Solvent-Exchange Drying of Cellulose Acetate Membranes for

33 Solvent-Exchange Drying of Cellulose Acetate Membranes for Separation of Hydrogen-Methane Gas Mixtures B. S. MINHAS, TAKESHI MATSUURA, and S. SOURIRAJAN

Downloaded by COLUMBIA UNIV on July 1, 2012 | http://pubs.acs.org Publication Date: January 1, 1985 | doi: 10.1021/bk-1985-0281.ch033

Division of Chemistry, National Research Council of Canada, Ottawa, Ontario, Canada K1A 0R9

Cellulose acetate reverse osmosis membranes of different surface porosities were prepared using different solvents in the solvent exchange drying method. These membranes showed a wide variation in the permeation rate and separation of hydrogen-methane gas mixtures. An attempt was made to develop a correlation between the membrane performance and the solvent used for drying. The membranes were characterized and their performance was predicted on the basis of the pore flow model developed in the previous work. The results obtained are discussed. Though the history of gas separations by membranes can be traced back to 1831 (jL_) when ivestigations were reported for the enrichment of oxygen in air by rubber membranes, membranes for industrial gas separations could not be used until recently. Low permeation rates and poor separations by earlier membranes made them no match for conventional processes such as cryogenics and adsorption (2_). Breakthrough in the formation of appropriate membranes with higher flux came with the development of asymmetric cellulose acetate membranes for reverse osmosis water desalination (3_)» The reverse osmosis membranes, when dried in a manner to preserve their porosity and the surface pore structure, showed higher permeation rates and significant separations for gaseous mixtures (4^ _5). Fundamental studies have been reported on the permeation of different gases through dry reverse osmosis membranes of different polymeric materials (£, 7) and the separation of hydrogen-methane gas mixtures by cellulose acetate membranes (8^, 9) establishing a valid means of treating gas permeations and separations through reverse osmosis membranes. In the above studies the gaseous flow through the pores in the surface layer of the asymmetric porous membrane is considered to be governed by Knudsen, slip, viscous and surface flow mechanisms. The contributions of different flows, defined by different flow mechanisms, to the total flow through the membrane are dependent upon the pore size and the pore size distribution in the 0097-6156/85/0281-0451$06.00/0 Published 1985, American Chemical Society

In Reverse Osmosis and Ultrafiltration; Sourirajan, S., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1985.


452

REVERSE OSMOSIS AND ULTRAFILTRATION

surface layer of the membrane. For achieving a membrane of appropriate pore size and pore size distribution which gives high separation factor and high permeation rate, a firm cause and effect relationship has to be established between the variables involved in the membrane formation process and the performance data of membrane produced. Among the many variables involved in the formation of cellulose acetate membranes, particularly, we have identified that the evaporation period, shrinkage temperature and the solvent used for replacement of water in the membrane during the drying process are some of the important factors affecting the ultimate pore size and pore size distribution of the membrane and consequently the membrane performance data (9). The objective of this investigation is primarily to study the effect of various solvents used in the solvent exchange drying process of membrane on its subsequent performance in the separation of hydrogen-methane gas mixtures. The effects of process variables such as operating pressure and hydrogen mole fraction in the gas mixture on the separation factor and the product permeation rate were also studied. An attempt was made to correlate the membrane performance with the solvents used for drying the membrane. The predictability of membrane performance is also attempted on the basis of gas transport mechanism referred above. Gas Transport Through Porous Reverse Osmosis Membranes Transport equations for the gaseous flow through pores on the surface of an asymmetric porous membrane have been reported earlier by Rangarajan et al (])• These equations have been developed assuming that the pores can be represented by a bundle of cylinderical capillary tubes running through the membrane surface layer. The gaseous flow through these capillaries is governed by four flow mechanisms, namely, Knudsen, slip, viscous and surface flows. The contributions from the first three mechanisms are categorized as pore flow which is free of any gas-polymer interaction. Surface flow, on the other hand, involves interaction forces between the gas and the membrane material. The surface flow mechanism is applicable to all pores irrespective of their sizes, whereas pore flow is governed by one of the first three flow mechanisms depending upon the pore size, pore size distribution and the mean free path of the gas. The mean free path of a gas can be calculated from the expression (10),

