I n d . Eng. C h e m . Res. 1993,32, 533-539
633
Separation of Volatile Organic Compound/Nitrogen Mixtures by Polymeric Membranes Xianshe Feng, S. Sourirajan, and F. Handan Tezel Industrial Membrane Research Institute, Department of Chemical Engineering, University of Ottawa, Ottawa, Ontario K I N 6N5, Canada
T. Matsuura' Institute for Environmental Chemistry, National Research Council of Canada, Ottawa, Ontario K 1 A OR6, Canada
B. A. Farnand Energy Research Laboratories, C A N M E T , Energy, Mines and Resources Canada, Ottawa, Ontario K I A OG1, Canada
Dry asymmetric poly(ether imide) membranes were prepared by the phase inversion technique under different conditions to investigate their performance for separating volatile organic compound/ nitrogen gas mixtures. The membrane performance was further tested under different operating conditions such as operating temperature, feed vapor concentration, and the pressure on the permeate side of the membrane. A set of transport equations was proposed for the membrane permeation of organic vapor/gas mixtures. The validity of the transport equations was tested by their applicability to describe the experimental data.
Introduction Volatile organic compounds produce a large amount of waste emissions which cause not only a severe environmental pollution problem but also a significant economic loss. The recovery of volatile organic compounds from loading, unloading, and other handling operationshas been under scrutiny from both environmental and economic points of view. However, most existing techniques for organic vapor emissions control have so far proved to be unsatisfactory in view of safety, performance, operating cost, and facility space (Theodore and Buonicore, 1988). Membrane technology,which has been successfullyapplied in seawater desalination and gas-gas mixture separations, is expected to provide an alternative to the conventional methods. The concept of organic vapor separation by membrane is not new, but only recently did interest increase in this process. The separation of organic vapor from waste gas streams is different from gas-gas separations. Besides permeability and selectivityrequirements, the membrane material has to be highly resistant to organic vapor attack. Though this process is a low-pressure membrane process and involves vapor components in the permeate, it also differs from pervaporation since the membrane is dry. Although air and vapor permeation experiments were conducted for various polymeric films by Baker et al. (1987),most of the experimental works reported so far on membrane vapor recovery have concentrated on composite silicone rubber membranes. Pinnau et al. (1988),Kimmerle et al. (19881,Strathmann et al. (1986),and Paul et al. (1988)tested silicone rubber membranes coated on a porouspolysulfone substrate. Behling (1986)and Behling et al. (1989)chose poly(ether imide) as the supporting material because it is much more stable to organic vapors than polysulfone. Buys et al. (1990)used polyhydantoin and polyimide as the porous support to silicone rubber coating layer in their study. A membrane system for the treatment of low-volume, high vapor concentration gas streams was tested, but no information was disclosed concerning membrane materials (Wijmans and Helm, 1989).
The resistance of siliconerubber to some organicvapors, for example gasoline, is however poor (Billmeyer, 1984). An attempt was made, therefore, by the authors to prepare membranes from a single polymeric material of high organic resistance. In our previous studies, asymmetric aromatic polyimide membranes were investigated for this purpose (Feng et al., 1990, 1991). It was shown that membranes of both high selectivity and reasonably high permeability could be produced by controlling the conditions of the membrane preparation properly. In this study aromatic poly(ether imide) membranes are also included. In order to reduce the resistance to the membrane permeation, asymmetry is introduced to the membrane structure using the phase inversion technique in the preparation of membranes. Membranes prepared under different conditionswere teated for various operating conditions and the results are reported. The mathematical modeling of the transport involved in the membrane permeation of gas-vapor mixtures is another objective of this study. In comparison with other membrane processes such as separations of liquid-liquid and gas-gas mixtures, the transport study for gas-vapor mixture permeation is in a far less advanced state primarily due to the lack of experimental data. For this purpose a new set of transport equations is proposed and its validity is examined by experimental data.
