Greenstein, J. P., Winitz, M., “Chemistry of the Amino Acids,” Vol 2 , p 1309, Wiley, New York, 1961. Howery, D. G., Vermeulen, T., Birmingham Unit. Chem. Eng., 15, 72 (1964). Hybarger, R., Tobias, C. W., Vermeulen, T., Ind. Eng. Chem. Process Des. Develop., 2, 65 (1963). Masson, M., Kavanaugh, M., Univ. of Calif., Berkeley. unpublished data, 1965. Morris, C. J. 0. R., Morris, P., “Separation Methods in Biochemistry,” Interscience, New York, 1963. , Nady, L., Univ. of Calif., Berkeley, unpublished data, 1965. Porath, J., Ark. Kemi, 11, 161 (1957). Pucar, Z., in “Chromatographic Reviews,” Vol 3, p 38. M. Lederer, Ed., Elsevier, Amsterdam, 1961. Ravoo, E., Gellings, P. J., Vermeulen, T., Anal. Chim. Acta, 38, 219 (1967). Schoen, H. M., Ed., “New Chemical Engineering Separation Techniques,” Interscience, New York, 1962.
Vermeulen, T., Tobias, C. W., “Annular-Bed Electrochromatography,” Report 1, N I H Grant GM-08042 (1962). Vermeulen, T., Tobias, C. W., ibid., Report 2 (1963). Vermeulen, T., Tobias, C. W., ibid., Report 3 (1964). Vermeulen, T., Tobias, C. W., ibid., Report 4 (1965). Vermeulen, T., Tobias, C. W., ibid., Report 5 (1966). Werum, L. N., Gordon, H. T., Thornburg, W., J . Chromatogr., 3, 125 (1960). RECEIVED for review March 30, 1970 ACCEPTED September 21, 1970
Initial phases of the work described were performed under the sponsorship of the National Institute of General Medical Sciences, IJnited States Public Health Service, through Research Grant GM-08042. Later phases were done under the auspices of the United States Atomic Energy Commission.
Reverse Osmosis Separation of Some Organic Solutes in Aqueous Solution Using Porous Cellulose Acetate Membranes Takeshi Matsuura and S. Sourirajan Division of ChemistT, National Research Council of Canada, Ottawa, Canada Reverse osmosis experiments were carried out for the systems glucose-water (0.1 to 1 S M ) , maltose-water (0.03 to 0.1 l M ) , lactose-water (0.04 to 0.22M), ethylene glycol-
-
water (0.2 to 1 S M ) , propylene glycol-water (0.2 to 0.8M), and ethylene glycol-propylene glycol-water (total molality
1 .OM) in the concentration ranges indicated. The correla-
tions of data for the single solute systems were similar t o those reported for the system sodium chloride-water. A relative scale of membrane selectivity i s given for the organic solutes in terms of solute transport parameters. The results obtained with the mixed solute system show that the prediction technique already developed for aqueous solution systems containing mixed inorganic solutes with a common ion i s applicable for systems containing nonionic mixed organic solutes. The performance of a typical film for the concentration of aqueous glucose solution i s illustrated.
T h e reverse osmosis separation, concentration, and fractionation of organic solutes in aqueous solution are of practical industrial interest. Detailed reverse osmosis studies on the systems glycerol-water, sucrose-water, and urea-water have been reported (Kimura and Sourirajan, 196813; Ohya and Sourirajan, 1969a; Sourirajan, 1967; Sourirajan and Kimura, 1967). This paper is concerned with similar studies for the systems glucose-water, maltosewater, lactose-water, ethylene glycol-water, propylene glycol-water, and ethylene glycol-propylene glycol-water, using the Loeb-Sourirajan type porous cellulose acetate membranes. Experimental
Reagent grade solute substances and porous cellulose acetate membranes (CA-NRC-18 type films) made in the laboratory were used. These films were made in accordance 102
Ind. Eng. Chem. Process Des. Develop., Vol. 10, No. 1, 1971
with the general method described earlier (Sourirajan and Govindan, 1965) using the following composition (weight per cent) for the film casting solution: acetone, 68.0; cellulose acetate (acetyl content = 39.8), 17.0; water, 13.5; and magnesium perchlorate, 1.5; the temperature of the casting solution and that of the casting atmosphere were the same a t -10°C. The film details, the apparatus, and the experimental procedure have been reported (Sourirajan, 1964; Sourirajan and Govindan, 1965). Membranes shrunk a t different temperatures were used to give different levels of solute separation a t a given set of operating conditions. The effective area of film used was 7.6 sq cm. All membranes were initially subjected to an initial pressure treatment a t 1700 psig with pure water for an hour prior to subsequent use in reverse osmosis experiments. All experiments were carried out a t the laboratory temperature (23-5” C ) . Unless otherwise
Table I. Osmotic Pressure a n d Molar Density D a t a for Aqueous Solutions of Some Organic Solutes Solute D-Glucose
Molality
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.2 1.4 1.6 1.8
Mole Fraction X loJ
0 1.798 3.590 5.375 7.154 8.927 10.693 12.453 14.207 15.955 17.696 21.160 24.600 28.016 31.408
Osmotic pressure, psi
0 37.5 75 112.5 150 187.5 220 253
Molar density, mole/cc.
