0
ION EXCLUSION A Unit Operation Utilizing Ion Exchange Materials
e
R. M. WHEATON
AND W .
Physical Research laboratory, The
C. BAUMAN
Dow Chemical C o . , Midland, Mich,
A
NUMBER of investigators into the theoretical behavior of ion exchange resins have, by various methods, shown that when a n ion exchange resin is placed in a dilute solution of an electrolyte, the concentration of that electrolyte is lower within t h e aqueous portion of the resin phase than in the surrounding solution. I n sulfonated cross-linked polystyrene resins, for example, the inside concentration may be only one tenth as great as the outside concentration for a strong electrolyte such as sodium chloride or hydrochloric acid. Bauman and Eichhorn ( 3 ) have shown this to be true with Dowex 50, hydrogen form, after equilibration with standardized hydrochloric acid. The commercial resin a t that t h e cont,ained about l2yO divinylbenzene. Since then the cross linkage has been reduced in several steps to 8y0,lowering the internal ionic concentration of fixed groups and decreasing to a sniall extent on equilibration the partition between the aqueous and resin phases. Comparative curves are shown in Figure 1. To date there has been far less reported on the distribution of a nonpolar material (or slightly ionized material) between aqueous and resin phases. One might assume from the Donnan mernbrane theory as it applies to elect,rolyte behavior that nonionic materials would be equally concentrated in the aqueous portion of the resin and the surrounding solution. K i t h the data presently available this applies for a large number of niat'erials though there is generally some additional adsorptionperhapscaused by the affinity of the nonelectrolyte for the hydrocarbon matrix of the ion exchange material. Gregor et al. ( 6 ) have made such studies on a sulfonated cross-linked polystyrene resin Dowex 50. They showed that the relative inside and outside concentration of nonionic material is dependent on the ionic form of the resin and equilibration temperature. Time is a factor to be conaidered i n reaching equilibrium and particle size has a small effect. The behavior of urea and ethyl acetate was studied most thoroughly. Urea, in the outside solution, in one study, was maintained a t 0.1 molal. Inside concentration then varied from 0.131 to 0.172 molal for different ionic forms a t 25" C. and from 0.117 to 0.152 molal for the same forins a t 43' C.; increasing tcmperatures decreaeed adsorption. At 25' C. ethyl acetate had a n outside concentration of 0.25 molal a t equilibrium wit,h the lit'hiuni form of Dowex 50 a t 0.297 molal. The sodium form showed 0.331 molal, the potassium form 0.383 molal, and the tctramethg-lammonium form only 0.090 molal. The greatest concentration gradient shown, h o ~ e v e r ,for the nonionic (outside; inside) is about 1:3, and in some cases this selectivity is reversed, whereas ionics are excluded by as much as 10: 1. Other examples of molecular adsorption of nonpolar or slightly ionized compounds on ion exchange materials have been reported by Barrer ( 1 , 2 ) for the zeolites, and by Cleaver and Cassidy (J), who showed that amino acids are sorbed by cation exchangers independent of true ion exchange. This brings up, then, the use of ion exchange resins as tools for fractionation of nonionic from ionic materials, by necessity in solution in mat,er or ot,her polar solvent. This fractionation has 228
been designated ' ion exclubion." I t promises greatly to broaden the utilitv of ion exchange rcsina and to improve many existing proreises and analytical methods. Theory of Ion Exclusion Ion esclusion t ? y e separations are visualized as taking place in a column the solut,ion of mixed electrolytes and nonelcctrolytes pasws don-nn-arti t>hroughthe column. As mentioned mr1it.r the inside of a resin particle contains a consitlei,able ainount of ~ a t e r . (The sodium form of Dowex *or 10
-
__
.~
IONIC CONCENTRATIONS /NSlOE Y S OUTSIDE
1
SULFONATED POLYSTYRENE
7 1 ,'y/
~
MOLALITY INSIDE RESIN
.IO
i
d /-/
.I'
/ /
.01
.01
O i
I O
IC
0
MOLALITY OUTSIDE R E S I N
Figure 1
ION EXCLUSION RESIN FEED
NORMALITY OF EFFLUENT
0.6
LEADING EDGE OF SEPARATION OOWEX S 0 , X S X 50-100 M E S H 100 Mi.. I15 N HCI, 0.66 N I C E T I G ACID
1
;1':
n
30
40
60
80
100
VOLUME EFFLUENT ( r n l . )
Figure 2
INDUSTRIAL AND ENGINEERING CHEMISTRY
January 1953
for 100 ml. of Dowex 50, X-S%, 50 to 100 mesh. been experimentally determined as follows :
ION EXCLUSION' TRAILING EDGE OF SEPARATION RESIN: DOWEX 54 X 8% 50-100 MESH F E E D : I O 0 ML I15 N HCl; 0.66 N ACETIC ACID
-
NORMALITY
.-.-.-.-.-.r.-.
