I n d . E n g . Chem. Res. 1990,29, 849-857 ninger, J. M., et al., Eds.; Elsevier: New York, 1985. Bartle, K. D.; Ladler, W. D.; Martin, T. G.; Snape, C. E.; Williams, D. F. Structural Analysis of Supercritical-Gas Extracts of Coals. Fuel 1979a, 8 (61, 413-422. Bartle, K. A.; Calimli, A.; Jones, D. W.; Mattlens, R. S.;Olcay, A.; Pakdel, H.; Tugrul, T . Aromatic Products of 340 "C Supercritical-Toluene Extractions of Two Turkish Lignites: an NMR Study. Fuel 197913, 58 (6), 423-428. Box, G. E. P.; Hunter, W. G.; Hunter, J. S. Statistics for Erperimenters; John Wiley and Sons: New York, 1978. Corbett, R. W.; Gir, S.; Janka, R. C. Developments in Critical Solvent Deashing. Presented at the 90th National Meeting of the American Institute of Chemical Engineers, Houston, T X , May 9, 1981. Ceylan, R.; Olcay, A. Supercritical Gas Extraction of Turkish Coking Coals. Fuel 1981, 60, 197-200. Desphande, G. V.; Holder, G. D.; Shah, Y. T. Effect of Solvent Density on Coal Liquefaction Kinetics. In Supercritical Fluids; Squires, T. G., Paulaitis, M. L., Eds.; Advances in Chemistry 329; American Chemical Society: Washington, DC, 1987; pp 251-265. Eisenbach, W. 0.; Niemann, K.; Gottsch, P. J. Supercritical Fluid Extraction of Oil Sands and Residues from Oil and Coal Hydrogenation. In Chemical Engineering at Supercritical Conditions; Penninger, M. L., Gray, R., Davidson, P., Eds.; Ann Arbor Science: Ann Arbor, MI, 1983. Fong, W. S.; Chan, P. C.; Pichaichanarong, P.; Corcoran, W. H.; Lawson, D. D. Experimental Observations on a Systematic Approach to Supercritical Extraction of Coal. In Chemical Engineering at Supercritical Conditions; Paulaitis, M. E., Penninger, M. L., Gray, R. D., Davidson, P., Eds.; Ann Arbor Science: Ann Arbor, MI, 1983. Herod, A. A,; Ladner, W. R.; Snape, C. E. Structural Studies of Coal Extracts. Philos. Trans. R. SOC.London 1981, A300 (1453), 3-14. Kershaw, J . R.; Jezko, J. Supercritical Extraction of South African Coals. Sep. Sci. Technol. 1982, 17 (l),151-166. Lee, B. I.; Kessler, M. G. Generalized Thermodynamic Correlations Based on Three Parameter Corresponding States. AIChE J. 1975, 21 (3), 510-527.
849
Maddocks, R. R.; Gibson, J.; Williams, D. F. Supercritical Extraction of Coal. Chem. Eng. Prog. 1979, 75 (6), 49-55. Martin, T. G.; Williams, D. F. The Chemical Nature of Supercritical Gas Extracts from Low-Rank U. K. Coals. Philos. Trans. R. SOC. London 1981, A300 (1453), 183-192. Penninger, J . M. Extraction of Oil from Wyoming Coal with Aqueous Solvents at Elevated Pressure. In Supercritical Fluid Technology; Penninger, J. M., et al., Eds.; Elsevier: New York, 1985. Ross, D. S.; Hum, G. P.; Miin, T.; Green, T. K.; Mansani, R. Isotope Effects in Supercritical Water: Kinetic Studies of Coal Liquefaction. In Supercritical Fluids; Squires, T. G., Paulaitis, M. L., Eds.; Advances in Chemistry 329; American Chemical Society: Washington, DC, 1987; p p 242-250. Scarrah, W. P. Liquefaction of Lignite Using Low Cost Supercritical Solvents. In Chemical Engineering at Supercritical Fluid Conditions; Paulaitis, M. E., Penninger, M. L., Gray, R., Davidson, P., Eds.; Ann Arbor Science: Ann Arbor, MI, 1983. Sunol, A. K. Supercritical Extraction of Coal. Ph.D. Dissertation, Virginia Polytechnic Institute and State University, Blacksburg, 1982. Sunol, A. K.; Beyer, G. H. Entrainer-Aided Supercritical Extraction of Coal. Presented a t the National Meeting of the American Institute of Chemical Engineers, San Francisco, CA, Nov 1984; Paper 133d. Tsekhanskaya, Y. V.; Iomtev, M. B.; Mushkina, E. V. Solubility of Naphthalene in Ethylene and Carbon Dioxide under Pressure. Russ. J . Phys. Chem. 1964,38, 1173-1178. Tu, S. T. Supercritical Steam Desulfurization of Coal. Presented a t the National Meeting of the American Institute of Chemical Engineers, Atlanta, GA, March 1984; Paper 23e. Whitehead, J. C. Development of a Process for the Supercritical Gas Extraction of Coal. Presented at the 89th National Meeting of the American Institute of Chemical Engineers, Philadelphia, PA, June 1980.
