poly(methyl methacrylate) blend gas separation membranes

poly(methyl methacrylate) blend gas separation membranes

Journal of Membrane Science, 94 (1994) 313-328 313 Elsevier Science B.V., Amsterdam Composite cellulose acetate/poly blend gas separation membranes...

4MB Sizes 0 Downloads 54 Views

Journal of Membrane Science, 94 (1994) 313-328

313

Elsevier Science B.V., Amsterdam

Composite cellulose acetate/poly blend gas separation membranes

(methyl methacrylate)

B. Bikson*v*,J.K. Nelson”, N. Muruganandamb BInnovative Membrane Systems, Inc. 189 Dean Street, Norwood, MA 02062, USA bPraxair, Inc., River Road, Bound Brook, NJ 08805, USA

(Received September 23,1993; accepted in revised form February 7,1994)

Abstract

The morphology of cellulose acetate/poly (methyl methacrylate), CA/PMMA, blends of various compositions was studied. Optical microscopy and DSC conclusively showed that the polymers form incompatible blends. However, partial miscibility between the phases was noted since the glass transition temperature of the PMMA phase increased with the increase in the amount of the CA fraction in the blend, while the glass transition temperature of CA decreased with the increase in the amount of the PMMA fraction in the blend. ATR-FTIR spectra and surface energy measurements showed that the CA/ PMMA film surfaces were comprised primarily of PMMA polymer. These results suggest that the CA/ PMMA blends form a layered morphology with surface layers comprised essentially of PMMA polymer and the interior layer comprised of phase-separated blend. He, 0, and Nz gas transport properties of CA, PMMA and CA/PMMA blends of several compositions have been measured at 35°C. The data were analyzed in terms of parallel and series resistance models. The high He/O, and He/N, gas separation factors were best described by a combination of the parallel and series models that is consistent with the proposed layered film morphology. A novel composite gas separation membrane was prepared by coating porous polysulfone hollow fibers with a solution of the CA/PMMA blend. CA/PMMA composite hollow fiber membranes with a broad range of coating thicknesses were prepared and their He and Nz gas transport properties were measured. The properties were compared to the gas transport properties of the CA and PMMA composite hollow fiber membranes of equivalent coating thicknesses. The CA/PMMA composite membranes exhibited a superior combination of He/N, gas separation factors and He permeation rates compared to the CA or PMMA composites. The superior properties were attributed to the unique layered morphology of the coating comprised of PMMA exterior layer and mostly CA interior layer. The formation of layered morphology was supported by surface energy measurements. The surface energy of the CA/PMMA composite hollow fiber membrane was measured and found to be essentially identical to that of the PMMA polymer. The multilayer coating morphology was observed directly under SEM examination of the hollow fiber cross sections. CA/PMMA composite membranes can be used to purify helium gas streams to produce a high purity product (less than 10 ppm contaminant level) with high helium recovery in a single-stage process. Key words: gas separations; membrane preparation

and structure; composite membranes; polymer blends

*Corresponding author.

0376-7388/94/$07.00

0 1994 Elsevier Science B.V. All rights reserved.

SSDZ 0376-7388(94)0047-3

314

1. Introduction Composite membranes are receiving increased attention in the gas separation membrane field [l-4]. They offer several advantages over the more widely utilized asymmetric membranes. The advantages include the ability to use unconventional, expensive materials as separation layers since only small amounts are required for membrane preparation and the ability to decouple gas permeation characteristics of the separation layer from membrane forming characteristics and load carrying requirements of the porous support. The present work deals with composite gas separation membranes prepared from polymer blends, specifically cellulose acetate/poly (methyl methacrylate), CA/PMMA, blend. In the literature, both miscible and immiscible polymer blends have been studied extensively as gas separation materials [ 5-141. In miscible blends, the gas transport properties are affected by the extent of polymer/polymer interactions [ 61. Miscible blends have been generally unattractive for gas separation due to the free volume contraction that typically occurs on mixing and results in decreased gas permeation rates. The homogeneous, miscible polymer blends that show a significant negative volume change on mixing will frequently show high gas separation factors but at a significant expense in gas permeation rates [ 71. The gas transport characteristics of immiscible blends are highly dependent on the specific blend morphology that in turn is affected by such factors as blend composition, method of sample preparation, etc. In extremes, the gas transport in such systems can be modeled by series resistance and parallel resistance transport models [ll]. Highly interconnected phases will lead to gas transport properties that can be modeled by some combinations of series and parallel resistances [lo].