Theoretical Section The molar permeation flux of nitrogen gas through the membrane, QN [mol/(m2~s)l, is obtained by V 273.15 1 (1) QN = 3 273.15 + 23 22400 X 60 where V is the volumetric permeation rate (mL/min) of nitrogen gas and S is the effective film area (m2). The permeability of nitrogen gas, JN [mol/(m2*s*Pa)l, is given by JN = QN/ AP (2) where A p is the pressure difference across the membrane (Pa). o a a a - ~ a a ~ ~ ~ ~ i ~ ~ ~0 ~1993 - oAmerican ~ ~ ~ ~ Chemical o ~ . o oSociety ~o
534 Ind. Eng. Chem. Res., Vol. 32, No. 3, 1993
The following transport equations are used for the analysis of membrane permeation data when the feed is a nitrogen gadorganic vapor mixture
2. Assuming that the contribution from the surface flow is far greater than that from the Knudsen flow in the pore flow mechanism, according to Gilliland (1958), 2
Qv
Q,
+ QN
where Qs are permeation fluxes; p’s are pressures and Y’s are mole fractions of the organic vapor. Subscripts N and v represent nitrogen and organic vapor, respectively. Subscripts 1 and 3 represent the feed and the permeate, respectively. It should be noted that, unlike our previous paper (Feng et al., 19911, eq 4 assumes that Qv is proportional to the decrease in the square of the partial vapor pressure across the membrane rather than the decrease in the partial pressure itself. This assumption was necessary because of the nonlinear dependence of the permeation flux on partial vapor pressure as shown later in the Experimental Section. Justification of eq 4 can be provided in two ways. Subscript v is used to express the vaporous permeant in both approaches. 1. According to solution-diffusion mechanism,
where Dv(cv)is diffusivity of the vaporous permeant given as a function of the local permeant concentration, c,. Assuming a linear dependence of D , on c,,
D , = D@c,
P3
5 dp, Pv
where x , is a local value for the amount of vapor adsorbed in a given amount of membrane material and CY is a parameter in which the effects of pore geometry,tortuosity, and permeant mobility in the proe are all included. Approximating the vapor adsorption by Henry’s law,
Y3 = Qv
= a$’:
(7)
then, dc, Q,=-D c OV dx and, therefore,
Q, dx = -Dove, dc,
(9) Integrating from the upstream side ( x = 0) to the downstream side ( x = 6) of the membrane,
eq 16 becomes
Setting
and using eqs 14 and 15, eq 19 becomes equal to eq 4. Thus, eq 4 can be derived by both approaches equally well. The validity of eq 4 will also be testified to by experimental data. It should be noted that the diffusivity and solubility coefficients involved in the first approach are considered to be intrinsic properties of the polymer. In other words, only one set of diffusivity and solubility coefficients, and consequently only one value of B, is allowed for a given polymer. On the other hand, by the second approach, B is not only the function of the polymeric material but also the function of the pore geometry. In other words, there may be many B values, depending on the membrane morphology,for a given polymeric material. Since a number of membranes with different morphologies are included in this work, the second approach seems to be more appropriate. Assuming further that the permeability of nitrogen gas is unaffected by the presence of organic vapors in the feed and vice versa, JN
A
(21)
(10) Assumingfurther that Henry’s law is valid for the permeant sorption, ‘v = khvpv (11) where k h v is Henry’s law constant. Combining eqs 10 and
11,
Setting
Plv = Ply1 and P3v = P3Y3 eq 12 becomes equal to eq 4.
Looking into eqs 3-5, Qv, PI, p3, Y1, and A are known quantities from the permeation experiment for pure nitrogen gas and from the permeation experiment for feed nitrogen gadorganic vapor mixtures. Consequently, three unknown quantities B, QN,and Y3 can be calculated from above three equations. Conversely, when p1, p3, YI,A, and B are given, QN, Qv, and Y3 can be calculated. Therefore, there are two possibilities of calculating Y3, either by giving experimental Qvor by giving a parameter B. It should be further noted that the quantity B remains constant for a given organic vapor and for a given temperature, if the above equations are valid.