It is assumed that Knudsen flow occurs in the pores of radii > 0 to 0.05X, slip flow occurs in the pores of radii 0.05X to 50X, and viscous flow occurs in the pores of radii larger than 50X. Further, it is assumed that the variation of the pore sizes on the surface of an asymmetric porous membrane is represented by a single equivalent normal distribution

The development of a transport equation, for the calculation of gas permeability coefficient, based on the four flow mechanisms


33. MINHAS ET AL.

Solvent-Exchange Drying of Cellulose Acetate Membranes

453


described above and the single equivalent normal distribution of pores on the surface of a membrane has been reported by Rangarajan et al (7_). According to this theoretical development a membrane can be characterized by the average pore radius, R, the standard deviation, a, and by two additional parameters A-, and A£, all of which can be determined from the permeability data of a reference gas. The quantity A, is related to the total number of pores and the effective membrane thickness, and should remain constant for a given membrane irrespective of the gas. On the other hand, the quantity Pi* is related to the adsorption equilibrium, the mobility of the sorbed gas species and the membrane pore structure. The value of A2 for various gases has been related to that of a reference gas under the assumption that different gases do not affect the pore structure of a given membrane.

The parameter j, called relative surface transport coefficient, for various gases has been reported by Rangarajan et al (7_) taking nitrogen as the reference gas. As stated by Rangarajan et al (7) the effective mean pore radius K is distinct from the geometrical mean pore radius R • The former mean pore radius is expected to depend upon the gas-membrane interaction and the mobility of the adsorbed gas molecule. According to the paper referred above, for any gas i, R, is computed from the following equation:

The quantities A^ called the radius correction factor, for several gases are reported in literature (7^) with respect to the reference gas, nitrogen. The standard deviation, a, is expected to remain constant for a given membrane irrespective of the gas. These four characteristic parameters can be used to compute the permeation rate and the separation factor of a gas mixture with respect to the characterized membrane. The method for the calculation of product composition have been reported by Mazid et al (8). According to this method the separation factor, Sj^, defined as:

can be obtained from the calculated compositions of product gases. The product permeation rate can be calculated from the fluxes, J^ , of each component of the gas mixture. The equations for the evaluation of flux, J^, has been reported in the literature (8).

Thus to predict the separation factor and the permeation rate first the membrane is characterized by evaluating £, a, A, and A2 using the


454


permeability of a reference gas through the membrane. These reference parameters are then used to evaluate the characteristic parameters with respect to each component of the gas mixture. These evaluated parameters are subsequently used to predict the separation factor and the permeation rate for a gas mixture for the characterized membrane.


Experimental The membranes used in the present investigation were cast using a solution of the following composition (wt. % ) : cellulose acetate (Eastman 398-3) 17, acetone 69.2, magnesium perchlorate 1.45 and water 12.35 (11). All membranes were cast to equal nominal thickness. The casting solution and the casting atmosphere temperatures were kept constant at 10° and 30°C, respectively. The relative humidity of the casting atmosphere was maintained at 65%. Membranes were gelled in ice cold water after 60 s of solvent evaporation time and then shrunk in hot water at 80 °C. These membranes were then dried, in order to use them for gas separation experiments by a multiple stage solvent exchange drying technique. In this technique the water in the membrane is first replaced by a water miscible solvent (called the "first solvent") which is a nonsolvent for the membrane material. Then, the first solvent is replaced by a second solvent which is volatile. The second solvent is subsequently air evaporated to obtain the dry membrane. A number of different solvents were used both as the first solvent and as the second solvent. First solvents used include isopropyl alcohol, tertiary butyl alcohol, ethylene glycol, diethylene glycol, triethylene glycol and ethylene glycol monoethyl ether. The solvents used as the second solvent include pentane, hexane, cyclohexane, benzene, toluene, carbondisulfide, triethylamine and isopropyl ether. The replacement of water in the membrane by the first solvent was done by successive immersion in first solvent-water solutions which were progressively more concentrated in the first solvent. For example in four stage replacement, which was the most common in our experiments, 25, 50, 75 and 100 vol. % aqueous solutions of the first solvent were used. When isopropyl alcohol was employed as the first solvent the replacement was carried out also in one, two and three stages in order to study the effect of number of stages on the membrane performance. In the cases where the first solvent used was any one of the glycols an intermediate solvent was also used, for membrane drying, between the first and the second solvents, as the glycols are not miscible with the second solvents. Intermediate solvents used were ethyl alcohol and n-butyl alcohol which are miscible with both the first solvents and the second solvents. Usage of different solvents in the drying procedure resulted in membranes of different average pore sizes and pore size distributions. Membranes dried by the combination of different solvents were numbered in the form of CA(K)-mn for the purpose of membrane identification, where CA indicates cellulose acetate material, K denotes the number of stages involved in the replacement of water in the membrane by the first solvent, and m and n are the numbers given to the first and to the second solvents, respectively, used in the membrane drying process. Numbers given to the first solvents are as



