Membrane Formation and Modification - American Chemical Society

Chapter 5

Influence of Surface Skin Layer of Asymmetric Polyimide Membrane on Gas Permselectivity Hiroyoshi Kawakami and Shoji Nagaoka

Downloaded by MIT on June 18, 2013 | http://pubs.acs.org Publication Date: December 29, 1999 | doi: 10.1021/bk-2000-0744.ch005

Department of Applied Chemistry, Tokyo Metropolitan University, Hachioji, Tokyo 192-0397, Japan

Asymmetric 6FDA-APPS polyimide membranes were made by a dry/wet phase inversion technique. The membranes had high gas selectivity without any additional coating of the skin layer. The apparent skin layer thickness depended strongly on evaporation time, which was of great importance for the formation of an asymmetric membrane with high gas permeance. The surface roughness of the membranes, analyzed from atomic force microscopy (AFM) micrographs, increased with decreasing the apparent skin layer thickness. The permeances of the asymmetric membranes for CO , O , N , and CH were measured at 35°C and 76 cmHg. The asymmetric membranes exhibited gas selectivities equal to or even higher than those determined for dense, isotropic 6FDA-APPS membranes. 2

2

2

4

Recently, aromatic polyimides have been identified as membrane materials with high gas selectivity for gas pairs such as H / N , 0 / N , and C 0 / C H (1-3). There have been many studies of novel, structurally modified polyimides to enhance gas permeance and selectivity. However, the permeance of most polyimides is low; therefore, it is highly desirable to develop a polyimide membrane that provides higher gas permeance. The most important factor for enhancing the gas permeance is to make the membrane as thin as possible without introducing defects or pinholes which will reduce the selectivity significantly. To minimize thickness, an asymmetric membrane is preferred with a defect-free ultrathin skin layer and a porous substructure to support the skin layer during highpressure operation (4,5). The phase separation method is the most widely used technique for fabricating asymmetric membranes, and such membranes are prepared by controlled phase separation of polymer solutions into two phases with high and low polymer concentrations (6,7). In recent years, Pinnau and Koros reported that 2

2

2

2

2

4

© 2000 American Chemical Society

In Membrane Formation and Modification; Pinnau, I., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1999.

79

80 defect-free asymmetric membranes with skin layer thicknesses as thin as 20 nm can be formed by a dry/wet phase inversion process and that phase separation is induced in the outermost region of the cast membrane during the evaporation step (8,9). In this method, an evaporation step prior to immersion in the coagulation medium is necessary for the formation of a defect-free skin layer. In this paper, we describe the effect of the surface skin layer of an asymmetric polyimide membrane on its gas transport properties. A number of asymmetric membranes were prepared by dry/wet phase inversion, and the permeances of the membranes to C 0 , 0 , N , and C H were determined at 35°C and 76 cmHg. To investigate the influence of the evaporation kinetics on the surface skin layer properties, a series of asymmetric polyimide membranes were prepared using different evaporation times. Downloaded by MIT on June 18, 2013 | http://pubs.acs.org Publication Date: December 29, 1999 | doi: 10.1021/bk-2000-0744.ch005

2

2

2

4

Experimental Preparation of Asymmetric Membranes. 6FDA-APPS was synthesized from 2,2'-bis(3,4-diearboxyphenyl)hexafluoropropane dianhydride (6FDA) and bis[4-(4aminophenoxy)phenyl]sulfone (APPS) (Figure 1). Asymmetric 6FDA-APPS membranes were made by a dry/wet phase inversion technique (10,11). The composition of the casting solution used for preparation of asymmetric membranes was 12 wt% polyimide, 55 wt% methylene chloride, 23 wt% 1,1,2-trichloroethane, and 10 wt% butanol. High volatility methylene chloride and low volatility 1,1,2trichloroethane were good solvents for 6FDA-APPS and butanol was used as a nonsolvent additive. The polymer solutions were filtered and subsequently degassed. The solutions were cast on glass plates with a knife gap of 250 μπι and then air-dried for 15-600 sec. After evaporation, the membranes were coagulated in methanol, washed in methanol for 12 hours, air-dried for 24 hours, and finally dried in a vacuum oven at 150°C for 15 hours to remove any remaining solvent. Structure of Asymmetric Membrane. The cross-sections of the asymmetric membranes were observed with a scanning electron microscope (SEM: JXP-6100P, JEOL). The membranes were cryogenically fractured in liquid nitrogen and then coated with Pt/Pd. The surface morphology of asymmetric membranes was visualized using an atomic force microscope (AFM: Seiko SPI3700) in air at room temperature. The cantilevers (Seiko SN-AF01), with a spring constant of 0.38 N/m, were microfabricatedfrom silicon nitride. The surface Was continuously imaged in the feedback mode with a scan area of500 nm χ 500 nm and a constant scan speed of 2 Hz. The surface roughness (R ) of the membranes was characterized by the average vertical peak to valley distance. z

Permeability Measurements. The gases used in this study (carbon dioxide, oxygen, nitrogen, and methane) had a purity of at least 99.99% and were used without further purification.



81

_ Ο _

Κ>θΚ>οΟ ο Figure 1.

Structure of 6FDA-APPS polyimide.