B. B&son et al. /J. Membrane Sci. 94 (1994) 313-328

Some of the complex morphologies formed in immiscible polymer blends can lead to an attractive combination of gas separation factors and permeation rates [ 10,12,13]. The gas transport properties of PMMA and miscible and immiscible blends of PMMA and other polymers such as poly (vinylidene fluoride), bisphenol chloral polycarbonate, and styrene/ acrylonitrile copolymers have been reported in the literature [8-10,141. PMMA exhibits exceptionally high He/N,, He/CH, and He/O2 gas separation factors [8] that would make PMMA-based membranes most attractive for helium extraction and purification. Attempts to prepare ultrathin PMMA films for gas separation have been reported in the literature [ 151, preparation of composite gas separation membranes from PMMA or blends of PMMA with other polymers have been reported by one of the authors [ 161. This work provides details on the structure, morphology, and unique gas permeation characteristics of CA/PMMA blends and composite hollow fiber membranes prepared therefrom. 2. Experimental 2.1. Materials CA and PMMA used in this study were GLF84 (Hoechst-Celanese, MW 55,300) and Elvacite 2140 (DuPont, MW 350,000), respectively. The materials were used without further purification. Flat-sheet films of CA, PMMA and CA/PMMA blends were prepared by casting 10% (weight by volume) solution of polymers in glacial acidic acid (VWP technical grade) on glass plates and drawing with a doctor blade. Thick films (20 pm) were dried at 50’ C for one day, followed by drying at 90°C in a vacuum oven for at least a week prior to testing.

B. B&son et al. /J. Membrane Sci. 94 (1994) 313-328

2.2. Preparation of composite hollow fiber membranes

Preparation of composite CA/PMMA hollow fiber membranes is described elsewhere [ 161. In short, predriedporous polysulfone hollow fibers prepared by a dry/wet spinning process were coated with dilute polymer solutions. The hollow fibers were drawn through the coating solution at a speed of 15 m/min and the solvent evaporated in a dryer oven at temperatures ranging from 50 to 125” C in the final curing stages. The coating solutions of CA, PMMA and CA/PMMA blends were prepared by dissolving polymers in a solvent system comprised of glacial acetic acid/isopropyl alcohol (50/50 by volume ). The polymer concentrations in the coating solutions ranged from 0.2 to 1 g per 100 ml. The solutions were filtered through 1.5 pm filters prior to use. In this paper, only the properties of the CA/PMMA composite membrane prepared from 50/50 by weight blend composition are discussed. The membrane samples used for gas permeation measurements were further coated with poly (dimethylsiloxane ) by immersing the composite hollow fibers into 1% solution of Sylgard 184, Dow Corning, in hexane followed by drying [ 171. The coating was necessary to repair defects in the PMMA composite membranes that only on few occasions formed defect-free membranes. The composite CA and CA/PMMA membranes exhibited less than 5% improvement in the gas separation factor after coating with poly (dimethylsiloxane) . This would indicate that CA and CA/PMMA blends formed essentially defect-free composite membranes. 2.3. Gas permeation measurements

The permeation apparatus and procedures that were utilized to measure permeability of flat-sheet films were the same as described by

315

Koros et al. [ 181. All measurements were made at 35’ C using an upstream pressure of l-2 atm while the downstream pressure was effectively zero. Films were heated in situ in the permeation cell at 45 ’ C for several days in vacuum to remove residual sorbed water before actual measurements at 35 ’ C were performed. The permeances and separation factors for composite hollow fiber membranes were measured as follows: small hollow fiber modules were prepared for gas permeation measurements by potting about 20 fibers in an epoxy resin tube sheet. The He/N2, 10/90, gas mixture that was used as a feed gas was obtained from Praxair, Inc. All measurements were performed at 25”C, 1045 kPa, and a stage cut of less than 2%. The composition of feed and permeate gas was measured with a Carle Hach gas chromatograph, Series 400, using a thermal conductivity detector and the flows with Porter Instruments Co. mass flow meters. At least 10 hollow fiber modules were prepared from each membrane sample and tested under the abovespecified conditions. The separation factors and permeation rates for each sample were calculated and the averages are reported. Error bars (one standard deviation ) is usually indicated with the reported data. 2.4. Helium purification test Purification of helium was carried out utilizing a hollow fiber module of the design described elsewhere [ 191. Helium feed gas that contained N 107 ppm of oxygen was obtained from Praxair, Inc. The gas composition of feed, nonpermeate and permeate gas was measured utilizing a Gemini Plus (Meeco Inc.) oxygen analyzer and the flow rates were measured using Porter Instruments Co. mass flow meters. The data were analyzed using a countercurrent flow model with permeate and feed-side pressure drops neglected. All measurements were

316

B. Bikson et al. /J. Membrane Sci. 94 (1994) 313-328

performed at 25°C and a feed gas pressure of 697 kPa.

scribed above. Only second thermograms reported.

2.5. Microscopy

2.9. Coating thickness &termination

Scanning electron micrographs were obtained using an IS1 DS 130. Cross sections were obtained by freeze fracture in liquid nitrogen. The samples were coated with 100 A of platinum in preparation for SEM. At least five different cross sections were used to measure coating thicknesses for every sample and the averages are reported. The optical transmission micrographs were obtained using phase contrast microscopy at 100x magnification on an American Optical Co. microscope.

Coating thicknesses were determined from micrographs (SEM) of hollow fiber cross sections and by a gravimetric technique. The latter technique involved measuring the weight increase upon coating N 400 m of polysulfone hollow fiber. The coating weight was determined by subtracting the weight of the polysulfone hollow fiber from the weight of the coated sample of equal length. The coated samples were dried at 110°C to a constant weight prior to weight measurements. The coating thickness was determined by dividing the volume of the coating by the total coated membrane area (circumference x length). The former was determined from the total weight of the coating utilizing an average coating layer density of 1.22 g/cm3. The coating thicknesses determined by two methods were generally within 30% and the averages of the two methods are reported.