Experimental Section Materials. Aromatic polyimide (PI) (2080 grade) powder, supplied by Upjohn Company, was dried a t 140 “C for 16 h before use. Aromatic poly(ether imide) (PEI) (lo00 grade) was supplied by General Electric Co. in a
Ind. Eng. Chem. Res., Vol. 32,No. 3, 1993 53s saturated by vapor, while the other was used to dilute the vapor-saturated stream By changing the flow rate ratio of the two nitrogen streams, different feed vapor concentrations were obtained. The permeation rate of pure nitrogen was determined in the following way. A bubble flowmeter was connected to the feed chamber outlet of the permeation cell and the inlet valve was closed as illustrated in Figure 2a. When the permeate side of the permeation cell was evacuated by a vacuum pump, nitrogen was sucked from the bottom of the bubble meter into the feed chamber, pushing a soap film in the buret upward. The nitrogen flow rate was determined from the speed of the movement of the soap film. Figure 1. Membrane preparation sequence.
pellet form. The pellets were dried at 150 "C for 4 h in an oven with forced air circulation as suggested by the supplier. Lithium chloride (LiC1) and lithium nitrate (LiNOs) from Fisher Scientific Co. were dried at 140 "C for 4 h before being used as additives in membrane preparation. All organic chemicalswere supplied by BDH Inc. and were of reagent grade. Nitrogen gas with a purity of 99.997% was obtained from Air Products. Membrane Preparation. Membranes were prepared using the phase inversion method. The details of the preparation procedure have been reported elsewhere (Feng et al., 1991). Four major steps, including casting, evaporation, gelation, and solvent exchange, were involved in the membrane preparation, as summarized in Figure 1. Briefly, membranes were cast from polymeric solutions, the compositions of which are summarized in Table I. After partial evaporation of solvent the cast film was immersed in ice-cold water for gelation. Then, the membrane was immersed in ethanol for solvent exchange before it was dried in air. Apparatus and Procedures. The flow diagram of the experimental setup is shown schematically in Figure 2.A mixture of organic vapor and nitrogen was produced by bubbling nitrogen gas from a porous sintered stainless steel ball immersed in a chosen organic liquid at room temperature. The permeation cell, whose structure is the same as that used in our earlier study (Feng et al., 19911, was housed in an isothermal chamber whose temperature was controlled within k0.5 "C. If necessary, the feed gas mixture was preheated to the experimental temperature in a heating coil before it was introduced to the feed side of the membrane. The permeate side of the membrane was connected to two cold traps, followed by a DuoSeal vacuum pump (Model 1400). The permeation of the organic vapor was induced by maintaining its partial pressure on the permeate side lower than the feed side. The membrane-permeated organic vapor was condensed and collected initially in one of the cold traps, and then the cold trap was switched to the other after the steady state was reached. The permeation rate was determined gravimetrically by weighing the sample collected for a predetermined period. The feed gas stream was connected to a Varian gas chromatograph (Model 3400)both at the inlet and at the outlet of the feed chamber of the permeation cell through bypass valves to determine the composition of the feed mixture. The feed flow rate was so high that the concentration change from the inlet to the outlet of the feed chamber was negligible. In order to change the feed vapor concentration,nitrogen gas supplied from a cylinder was divided into two streams in such a way that only one stream was bubbled through organic liquid for being