Received for review December 15, 1989 Accepted J a n u a r y 5 , 1990
Equilibrium Sorption of Amino Acids by a Cation-Exchange Resin Susan R. Dye,+Joseph P. DeCarli, II,l and Giorgio Carta* Department of Chemical Engineering, University of Virginia, Charlottesuille, Virginia 22903
The equilibrium sorption of representative amino acids by a strong-acid cation-exchange resin, Dowex 50W-X8, has been investigated. T h e uptake of an amino acid by the hydrogen form of the resin occurs primarily as the result of the stoichiometric exchange of amino acid cations for hydrogen ion. T h e solution dissociation equilibria of the amino acid and the solution composition determine the ionic fraction of amino acid cations in solution. In multicomponent systems, the uptake equilibrium is determined by the competitive interaction of the amino acid cations. A model that accounts for both solution and ion-exchange equilibria is developed to correlate binary uptake data and predict the uptake of amino acids in multicomponent systems. Ion-exchange resins are commonly used for the separation of amino acids both on an analytical scale and in industrial manufacturing processes. Because amino acids can exist both as cations and as anions depending upon the solution pH, ion-exchange resins offer a variety of opportunities. For example, with the proper choice of pH, it is possible to isolate in separate fractions the acidic, basic, and neutral amino acids present in a mixture. Individual members of a group of amino acids of the same charge type can also be separated by using ion-exchange resins, either in the same operation or in a separate step,
* Author t o whom correspondence should be addressed. Current address: Department of Chemical Engineering, University of Massachusetts, Amherst, MA 01003. * C u r r e n t address: Dow Chemical Company, Midland, M I 48667. 0888-5885/90/ 2629-0849$02.50/0
exploiting differences in the ionization constants of the various amino acids as well as differences in the specific affinities of the charged amino acid molecules. A voluminous amount of literature exists on analytical uses of ion-exchange resins for the separation of amino acids (Blackburn, 1983). Less quantitative information, on the other hand, is available on the basic equilibrium and mass-transfer phenomena that determine the performance of industrial processing operations. Saunders et al. (1989) have reviewed the earlier literature on the subject. Recent work includes the studies of Yu and Wang (19861, Yu et al. (19871, and Carta et al. (1988) on the dynamics of multicomponent chromatography for the separation of amino acids using cation-exchange resins and the study by DeCarli et al. (1988) on the continuous displacement separation of amino acid mixtures. Further work on the equilibrium and rate of adsorption and de0 1990 American Chemical Society
850
Ind. Eng. Chem. Res., Vol. 29, No. 5, 1990
sorption of amino acids from ion-exchange resins has been reported by Saunders et al. (1989) and Carta et al. (1989). In this report we provide the results of a study on the equilibrium sorption of several amino acids by a commercial strong-acid cation-exchange resin. The amino acids studied include five neutral amino acids (L-alanine, L-valine, L-isoleucine,L-leucine, and L-phenylalanine), an acidic amino acid (L-glutamic acid), and a basic amino acid (Llysine). This list is sufficiently broad to allow some generalization of the results. The equilibrium uptake of these amino acids by the hydrogen form of the resin was determined by means of batch experiments. Both singlecomponent and multicomponent measurements were obtained. Based on the experimental results, we show that the equilibrium uptake can be predicted by taking into account the solution dissociation equilibria and the ionexchange reactions. An ion-exchange model that assumes heterogeneity of the resin functional groups developed by Saunders et al. (1989) is then used to fit binary exchange data and predict the equilibrium in multicomponent systems.