2.6. ATR-FTIR

Attenuated reflectance Fourier transform infrared spectra were obtained on a Bio Rad FTS-40 Digilab. A zinc-selenium crystal at a 45 ’ angle was used for the ATR measurements. 2.7. Surface energy measurement

3. Results and discussion

Dynamic contact angle data were obtained on a Cahn DCA322. The surface energy was calculated from the advancing contact angle data of probe liquids (deionized water, methyl iodide and ethylene glycol) using the method of Kaelble [ 201. At least three measurements were made for each sample and the averages are reported.

3.1. Morphology and gas permeation characteristics of CA/PMMA blend

2.8. Differential scanning calorimetry

The measurements were performed on a DuPont 1090 Thermal Analyzer at a heating rate of lO”C/min. Five mg-size samples used for DSC measurements were cut from CA, PMMA and CA/PMMA films prepared as de-

are

We have studied the morphology and gas permeation characteristics of CA/PMMA blends. Fig. 1 shows the morphology of a 20 p thick CA/PMMA film of 50/50 blend composition as observed under an optical microscope. The phase separation is clearly visible and is consistent with the incompatibility of CA and PMMA reported in the literature [21]. The DSC measurements further confirmed the incompatible nature of CA/PMMA blends. In Fig. 2, DSC scans of several CA/PMMA blend compositions are shown together with DSC scans of PMMA and CA polymers. Two transitions

B. Bikson et al. /J. Membrane Sci. 94 (1994) 313-328

317

Fig. 1. Optical micrograph of CA/PMMA blend (50/50 composition)

at x 100 magnification.

TABLE 1 Glass transition temperature of CA, PMMA and blends of CA/PMMA CA/PMMA (ratio by weight)

TB (“C) PMMA phase

20

60

100

140

160

220

o/100 25115 50150 15125 100/o

110 114 120 124

CA phase

176 181 180 193

Temperature (” C)

Fig. 2. Differential scanning calorimetry thermograms of CA, PMMA and CA/PMMA (50150 composition) blends.

are observed for CA/PMMA blends consistent with the presence of two phases. The measured glass transition temperatures are further summarized in Table 1. It is interesting to note that a consistent shift in the Tg of PMMA phase is

observed with the increase in CA fraction. Conversely, a decrease in T, of the CA phase is observed with the increase in PMMA fraction. It is possible that the shift in the glass transition temperatures of the two phases is due to partial mixing between the polymers. The low molecular weight fraction of CA may be miscible with PMMA and low molecular weight fractions of

B. Bikson et al. /J. Membrane Sci. 94 (1994) 313-328

318

PMMA may be miscible with CA leading to shifts in TB’It should be noted that both polymers used in this study are highly polydispersed. Alternatively, the observed properties may be the result of limited miscibility in volume ratio. The gas permeation characteristics of CA and PMMA polymers and of CA/PMMA blends are summarized in Table 2. Figs. 3-5 show gas permeability coefficients P for He, O2 and N, versus volume fraction of CA in the blend. In addition to experimental data, the calculated gas permeability coefficients based on limiting simple series and parallel laminate models are also presented. Following Robeson et al. [ 111, the limiting series and parallel resistance models for laminates comprised of two phases designated by subscripts 1 and 2 can be expressed as follows: p, = p,pJ ( q5,P2+ q&PI ) for series laminate resistance, and I’, =p,$, + pZq& for parallel laminate resistance, where p, is the permeability coefficient of the blend sample, and I’, and p2 the permeability coefficients of the respective phases, and @Iand q&their volume fractions. As can be seen, the measured permeability coefficients are substantially lower than would be expected by the parallel resistance model and close to the values predicted by the series resistance model. Figs. 6 and 7 show He/O, and He/N2 ideal gas separation factors versus blend compositions. The mea-

sured ideal gas separation factors are significantly higher than predicted by the parallel resistance model and closer to but lower than predicted by the series resistance model. The gas permeation characteristics of CA/ PMMA blends, specifically the high He/N2 and He/O, gas separation factors suggest significant contribution of series resistance to the overall transport mode and thus formation of some form of layered morphology. To investigate the possible formation of layered morphology in CA/PMMA films, ATR-FTIR spectroscopy was employed. Fig. 8 shows the attenuated reflectance IR spectrum of the 50/ 50 CA/PMMA blend film as well as the spectra of the PMMA and CA films. It is clear that the ATR-FTIR spectrum of the blend and thus the surface composition of the CA/PMMA film is essentially identical to that of PMMA. Only a minor contribution of CA to the surface spectra of the CA/PMMA blend is noted. For example, the shoulder at 1033 cm-‘, pointed to with an arrow in Fig. 8, can be assigned to the CA contribution. The ATR-FTIR spectrum of the CA/ PMMA film in Fig. 8 is that of the polymer film/ air interface; however, it should be pointed out that the spectrum of the CA/PMMA film surface from the film/glass plate interface was essentially identical to the spectrum of the film/ air interface. The ATR-FTIR spectra not only indicate that the film surface is comprised es-

TABLE 2 Permeation data for CA, PMMA and CA/PMMA blends at 35°C (flat films) PMMA (ti%)

p”

(Y

He

0,

N*

He/&

He/G

0 25 50 75 100

16.0 13.2 11.0 9.5 9.1

0.78 0.42 0.21 0.14 0.11

0.139 0.069 0.029 0.018 0.014

115 191 379 516 640

21 31 52 68 81

“Permeability units: lo-”

cm3( STP) cm/cm’ cmHg s.