x lo2
Lactose Osmotic pressure, psi
5.535 5.494 5.446 5.394 5.342 5.291 5.245 5.195 5.152 5.105 5.065 4.979 4.898 4.822 4.746
0 31 66 105
Maltose
Molar density, mole/cc.
x lo2
5.535 5.432 5.335 5.238 5.149 5.060 4.970 4.886 4.809 4.733
stated, the experiments were of the short-run type each lasting for about 2 hr. The reported product rates are those corrected to 25°C. using the relative viscosity and density data for pure water. I n all experiments, the terms “product” and “product rate” refer to membrane permeated solutions. I n each experiment involving only one solute, the solute separation, f , defined as
f=
molality of feed (m:) - molality of product (m3) molality of feed (mJ
(1)
and the product rate [PR], and the pure water permeation rate [PWP], in grams per hour per given effective area of film surface were determined at the specified operating conditions of pressure, feed concentration, and feed flow rate. The solute concentrations in feed and product solutions were determined from refractive index readings using a Bausch and Lomb refractometer. The accuracy of the separation data is within 1%, and that of [PWP] and [ P R ] data is within 2%. In each experiment involving the mixed solute system ethylene glycol-propylene glycol-water, the data on [PWP], [PR], and f were obtained expressing [PWP] and [ P R ] as for single solute systems, and the values of f as calculated from the relation
Osmotic pressure, psi
0 31 66 05 44 85
Molar density, mole/cc.
x lo2 5.535 5.432 5.335 5.238 5.149 5.060 4.970 4.886
Ethylene Glycol Molar density, mole/cc.
Osmotic pressure, psi
x lo2
Propylene Glycol Osmotic pressure, psi
5.535 5.515 5.495 5.476 5.456 5.437 5.419 5.400 5.381 5.363 5.345 5.310 5.275 5.246 5.207
0 35 72 109 145 181 216 254 291 328 363 435 506 578 647
0 31 64 99 136 173 211 251 291
Molar density, mole/cc. x 102
5.535 5.505 5.417 5.448 5.421 5.393 5.366 5.339 5.312 5.286 5.261 5.210 5.161 5.118 5.067
Results and Discussion
Physicochemical Data. The data on osmotic pressure and molar density for the solution systems studied are listed in Table I. These data were obtained from the osmosity data of Wolf and Brown (1969), osmotic pressure data of Sourirajan (1970) for the system NaC1-H20, and the density data taken from Timmermans (1960). Wherever necessary for purposes of reverse osmosis calculations, the osmotic pressure a t a higher concentration (mole fraction) was obtained by linear extrapolation. Membrane Specifications. Assuming constant molar density, c, for the feed and product solutions, the KimuraSourirajan analysis (Kimura and Sourirajan, 1967) leads to the following basic reverse osmosis transport equations which have been derived (Sourirajan and Agrawal, 1969):
A=
[PWPI (MB x S x 3600 x P )
(3)
(4) (5)
f= molality of feed (ml)Ai - molality of product molality of feed ( m l ) ~ i
(m3)Ai
(2)
where ( m , ) ~ and i (m3)Ai represent the molality of the particular solute Ai (ethylene glycol or propylene glycol) in feed and product solutions respectively. The concentrations of ethylene glycol and propylene glycol in the feed and product solutions were determined by gas chromatography using Beckman GC 2A, Porapak Q column (6-ft x 0.25-inch od) a t 220” C, helium flow rate of 33.7 ccimin, and water as internal standard; the accuracy of these determinations was within 270. Reverse osmosis experiments were carried out with a t least six different membranes covering a wide range of solute separations and only a few typical results are reported.
The symbol k refers to the mass transfer coefficient on the high pressure side of the membrane. The parameter (DaM/K6)plays the role of a mass transfer coefficient with respect to solute transport through the membrane; hence it is treated as a single quantity for purposes of analysis. For isothermal reverse osmosis operation, the following correlations have been established for the variations of A, and ( D A ~ / K dand ) k for the system sodium chloridewater (Sourirajan and Agrawal, 1969).