e-**
---__
HCI HAC THEORETICAL
0.6 0.4
,0.2
0
100
I20
140
4
160
200
I80
Figure 3
ION EXCLUSION: COMPLETE SZPARATION RESIN: DOWEX 50 V. BY 50-100 MESH FEED.' I 5 ml. of / . I 7 N HCl #' 0.66 N NORMALITY OF ACETIC ACID EFFLUENT
0-0-0
. . -----
Similarly, the rinse curve shows the removal of all of the ionic material with V I volume of rinse while the nonionic component requires VI V Zrinse volumes (Figure 3). By combining the curves of Figures 2 and 3 and maintaining the feed solution a t V 3which must be less than V Z we , should get a complete separation of components A and B as shown in Figure 4, again observing the dotted line. If the cycle is repeated adding alternately a volume, V3, of solution and a volume, Va, of water, which volume must be greater than V 2 ,we should obtain an effluent curve indicating alternate and equivalent quantities of components A and B. Such a n operating curve is shown in Figure 5 and will be discussed later.
+
V O L U M E E F F L U E N T (ml..)
cw4
VI and Vz have
A 50-ml. sample of resin (bulk settled volume under water) was placed in a fritted glass filter funnel after which the water was drained t o the top resin level and discarded. Suction was then applied briefly to collect most of the interstitial water after which some additional water was collected by use of the centrifuge. The total volume represented VI. V Zwas then obtained from the loss in weight after drying the resin for 16 hours a t 110" C. These results represent the approximate averages of three determinations with individual deviations of 5 1%. These interstitial volumes check very closely with that reported by Brown (4). H e uses the term "porosity" which is plotted against the ratio, diameter of spheres t o diameter of bed. As this ratio approaches zero, which applies in these studies, the porosity is about 31% for smooth mixed spheres randomly packed. The porosity of systematically packed uniform beads may vary from 26 t o 48% depending on the geometric arrangement.
0.8
so
229
HCI HAC THEORETICAL
ION EXCLUSION: SEMi- CONTINUOUS CYCLES FLOW RATE: 0.6P g p m / f t P 1 VOL. FEED: 3 Voi. Rinse DENSITY
T
1.024
0.2
20
60
40
VOLUME
BO E F F L U E N T (ml.)
I00
I
Figure 4
50, 870 cross-linked, and hereafter written X-SY,, contains about 45% water.) I n addition to this, in usual operations, the void spaces between the resin particles are also filled with water. I n a uniform spherical product, this volume is about 30% of the total bed volume. This is hereafter referred to as interstitial volume. Assuming complete exclusion of the ionic component of an aqueous solution by the resin particle and no exclusion of the nonionic constituent (neither condition, of course, may be actually realized), the authors suggest the behavior of a n ion exchange column should be as follows: VI = interstitial volume of the resin column = solvent volume inside resin particles in total column C A O = initial concentration of ionic component A CB' = initial concentration of nonionic component B
Vt
+
If, then, a solution of A B is passed through the column the effluent curve as shown by the dotted lines of Figure 2 is obtained. The ionic material appears as soon as the influent volume equals the interstitial volume, VI, and soon reaches its initial concentration, CA'. The nonionic material, however, does not appear unVZ. It soon climbs to its til the influent volume reaches VI peak of CB'. As will be discussed later, a comparison is made here of the theoretical curve with a n actual operating curve. VI and V Zhave been experimentally determined t o be 30 and 42 ml., respectively,
+
1.016
1.008
1.000
I 0
20
I 40
I
60
I 80
I I00
I I20
I 140
I
160
I IS0
200
E L U T I O N TIME (M1na.l
Figure 5
Actually t o keep the boundaries as sharp as possible, V8 must be much less than VZand Vd much greater than V Zt o complete the separation of A and B. Furthermore, if the boundaries become too diffuse, a complete separation in a single pass will be impossible. Application of the Principle
of Ion Exclusion
Experimental evidence of the development of this theory may easily be shown in the hydrochloric acid-acetic acid system. As will be shown later, acetic acid ( K = 1.75 X 10-5) is typically nonionic with respect to this system-the more so in the presence of the strongly acidic Dowex 50, Hf. For the purposes of these experiments and others described here, unless otherwise specified, a 100-ml. buret was used. Resin bed dimensions were approximately 62 X 1.5 cm., with a fritted glass filter support. Feed solution was percolated through the resin bed up to the desired volume followed by a rinse with fresh water, all a t the same flow rate. Samples were collected by use of a Technicon fraction
INDUSTRIAL AND ENGINEERING CHEMISTRY
230
An example of a n experiment in which successive feeds and rinses were used, is shown in Figure 5, again with Dowex 50, X-S%, 50- to 100-mesh, but this time in a larger bed, 6.25 inches diameter by 60 inches. Here a mixture of ionic compound A and a slightly dissociated compound B were fed at constant flow rate in alternate shots of 1.57 and 4.71 gallons, respectively. Actual measurements were made with a density recorder. As is apparent, separations are good and reasonably reproducible. The possibilities of a timer-controlled cycle are obvious.