B. B&son et al. /J. Membrane Sci. 94 (1994) 313-328

319 1

.o

I’

/I

.’

,’

.’

.’ I’

.’

.’

.’

.’ .’

,’

. /

.’

/’

.’

.

.’ F

J

I

0.0

0.5

0.0 PMMA

Volume

Fraction,

CA

CA PMMA

Fig. 3. Helium permeability as a function of CA/PMMA blend composition.

z”

0.10

> C =

0.08

8;

0.06

#

I 0.0 PMMA

1

0.5 Volume

t

,’

I’

.’

.’

1

Fraction,

1

Fraction.

CA

CA

CA

Fig. 5. O2 permeability as a function of CA/PMMA blend

I 1 .o

CA

Volume

1 .o

1

0.0 PMMA

0.5 Volume

Fraction,

1 .o CA

CA

Fig. 4. N2 permeability as a function of CA/PMMA blend composition.

Fig. 6. He/N, separation factor as a function of CA/PMMA blend composition.

sentially of PMMA polymer but that this PMMA polymer-rich layer extends up to 2 pm depth. The ATR-FTIR data were collected utilizing a zinc-selenium crystal at a 45 ’ angle (see section 2.6) and span wavenumbers in the 2000 to 800 cm-’ range. Under this experimental condition, the spectra reflect surface composition of up to 2 pm in depth [ 221. The surface characteristics of CA/PMMA blends were further studied utilizing dynamic contact angle measurements. The measured surface energy of the 50/50 CA/PMMA blend

and the measured surface energies of CA and PMMA films are summarized in Table 3. The total surface energy ySof the CA/PMMA film is closer to the surface energy of the PMMA film and is substantially lower than that of the CA film. This finding is consistent with the ATR-FTIR data, which show that the surface of the CA/PMMA film is comprised essentially of the PMMA polymer phase only. It should be noted, however, that the J$ and ydvalues of the CA/PMMA film differ somewhat from the ri, and yd values of the PMMA film. yp and yd are

B. Bikson eb al. / J. Membrane Sci. 94 (1994) 313-328

320

TABLE 3 Surface energy of PMMA, CA and CA/PMMA blend flat-sheet films Polymer

PMMA CA CA/PMMA

,ol

I

I

PMMA

1

Volume

1600

(dyne/cm

30.4 29.5 28.9

4.7 10.6 7.5

Y. )

(dyne/cm)

35.1 40.1 36.4

1 .o

Fraction,

CA

CA

Fig. 7. He/O2 separation factor as a function of CA/PMMA blend composition.

2000

&I

(dyne/cm)

yd= surface tension dispersive force component; yr,= surface tension polar force component; y. = total surface energy.

J

I

!

0.5

0.0

yd

(50/50)

1200

Wavenumbers

800

(cm-‘)

Fig. 8. ATR-FTIR spectra of CA, PMMA and CA/PMMA (50/50 composition) films.

the polar and the dispersive surface energy components of the total surface energy Ye. The surface composition of the CA/PMMA blend suggest the formation of layered morphology. At the same time, the films are opaque as shown in Fig. 1 and clearly show interdispersion of CA and PMMA phases. In combination, these observations suggest a layered film morphology for the CA/PMMA blend com-

prised of exterior PMMA layers and an intermediate phase-separated CA/PMMA layer. The proposed CA/PMMA film morphology is shown schematically in Fig. 9. The exterior layers, e.g., the film/air interface layer and the film/glass plate interface layer, were shown by ATR-FTIR to be identical in composition and comprised essentially of PMMA polymer. The layers in Fig. 9 are not drawn to scale and the intermediate phase-separated layer may represent a major part of the overall thickness particularly for 20 ,um thick films. The film morphology shown in Fig. 9 is consistent with the gas permeation results (Figs. 3-7). As previously discussed, these results show that the measured gas separation characteristics are consistent with a combination of series and parallel resistance models. The morphology delineated in Fig. 9 should be modeled by a combination of series and parallel resistances. Such model was developed by Chiou and Paul [lo] based on the isotropic model by Kraus and Rollmann for polymer mixtures [ 231. However, since the thicknesses of the laminate layers in the present case were not known as well as the precise composition and the gas permeation characteristics of the phases, such detailed modeling was not possible. The relative contributions to series and parallel resistance modes by layers in Fig. 9 will depend, among other factors, on layer thicknesses. The thickness of the exterior PMMA