A
(7)
a
and
(%)
cc
p-0
Ind. Eng. Chem. Process Des. Develop., Vol. 10, No. 1, 1971
103
Table II. M e m b r a n e Specifications. System: Sodium Chloride-Water
* 2
4
Operating pressure: 102 atm Film No
1 2
8
A X lo6, Gmol H*O/ Sq C m Sec Atm
(Dn,v/K6) X 1 0 ,
2.00 2.88 3.40 6.77 6.34 7.93 1.63 2.07
8.8 12.6 176.0 643.8 535.3 934.1 7.2 29.9
2
Cm/Sec
loot
-1
F I L M TYPE C A - N R C - 1 8 SYSTEM
GLUCOSE-WATER
FEE0 MOLALITY
0.5M
pl
W
I-
2 At a given operating pressure, (DAMIK6) is independent of X A and ~ k. The mass transfer coefficient is a function of feed concentration and feed flow rate, Q, and for practical purposes, essentially independent of operating pressure and solute separation. The values of h can be correlated by the relation
60
t /
7.6 c m 2
F O R F I L M NO. 3
DATA
/
OPERATING PRESSURE a t m
(9) On the basis of the above correlations, it has been shown (Sourirajan and Agrawal, 1969) that a t a given operating pressure, the values of A and (DAM/K~),,,, specify a film completely. Such specifications for the films used in this work are given in Table 11. Analysis of Reverse Osmosis Data for Some Organic Systems. Reverse osmosis experiments were carried out for the systems glucose-water (0.1 to 1.5M), maltose-water (0.03 to O.llM), lactose-water (0.04 to 0.22M), ethylene glycol-water (0.2 to 1.5M), and propylene glycol-water (0.2 to 0.8M) in the concentration ranges indicated a t the operating pressure of 102 atm and feed flow rate of 500 ccimin. The effects of operating pressure (in the range 17 to 102 atm) and feed flow rate (in the range -100 to 700 ccimin) on membrane performance were also studied for the system 0.5M glucose-water. The reverse osmosis data for six different films covering a wide range of solute separations, were analyzed using Equations 3 t o 6. One set of results is illustrated in Figures 1 to 6, and the other results were similar. The results show that P us. log A is a straight line for each film (Equation 7) and increase of operating pressure increases both solute separation and product rate (Figure 1). The log-log plot of P us. (DA,v/K6)is a straight line (Equation 8, Figure
Figure 2. Effect of operating pressure on solute transport parameter for glucose
500 cm3/mln
FEED RATE F I L M AREA
40
Figure 1. Reverse osmosis data for system glucosewater. Effect of operating pressure
2). At a given operating pressure, product rate decreases, and solute separation slightly decreases or passes through a slight maximum, with increase in feed concentration (Figures 3 and 4). Similar results have been reported (Kimura and Sourirajan, 1967; Sourirajan and Govindan, 1965). Both solute separation and product rate increase with increase in feed rate (Figure 5). The solute transport parameter is independent of feed concentration and feed flow rate for all the systems studied (Figures 3-5). In the concentration ranges studied, the values of k varied only within 10% for each system, and this variation was no more than that obtained with different films. At the feed flow rate of 500 ccimin, the average values of k obtained for the systems ethylene glycol-water, propylene glycol-water, glucose-water, maltose-water, and lactosewater were 29, 27, 25, 19, and 19 ( X cm/sec respectively. The average values of k and ( D A M I Kwere ~ ) used to back calculate solute separation and product rate as a function of feed molality in the manner of Kimura and Sourirajan (1968a), and the calculated values agreed well with the experimental values. The variation of average
(3)
105 60 -
crn/sec
-
*\
500 c m 3 / m l n DATA F O R F I L M NO 3
FEED RATE
40 -
*\ 30
104
Ind. Eng. Chem. Process Des. Develop., Vol. 10, No. 1, 1971
I
I
1
I
I
I
*\ I
l
l
FILM T Y P E : CA-NRC-I8 SYSTEM:GLUCOSE-WATER E T H Y L E N E GLYCOL-WATER 0 P R O P Y L E N E GLYCOL-WATER 0 F E E 3 RLTE 5 0 0 c m 3 / m n
.
-
I 3001
IO-
z 5
:
c
0.8+-..-
e I
I
-
no
-
e-
"
1
I
I
0.6'
I
I
&.-
0
:200 \ E
A
0
s5
OPERATING PRESSURE 1 5 0 0 ~ E 1 g DLTA FOR F I L M NO 3
I
120 F I L M A R E A : 7.6 cm2
a -*-
100
-
.;ye-
I
Figure 4. Reverse osmosis data for systems lactose-water and maltose-water. Effect of feed concentration 0.2
b
FILM TYPE: CA-NRC - I8 SYSTEM:GLUCOSE-WATER
c ZF it€
F I L M L R E A : 7.6 c m 2
I
I
0.5 FEED
I
1.0 MOLALITY
I
e
30--*
e.