82 Asymmetric 6FDA-APPS membranes were mounted in a permeation cell with a surface area of 1.0 cm . Gas permeances at 76 cmHg were determined with a high vacuum apparatus (Rika Seikilnc, K-315-H). The pressures on the upstream and downstream sides were measured using a Baratron absolute pressure gauge. The error in the permeance was estimated to be ±0.1 to 0.5%. The apparent skin layer thickness of a defect-free asymmetric membrane was calculated from the known permeability coefficient and permeance using 2

L = P/Q

where L is the apparent skin layer thickness, Ρ is the gas permeability coefficient as determined for a dense, isotropic film of known thickness (cm (STP)*cm/cm *s*cmHg), and Q is the gas permeance of the asymmetric membrane (cm (STP)/cm «s»cmHg). 3


(1)

2

3

2

Results and Discussion Thermogravimetric analysis (TGA) was used to test the membranes for residual solvent. The membranes showed no weight loss below 450°C in nitrogen. This result indicates that the membrane preparation protocol completely removed all solvent. A representative scanning electron micrograph of the structure of an asymmetric 6FDA-APPS polyimide membrane prepared after an evaporation time of 15 sec is shown in Figure 2. The photomicrograph shows that a membrane prepared by the dry/wet phase inversion process consisted of a thin skin layer and a porous substructure. The apparent skin layer thickness of the asymmetric membrane was calculated from equation 1. The thickness was determined based on the oxygen permeability coefficient of a dense 6FDA-APPS polyimide film. The thickness of membranes coagulated after an evaporation time of 15-600 sec increased with an increase in evaporation time. Pinnau et al. proposed a mechanism for the formation of the surface skin layers of asymmetric membranes made by dry/wet phase inversion (12). It was suggested that the outermost region of the membrane undergoes phase separation by spinodal decomposition during the initial stages of the evaporation process. Furthermore, the evaporation step prior to immersion of the membrane in the nonsolvent strongly influences the surface skin layer thickness and is also necessary for the formation of the defect-free skin (12). Differences in the morphology were evaluated by the surface roughness (R ), which is the mean difference between the ten highest peaks and the ten lowest valleys as measured by A F M . Figure 3 shows the relationship between the apparent skin layer thickness and R for asymmetric membranes prepared by dry/wet phase inversion. R decreased as the skin layer thickness increased. Interestingly, the surface of a dense membrane showed a uniform nodular structure (R = 0.89 nm). The average nodule size and surface morphology of the dense membrane were similar to that of the asymmetric membranes coagulated after z

z

z

z



83

Figure 2. Cross-section of an asymmetric 6FDA-APPS membrane made by dry/wet phase inversion. Evaporation time: 15 sec.

0

1000

2000

3000

Apparent skin layer thickness(nm) Figure 3. Dependence of R on apparent asymmetric 6FDA-APPS membranes. z

skin layer thickness


for

84 evaporation times longer than 120 sec. The surface of the asymmetric membrane prepared using an evaporation time of only 15 sec was three times as rough as the membrane surface with an evaporation time of 600 sec. A faster exchange rate between solvent and nonsolvent at the membrane surface with shorter evaporation time appears to be responsible for the higher roughness of the membrane. We believe that the wet process in the nonsolvent determines the surface roughness of the membrane. However, the influence of the exchange rate between solvent and nonsolvent on the surface morphology of the membrane must be elucidated in a future study. Table I shows gas permeances and selectivities of the 6FDA-APPS membranes. The selectivities for 0 / N and C 0 / C H in the asymmetric membrane with an apparent skin layer thickness of 25 nm were 6.3 with an 0 permeance of 7.7 χ 10" [cm (STP)/(cm -S'cmHg)] and 40 with a C 0 permeance of 3.5 χ 10" [cm (STP)/(cm »s»cmHg)j, respectively, without the necessity of an additional coating to seal any possible defects. These gas permeances were approximately 2000 times higher than those determined for the dense, isotropic 6FDA-APPS film. 2

2

2

4

5

2

3

2

4


2

3

2

Table I. Gas permeance and selectivity of 6FDA-APPS polyimide membranes at 35°Cand76 cmHg. Membrane type Asymmetric Dense film

Permeance Q* C0 0 350 77 0.182 0.038 2

2

Selectivity C0 /CH4 0 /N 40 6.3 36 6.1 2

2

2

Thickness (nm) 25 50,000

Figure 4 shows the relationship between the C 0 / C H selectivity and the apparent skin layer thickness. The gas selectivity and permeance of the asymmetric membranes is presented in Figure 5. The selectivity of the asymmetric membranes increased with decreasing thickness, and the membranes exhibited gas selectivities equal to or even higher than those determined for the dense membrane. This result indicates that the surface skin layer of the asymmetric polyimide membrane was essentially defect-free. Therefore, the gas transport of the asymmetric membranes occurred predominantly by a solution/diffusion mechanism. Both structure and dynamics of aromatic polyimides are influenced by intermolecularand intramolecular charge transfer interactions between the polymer chains, because aromatic polyimides contain an alternating sequence of electron-rich donor and electron-deficient acceptor moieties (13,14). Recently, we reported that asymmetric polyimide membranes prepared by dry/wet phase inversion exhibit stronger charge transfer complex formation than a dense polyimide film (10,11). This means that the packed polyimide structure in the skin layer is formed by the charge transfer complex and provides high size and shape discrimination for gas molecules. The skin layer of the asymmetric membrane coagulated after a short evaporation time might form a more efficiently packed structure relative to that of membranes prepared with longer evaporation times. However, these arguments 2

4


85

43

·· ··

41 •Ρ '•8

• Η. οα 37 ··• · • α>

39


• ··• •%

•

S?

35 .01

Dense 1embrane

ft 10

Thickness^m) Figure 4. Relationship between C 0 / C H selectivity and apparent skin layer thickness for asymmetric 6FDA-APPS membranes at 35°C and 76 cmHg. 2

4

43

•ft 41

•ft

ι

ft

39

% · «ι

•

··· ο 37 ··•· υ

aft

u

ft

35 10-

6

• ft ft

·

Dense membrane 5

10-

1

Membrane Formation and Modification - American Chemical Society

Recommend Documents