B. Bikson et al. /J. Membrane Sci. 94 (1994) 313-328

321

Fig. 9. Model of layered morphology of CA/PMMA film. (m) MMA phase, (0 ) CA phase.

layer may be independent of the film thickness and thus provide a more significant contribution to the overall gas transport in thin films. We have not studied the effect of film thickness on the gas transport characteristics of the CA/ PMMA blends. The film morphology including layer thicknesses can be further potentially affected by kinetic factors such as solvent evaporation rates. As was discussed previously, some mixing between the CA and PMMA polymers does take place as indicated by the DSC measurements. This mixing could lead to some collapse of the free volume in the polymer phases and to a decrease in the gas permeation coefficients. The low gas permeation coefficients of the CA/PMMA blends, in particular, the helium permeability coefficient, may indeed be a consequence of such free volume collapse. 3.2. Gas permeation characteristics of composite membranes

In Fig. 10, the He/N, gas separation factors of the composite CA, PMMA, and CA/PMMA hollow fiber membranes are shown as a function of the coating solution solids content. Error bars have been indicated for each case. In addition to coating solution solids content, the corresponding measured coating layer thicknesses are also indicated. As can be seen from the data in Fig. 10, the composite PMMA membranes exhibited poor He/N2 gas separation characteristics, e.g., substantially below the intrinsic separation factor of 700-800 reported for

Composite 500

I

350

1

300

$-

250

IF

i3

200 1

s = m g lE

150. ioo.I

Layer Thickness

(A) 1600

600

I

ICA

I

50

I

PMMA

0 0.2

04 Coating Solutmn

0.6 Wads Content (wt.%)

Fig. 10. Selectivity of composite CA, PMMA, and CA/ PMMA hollow fiber membranes as a function of coating solution solids content.

the PMMA polymer [ 8). The separation characteristics were poor throughout the entire range of the coating solution solids content, i.e. the coating thicknesses studied. Occasionally, it was possible to form PMMA composite membranes with separation factors as high as 400. This fact supports the notion that the poor separation characteristics of PMMA membranes can be attributed to defects formed in the coating layer that could not be repaired by poly (dimethylsiloxane ) coating. The composite CA hollow fiber membranes exhibited He/ N2 separation factors from 90 to 110 which are slightly below but close to the intrinsic He/N2 separation factor of the CA polymer. Of particular interest are He/N, separation factors exhibited by the CA/PMMA composite hollow fiber membranes. Separation factors

B. Bikson et al. /J. Membrane Sci. 94 (1994) 313-328

322

above 200 were attained in the range of the coating solution solids content studied. The high gas separation factors exhibited by the CA/ PMMA composite membranes become even more attractive when the helium permeation rates (permeances) of these membranes are compared to the permeation rates of the CA composites. In Fig. 11, the ratio of helium permeation rates (permeability coefficient/thickness) for the CA/PMMA versus CA composites is plotted as a function of the coating solution solids content. Again, error bars have been indicated for each case. The results indicate that the CA/PMMA composite membranes of equivalent thicknesses exhibit only marginally lower helium permeation rates than the CA composite membranes. Thus more than a two-fold increase in He/N, gas separation factors is attained for the CA/PMMA composite membranes as compared to the CA composite membranes with only a modest decrease in helium permeation rate. The separation factors exhibited by the CA/PMMA composite membranes though higher than that of the CA composite were significantly lower than measured for CA/PMMA flat-sheet films. A number of factors could account for this discrep-

I/(P/t)composite

Composite Layer Thickness (A)

500

0.2

SO0

0.4

1600

0.6

ancy: the presence of defects in the separation layer of composite membranes, significant differences in coating layer morphology as compared to flat-sheet films, or significant contribution of polysulfone substrate to the overall helium transport. It is known that the intrinsic selectivity of composite membranes can be limited by the resistance of the porous support layer when the coating layer is thin and the material of the support layer has a low instrinic gas separation factor [ 31. The resistance model [ 241 is frequently used to elucidate the respective contribution of the coating layer and the substrate layer to the overall gas transport. We thought it may be instructive to conduct such analyses for a case where the substrate resistance was expected to be most significant; namely, for the case of a 500 A thick coating layer. We measured a helium permeability of 21.0 x 10m5cm3 (STP) /cm” cmHg s for the 500 A thick CA/PMMA composite membrane at 35°C. If one assumes an extreme case of low surface porosity and complete occlusion of surface pores by coating material, the overall resistance can be presented as the sum of the coating layer resistance and the substrate layer resistance:

0.6

Coating Solution Solids Content (wt%)

Fig. 11. Ratio of helium permeation rates of CA/PMMA and CA composite hollow fiber membranes as a function of coating solution solids content.

l/ (P/t)coating+

= I/ (P/t)substrate

The coating layer resistance to helium permeability is 4545 cm2 cmHg s/cm”(STP) (the helium permeability of the CA/PMMA blend is 11.0 barrers and the coating thickness is 500 A). The overall resistance for the composite is 4762 cm2 cmHg s/cm” (STP), by difference the resistance of the polysulfone substrate is calculated to be 217 cm2 cmHg s/ cm3 (STP) . Namely, the substrate represents only 5% of the overall resistance to the helium transport. The analysis indicates that the assumption of complete surface pore occlusion by the coating material cannot be valid. To satisfy