Experimental Section The resin used in these studies, Dowex 5OW-XS (Dow Chemical Company, Midland, MI), is a sulfonated, styrene-divinylbenzene copolymer, with a nominal degree of cross-linking of 8%. A single lot of commercial resin was sieved to recover a fraction with particle size of approximately 50-60 pm. This small size was chosen to speed up equilibration. The total ion-exchange capacity of the resin, qo, was determined by equilibrating a sample of the resin in the hydrogen form with an excess volume of 0.1 mol/L NaOH containing 50 g/L of sodium chloride. At equilibrium, the excess NaOH was titrated with 0.1 mol/L HCl and the capacity determined from a material balance. The resin capacity was found to be 5.6 f 0.2 mmol/g of dry resin. The dry weight of fully swollen resin was obtained by determining the weight loss of a hydrated sample in the hydrogen form upon drying in an oven at 110 "C. Prior to drying, the resin sample was vacuum filtered in a Buchner funnel to remove the interstitial water. Several experiments were repeated, providing a dry weight, ps, of 0.45 f 0.04 g of dry resin/g of hydrated resin. The equilibrium uptake of amino acids was determined as follows. Samples of the resin in hydrogen form (1-2 g of hydrated resin) were placed in sealed Erlenmeyer flasks in contact with solutions (100-250 mL) containing various initial concentrations of amino acids, sodium, and chloride ions. The flasks were sealed and shaken for 12-24 h in a constant-temperature bath (Fisher, Versabath, Model 236) at 25 f 1 "C. After this period, the solutions in the flasks were sampled to determine the equilibrium concentrations of amino acids and sodium and the pH. The concentration of amino acids in the resin phase, qA,(millimoles/gram of dry resin), was calculated from
v(cx,
-
qA, =
cAa)
M~~
(1)
were V (milliliters) and M (grams of hydrated resin) are the volume of solution and the mass of hydrated resin, respectively, and C and CA, (moles/liter) are the initial and equilibrium concentrations of amino acid A, in solution. The sodium concentration in the resin was also determined by using eq l. Achievement of equilibrium in these experiments was determined by taking repeated samples as well as with a few experiments in which the system was allowed to reach equilibrium from the opposite
Table I. Solution Equilibrium Parameters" amino acid pK, PK? DK~ L-G~u 2.19 4.25 9.67 L-Ala 2.34 9.69 L-Val 2.32 9.62 L-Isoleu 2.36 9.68 2 36 9.62 L-Leu L-Phe 2.11b 9.13* 2 18 8.95 10.53 L-LYS
PI
3.22 6 02 5 97 6.02 5.99 5.62 9.47
'Values from Meister (1965),unless indicated. *Saunders et a1 (1989). 10
~
08L
> ~~__
' experiment --model Smith Pieronl 8
0
02
0A
06
38
1C
x Figure 1. Ion-exchange equilibrium for Na'/H+ binary. mol/L.
cc,-= 0.01
direction, i.e., starting with resin initially loaded with solutes. Prior to any experiments, the resin was pretreated by repeated washes with 1 mol/L HCl and 1 mol/L NaOH, converted to the desired ionic form, and then thoroughly rinsed with deionized distilled water with a resistivity greater than 10 MQ-cm. The concentrations of amino acids were determined by HPLC analyses with a Waters apparatus. A reverse-phase column (Fisher, Model 06-650-6) with 10-pm silica particles was used to perform the analyses. The mobile phase was a 0.1% phosphoric acid solution with UV detection at 200 nm. The concentration of sodium was determined with the same HPLC apparatus using a high-performance ionexchange column (Waters, IC-PAK) with a conductivity detector (Waters, Model 430). The mobile phase in this case was a 0.002 mol/L HNO, solution. Finally, the solution pH was determined with a Ross pH electrode (Orion, Model 8103). The chloride concentration was assumed to remain constant in the batch equilibrium experiments and was not measured. This species is in fact essentially completely excluded from the resin by the Donnan potential effect at the relatively low ionic strength (10.17 mol/L) used in this work (Helfferich, 1962). This was confirmed indirectly by the fact that experimental pH values and calculations based on this assumption agreed quite well. The amino acids used in these studies were all L form, obtained with purity greater than 99% from Sigma Chemicals and Ajinomoto U.S.A., Inc. The ionization constants of the amino acids used are given in Table I. Other chemicals were analytical reagent grade from Fisher.
Results and Discussion Na/H Equilibrium. Experimental data for the exchange of sodium and hydrogen on the resin are shown in Figure 1 along with two sets of data from the literature. The data are given in terms of ionic fractions of sodium in the resin, YNa = qNa/qO, and in the solution, XNa = CNa+/(CNa++.C,+). There is considerable discrepancy with the data of Pieroni and Dranoff (1963). However. the resin
Ind. Eng. Chem. Res., Vol. 29, No. 5, 1990 851
0
0
20
10
30
LO
60
50
70
80
90
100
0" 0
10
20
30
50
40
mM
CA,
CA
Figure 2. Uptake of Leu as a function of total Leu concentration in solution at various chloride concentrations.
1
I
60
70
20
10
30
LO
50
60
70
Cl: M
001 010
' -2 80
90
I
100
0
X ) M r ) 4 0 5 0 6 0 7 0 8 0 9 0 W X ) CA. mM
Figure 3. Uptake of Isoleu as a function of total Isoleu concentration in solution a t various chloride concentrations.
ov 0
10
'A
20
30
40
50
1
017
CA, mM
' f /'
100
Figure 5. Uptake of Ala as a function of total Ala concentration in solution a t various chloride concentrations.