ION EXCLUSION: STRONG BASE TYPE RESIN R E S I N : DOWEX I . GI Y. 10% ' 50-100 M E S H FEED: 15 ml. 2 % N o C l - P% ETHANOL
4 35.
1.335
1.331
v
Scope of Ion Exclusion
I
I
I
I
30
40
50
60
I 70
EFFLUENT
I
I
80
90
VOLUME
I
100
I
I 110
(ml.)
Figure 6
collector, which collects a predetermined number of drops in successive test tubes. I n these fist runs with hydrochloric and acetic acids in mixture, the resin used was Dowex 50, X-8%, 50- t o 100-mesh hydrogen form. (Dowex 50 is a sulfonated cross-linked polystyrene type cation exchange resin. X-8% refers t o cross linkage as percentage divinylbenzene in the resin matrix.) The water content of the resin used here (Vz) was 42% of the total bed volume or 42 ml. Interstitial water volume (VI)was 30% of the total bed volume or 30 ml. These quantities change slightly in the presence of the strong electrolyte, hydrochloric acid, which causes resin shrinkage. Figure 2, solid lines, shows the separation of hydrochloric from acetic acid a t the early part of a run. A fraction of hydrochloric acid, entirely free of acetic acid, is attained before the acetic acid appears in the effluent and approaches the feed concentration. ION EXCLUSION: WEAK BASE (POLYAMINE) TYPE RESIN RESIN: DOWEX 3,- HCI FORM, 50-100 MESH FEED: 15 ML, 2 % NOCI, 2% CoHsOH
3 5.
1.334
The process is not limited t o strongly acidic cation exchange resins-for example, hydrochloric and acetic acids may be completely separated on the chloride form of a strongly basic anion exchange resin, such as Dowex 1,50-100 mesh (Figure 6). Similarly sodium chloride and ethanol have been completely separated on the chloride (hydrochloride) form of 50- to 100-mesh Dowex 3-a weakly basic anion exchanger (Figure 7). -4fair degree of separation was even realized with the basic form of 50- to 100-mesh Dowex 3 which is certainly much less ionized and hydrophyllic in
t
1.333
1.332
1.331
I 30
I 50
To this point most of the examples described have been with the hydrochloric acid-acetic acid system using Dowex 50. But one may logically wonder as to just how broad this method may be. What other separations may he carried out? What constitutes a n ionic material? Nonionic? What choice of resins is required? Several examples of separations thus far carried out are sodium chloride and ethylene glycol (and higher glycols), sodium chloride and formaldehyde, sodium chloride and glycerol, sodium chloride and ethylene diamine (and higher amines), sodium chloride and several alkanol amines, and others-all on the sodium form of Dowex 50. Cross linkage and mesh size were varied somewhat as described later. Other ionic forms of the resin may also be used. The resin should be in the same ionic form as the ionic portion of the materials to be separated. In actual practice this will, of course, adjust itself after the required number of cycles as the resin is converted or comes to equilibrium with the surrounding solutions. Though every example thus far of ion exclusion has shown the nonionic fraction to be an organic compound, ion exclusion is by no means thus limited. The separation of hydrochloric and boric acids, for example, may be successfully carried out by these techniques amplifying the theoretical explanation of the behavior.