33. MINH AS ET AL.


455

follows: tertiary butyl alcohol-1, isopropyl alcohol-2, ethylene glycol monoethyl ether-3, ethylene glycol-4, diethylene glycol-5, triethylene glycol-6; whereas the second solvents were numbered as follows: pentane-1, hexane-2, cyclohexane-3, benzene-4, toluene-5, triethylamine-6, isopropyl ether-7 and carbondisulfide-8. The equipment used in the present investigation and the details of the experimental procedure have been reported previously (4^ _7). Air in the reverse osmosis cells and in the feed gas line was removed by flushing them with the feed gas mixture of hydrogen and methane. The mole fraction of hydrogen in the feed gas was changed from 0.883 to 0.116. All the experiments were conducted at room temperature and the feed pressure was varied in the range of 400 to 2400 kPa abs. The permeate flow rate was measured by a soap bubble meter. The composition of the gas was measured by gas chromatography using Tracor MT 160/220 model equipped with a Porapak Q column. The accuracy involved in gas composition analysis was ±1%. All the gases (pure and mixed) were obtained from Matheson Canada with a specified purity of 99.9%. The liquid chromatography experiments for measuring the retention volume of various second solvents, in the presence of first solvent was conducted using the equipment model ALC 202 from Waters Associates. The column used was 0.16 cm inner diameter and 60 cm in length and packed with commercially available cellulose acetate powder. The solutions of various second solvents (1 vol. %) in a first solvent were injected into the stream of the first solvent flowing through the column as a carrier solvent in order to measure the retention volume of the second solvent. Either tertiary butyl alcohol or isopropyl alcohol was used as the first solvent. When tertiary butyl alcohol was used as the carrier solvent, the entire chromatography system including the column was maintained at 27°C in order to avoid the freezing of tertiary butyl alcohol. All the solvents used were of reagent grade. Results and Discussion Membrane Performance. Cellulose acetate membranes dried by solvent exchange technique using various combinations of first, second and intermediate solvents showed a wide variation in separation factor ranging from 1 to 28 for a feed mixture of hydrogen and methane. A wide variation in permeation rate was also found ranging from 1.25 x 10" 4 to 9.71 x 10" 7 kraol/m2s, for the feed containing 0.883 mole fraction of hydrogen at the operating pressure about 2200 kPa abs. These wide variations in the separation factor and in the permeation rate emphasize the importance of solvent used in the solvent exchange technique of membrane drying. The membranes dried by the combination of isopropyl alcohol and hexane as the first and the second solvent, respectively, usually gave higher separation factors and permeation rates compared to the membranes dried using other combinations of solvents. It was also found that when membranes were dried using the combination of isopropyl alcohol and hexane solvents, the number of stages involved in the replacement of water in the membrane by isopropyl alcohol significantly affected the membrane performance. It may be seen in Table I that the membrane (CA(2)-22) dried by two stage process (replacing water with 50 vol. % aqueous solution of isopropyl alcohol which was then replaced by


456



Table I. Effect of Number of Stages Involved in the Replacement of Water in the Membranes by the First Solvent on the Separation Factor and the Permeation Rate

Xj2

Membrane

Pressure, kPa

0.883

CA(l)-22

487.4 1411.3 1756.1 446.1 652.9 1135.5 1494.1 2224.9 487.4 914.9 1411.3 1756.1 2169.7 515.0 1410.0 2310.0 452.9 1349.3 1756.1 446.1 790.8 1101.1 1549.2 2204.2 452.9 859.7 1756.1 515.0 1410.0 2310.0

