carbon dioxide sorption and transport in polycarbonate, J. Polym. Sci., Polym. Phys. Ed., 14 (1976) 687-702. [26]G.J. Van Amerongen, Diffusion in elastomers, Rubber Chem. Technol., 37 (1964) 1065-l 152. [27]J. Crank and G.S. Park, Diffusion in Polymers, Academic Press, New York, 1968. [ 281 J. Crank, The Mathematics of Diffusion, Oxford University Press, Oxford, 1956.