Results and Discussion Conditions of Membrane Preparation and Membrane Performance. Some results from experiments with feed vapor concentrations corresponding to 97-99 % of saturation vapor pressures are illustrated in Figure 3. The conditions of the membrane preparation are listed in Table I. From Figure 3 and Table I we are able to draw several important conclusions concerning the effect of membrane preparation on membrane performance. 1. Solvent evaporation was at room temperature for membranes PEI-1, PEI-7, and PEI-9. Since the boiling point of dimethylacetamideis 165"C, there was practically no evaporation of solvent. Membranes prepared under such conditions are near the left-upper corner on Figure 3, meaning high fluxes and low selectivities. 2. Comparing membranes PEI-9, PEI-5, and PEI-8, the evaporation temperature is increased in the above order. Although the evaporation time of PEI-9 is shorter than other membranes, it does not invalidate the above comparison since there is no solvent evaporation anyway at room temperature. The performance data shifts on Figure 3 from the left-upper corner to the right-lower corner, meaning that a higher evaporation temperature resulted in a lower flux and a higher selectivity. 3. Comparing membranes PEI-2, PEI-3, and PEI-4, the evaporation time increases in the above order. The performance data on Figure 3 shift9 also from the leftupper corner to the right-lower corner, meaning that a longer evaporation time resulted in a lower flux and a higher selectivity. 4. Finally, comparison of membranes PEI-3 and PEL5 indicates that a decrease in the polymer concentration in the casting solution resulted in a lower flux and a lower selectivity, although the change was not significant. Permeability and Permselectivity. Looking at the data on Figure 3 more in detail, the permeability, J, [mol/ (m2*e.Pa)], is plotted versus membrane selectivity,defined as the permeability ratio, J J J N , of organic vapor and nitrogen, for five different organic vapors. Looking at the data with the same symbol, for example 0 for n-pentane, there is a trend that J, decreases with an increase in J,/ JN ratio, indicating that there is a trade-off between permeability and selectivity. Some data, however, clearly show that it is possible to obtain one membrane that is higher both in permeability and selectivity than another by adjusting membrane preparation conditions properly. Although it is generally considered that elastomeric polymers such as silicone rubber should be used for the separation of organic vapors from gas streams, the experimental results here show that the asymmetric membranes prepared from glassy poly(ether imide) are more permeable to organic vapors. The selectivity of the membrane is even higher than the silicone rubber mem-
536 Ind. Eng. Chem. Res., Vol. 32, No. 3, 1993 Table I. Membrane Preparation Conditions PI-1 PI-2 casting soln composn, w t % 25 25 polyimide (2080) poly(ether imide) (1OOO) 72.5 72.5 dimethylacetamide 2.5 2.5 lithium chloride lithium nitrate 45 22.5 temp of casting soln, 'C a a casting atmosphere 22.5 22.5 temp of casting atm, O C b b humidity of casting atm 95 70 solvent evapn temp, OC 40 10 solvent evapn time, min C C gelation medium 20 20 gelation period, min d d solvent for replacing water in film 24 30 solvent exchange time, h e e drying of membrane a Ambient air. 60-65%. c Ice-cold water (1-3 "C).
PEI-1
PEI-2
PEI-3
PEI-4
PEI-5
PEIB
PEI-7
30.95 69.05
30.95 69.05
30.95 69.05
30.95 69.05
25 74
25 74
25 75
25 74
25 74
1
1
1
23 a 23 b 23 0.05
23
23
22.5
24
24
23
a
a
a
a
a
a
23 b 100 1.33
23 b 100 2
22.5 b 100 8
24 b 100 2
24 b 100 0.5
23 b 23 0.1
22.5 a 22.5 b 110 1.8
1 23 a 23 b 24 0.6
C
C
C
C
C
C
C
C
C
20
20
24
20
d
d
d
d
30
30
20 d 48
20
d
20 d 30
20
30
20 d 30
30
30
24
e
e
e
e
e
e
e
e
e
d
PEI-8
PEI-9
Ethanol. e Air-dried. Table 11. Comparison of Permselectivities of Poly(ether imide) a n d Silicone Rubber Membranes poly(ether imide)O organic vapor methanol n-pentane cyclohexane n-heptane toluene TRAPS
VACIJ~
silicone rubber
JvbX lo9
JJJN
20.9 1.13 7.52 10.8 19.9
1024.3 65.2 434.5 621.5 1147.1
Jv/J N ~ ~~
46.4 66.8 30.4
a P E I d membrane. In mol/(m2s Pa). Calculated from General Electric Bulletin GEA-8685B.
VACUUM REGULATOR
10.~ I
I
1
I
l
1
I
1
PUMP
Figure 2. Schematic diagram of experimental setup. L T
0=
Id8
a
Methanol
m
0 = n-Pentane
"E
Cyclohexane + = n-Heptane X = Toluene A =
'1 0
E
I
,
>
10.' I
I
ELL333 10 = 0= A = 10 = H=
toluene methanol n-heptane cyclohexane) n-pentene
lo-'o
0
20
10
60
EO
100
Temperature, "C Figure 4. Temperature dependence of permeability of organic vapors. Membrane, PEI-8; feed, nitrogen + organic vapor (97-99% saturated); permeate pressure, 0.18-1.1 kPa.