CA(2)-22

CA(3)-22

CA(4)-22 0.781

CA(l)-22 CA(2)-22

CA(3)-22 CA(4)-22

s

xz

9.3 4.3 4.2 9.0 15.0 21.0 25.0 28.0 6.1 7.0 8.6 13.3 19.1 9.4 14.5 17.5 6.8 2.2 1.6 15.0 20.0 21.0 18.5 17.0 13.0 11.0 7.2 7.3 10.0 12.7

[PR], kmol/m . s 0.144 0.261 0.296 0.330 0.340 0.700 0.960 0.150 0.254 0.497 0.793 0.990 0.606 0.947 0.104 0.172 0.205 0.170 0.360 0.530 0.780 0.120 0.225 0.411 0.832 0.110 0.120

X lo 5

X

: 5 -5 10 -5 10 -5

X

10 I

X

10

L

1 0

f.

X X X

X

10

X

10"° 10"° 10"° 10"* 10~ 5

X

10"f

X X X X

X lo X 10 X

:

10- 5I

X 10 X X X X X X X X X

10'-5I 10-5

10

i o"i

i

10"° 10"? 10-6 10"* 10 " 6

Note: The first solvent is isopropyl alcohol; the second solvent is hexane.



33. MINHAS ET AL.


457

nonaqueous isopropyl alcohol) gave higher separation factor as well as the permeation rate for hydrogen methane gas mixtures compared to the membranes dried by one (CA(l)-22), three (CA(3)-22) and four (CA(4)-22) stage processes. The highest separation factor obtained for the membrane CA(2)-22 was 28 at the operating pressure of 2225 kPa abs for a feed gas mixture containing 0.883 mole fraction of hydrogen. This result corresponds to the hydrogen permeation rate of 99 relative to methane, which is significantly higher than the values reported in the literature for various membranes. These results indicate that significant separation factors are obtainable by gas permeation under pressure through porous cellulose acetate reverse osmosis membranes if dried by suitable combination and sequence of solvents. The present investigation also revealed that a membrane with a lower permeation rate does not necessarily yield a higher separation factor. The permeation rates and the separation factors obtained in the present study for some of the membranes dried using different combinations of solvents are presented in Table II. It may be seen in Tables I and II that the permeation rates and the separation factors obtained at all feed compositions and at all feed gas pressures were lower for the membrane dried using tertiary butyl alcohol as the first solvent and benzene as the second solvent (CA(4)-14) than those for the membrane dried using isopropyl alcohol as the first solvent and hexane as the second solvent (CA(2)-22) at corresponding feed compositions and feed gas pressures. Correlation Between Membrane Performance and Drying Solvent. An attempt was made to develop a correlation between the properties of the solvent used for membrane drying and the separation factor. In order to develop such a correlation the retention volumes of various second solvents were measured in a liquid chromatography system in which the column was packed with cellulose acetate powder. The retention volume is considered to be a measure of the magnitude of the interaction force working between the second solvent and the cellulose acetate polymer in the presence of the first solvent. As described in the experimental section two first solvents, namely isopropyl alcohol and tertiary butyl alcohol, were used as the carrier solvent. Figure 1 shows the results of the chromatography experiments. In the figure retention volumes of various second solvents are plotted against the separation factor obtained from the membrane dried using the corresponding second solvent. It may be seen in Figure 1 that the membranes which were dried using isopropyl alcohol as the first solvent gave the highest separation factor at the retention volume corresponding to the second solvent hexane, irrespective of the operating pressure. On the other hand, when tertiary butyl alcohol was used as the first solvent, the second solvent which gave the maximum separation factor, shifted towards the higher retention volume, suggesting a stronger interaction force between the second solvent and the membrane material is needed. The second solvent corresponding to the maximum separation factor was found to be isopropyl ether. Though Figure 1 includes points for the membranes dried using different stages when isopropanol was used as the first solvent, the correlation emerging between the separation factor and the retention volume of the second solvent is quite clear. The above results indicate that for a given first drying


458


Table II. So^e Experimental Data on Permeation Rate and Separation Factor for Different Membranes at Various Pressures and Feed Compositions

X12

Membrane

Pressure, kPa

S-,?