Choice of Resins
?D 1.335
Vol. 45, No. 1
I
I
I
I
70
90
I10
I30
ION EXCLUSION: CARBOXYLIC TYPE RESIN FEED: 1 5 m l . 2 % NoCI, 2 % HCHO
EFFLUENT VOLUME ( m l )
Figure 7
Similarly Figure 3 shows the result of rinsing this bed with water. The rinse was started after 100 ml. of effluent were collected from the loading cycle and so VI is determined from t h a t point. The hydrochloric acid is essentially rinsed from the bed before the acetic acid and we have a fraction of acetic acid free of hydrochloric. I n each run the quite different slopes of the ionic and nonionic fractions are apparent. By introducing only 15 ml. of this feed into the resin bed followed by a water rinse the curve shown in Figure 4 was obtained. Here the feed volume, Va or 15 ml., is much less than the solvent volume inside this resin particle, Vz or 42 ml., and separation of hydrochloric and acetic acids is complete with no chemical consumption.
1334
1.333
I332
1.331 30
40
50
60
70
EFFLUENT
80
90
VOLUME ( m l )
Figure 8
100
110
120
130
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
January 1953
x
*
properties than the salt form. Finally the sodium salt form of a 50- t o 100-mesh ground carboxylic exchanger (Amberlite IRC 50) has also been used for the separation of hydrochloric acid and formaldehyde with good results (Figure 8). From these several runs, it is apparent that a n y of the ion exchange materials may be used with the provision that the resin be nonreactive with the components to be separated. The more highly ionized resins give considerably better results than those slightly ionized. I n the process evaluation t h a t follows all runs were made with Dowex 50. This resin is highly ionized, stable, gives low color throw, and is relatively cheap compared to the other resins mentioned previously. Though the Dowex 50 type of ion exchange resin was used as the basis for these studies, and in fact appears most promising for commercial application, the name is somewhat generic in that it applies to a family of cross-linked sulfonated polystyrene resins having quite different properties. Inasmuch as cross linking may be controlled prior to sulfonation by the addition of predetermined amounts of divinylbenzene, an accurately predetermined degree of cross linking is possible. Copolymer spheres have been prepared containing from 0.25% to greater than 50% divinylbenzene. The effects of these differences have been shown by a large number of investigators. I n addition to cross linkage it is also possible to vary particle size by controlling the conditions of suspension polymerization. The importance of these variations is realized in reviewing the principle of ion exclusion. It is necessary (1)that the fixed ionic concentration inside the resin particles be great; (2) that water content of the resin be high enough to give appreciable capacity; and (3) that equilibrium be rapidly attained-Le., that diffusion rates be high as possible. Requirement (1) favors a high crosslinked resin; (2) favors a low cross-linked resin; and (3) implies either low c r m linkage or small particle size or both. Thus, a compromise must be made depending on the requirement of a particular job. No one combination has been found which is most suitable in every case. In this series of experiments, Dowex 50, 50-100 mesh prepared with 1,2,4,8,12, and 16% divinylbenzene wasinvestigated for the separation of 15 ml. of solution containing 4Yo hydrochloric and 4% acetic acid by weight. Flow was maintained a t approximately 1 ml. per minute from the 100-ml. buret and all runs were carried out a t room temperature (20" to 30" C.). The runs as made with the 1, 4,and 167' cross-linked resins are shown in Figure 9. The dissimilarity is apparent. All the runs are summarized in Table I. These data and analyses of all the curves indicate that the best separations are attained with the 4 and 8yocross-linked resins; however, i t is equally noteworthy that separations are nearly complete in every case except with the 1% resin, and even
there it seems evident that reducing feed quantity or otherwise improving the factors for separation would make a complete separation possible.
35.
nQ I341
bx-a
"D
1.340 1.338 1.336
1.330
-
-a
I
I
I
I
I
I
I
40
48
56
64
72
EFFLUENT
80
58
96
VOLUME f m l )
Figure 9
104
I12
I20
128
I
What constitutes ionic and nonionic fractions? It soon becomes apparent that no sharp line of demarcation is possible. The most exhaustive study was made of the system of separations
Effect of Cross Linkage on Ion Exclusion Separation
Feed: 15 ml. 4% HC1-4% acetic mixture Resin: Dowex 50, 50-100 mesh, I-I *form Hydrochloric Acid, % Acetic Acid, % DVB Peak Av. Peak Av. 1 3.70 0.9 3.15 Und.0 2 3.75 1.2 3.90 2.0 4 3.54 1.7 4.02 2.0 8 3.95 2.1 4.20 2.0 12 4.00 2.3 4.55 2.0 16 4.05 2.5 4.30 2.0 Undetermined.