L-1 0
90
mM
. @
80
60
70
80
90
100
CA , mM
Figure 4. Uptake of Val as a function of total Val concentration in solution at various chloride concentrations.
capacity obtained by these investigators was only 4.74 mmol/g of dry resin, while the reported selectivity for the Na-H exchange was abnormally high (5' = 2.67) for a resin with this degree of cross-linking. On the other hand, the agreement of our data with the data of Smith (1960) and with selectivity values commonly found for resins of this type (Helfferich, 1962) is quite good. Uptake of Neutral Amino Acids. The equilibrium uptake of leucine, isoleucine, valine, alanine, and phenylalanine by the hydrogen form of the resin is shown in Figures 2-6. q A is the total uptake of amino acid determined from eq 1, and C A is the analytical concentration of the amino acid in solution. The uptake curves are similar for the different amino acids. In each experiment, the co-ion concentration, Ccl-, remained approximately
Figure 6. Uptake of Phe as a function of total Phe concentration in solution at various chloride concentrations.
equal to the initial values that are reported in these figures. Thus, the solution pH varies from point to point on each curve. If the chloride concentration is increased, while keeping the amino acid concentration constant, the p H decreases. As a result, the uptake of amino acid by the resin is reduced because of the increased competition by hydrogen ions for the resin functional groups. Conversely, if the chloride concentration is small or zero, the solution pH approaches the isoelectric pH of the amino acid. In this case, the zwitterionic form of the amino acid is predominant, and only a very small fraction of the amino acid is present in the exchangeable cationic form. Nevertheless, the uptake of amino acid approaches a maximum for these conditions, because competition from hydrogen ion is minimized. In the limit of high C,, the uptake approaches the total exchange capacity of the resin of 5.6 mmol/g of dry resin. The observed behavior may be easily explained quantitatively by taking into account the dissociation equilibria of the amino acids, as shown, for example, by Saunders et al. (1989). The concentration of amino acid, Ai, in solution is given by cA,
where
=
CA,+
+ C A , f + CA,-
(2)
852 Ind. Eng. Chem. Res., Vol. 29, No. 5 , 1990 1.0 I
I
1
0
20
40
60
80
0
-
loo
0
02
08
06
04
10
X
Ca, mM
Figure 7. Calculated ionic fraction of Leu cations and solution pH as a function of total Leu concentration. Parameter is the chloride concentration.
Figure 8. Ion-exchange equilibrium for Leu+/H+ binary. 10
v---I
Here Kli and KZiare the dissociation constants of the amino acid. The concentrations of positively and negatively charged species in solution are related by the electroneutrality condition. In general, for a solution containing several amino acids, sodium, and chloride ions, this condition may be expressed as CCAi++ CH+ CNa+ = CCAi- + KW/CH+ + Ca- (5) i
+
i
~0 0 I @ 001
i
where Kw is the ionic product of water. Equations 2-5 can be used to compute the ionic composition of an amino acid solution. If the pH and C, are known, eqs 2-4 can be used directly to compute C&+,C,+, and C,-. Conversely, if CNn+, ~ known, eqs 3 and 4 can be inserted into Ccl-, and C A are eq 5 , which, in turn, can be solved to yield CH+. Next we consider the exchange of an amino acid cation with hydrogen ion. The process may be represented by the reaction R'H
08 t
+ Ai+ + R'AI + H+
01 ,
0
02
-
--L
08
06
04
~
-10
X
Figure 9. Ion-exchange equilibrium for Isoleu+/H+ binary.
08
i
(6)
A selectivity coefficient for this exchange reaction may be defined as
061 >
(7) 02
A 001 A
By use of eq 7, if no other counterions are present, qA,may be expressed as
0
02
04
010
08
06
10
X
Figure 10. Ion-exchange equilibrium for Val+/H+ binary.
where #I
- o 08C l
The ionic fraction, X can be readily calculated from eqs 3-5. As an example,%gure 7 shows the ionic fraction of leucine cations and the solution pH as a function of the total amino acid concentration, CA,for different values of the co-ion concentration, Ccr. Note that the solution pH approaches the isoelectric pH of the amino acid as CA is increased or when Ccr is reduced to values approaching zero. For these conditions, XAiapproaches a maximum, and the uptake by the resin should be maximum as per eq 8. At high chloride concentrations, the pH is low, and X, increases in an approximately linear manner with the total amino acid concentration. The experimental uptake data of Figures 2-6 are replotted in Figures 8-12 in the form suggested by eq 8, as
y 1 I
f
0.4
02
I
4 001 01
n 0
02
04
06
08
10
X
Figure 11. Ion-exchange equilibrium for Ala+/H+ binary.