0.883

CA(4)-14

790.8 1114.8 1818.1 2176.6 466.7 790.8 1114.8 1818.1 2176.6 466.7 790.8 1114.8 1818.1 2176.6 466.7 790.8 1114.8 1818.1 2169.7 446.1 859.7 1342.4 1811.2 2169.7 446.1 859.7 1342.4 1811.2 2169.7 1342.4 1825.0 2169.7 859.7 1342.4 1825.0 2169.7 459.9 859.7 1342.4 1825.0 2169.7 459.9 859.7 1342.4 1825.0 2169.7

1.9 1.74 1.7 1.7 4.6 3.3 2.5 1.5 1.1 3.2 2.5 2.1 1.6 1.3 1.2 1.0 1.0 1.0 1.0 1.9 1.2 1.0 1.0 1.0 1.9 1.9 1.3 1.2 1.2 1.8 1.8 1.84 3.5 2.5 1.5 1.2 2.8 2.3 1.8 1.6 1.4 1.2 1.1 1.1 1.0 1.0


CA(4)-17

CA(4)-27

CA(4)-32

CA(4)-47a

CA(4)-62b

0.781

CA(4)-14

CA(4)-17

CA(4)-27

CA(4)-32

o [PR], kmol/m .s 0.19 0.33 0.72 0.97 0.47 0.87 0.14 0.27 0.35 0.43 0.78 0.12 0.23 0.28 0.15 0.42 0.61 0.10 0.13 0.19 0.40 0.73 0.11 0.14 0.24 0.57 0.12 0.17 0.21 0.39 0.63 0.84 0.78 0.14 0.22 0.36 0.33 0.72 0.13 0.19 0.22 0.19 0.37 0.61 0.84 0.99

x 10~* x 10"7 x 10"° x 10"° x 10~5 x 10"^ x 10~* x 10"; x 10""; x 10"5 x 10"^ x 10"; x 10"; x 10"; x 10"; x 10"J x 10"* x 10 ^ x 10"; x 10"; x 10"; x 10"* x 10";? x 10"3 x 10";? x 10"; x 10"; x 10 ; x 10"; x 10"6 x 10"*? x 10 6 x 10"; x 10"; x 10~* x 10"; x 10"^ x 10 5 x 10"* x 10"; x 10~* x 10"; x 10"* x 10"J x 10 ; x 10 4


33.

MINHAS ET AL.


Table I I . X^2

Membrane

0.781

CA(4)-47 a


CA(4)-62 b

0.484

CA(4)-14

CA(4)-17

CA(2)-22 CA(4)-27

CA(4)-32

CA(4)-47 a CA(4)-62 0.238

CA(4)-14

CA(4)-17

CA(2)-22

b

459

Continued

P r e s s u r e , kPa

hz

473.6 859.7 1342.4 1818.1 2238.7 473.6 859.7 1342.4 1818.1 2238.7 859.7 1342.4 1804.3 2204.2 446.1 859.7 1342.4 1804.3 2204.2 1135.5 1549.2 1997.4 446.1 859.7 1342.4 1804.3 2204.2 446.1 859.7 1342.4 1804.3 2190.4 501.2 873.5 501.2 873.5 446.1 1342.4 1790.5 2204.2 473.6 859.7 1342.4 1790.5 2238.7 446.1 790.8 1135.5 1549.2

1.5 1.2 1.0 1.0 1.0 2.3 2.0 1.5 1.3 1.3 1.8 1.8 2.0 1.9 4.2 3.3 2.2 1.6 1.3 14.7 13.7 12.8 2.8 2.5 1.9 1.8 1.6 1.1 1.1 1.0 1.0 1.0 1.8 1.1 2.4 1.8 2.0 2.0 1.9 1.8 3.4 2.7 2.2 1.6 1.3 11.0 10.5 10.2 10.0