B. Bikson et al. /J. Membrane Sci. 94 (1994) 313-328

the low value of substrate resistance, the skin thickness of the polysulfone substrate has to be only 28 A thick (the helium permeability coefficient for polysulfone at 35” C is 13.0 barrers ). The skin thickness of the polysulfone substrate is close to 500 A (see Figs. 12-15). By implication the coating layer is substantially nonocclusive in nature or the resistances attributed to the coating and substrate layer are incorrect. Some combination of both is most plausible. Applicability of the resistance model for the present case analyses may be limited since the intrinsic separation characteristics of very thin up to 500 A coating layer can differ from that measured for thick films. The morphology of phase-separated ultrathin CA/ PMMA films can differ from that of a thick film, and the gas permeation characteristics in thin separation layers are known to frequently deviate from the permeation characteristics of polymeric materials in isotropic thick films

323 [ 251.

Combined with substantial uncertainty in coating thickness measurements and limited information on the extent of coating occlusion into surface pores, the value of the resistance analyses may be limited. Nevertheless, it can be safely stated that the contribution of the polysulfone substrate resistance to the helium transport in the present case is limited and cannot account for the observed permeation/ separation characteristics of the composite CA/ PMMA membranes. We have characterized the CA/PMMA and CA composite membranes as essentially defect free, since there was little improvement in gas separation factors after coating with poly (dimethylsiloxane) rubber. However, we cannot rule out the possibility that some defects do exist that were not repaired by the coating. The separation factors exhibited by the CA/PMMA composites were only modestly affected by the coating layer thicknesses. Some decreases in separation factors observed

Fig. 12. Cross eection of CA composite hollow fiber membrane (0.4% coating solution solids content).

B. B&son et al. /J. Membrane Sci. 94 (1994) 313-328

324

Fig. 13. Cross section of CA composite hollow fiber membrane (0.8% coating solution solids content).

at the lower range of coating thicknesses studied may be attributed to difficulty in forming exceptionally thin defect-free coating layers or to a minor contribution of the porous support structure to the overall gas transport. Substrate contribution to the overall gas transport will become more noticeable in the range of low coating thicknesses. The morphology of the composite CA/ PMMA hollow fiber membranes was studied to determine if the coating layer formed a layered morphology analogous to that formed in the flat-sheet CA/PMMA films. The surface energy of the CA/PMMA composite hollow fiber membrane was measured and compared to the measured surface energies of the CA and PMMA composite. The values of the total surface energy, Ye, and the polar and dispersive components, )$ and &, are listed in Table 4. The values of the total surface energy as well as the values of yd and rp of the CA/PMMA, CA and

PMMA composite membranes are consistent with the 1$, yd and &, values of the flat-sheet films of the respective polymers as summarized in Table 3. However, in the case of composite membranes, the surface energy of the CA/ PMMA composite membrane is even somewhat lower than that of the PMMA composite membrane. In fact, it is the surface energy value of the PMMA composite that is high as compared to the surface energy value of the PMMA film. As pointed out previously, the PMMA composite membranes do not form defect-free coatings. The uncoated or exposed areas of the polysulfone substrate would lead to an increase in the measured surface energy since the surface energy of polysulfone is significantly higher than that of the PMMA polymer. As in the case of the flat-sheet films, the y8of the CA/PMMA composite is very close to the l/s value of the PMMA polymer. This result suggests that the surface layer of the CA/PMMA composite hol-

B. Bihon et al. /J. Membrane Sci. 94 (1994) 313-328

325

Fig. 14. Cross section of CA/PMMA composite hollow fiber membrane (0.2% coating solution solids content).

low fiber membrane, in analogy with the flatsheet CA/PMMA films, is comprised almost entirely of PMMA polymer. Because the coating layers of the composite membranes are in the 500 to 1500 A range, the coating layers can be viewed as a surface. This surface is comprised of a PMMA-rich phase (several hundred angstroms thick exterior layer) and a PMMAlean phase (several hundred angstroms thick interior layer). As a consequence, a sharp concentration gradient between the layers must be expected leading to a layered morphology. SEM examination of cross sections of the composite CA/PMMA hollow fiber membranes directly confirmed the formation of the layered morphology in the coating. We have examined and compared the coating morphology of the CA, PMMA, and CA/PMMA composite membranes by SEM. Samples used in this study did not contain poly (dimethylsiloxane) coating to simplify the analyses. All samples studied were prepared by freeze fracture. It was

found that during freeze fracture of the CA/ PMMA hollow fiber samples delamination in the coating layer occurred revealing the presence of a layered morphology. Namely, the formation of two distinct layers was frequently observed on examination of the CA/PMMA composite hollow fiber cross sections. Only one layer was found on examination of the cross sections of the CA or PMMA composite hollow fiber membranes. The cross sections of two CA composite hollow fiber membranes of different coating thickness are shown in Figs. 12 and 13. The CA coating layer is clearly visible as it is superimposed over the porous polysulfone support structure. The cross sections of the CA/ PMMA composite hollow fiber membranes are shown in Figs. 14 and 15. The layered morphology of the CA/PMMA composite is visible. The layers have delaminated during freeze fracture and, in the case of the sample in Fig. 15, are partially folded (see the upper section of the picture).