Y, vs X4. Note that for these amino acids the uptake data obtained a t different pH fall on a single line. This indi-
Ind. Eng. Chem. Res., Vol. 29, No. 5, 1990 853 7 1
I
6-
02
P 0
002 007 0 017 @
1
~
, I
I 02
06
OL
08
0
10
Figure 12. Ion-exchange equilibrium for Phet/Ht binary. 1
0
10 Relative
20
30
40
50
60
70
80
90
100
CA, mM
X
----ol
10
20
30
Hydrophobicity
Figure 13. Limiting selectivity coefficient as a function of the relative hydrophobicity defined by Nozaki and Tanford (1971). Hydrogen is the reference counterion.
cates that the uptake process depends only upon the ionic fraction of amino acid cations and not upon the total amino acid concentration. The occurrence of nonionic adsorption can thus be discounted, at least for the relatively high concentration levels used in the experiments. To obtain further confirmation of this finding, we carried out a few experiments at pH values higher than the isoelectric pH. For these conditions, which correspond to a value of XA, of essentially zero, no uptake was observed and any amino acid previously loaded on the resin was completely removed. Similar conclusions were obtained by Saunders et al. (1989) for the uptake of phenylalanine and tyrosine by the resin Amberlite 252. The Y-X diagrams show that the selectivity coefficient is not constant during the exchange process, generally decreasing from values well above unity at low resin loadings to values that, in some cases, are below unity when the resin is almost fully loaded with amino acid. A comparison of the initial, low-loading selectivities can be made for the various amino acids. The selectivity increases in the order Ala C Val C Isoleu C Phe Leu. Except for leucine, this order correlates well with the hydrophobicity scale defined by Nozaki and Tanford (1971) as illustrated in Figure 13. It is apparent, thus, that the hydrophobic interactions of the amino acid side chain with the organic backbone of the resin play an important role in determining the affinity of amino acid cations for the resin’s functional groups, without, however, significantly contributing to the maximum uptake capacity. [Interestingly, a similar correlation with the hydrophobicity scale has been found by Thien et al. (1988) for the flux of various amino acids through hydrophobic liquid membranes]. Although stoichiometric ion exchange would seem to involve only electrostatic interactions, hydrophobic contributions and the size of the exchanged molecule influence activity
-
Figure 14. Uptake of Glu as a function of total Glu concentration in solution at various chloride concentrations.
coefficients within the resin matrix and determine the selectivity coefficient. As shown in Figure 13, leucine appears to deviate somewhat from the correlation. On the other hand, since data for leucine were not available at very low Y values, the discrepancy might be the result of a faulty extrapolation to a zero Y value. Uptake of Glutamic Acid. The equilibrium uptake of glutamic acid is shown in Figure 14, which indicates a general behavior similar to that of the neutral amino acids. The total amino acid concentration in this case is given by CA = CAt + CA&+ CA- + c A 2 (10) where
Here K 2 is the dissociation constant of the carboxylic acid group on the side chain. The electroneutrality condition becomes CAt + CH+ + CNst = CA- + 2CAZ- + Kw/CH+ + CCl- (14) Examination of the pK values in Table I shows that dissociation of the a-carbon carboxylic acid group (pK2 = 9.67) will not occur until pH values above 8. For this pH, however, the amino acid is completely in negatively charged form and XA is essentially zero. As for the neutral amino acids, XAis maximum at the isoelectric pH (3.2), when the concentrations of glutamic acid having a protonated amine group and a dissociated R group are in balance. The Y-X diagram for glutamic acid is shown in Figure 15. Data at different co-ion concentrations and pHs fall again essentially along the same line. The uptake process, thus, appears to be still dominated by the stoichiometric exchange of amino acid cations and hydrogen ion. The selectivity coefficient for this exchange is close to unity, although there appears to be again some variation with resin loading. Uptake of Lysine. The equilibrium uptake of lysine is shown in Figure 16 for a concentration of the chloride co-ion of 0.1 mol/L. The uptake curve has in this case a
854
Ind. Eng. Chem. Res., Vol. 29, No. 5 , 1990 '3
,/ / j
06 >-
The electroneutrality condition becomes 2C~2++ CA++ C H t + C N ~=+ CA- + KW/CH+ + CCl- (19) Two ion-exchange reactions must be considered, one for divalent lysine cations and one for monovalent cations 02
0
06
Ob
08
10
R'H
X
Figure
2R'H
15. Ion-exchange equilibrium for Glu+/H+ binary.
+ AH2+ + AH32+
R'AH2 + H+ R,'AH,
+ 2H+
(20) (21)
The corresponding selectivity coefficients are
- -~
with the conditions 4A =
It
YO
i
0' 0
=
4AH2
(?AH?
+
(24)
+ qAH,
2qAHl
+
qH
(25)
Combining eqs 22-25, one obtains A
20
60
LO
Ca
80
100
mM
Figure 16. Uptake of Lys as a function of total Lys concentration a t Ccl- = 0 1 mol/L.