lo-Q 1
10
100
1000
10000
Selectivity, J " / J ~ Figure 3. Permeability vs selectivity for organic vapors through P I and PEI membranes. Feed, nitrogen + organic vapor (97-99% saturated); temperature, 23 "C; permeate pressure, 0.18-1.1 kPa depending on the membrane permeability.
brane as shown in Table 11. Presumably, this high selectivity is caused by the transport of organic vapors by the surface flow mechanism; i.e., the organic vapor is adsorbed on the surface of the membrane pore. The
multilayer of the organic vapors so formed is driven by the spreading pressure on the surface to the low-pressure side (Gilliland, 1958). When the contribution of the surface flow surpasses that from the Knudsen flow, the permeation rate of the organic vapor can become much higher than that of nitrogen which is not adsorbed to the surface. Effects of Some Operating Variables. The permeabilities of five organic vapors are plotted versus temperature in Figure 4 in the temperature range of 23-83 ' C with respect to the PEL8 membrane. The permeability decreases with an increase in temperature except for toluene. The permeability change in the above temper-
Ind. Eng. Chem. Res., Vol. 32, No. 3, 1993 537 l2
toluene n-heptane 0 cyclohexane 0 methanol in-pentane
0
v "E 'L:
I
A
2 -
10
c ? a
"E
LE
6
v3
$: X
CP.'
4
2
0 0
Figure 5. Effect of feed vapor concentration on vapor permeation flux. Membrane, PEI-&temperature,23 "C;permeatepressure,O.l% 1.1 kPa.
k 0
3
n-pentane cyclohexane n-heptane = toluene
,
I
I
6
9
12
2
3
4
5
6
7
[(P,YJ - ( P ~ Y ~ ) ' ] / ' I OPa2 ~, Figure 7. Linear relationship between QY and [@1Y# for n-pentane.
zero when the permeate pressure approaches their respective saturation vapor pressures of 5.8 and 3.6 kPa.
0= 0= A = X
1
I 15
Permeate pressure, kPa Figure 6. Effect of permeate pressure on vapor permeation flux. Membrane,PEI-8; feed,nitrogen + organicvapor (97-99% saturated); temperature, 23 O C .
ature range is, however, below 50% in contrast to a sharp decrease reported in the literature with respect to silicone rubber (Behling et al., 1989;Buys et al., 1990). Figure 5 shows the effect of feed vapor concentration on the vapor permeation flux. The pressure on the permeate side was kept below 1.1 kPa throughout the experiments. The permeation flux increases with an increase in feed vapor concentration in nonlinear fashion as observed by Buys et al. (1990). The effect of permeate pressure on the vapor permeation flus is shown in Figure 6. The feed vapor concentration was kept between 97% and 99% of saturation vapor pressure throughout the experiments. The permeation flux decreases with an increase in the permeate pressure in nonlinear fashion and approaches zero as the permeate pressure approaches the saturation vapor pressure. For example, the flus for n-heptane and toluene approaches
Transport Study Equations 3-5 were used to analyze experimental data obtained at 23 "C. In particular, eqs 3 and 21 assume that nitrogen gas permeability is unaffected by the presence of organic vapors in feed gas mixtures. The same assumption was made in our earlier work (Feng et al., 1990,1991).In order to test this assumption, the permeability of nitrogen gas saturated with methanol and n-pentane at 23 OC was determined by experiments. Methanol and n-pentane were chosen for this test since the saturation vapor pressures of these organic compounds are the highest among all organic compounds involved in this study. Therefore, the strongest effect is expected from vapors of the above two organic compounds. Compared with pure nitrogen, 13% and 18.6% increase in nitrogen permeability were observed in the presence of methanol and n-pentane vapor, respectively, indicating that eqs 3 and 21 are valid approximations for practical purposes. Under the above assumption, A in eq 3 was evaluated to be 1.73 X lo-" mol/(m2.s.Pa) using eqs 2 and 21. This A value will be used throughout the rest of this paper. All the data from permeationexperimentswere analyzed by using eqs 3-5 and three unknown quantities, QN,B, and Y3,were calculated. Figures 7 and 8 illustrate the relationship between Qv and [@1Yd2 - @3Y3l21. Note that eq 4 implies that the relationship should be linear and the slope should be equal to the quantity B. The scatter of the data along the straight line measures the scatter in the numerical values of B. Thus, the straightline relationship presented in Figures 7 and 8justifies the use of eqs 3-5 for the analysis of permeation data. The slopes of straight lines in the figure are listed in Table I11 with the parameter B. It is interesting to note that there is a strong correlation between the boiling point of hydrocarbons and the B value. The methanol B value is higher than expected from the boiling point probably due to a stronger interaction between alcohol and poly(ether imide) membrane material. Using the B values listed in Table 111,instead of Qv values, the vapor mole fraction in the permeate, Y3,was recalculated by eqs 3-5. The results
538 Ind. Eng. Chem. Res., Vol. 32, No. 3, 1993 10
sc
U= O = A =
08
0
c
O =
L
c
e
06 I e
i
P
Y
2
04
u e
I
i F:
-g
02
n
C.30
I
00 30
1
c2
04
0.15
0.10
3.35 .