[PR], kraol/m 2 .s 0.16 0.30 0.50 0.78 0.10 0.25 0.54 0.10 0.15 0.19 0.13 0.24 0.38 0.52 0.15 0.38 0.69 0.12 0.18 0.20 0.27 0.36 0.16 0.38 0.68 0.96 0.12 0.65 0.24 0.38 0.51 0.63 0.86 0.14 0.16 0.32 0.35 0.13 0.24 0.34 0.74 0.20 0.42 0.94 0.14 0.18 0.33 0.51 0.78

x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x

i°1

10 10

4

10- 3i

i°: 1°_4 1° A

1°1 10 ° 10"° 10"° 10"° 10 i

10

'-l

1° A li oo - ; :

10 I 10-5 lo

:

10 I 10-5 io-;

1°1 1°1 1°1 10

'-5

1 0

A

io-;

i°: 10

-7 1° 6 10° 10"° 10"° 10"° 10-5 IO"! l°"i 10

"2

10"° 10"° 10"° 10" 6

Continued on next page. In Reverse Osmosis and Ultrafiltration; Sourirajan, S., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1985.

460

REVERSE OSMOSIS AND ULTRAFILTRATION Table I I .

X, 2

Membrane

0.238

CA(4)-27

Pressure, kPa


CA(4)-32

CA(4)-47 a CA(4)-62 b 0.116

CA(4)-17

CA(2)-22 CA(4)-27

CA(4)-32 CA(4)-47 a CA(4)-62

a) b)

Continued

b

1997.4 473.6 859.7 1342.4 1790.5 2238.7 473.6 859.7 1342.4 1790.5 2204.2 459.9 825.3 1328.6 459.9 825.3 1328.6 563.3 914.9 1411.3 1825.0 480.5 859.7 1342.4 563.3 914.9 1411.3 1825.0 563.3 914.9 1411.3 452.9 804.6 1328.6 452.9 804.6 1328.6 1756.1 2204.2

s

iZ

10.9 2.4 2.2 2.0 1.73 1.7 1.1 1.0 1.0 1.0 1.0 1.9 1.6 1.3 2.2 2.2 1.9 3.7 2.6 2.2 1.6 11.7 11.7 11.3 3.0 2.2 2.2 1.9 1.3 1.1 1.0 2.0 1.4 1.1 2.1 2.0 1.73 1.6 1.53

[PR],, kmol/m . s 0.12 0.96 0.21 0.43 0.61 0.77 0.45 0.12 0.22 0.38 0.48 0.44 0.10 0.17 0.89 0.22 0.45 0.76 0.16 0.38 0.69 0.98 0.19 0.36 0.98 0.18 0.37 0.49 0.48 0.10 0.21 0.38 0.74 0.14 0.73 0.16 0.35 0.52 0.72

X X X

10 f 10"5

X X X X X X X X X

10 \ 10-5 1O

:A

10 1 0

A

10 1

'-t

10

-5

io A 1 0

X

i

X X X

1 0

10"*f.

X

10 -5 10

1

X

10"°

X

10

X

10

X X X X X X

-5 -5 10 -7 10 -6 10 ° 10"° 10"° 10

X

10- 5I

X

10

X X X X X

~5

10 1 0

A

1 0

A

1 10-; w:5

X X X X X X

ethylene g l y c o l followed by ethyl alcohol t r i e t h y l e n e glycol followed by n-butyl alcohol


10_

6 10 -5 10 -5 10 -5 10 _5 10 >

MINHAS ET AL.



33.

Figure 1. Retention volume of various second solvents versus separation factor of H 2 /CH^ gas mixture. Liquid chromatography column, packed with cellulose acetate powder. Membrane gas separation conducted at the hydrogen mole fraction of 0.883 in feed gas and various operating pressures.