B. B&son et al. /J. Membrane Sci. 94 (1994) 313-328

326

Fig. 15. Cross section of CA/PMMA

composite hollow fiber membrane (0.4% coating solution solids content).

TABLE 4 Surface energy of PMMA, CA and CA/PMMA composite hollow fiber membranes Polymer

PMMA CA CA/PMMA

(50/50)

yd

YP

Ys

(dyne/cm)

(dyne/cm)

(dyne/cm

31.4 31.8 31.5

5.6 9.6 4.6

37.0 41.4 36.3

)

The multilayer morphology of the CA/ PMMA composite membranes is the primary reason for the high gas separation characteristics of this membrane (see Fig. 10). The layered morphology leads to a substantial series resistance contribution to the overall transport mode and thus to a beneficial combination of separation characteristics. The nature of the interface between the CA/PMMA coating and the polysulfone hollow fiber could not be stud-

ied in any detail. Thus it is not clear if the layered morphology has additional features at the coating hollow fiber interface. The high gas separation factors of the CA/PMMA composite membranes attributed to the unique layered morphology are of significant interest since they are combined with a relatively high fast gas permeation rate as compared to the CA composites. The fast gas permeation rates of the CA/PMMA composites in the 500-800 A thickness range are somewhat high as compared to the respective permeation rates of the CA composites when considering the differences in the helium permeability coefficients for these materials and the relative equivalency of the coating thicknesses of the CA/PMMA and CA composites prepared under equivalent experimental conditions. Somewhat higher than expected helium permeation rates may be related to the fact that the separation factors of the CA/PMMA composites are lower than the

B. B&on

et al. /J. Membrane Sci. 94 (1994) 313-328

respective flat-sheet values. The potential reasons for the differences in the separation factors were discussed above. In addition, it is possible that the unique features of the coating substrate interface such as the extent of coating occlusion into surface pores contribute to the observed gas permeation rates (in the case of the CA composites, a higher degree of occlusion can lead to a reduced permeate rate). Furthermore, it should be pointed out that the uncertainty in the coating layer thicknesses measured is large and the differences in the coating thicknesses can thus easily account for the observed permeation rates. The formation of the laminate structure in immiscible CA/PMMA blends and in composite CA/PMMA membranes is favored on thermodynamic grounds. The PMMA polymer has substantially lower surface tension than the CA polymer. The enrichment of the surface with PMMA polymer would thus lower the overall surface energy of the system and would be favored in thermodynamic terms. The proposed mechanism would also explain the formation of the multilayer morphology in the CA/PMMA composite membranes that was accomplished in a single coating step. It was reported by Chiou and Paul [lo] that the gas transport behavior of the phase-separated PMMA/SAN blend is well described by a parallel-series or seriesparallel resistance model. The low surface tension of the PMMA polymer as compared to the SAN polymer could lead to surface enrichment with PMMA polymer and to a rise in series resistance in analogy with the CA/PMMA blends. 3.3. Helium purification test

The high He/N, and He/O, gas separation factors of the CA/PMMA composite hollow tiber membranes can be utilized to purify helium from air contaminants with high helium recovery. The advantageous separation features of the CA/PMMA membrane were demonstrated

327 107 ppm 0,

I

10ppm02

Balance He

Fig. 16. Separation performance of CA/PMMA composite hollow fiber membrane module.

by conducting purification of a helium gas stream that contain N 100 ppm of oxygen impurity. The objective was to produce a helium product stream with an impurities content of 10 ppm or less and with a high helium recovery rate. As can be seen from the results summarized in Fig. 16, a helium product stream with an impurity content of 10 ppm was produced combined with 90% helium recovery. The separation results exhibited by the CA/PMMA membrane module correspond to a He/O, separation factor of 32. The result demonstrates the utility of gas separation membranes with high separation factors not only for bulk gas separations but also for the production of high purity gases. 4. Conclusions

CA/PMMA blends exhibit phase separation with limited intermiscibility between the two polymers. The CA/PMMA flat-sheet films exhibit complex morphology comprised of an exterior PMMA surface layer and a phase-separated CA/PMMA blend as an interior layer. The gas separation characteristics of the CA/ PMMA blends can be best described by a combination of series and parallel resistance models. It has been shown that a multilayered composite CA/PMMA gas separation membrane comprised of an exterior PMMA layer and mostly CA interior layer can be formed in a single-step coating process. The high He/N, and