,o
-
_-
.
--
/
xa.
I/l
/
where
Ca, mM
Figure 17. Calculated ionic fraction of Lys divalent and monovalent cations and solution pH as a function of the total Lys concentration. Ccl- = 0.1 mol/L.
sigmoidal shape; initially the uptake increases rapidly as a function of CA, then approaches a plateau at about one-half of the resin's ion-exchange capacity for intermediate CA values, and finally increases again at higher values. Since lysine possesses an amine group on the side chain, the total amino acid concentration is given by where
Calculated values of xA+, X A H , and solution pH are shown in Figure 17 as a function of the total lysine concentration for Ccl- = 0.1 mol/L. The calculations were carried out by inserting eqs 16-18 in the electroneutrality condition (eq 19) and solving the resulting equation for CH+ by trial and error. The results show that, up to about CA = 40 mM, the pH is low and lysine is predominantly in the divalent cationic form. For these conditions, uptake of lysine occurs largely according to eq 21, and the maximum uptake is limited to one-half of the total resin capacity as shown in Figure 16. A t approximately 50 mM, XA2+reaches a maximum of about 0.7 and then decreases to zero as CAis increased further. A t the same time, XA+ increases rapidly, approaching 1when CA = 0.1 mol/L. For these conditions, the pH is about 5.57 [=(pK, + pK2)/2] and lysine is predominantly in the hydrochloride form. The uptake of lysine, however, approaches the resin ca-
Ind. Eng. Chem. Res., Vol. 29, No. 5 , 1990 855 Table 11. Ion-Exchange Eauilibrium Parameters” species Na+ L-G~u L-Ala L-Val L-Isoleu L-Leu L-Phe
s i . jb
Wi.ib
1.54 1.01 1.05 1.00 2.31 3.01 5.04
5.00 1.70
1.oo 0.35 0.25 0.20 4.00
“ q o = 5.6 mmol/g of dry resin. *Relative to hydrogen ion (j =
H+). 0
L-.__. 0
02
06
04
08
10
XA++
Figure 18. Ion-exchange equilibrium for divalent Lys cations with hydrogen ion. Ccl- = 0.1 mol/L.
pacity since competition from hydrogen ions is minimized. The uptake data up to CA = 0.05 mol/L are reproduced in Figure 18 in terms of the equivalent fraction of divalent lysine cations in the solution, xA2+,and in the resin, YA2+ = 2qA/qO. Since up to this value XA+is quite small and the divalent form is largely preferred because of the electroselectivity effect, qA is essentially equal to qAH3.
Correlation of Equilibrium Data Uptake of Monovalent Amino Acid Cations. The Y-X diagrams obtained for the different amino acids indicate that the selectivity coefficients for the exchange of amino acid cations with hydrogen ion are nonconstant and depend on the resin composition. Similar behavior was also observed by Saunders et al. (1989) for the uptake of phenylalanine and tyrosine by Amberlite 252. More generally, variable selectivities are found for the exchange of various ions (including Na, K, etc.) typically with highly cross-linked resins (Helfferich, 1962). Such variability can often be attributed to inhomogeneities in the resin structure and in the effective “strength” of the functional groups. The degree of cross-linking in a resin bead, for example, can be expected to be higher near the center of the bead than at its outer surface. Similarly, a variation in the acid strength of the functional groups may result from the sulfonation process, which may cause a partial oxidation of the resin cross-links to form weakly acidic groups. In addition, variable selectivity coefficients may be caused (although to a less significant extent) by nonidealities resulting from interactions among the counterions in the resin. Saunders et al. (1989) have developed a model originally proposed by Myers and Byington (1986) to represent variable selectivities in ion exchange. The model neglects interactions among the counterions and assumes instead the existence of a continuous or discrete distribution of functional groups having different specific selectivities. The apparent selectivity exhibited by the resin is thus the result of the combination of the individual ion-exchange reactions occurring on each different functional group. Consider, for instance, the exchange of counterions A and B, with a resin initially in the B form. Initially A, which is assumed to be the “preferred ion”, will be exchanged for B by the functional groups with high selectivity, resulting in a high apparent selectivity. Then, as the resin becomes saturated with A, the exchange reaction will occur on the less selective groups, yielding a reduced apparent selectivity. Thus, in general, the apparent selectivity for the preferred ion will decrease as the loading of that ion on the resin is increased. Clearly, since the specific selectivities are determined by the nature of the counterions and