1
08
06
,
10
Feed vapor mole fraction 0
1
2
3
[(plyJ - ( P ~ Y ~ ) ' ] / I O ~Pa' ,
Figure 9. Permeate vapor concentration, Y3,as a function of feed vapor concentration, Y1. Solid lines correspondto Y3 values calculated by using average E; symbols correspond to Y3 values calculated by using experimental Qv.
Figure 8. Linear relationship between Qv and [@1Yd2 - @~Y3)~1 for methanol, cyclohexane, n-heptane, and toluene. 3
Table 111. Boiling Point of Organic Vapor and its B Value for Permeation through PEI-8 Membrane at 23 OC boiling point," O C E,mol/(m2s Pa2) organic vapor 1.1 x 10-12 methanol 64.7 n-pentane 36.3 1.9 x 10-14 cyclohexane 80.5 6.0x 10-13 n-heptane 98.4 2.8X IO-'* toluene 110.8 6.6X 10-l2 ~~
a
5
Perry et al., 1984.
are plotted versus feed vapor mole fraction, Y I ,as solid lines in Figure 9. Solid lines of Figure 10 illustrate similar plots but versus the permeate pressure, p3. Symbols indicate Y3 values obtained from calculations using experimental Qv data. The agreement between the symbols and the lines indicates that average B values represent the experimental data reasonably well. From Figure 10we claim the following conclusions. The vapor mole fraction in the permeate, Y3, is above 0.87 for allexperiments. The permeate pressure, p3, has little effect on Y3,when the liquid is as volatile a4 methanol and n-heptane.
Conclusions The following conclusions can be drawn from this work 1. Dry asymmetric poly(ether imide) membranes prepared by the phase inversion technique and dried by the solvent exchange method can be used to separate organic vapors from organic vaporhitrogen gas mixtures. 2. The membrane selectivity increases and the flux decreases with an increase in evaporation temperature and an increase in evaporation time. 3. Membrane performance data are affected by operating conditions such as temperature, permeate pressure, and feed concentration. 4. The proposed transport equations are applicable to analyze membrane permeation data. For a given temperature these equations involve only two adjustable parameters which are obtainable from one experiment for pure nitrogen permeation and one experiment under a single set of permeate pressure and feed vapor concentration. These parameters enable the prediction of the
1. 0 = 2.0 =
4
methanol n-pentane
3 . 0 = cyclohexane 4, A = n-heptane 5 , x = toluene
0
5
10
15
Permeate pressure, kPa Figure 10. Permeate vapor concentration, Y3, as a function of permeate pressure, p3. Lines and symbols have the same meaning as in Figure 9.
membrane performance for different permeate pressures and feed vapor concentrations.