0.4'
0
OPERATING PRESSURE,' I 5 0 0 p DATA FOR F I L M N O , 3
5 I
g
a-.
I
I
I
I
I
_
I
I
1
I
I
I
I
200
300
400
500
600
700
d
]
1.5
a n
Figure 3. Reverse osmosis data for systems glucose-water, ethylene glycol-water, and propylene glycol-water. Effect of feed concentration
h with feed flow rate for the system glucose-water is illustrated in Figure 6. The correlation given in Figure 6 is applicable only for the particular apparatus used. The correlations of reverse osmosis data illustrated in Figures 1 to 6 are similar to those reported for the system sodium chloride-water (Sourirajan, 1970; Sourirajan and Agrawal, 1969). Relative Scale of Membrane Selectivity. Figure 7 gives a relative scale of membrane selectivity for the solutes ethylene glycol, propylene glycol, glucose, maltose, and lactose in terms of their respective (Da,/ K6) values using sodium chloride as the reference solute. The log-log plots of ( D a ~ / K f values i) are similar to those given by Agrawal and Sourirajan (1969) for several inorganic and organic solutes. I n the range of (D4,V/K8)v,~1values 10 to 1000 (x10-j cm/sec)-which is a measure of the pore size on the membrane surface-the corresponding values for the solutes studied are in the order ethylene glycol > propylene glycol > glucose > maltose > lactose. (DaM/K8) for lactose is less than that for maltose even though their molecular weight and osmotic pressure and molar densities of their aqueous solutions are not different. The selectivity data of the type shown in Figure 7 are of practical interest for parametric studies. Systems Involving Mixed Organic Solutes. The predictability of reverse osmosis membrane performance for feed
100
FEED RATE
cm3/min
Figure 5. Reverse osmosis data for system glucose-water. Effect of feed flow rate
F I L M TYPE SYSTEM
..
CA-kRC- '8
GLUCOSE- WATER
FEED MOLALITY
0.5M
e'
.E *2
20.
x x
I_ ,c-
I
I
I
I
I
200
330
400
500
700
FEED RATE
crn3/min
Figure 6. Effect of feed flow rate on mass transfer coefficient for system glucose-water
solution systems containing mixed inorganic solutes with a common ion has been illustrated (Agrawal and Sourirajan, 1970). The system ethylene glycol-propylene glycolwater, AI-AB-HyO, a t a total feed molality of -1.OM was studied with a view to find out whether the same prediction technique was applicable to nonionic organic mixed solute systems. The study included three different membranes, three different feed concentrations, two Ind. Eng. Chem. Process Des. Develop., Vol. 10, N o . 1, 1971
105
H 2 0 (where A1 and A2 represent the solutes ethylene glycol and propylene glycol respectively) the graph of the product rate at total molality us. fractional molality of solute (molality of A2ltotal molality) is a straight line for all the films tested. Defining the fractional molalities zl, and x z as
and
the correlations given in Figure 8 may be expressed as
[PR]mix = XI[PR]AI+ x? [ P R ] A ~
5
100
50
IO
500
1000
(h) x10 5 cm/sec Kb
NaCl
Figure 7. A relative scale of membrane selectivity for organic solutes different feed flow rates, and two different operating pressures. The results obtained are illustrated in Figure 8 and Table 111. Figure 8 shows that a t a given set of operating conditions of pressure, total feed molality, and feed flow rate, the product rate obtained with the mixedsolute system is intermediate between those obtained for the respective single-solute systems at the same total feed molality. Representing the mixed-solute system as Al-A2-
(12)
where [PR]fix, [PR]A1,and [PR]Ag represent, respectively, the product rate for the mixed-solute system, and the product rates for the single-solute systems A1-H20 and A2-H20 at the same total feed molality. From the correlation of product rates represented by Equation 1 2 , and the membrane performance data for the single-solute systems a t the given total molality, solute separations and product rates for the mixed-solute system were calculated following the procedure of Agrawal and Sourirajan (1970), and they were compared with the experimental values. The results, given in Table 111, show good agreement between calculated and experimental values illustrating the applicability of the prediction technique for aqueous solutions involving mixed nonionic organic solutes of similar chemical nature. Concentration of Glucose in Aqueous Solutions. The reverse osmosis concentration of dilute glucose solutions, and solute recovery and processing capacity of membrane obtainable during such concentration are of practical interest. The effect of feed concentration on solute separation and product rate for the system glucose-water is illustrated in Figure 9 for a typical film suitable for the
Table Ill. Reverse Osmosis D a t a for the System Ethylene Giycol-Propylene Glycol-Water
Film type: CA-NRC-18; System type: A1-A2-H20;Solute molality in feed =
Total Film No,
(mI)A~/z(ml)A