461

462



solvent (carrier solvent in liquid chromatography), there is an optimum retention volume for the second solvent which corresponds to the highest separation factor for the gas mixture studied; and the above retention volume is independent of the operating pressure used in the gas separation experiment. Thus Figure 1 shows a useful correlation between the retention volume of the second solvent and the resulting porous structure of the membrane surface in terms of the gas separation experiment involved. Consequently, liquid chromatography data on retention volumes for different second solvents for a given first solvent offer a basis for the choice of the appropriate second solvent for obtaining the desirable porous structure on the resulting membrane capable of giving the highest separation factor for a given gas mixture. Prediction of Membrane Performance. All the membranes formed in the present investigation were characterized in terms of K, a, A, and A2 by using the transport equation described by Rangarajan et al (7_) together with the permeability data of helium through the membranes. The results are shown in Table III. The table shows that there are over 8 fold variation in ( R ) ^ and 11 fold variation in a. Furthermore, there are 380 fold variation in the value of A-. and 500 fold variation in (k^^Ea* These variations indicate that the membranes tested have a wide range of porosity and also a wide range in surface flow contribution, which should result in a wide variation in permeation rates and separation factors of a gas mixture. The parameters evaluated in membrane characterization were used to predict the permeation rate, [PR], for a given membrane under a given set of operating conditions such as hydrogen mole fraction in the feed gas and the operating pressure. The details of the prediction method were described previously (8). The comparison of calculated and experimental permeation rates is shown in Figure 2. The agreement between calculated and experimental values is excellent in some cases and unsatisfactory in many cases. Further, the product compositions and the separation factors were also predicted, using the approach developed previously (&), for all the membranes studied under all the combinations of operating pressures and feed compositions employed in the experiment using the membrane parameters given in Table III. The comparison of experimental and calculated values of product compositions is shown in Figure 3. These results also show excellent agreement in some cases, and unsatisfactory agreement in many cases. Such observed agreements and disagreements between the calculated and experimental data are similar to those reported earlier (80. These results indicate that while the transport equations employed are basically valid, the analytical expressions involved need improvement based on the physical chemistry of the gas adsorption process taking place in the system, and the porous structure of the membranes employed. In particular, it is necessary to examine the need for (i) alternative expressions for the individual and competitive gas adsorption processes, and (ii) bimodal pore size distribution on the membrane surface, for incorporation in the gas transport equations.


33.

MINHAS ET AL.

Table I I I ,


Membrane CA(4)-11 CA(4)-12 CA(4)-13 CA(4)-14 CA(4)-15 CA(4)-16 CA(4)-17 CA(4)-18 CA(4)-21 CA(l)-22 CA(2)-22 CA(3)-22 CA(4)-22 CA(4)-24 CA(4)-25 CA(4)-27 CA(4)-28 CA(4)-32 CA(4)-42 a CA(4)-47 a CA(4)-52 b CA(4)-62 C a) b) c)


463

Membrane Characterization by Using Helium Permeation Data

(R>He

x

10l0

8.0 8.0 12.0 4.0 7.0 4.0 24.0 7.0 26.0 7.0 5.0 6.4 3.2 5.0 7.0 5.0 24.0 7.0 5.8 5.0 5.0 24.0

>m

o x 10 1 0 ,m

(A 1 ) He> m

4.3 5.5 3.3 1.7 0.8 2.7 0.7 0.9 0.5 1.1 3.7 1.3 2.3 1.0 0.7 1.1 0.7 0.9 0.6 3.0 1.5 0.7

6.24 6.45 1.01 3.40 3.15 5.63 1.64 9.21 1.85 1.36 2.47 2.25 4.24 7.87 6.99 6.79 5.37 1.22 6.15 3.15 6.31 2.15

3 2 (A2)^ e ,kmol/m sPa

3

x 1019 x 101® x 10^

x 10J; lf

x 10 x H)18

x 10J; x 10 ]l

x 10 l f x 10 }« x 10^

x io|; x 101; x 101' x 1017

x 10J! x 101;

x 10f X l 0

Q

19

x 10 x 10 1 9 x 1018

5.15 1.46 4.38 1.02 3.15 1.24 5.55 6.26 4.18 9.83 1.30 1.18 1.15 1.53 5.83 2.26 2.53 7.07 1.68 8.01 1.99 1.59

ethylene glycol followed by ethyl alcohol diethylene glycol followed by n-butyl alcohol triethylene glycol followed by n-butyl alcohol


X X

io~! 10

"Q

X 1 0 X X X X

'\ 10-w 10-1° 10-9 10"10,

X 10 ~io X 10

~m

X X

10

X

10-1°

X

?n "?n 10" 10

X

io-i°

X

10

X

10

X

10-1°

X

10

"Q

10

"Q

X

"?o

X 10 X X

"« 10"° 10-9


464


Figure 2. Comparison of experimental and calculated permeation rates of H2/CH4 gas mixture. Hydrogen mole fraction in feed gas mixture, 0.116-0.883; operating pressure, 450-2300 kPa abs; membrane material, cellulose acetate; membrane porosity given in Table III; operating temperature, room.