328

He/O2 gas separation factors exhibited by the CA/PMMA composite hollow fiber membrane can be, to a large extent, attributed to the layered morphology. The CA/PMMA composite membranes can be successfully used to produce high purity helium gas streams combined with high helium recovery. Acknowledgements The authors to thank Dr. Y. Yuan from SUNY, Syracuse, NY for performing surface energy measurements and for very valuable discussions; Dr. Y. Ozcayir for performing some gas permeation measurements on polymer blends, and S. Giglia for performing helium purification experiments. References [ 11K.A. Lundy and I. Cabasso, Analysis and construction of multilayer composite membranes for the separation of gas mixtures, Ind. Eng. Chem. Res., 28 (1989) 742. [2] W.J. Ward III, N.R. Browall and R.M. Salemme, Ultrathin silicon/polycarbonate membranes for gas separation, J. Membrane Sci., 1 (1976) 99. [ 31 I. Pinnau, J.G. Wijmans, I. Blume, T. Kuroda and K.V Peinemann, Gas permeation through composite membranes, J. Membrane Sci., 37, (1988) 81. [4] S.C. Williams, B. Bikson, J.K. Nelson and R.D. Burchesky, Method for preparing composite membranes for enhanced gas separation, US Pat. 4,840,819 (1989). [5] H.B. Hopfenberg and D.R. Paul, Transport phenomenon in polymer blends, in D.R. Paul and S. Newman (Eds.), Polymer Blends, Academic Press, New York, 1978, p. 445. [ 61 D.R. Paul, Gas transport in homogeneous multicomponent polymers, J. Membrane Sci., 18 (1984) 75. [7]R.L. Stalling, H.B. Hopfenberg and V. Stannet, Transport of fixed gases in blends of polystyrene and poly(phenylene oxide), J. Polym. Sci., Symp. Ed., 41 (1973) 23. [8]K.E. Min and D.R. Paul, Effect of tacticity on permeation properties of poly (methyl methacrylate ) , J. Polym Sci., Part B, Polym. Phys., 26 (1988) 1021. [9] J.S. Chiou and D.R. Paul, Sorption and transport of inert gases in PVF,/PMMA blends, J. Appl. Polym, Sci., 32 (1986) 4793.

B. B&son et al. 1 J. Membrane Sci. 94 (1994) 313-328 [lo] J.S. Chiou and D.R. Paul, Gas permeation

in miscible homopolymer-copolymer blends. I. Poly(methy1 methacrylate) and styrene/acrylonitrile copolymers, J. Appl. Polym. Sci., 34 (1987) 1037. [ 111 L.M. Robeson, A. Noshay, M. Matzner and C.N. Merriam, Physical property characteristics of polysulfone/poly(dimethylsiloxane) block copolymers, Angew. Makromol Chem., 29-30 (1973) 47. [ 12lM.S. Lerma, K. Iwamoto and M. Seno, Structure and of poly (vinyl chloride) / gas permeabilities oligo(dimethylsiloxane) blend membranes, J. Appl. Polym. Sci., 33 (1987) 625. [ 131 J.Y. Lai and G.J. Wu, TPX/siloxane blend membrane for oxygen enrichment, J. Appl. Polym. Sci., 34 (1987) 559. [ 141 J.S. Chiou and D.R. Paul, Gas permeation in miscible blends of poly (methyl methacrylate ) with bisphenol chloral polycarbonate, J. Appl. Polym. Sci., 33 (1987) 2935. [ 15lR.S. Kohn, Process for preparing ultrathin poly(methy1 methacrylate) polymer films, US Pat. 4,765,939 ( 1988 ) . [ 161 J.K. Nelson, Composite membranes of poly(methy1 methacrylate) blends, their manufacture and their use, US Pat. 4,919,865 (1990). [ 17]B. Bikson, S. Giglia and G. Karas, Composite separation membranes and the preparation anduse thereof, US Pat. 4,767,422 (1988). [ 181 W.J. Koros, D.R. Paul and A.A. Rocha, Carbon dioxide sorption and transport in polycarbonate, J. Polym. Sci., Part B, Polym. Phys. Ed., 14 (1976) 687. [ 191B. Bikson and S. Giglia, Method for gas separation using helically wound hollow fibers permeable membrane cartridge, US Pat. 4,881,955 (1989). [ 201 D.H. Kaelble, Physical Chemistry of Adhesion, WileyInterscience, New York, 1971, Chaps. 5,13. [21]R.J. Petersen, R.D. Cornelicessen and L.T. Rozelle, Polymer compatibility, Polym. Prepr., Am. Chem. Sot., Div. Polym. Chem., 10 (1969) 385. [22]A.L. Smith, Applied infrared spectroscopy, in P.J. Elving, J.D. Winefordner and I.M. Kolthoff (Eds.), Chemical Analysis, Vol. 54, Wiley, New York, 1979, pp. 84-94. [23]G. Kraus and K.W. Rollmann, Dynamic and stressoptical properties of polyblends of butadiene-styrene copolymers differing in composition, Adv. Chem. Ser., 99 (1971) 189. [24] J.M.S. Henis and M.K. Tripodi, Composite hollow fiber membranes for gas separation: the resistance model approach, J. Membrane Sci., 8 (1981) 233. [ 251 P.H. Pfromm, I. Pennau and W.J. Koros, Gas transport through integral-asymmetric membranes: a comparison to isotropic film transport properties, J. Appl. Polym. Sci., 48 (1993) 2161.