Effect of Particle Size on Ion Exclusion Separations
Feed: 15 ml. 4% HC1-4% acetic mixture Resin: Dowex 50. X-8% H form Hydrochloric Acid, o/c Acetic Acid, % Particle Size Peak Av. Peak Av. 16-20 3.24 1.5 1.80 Und.a 30-40 3.90 2.5 3.43 1.7 50-100 3.95 2.1 4.20 2.0 100-200 3.96 2.5 4.92 2.5 200-400 4.05 2.5 4.68 2.5 a Undetermined, +
32
I
Ionic versus Nonionic
Table 11.
332
I
The effect of particle size is very pronounced, pointing up the significance of rapid diffusion. An analysis of the elution curves makes possible a determination of the separation factor (theoretical plates) for a given resin by the method of Mayer and Tompkins (8). Running a t constant flow of 1 ml. per minute with a 4% acetic acid4Yo hydrochloric acid mixture the results shown in Table I1 were obtained. All these runs are on Dowex 50, X-8% H + form with 15 ml. of feed. The runs with 16- to 20-, 30- to 40-, and 100- to 200-mesh resinp are shown in Figure 10. All the resins used here were carefully screened to the desired particle range. The chief draw-back of the finer resins are the pressure drop through the beds and, to a lesser extent, the difficulty of proper backwashing and classification.
a
4.333
16-20 MESH 3 0 - 4 0 MESH 100-200 MESH
-.-. I
16% DVB
6 334
r-a-x
-
1.332
I % DVE
d 335
*-*-e
-
-
1.334
4 1 DVE
*-e-*
-
350
ION EXCLUSION: EFFECT OF RESIN CROSSLINKAGE RESIN: DOWEX 50, 50-100 MESH
-
ION EXCLUSION: EFFECT OF PARTICLE SIZE RESIN: DOWEX 5 0 . X B O FEED: I 5 m I 4 %HGI 4 % ACETIC AGIO
--
Table 1.
-
231
INDUSTRIAL AND ENGINEERING CHEMISTRY
232 Table 111.
Dissociation Constant ( K ) and Degree of Ionization of 0.25 N Acids
Table IV.
Degree of Ionization, Yo 100 58.0 35.8
K (7)
Acid HC1 ClaCCOOH ClzCHCOOH ClCHaCOOH CHsCOOH
2
x
10-1
5 x 10-1 1 . 4 X 10-2
Separation of Various Strength Acids by Ion Exclusion
Resin: Dowe x 60,X-8%, 50-100 mesh, H * form Feed soln.: 0.25 hr with respect to each acid Ionic h'onionic Degree of Separation Hvdrochloric Acetic Excellent Chloroacetic Excellent Dichloroacetic Excellent Acetic Good Chloroacetic Good Dichloroacetio Good Trichloroacetic Fair Hydrochloric Acetic Fair Dichloroacetic Chloroacetic Fair Dichloroacetic Acetic None Chloroacetio
involving every combination of hydrochloric, trichloroacetic, dichloroacetic, monochloroacetic, and acetic acids, all a t 0.25 N . The dissociation constants for all these acids are known and the degree of ionization is calculated in each case for 0.25 N (Table 111). The dissociation of the acids is less a t the start of the run in the presence of each other, but during the separation, when it takes place, the condition is changed, A more thorough study of the change of ionization during the run could certainly be used to good advantage. DENSITY
I
ION EXCLUSION: EFFECT OF FLOW RATE
1.024
1.020
1.016
1.012
1.008
1.004
1.000
I
I
I 20
10
ELUTION T I M E
I
I
30
40
I
(Minutes1
Because of the large number of runs involved in this series, the graphs of the elut,ion curves are not shown. Instead the results are summarized in Table IV. Although no sharp definition can be made between ionic and nonionic, the break point for this sytem seems t o be about K = 10-1 (between trichloroacetic and dichloroacetic acids). From these runs the following generalization may be made as to ionization requirements for separation:
Nonionic fraction
K K K K
>5 >2