Figure 3. Comparison of experimental and calculated product compositions of H2/CH4 gas mixture. Hydrogen mole fraction in feed gas mixture, 0.116-0.883; operating pressure, 450-2300 kPa abs; membrane material, cellulose acetate; membrane porosity given in Table III; operating temperature, room.


33.

MINHAS ET AL.


465


Conclusion It may be concluded from the present study that the performance of reverse osmosis cellulose acetate membranes for the separation of hydrogen-methane gas mixtures is strongly influenced by the solvent systems and operational sequence used with solvent exchange drying of the membrane. By selecting a proper combination of solvents and proper operational sequence a cellulose acetate membrane of desired pore size and pore size distribution can be made to give higher separation factor and permeation rate. The selection of a solvent system can be facilitated by measuring the retention volume of the solvent in a column packed with the membrane material in liquid chromatographic study where the carrier is the first solvent. It is also concluded from this study that further refinements in the prediction method are necessary in order to improve the agreement between calculated and experimental gas permeation and separation data in a wide range of operating conditions. Nomenclature

Ax, ( A ^ A2. ( A 2 ) r e f j (A 2 )^

= constant for a given membrane related to the -.3 porous structure, A-^ for gas i, m = constant related to surface transport, A2 for reference gas and gas i, respectively, kmol/m .s. Pa = collision diameter, m = flux of gas i, kmol/m .s

d K m n N N(R)

* P [PR] R

K, E r e f , £j R T i» Xi2> Xi3

X

= number of stages involved in the replacement of water in the membrane by the first solvent = number given to the first solvent = number given to the second solvent = Avogadro number = number of pores having a radius R, m~ • total number of pores = pressure, Pa = mean pressure, Pa = total product permeation rate of the gas mixture, kmol/m *s - pore radius = mean pore radius, E for reference gas and gas = • =

i, respectively, m geometrical mean pore radius, m gas constant m.Pa/K.kmol separation factor for gas mixture absolute temperature, K mole fraction of gas i, X. on the high pressure side, and on the permeate side, respectively

Greek Letters Aj X

=» radius correction factor, a constant for gas i for a given membrane material, m = mean free path of gas, m


466


a

= standard deviation for the pore size distribution, m = characteristic parameter, called the relative surface transport coefficient, related to gas membrane interaction

^

Acknowledgments The authors are grateful to the NRC Bioenergy Project for supporting this work. One of the authors (BSM) thanks the NSERC for a visiting fellowship. This paper was issued as NRC No. 24035.


Literature Cited 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

Mitchell, J.V. J. Roy. Inst. 1831, 2, 101, 307. Rain, C. High Technol. 1983, Nov., 69-76. Loeb, S.; Sourirajan, S., Department of Engineering, University of California, Los Angeles, Report No. 60-60, 1960, July. Agrawal, J.P.; Sourirajan, S. J. Appl. Polym. Sci. 1969, 13, 1065. Agrawal, J.P.; Sourirajan, S. J. Appl. Polym. Sci. 1970, 14, 1303. Rangarajan, R.; Mazid, M.A.; Matsuura, T.; Sourirajan, S. Proc. Fourth Bioenergy R & D Seminar, National Research Council of Canada, Ottawa, 1982, pp. 435-40. Rangarajan, R.; Mazid, M.A.; Matsuura, T.; Sourirajan, S. Ind. Eng. Chem. Process Des. Dev. 1984, 23, 79-87. Mazid, M.A.; Rangarajan, R.; Matsuura, T.; Sourirajan, S. Ind. Eng. Chem. Process Des. Dev., in press. Minhas, B.S.; Mazid, M.A.; Matsuura, T.; Sourirajan, S. Proc. Fifth Bioenergy R & D Seminar, National Research Council of Canada, Ottawa, 1984. Metz, C.R. "Physical Chemistry"; McGraw-Hill: New York, 1976; p. 11. Pageau, L.; Sourirajan, S. J. Appl. Polym. Sci., 1972, 16, 3185.

RECEIVED April 22, 1985


Solvent-Exchange Drying of Cellulose Acetate Membranes for

Recommend Documents