by the environment surrounding each individual functional group, the apparent inhomogeneity of the resin, as revealed by the sigmoidal shape of the Y-X diagrams, will be different for different counterions. The Myers and Byington model is based on a binomial distribution of functional groups with arbitrary skewness. In the simplest formulation, one may assume the existence of only two types of functional groups with different selectivities, each with the same number of sites. This corresponds to a symmetrical binomial distribution with two elements ( n = 1, p = 0.5). In this case, considering N counterions and selecting ion j as the reference counterion, the apparent selectivity coefficient for ion i is given by
(29) For each N - 1 binary, there _are two parameters: an average selectivity coefficient, Si,j , and a heterogeneity parameter, Wi,j(Saunders et d.,1989). The latter parameter is a function of the energetic heterogeneity of the resin’s functional groups for the exchange of ions i and j . Note that, when W i jis unity for all the binaries, the apparent selectivity coefficient is constant. The resin composition can be determined from the Si, values from the following equation: (30)
It should be noted that, while more complex and skewed distributions can be used, the assumption of a two-site symmetrical distribution has the advantage of simplicity and provides sufficient flexibility to fit many experimentally observed ion-exchange isotherms. It also allows a reasonable prediction of multicomponent isotherms. Equations 29 and 30 were used to fit the binary exchange data obtained with hydrogen as the reference counterion. For each binary consisting of an amino acid cation (or sodium ion) with hydrogen ion as the counterion, Si,and Wi, were obtained by a nonlinear least-squares fit of the data. The resulting parameter values are given in Table I1 and calculated curves are shown in Figures 1-12 and 14-15. The fit is generally well within the experimental error. Multicomponent uptake measurements were also carried out for the systems Leu/Na/H, Leu/Val/Ala/H, and Leu/Val/Glu/H using the shaker-flask method described earlier, and the results are shown in Tables 111-V. Calculations also shown in these tables were based on eqs 2-5, 10-14, 29, and 30, using the parameters given in Tables I and 11.
856 Ind. Eng. Chem. Res., Vol. 29, No. 5 , 1990 Table 111. Leu/Na/H Equilibrium" Ccl10
CLeu
28.5 30.5 31.5 33.5 37.0 37.5 36.5 37.0 38.5 44.0
100
CNa'
0.58 2.13 4.46 5.72 4.51 8.57 21.4 40.7 80.4
pHnxP 2.86 2.89 2.95 3.13
pHCd 2.75 2.82 2.93 3.14 3.30 1.22 1.24 1.35 1.61 2.57
3.90
1.40 1:42 1.52 I .74 2.77
qLeuexP
4LeUCd
9NPp
qNCd
4.47 4.10 3.84 3.46 2.74 2.69 2.83 2.78 2.45 1.84
4.82 4.60 4.01 3.30 2.95 3.15 3.08 3.01 2.93 2.07
0.40 0.85 1.16 1.96 0.11 0.30 0.77 1.98 3.67
0.37 1.17 2.08 2.50 0.25 0.47 1.11 1.91 3.45
" C in mmol/L, 9 in mmol/g of dry resin. Table IV. Leu/Val/Ala/H Equilibrium" cCl-
0.1
10
CAla
13.3 14.1 15.3 3.51 13.0 22.0 44.0 3.10 14.0 16.7 20.6 27.5
Cvd 13.5 8.00 31.1 3.70 13.0 22.1 41.5 2.70 10.6 13.8 31.1 13.3 13.7
, ,C 9.60 15.8 18.6 2.50 10.5 18.5 38.0
8.00 10.2 18.6 4.70 5.10
" C in mmol/L, q in mmol/g of dry resin.
0.1
8.90 6.20 43.9 9.00 32.0 15.1 8.60
16.3 4.15 39.5 22.1 7.50 17.6 13.8
22.6 24.4 15.8 4.10 15.8
pHe'P b b b b b b b b b 2.89 3.12 2.89 3.00
pHCd 4.95 4.97 5.17 4.49 4.95 5.17 5.42 4.35 4.70 2.87 3.13 2.88 2.97
9Vdd
qLeuelP
QLeu
cal
qAIFp
qAlial
9valexp
1.42 1.24 0.98 1.32 1.54 1.78 1.33 1.53
1.26 1.52 0.82 1.12 1.25 1.26 1.31 1.58
2.74 2.94 2.81 2.53 2.78 2.82 2.85
1.26 0.87 1.69 1.91
1.42 0.97 1.87 1.33 1.41 1.42 1.38 1.43 2.40 1.40 1.82 1.21 1.05
2.19 2.98 2.35 2.65 2.09 2.56 2.68
1.24 0.65 1.85 2.48
1.39 0.42 1.84 1.39 1.54 1.78 1.69 1.62 2.10 1.30 1.77 1.34 1.22
2.68 2.05 2.31 2.08 1.97
2.89 2.73 2.79 2.40 2.39
0.36 1.30 2.07 0.21 1.69 0.99 0.28
0.37 1.41 2.04 0.30 1.57 0.75 0.44
1.16 2.01 3.57 1.63 0.53 2.46 1.33
1.99 1.86 2.89 2.15 0.94 2.12 1.98
2.53
3.08
3.25 3.04 2.19 3.04
3.03 2.91 2.43 2.99
*Not determined.