Acknowledgment The authors are grateful to Energy, Mines and Resources Canada for their financial support to this project under DSS Contract No. 23440-9-9237/01-88. X.F.also wishes to thank the School of Graduate Studies and Research, University of Ottawa, for an Academic Travel Award. Nomenclature A = parameter characterizing nitrogen permeation through membrane, mol/(m2 s Pa) B = parameter characterizing vapor permeation through membrane, mol/(m2 s Pa*) cv = local vapor concentration in the membrane, mol/m3 Dov = diffusivity of vapor in the membrane at unit vapor concentration, m2/s D, = diffusivity of vapor in the membrane, m2/s J = permeability, mol/(m2s Pa) k h v = Henry's law constant applicable to vapor sorption, mol/ (m3Pa)
Ind. Eng. Chem. Res., Vol. 32, No. 3, 1993 539 = Henry’s law constant applicable to vapor adsorption, kg/(kg Pa) p = pressure, Pa Q = molar permeation flux, mol/(m28 ) S = effective membrane area, m2 V = volumetric permeation rate, mL/min 3c = distance to the flow direction from the feed-membrane boundary, m 3cv = amount of vapor adsorbed in the membrane material, kdkg Y = mole fraction of organic vapor k’hv
Greek Letters a = constant defined by eq 16 6 = effective membrane thickness, m Subscripts N = nitrogen v = organic vapor 1 = upstream (feed) side of the membrane 3 = downstream (permeate) side of the membrane
Literature Cited Baker, R. W.; Yoshioka, N.; Mohr, J. M.; Khan, A. J. Separation of Organic Vapors from Air. J. Membr. Sci. 1987, 31, 259. Behling, R. D. In Proceedings of the Sixth Annual Membrane Technology Planning Conference, Session V-4: Cambridge, MA, Business Communications Co., Inc.: Stamford, CT, 1986. Behling, R. D.; Ohlrogge, K.; Peinemann, K. V.; Kyburz, E. The Separation of Hydrocarbons from Waste Vapor Streams. AZChE Symp. Ser. 1989, 85 (No. 272), 68.
Billmever. F. W., Jr. Textbook of. Polymer Science, 3rd ed.; Wiley: New York, NY, 1984; p 520. Buvs. H. C. W. M.: Martens. H. F.: Troost. L. M.: Van Heuven. J. W:; Tinnemans, A. H. A. New Intrinsic Separation Characteristics of Poly(dimethy1 siloxane) Membranes for Organic Vapour/Nz Gas Mixtures. In Proc. 1990 Znt. Conf. Membr. Proc.: Chicago, 1990; p 833. Feng, X.; Tezel, H.; Sourirajan, S.; Matauura, T. Organic Vapour Separation From Air by Aromatic Polyimide Membranes. Presented at the 1990 Canadian Institute of Chemistry conference, Halifax, Nova Scotia, July 15-20, 1990. Feng, X.; Sourirajan, S.;Tezel, H.; Matauura, T. Separation of Organic VaDor from Air by Aromatic Polyimide Membrane. J. A _p _p l . Poiym. Sci. 1991, 43, 1071. Gilliland. E. R.: Baddour. R. F.: Russel. J. L. Rates of Flow Through Microporous Solids. AZChE J. 1958,4, 90. Kimmerle, K.; Bell, C. M.; Gudernatach, W.; Chimel, H. J. Membr. Sci. 1988, 36, 477. Paul, H.; Philpsen, C.; Gerner, F. J.; Strathmann, H. J. Membr. Sci. 1988, 36, 363. Perry, R. H., Green, D. W., Maloney, J. O., Eds. ChemicalEngineers’ Handbook, 6th ed.; McGraw-Hill: New York, NY, 1973; Section 3, pp 3-1-3-44. Pinnau, I.; Wijmans, J. G.; Blume, I.; Kuroda, T.; Peinemann, K. V. J . Membr. Sci. 1988, 37, 81. Strathmann, H.; Bell, C. M.; Kimmerle, K. Pure Appl. Chem. 1986, 58, 1663. Theodore, L.; Buonicore, A. J. Air Pollution Control Equipment; CRC Press: Boca Raton, FL, 1988; Vol. 2, Gases. Wijmans, J. G.; Helm, V. D. A Membrane System for the Separation and Recovery of Organic Vapors from Gas Streams. AZChE Symp. Ser. 1989, 85 (No. 272), 74.
-
Receiued for reuiew April 3, 1992 Revised manuscript received July 30, 1992 Accepted December 7, 1992