A number of factors of operations, together or separately, determine the limitations of the method; these include flow rate, volume of feed, ionic concentration, nonionic concentration, temperature, and size of nonpolar molecule. A thorough study of all these factors would involve a tremendous amount of research, and this work may yet be required in choosing the optimum conditions for a given system. Studies on the effect of flow rate have been made in a pilot plant column, containing a resin bed 6.25 X 60 inches of Dowex 50, X-87,, 50- to 100-mesh. I n this series a volumeof feed solution equivalent to approximately 15y0of the bed volume and containing an ionic component ( A ) and a nonionic component ( B ) in aqueous solution was run through at increasingly rapid flow rates until the degree of separation was less than desired for a given process. Even a t the most rapid flow rate (1.74 gallons per square foot per minute) and separating the fractions a t the minimum, fraction A contained essentially all of component A , and fraction B contained 88% of component B. Comparative curves are plotted in Figure 11 for runs at flows of 0.63, 1.08, and 1.74 gallons per square foot per minute. By comparison the flows described in this paper through the buret a t 1ml. per minute are equivalent to 0.14 gallon per square foot per minute. To illustrate cycle time a t these various flow rates-Le., time required from first appearance of ionic fraction t o last of nonionic-only 18 minutes are required a t the flow rate of 1.74 gallons per square foot per minute. This is in sharp contrast to some of the very slow chromatographic separations described in the literature. With such accelerated rates, chromatography becomes more feasible for commercial installation. The volume of feed (V,)which may be separated in a column is known to depend on the other variables discussed here and ia limited theoretically in that it will be less than the free volume inside the resin beads ( V z ) which , is 42 ml. for the 100-ml. column. As stated earlier, Va must be much less than Vz. Inasmuch as ion exclusion depends on the Donnan membranr equilibrium discussed early in this paper, it may be assumed that the separations will be better at low ionic concentration. Again this has been studied with the hydrochloric-acetic acid system, maintaining acetic acid concentration at 4% and increasing hydrochloric acid stepwise starting a t 2%. The results of these runs are shown in Table V.
50
Figure I 1
Ionic fraction
ric acid is less than three times as strongly ionized as dichioroacetic acid and gives good separation; trichloroacetic acid is eight times as strongly ionized as chloroacetic acid and gives s fairly good separation; chloroacetic acid and acetic acid do not separate a t all even though the difference is ninefold. Operating Conditions and limitations
7.2 0.8
1.75 X 10-6
Vol. 45, No. 1
required X IO-' preferred < 2 X 10-1 reauired < 5 X 10-2 priferred
Table V.
Effect of Ionic Concentration on Ion Exclusion
Resin: Dowex 50, 50-100 mesh, 8% DVB Dowex 50, 50-100 mesh 80/ DVB column 23.5 X 0.6 inches diameter F I O ~r;te:'appro;imately I ml./min. Feed: 15 ml. a t 4% acetic acid with HC1 variable HCl Concn., % Separation Max. HC1, % Av. HC1, % 2 Excellent 2.1 1.1
a
Separation too poor t o determine average concentration.
X
The differences in ionization required for separation apparently become greater as one proceeds down the scale-e.g., hydrochlo-
The dependence of the separation on ionic concentration is readily apparent, though it would be impossible to draw any sharp line of demarcation. The results of runs at 2, 8, and 28% hydrochloric acid are plotted in Figure 12 and reemphasize the
I N D U S T R I A L Ah’D E N G I N E E R I N G C H E M I S T R Y
January 1953
because of the relative immobility and size of these large hydrated molecules. Inasmuch as there is virtually no separation, however, this appears to be a case of true equilibrium rather than one of slow diffusion.
35.
“D
1-36
0-0
2 % HCI- 4 % H Ac
“-I
8% H C I - 4 % H Ac 28%HC1-4%H Ac
Application of Process
1.35 50-100 MESH
1.34
-
1.33 30
62
46
78
EFFLUENT
I10
94
VOLUME
I26
(ml.)
Figure 1 2
change. Maximum acid concentrations in’Table V are determined from the peak refractive indices; average concentrations were calculated as approximation from the total volume of the respective cuts. The concentration of the nonionic fraction is of muGh less significance than that of the ionic fraction as is shown in Table VI and Figure 13. This is exactly as predicted from theory and makes possible the deionization of strong solutions of nonelectrolytes without the use of regenerants. The runs were carried out as described above with the exception that the ionic concentration was varied. Figure 13 illustrates the results a t 2 and 42y0 acetic acid. -r
Ion exclusion offers wide opportunities in chemical manufacturing processes as well as to the analytical chemist. Several processes using this technique are now being considered, and one plant unit, consisting of a 30-cubic-foot-resin bed, is already in operation. The value of ion exclusion to the operator rests mainly in its simplicity and the lowered chemical cost as compared t o the usual ion exchange. It is most applicable a t the higher ionic concentrations where ion exchange is prohibitive in cost. And, like ion exchange, it may be a substitute for fractional crystallization, distillation, etc. In addition, the nonionic fraction of ion exclusion may be too strongly ionized (weak acids and bases) to b e purified by ion exchange deionization, and in some cases t h e nonionic fraction is unstable a t either the high or low p H of a two-resin deionization system.
Table VI.
Effect of Nonionic Concentration on Ion Exclusion
Resin: Dowex 50, 50-100 mesh 8% DVB FIOW rate: ap roximately 1 h / m i n . Feed: 15 ml. a t 4% %IC1with acetic acid variable Acetic Acid Hydrochloric Acid, % Acetic Acid, % Separation Concn., % Max. Av. Max. Av. 2 Excellent 4.1 2.6 1.8 0.9 Excellent 4 4.0 1.8 4.1 2.1 Excellent 8 2.0 3.7 8.8 3.2 12 Excellent 2.1 3.9 12.2 4.3 Excellent 3.8 19.6 18 2.1 5.9 Excellent 3.8 24 2.1 24.6 7.6 Excellent 3.8 31.0 30 2.1 8.6 Excellent 3.6 40.0 38 1.8 11.4 Good 3.4 46 2.0 40.0 12.0
ION EXCLUSION: EFFECT OF NON IONIC CONCENrRATiON RESIN: DOWEX 50, I BY, S O - 1 0 0 MESH FEED: I S ML.
3s. “0
233
-
1.359
.-e-.
f\
1.355
4%HCI- 2% H Acetate 4 3 HCI-42% H Acetate
1.351
i
1.347
\. ‘J\
1.343 1.339 1.335
1.331
-
-L
40
48
56
64
72
80
88
96
104
I12
I20
I28
136
Conclusion Ion exclusion offers chemists and chemical engineers a new unit operation for separation. I n general, aqueous solutions of ionic and nonionic fractions may be separated into the two primary components without heat or use of regenerants other than water for elution. A wide variety of resins may be used, though a medium cross-linked (4to 12%0), fine mesh resin is most satisfactoryPresent economics and resin qualities suggest the use of a sulfonated styrene type resin, unless the higher priced anion exchangers are required for certain separations. Low ionic concentrations, small feed volumes, low flow rates, and elevated temperatures all contribute, t o greater or lesser extents, t o good separation.
EFFLUENT VOLUME f m / J
Figure 13
The temperature variable has been investigated a t low flow rates in the range 0’ to 80’ C. with some small improvement as temperature is raised, though not so much as might be expected. It seems that a t more rapid flows or with coarser resins, where diffusion rate becomes more of a factor, elevated temperatures would be more controlling, but this remains to be investigated. One decided advantage of the elevated temperatures, however, is the greater flow rates that may be accomplished with the same head pressure. One of the first conceived applications for this exclusion process was the separation of sugar and salt. However, sodium chloride cannot be separated from either sucrose or d-glucose, probably
literature References (1) Barrer, R. M., J. SOC.Chem. Ind., 64,130T,133T (1945). (2) Barrer, R. M.,Trans. Faradau Soc., 40,374, 555 (1944). (3) Bauman, W.C., and Eiohhorn, J., J. Am. Chem. SOC.,69, 2830 (1947). (4) Brown, G. G.,et al., “Unit Operations,” pp. 213 and 215, New York, John Wiley and Sons, 1950. (5) Cleaver, C. S., and Caasidy, H. G., J . Am. Chem. SOC., 72, 1147 (1950). (6) Gregor, H. P., Collins, F. C., and Pope, M., J. Colloid Sci., 6, 304 (1951). (7) Lange, N. A., “Handbook of Chemistry,” 7th ed., Sanduaky, Ohio, Handbook Publishera, Inc., 1949. (8) Mayer, S. W., and Tompkins, E. R., J . Am. Chem. Soe., 69,2760 (1947). RECEIVED for review August 16, 1952.
ACCEPTED October 23, 1953.