3.69 3.52 3.49 3.77 3.77 3.55 :3.70
3.71 3.45 3.42 3.74 3.70 3.50 3.70
C in mmol/L, q in mmol/g of dry resin.
Theoretical predictions are in good agreement with the experiments, with errors generally in the &20% range. Some more significant deviations, however, appear to occur for sodium at low loadings. On the other hand, these measurements were affected by a greater error caused by inaccuracies in the analyses of very dilute solutions. Uptake of Lysine. To correlate the uptake of lysine, we first fit the data shown in Figure 18. For these data the ionic fraction of monovalent lysine cations is negligibly small. The data could be fit with a constant value of the selectivity coefficient, S " (eq 23), of 1.9 g of dry resin/mL. The remainder of the uptake curve in Figure 16, above a lysine concentration of 0.05 mol/L, was obtained from eqs 26 and 27, using S " = 1.9 and a value of S ' determined from a nonlinear least-squares fit of the entire set of data. The resulting value of S ' was 1.0.
Conclusions We have carried out a comprehensive study of the uptake of various amino acids by a cation-exchange resin. For all the systems investigated, the equilibrium uptake of an amino acid can be interpreted as the stoichiometric exchange of amino acid cations for hydrogen ion. Multicomponent equilibrium is then determined by the competitive exchange of the different counterions present in the system. The nature of the side chain of the amino acid influences the selectivity of the resin for the amino acid cations:
generally more hydrophobic chains result in greater selectivity. However, even for phenylalanine, which has a strongly hydrophobic aromatic side chain, there is no evidence of significant nonionic adsorption. The uptake of amino acid by the resin is thus limited by the total ionexchange capacity. The equilibrium data can be correlated by taking into account solution and ion-exchange equilibria. We have successfully used an exchange equilibrium model that accounts for heterogeneity in the selectivity of individual functional groups in the resin. Multicomponent equilibria for monovalent amino acid cations could be predicted approximately on the basis of this model. For lysine, which can form both monovalent and divalent cations, we have developed a simple model for the simultaneous competitive exchange of both ionic forms, which allows an approximate prediction of the overall uptake isotherm.
Acknowledgment This research was supported by National Science Foundation Grant CBT-8709011 and by Ajinomoto Co., Inc. The experimental participation of Robert A. Elam and Joseph C. Kurek is greatly appreciated.
Nomenclature C = equilibrium solution concentration, mol/L = initial solution concentration, mol/L
I n d . Eng. C h e m . Res 1990, 29, 857-861
Ki = dissociation constant, mol/L Kw = ionic product of water, mo12/L2 M = mass of hydrated resin, g n = (number of functional group types) - 1 N = number of exchangeable counterions p = skewness of binomial distribution function q A = solute concentration in resin phase, mmol/g of dry resin qo = resin ion-exchange capacity, mmol/g of dry resin Si,= selectivity coefficient for exchange of ion i with ion j Si,! = average selectivity coefficient for exchange of ion i with ion j S ’ = selectivity coefficient for exchange of monovalent cations with hydrogen ion S ” = selectivity coefficient for exchange of divalent cations with hydrogen ion V = solution volume, mL Wi,,j= heterogeneity parameter for exchange of ion i with ion J
XA= ionic fraction of cation A in solution Y A = ionic fraction of solute A in the resin Greek S y m b o l
dry resin density, g of dry resin/g of hydrated resin Registry No. L-Ala, 56-41-7; L-Val, 72-18-4; L-Isoleu, 73-32-5; L-Leu, 61-90-5; L-Phe, 63-91-2; L - G ~ u56-86-0; , L-LYS,56-87-1; Dowex 50W-X8, 11119-67-8.
ps =
Literature Cited Blackburn, S. Amino Acids and Amines. In Handbook of Chromatography; Zweig, G., Sherma, J., Eds.; CRC Press, Inc.: Boca Raton, FL, 1983. Carta, G.; Saunders, M. S.; DeCarli, J. P., 11; Vierow, J. B. Dynamics of Fixed Bed Separations of Amino Acids by Ion Exchange.
857
AIChE Symp. Ser. 1988,84, 54-61. Carta, G.; Saunders, M. S.; Mawengkang, F. Studies on the Diffusion of Amino Acids in Ion Exchange Resins, Presented at the Third International Conference on Fundamentals of Adsorption, Sonthofen, FRG; Engineering Foundation: New York, 1989. DeCarli, J. P., 11; Carta, G.; Byers, C. H. Displacement Separations by Continuous Annular Chromatography. Presented a t the AIChE Annual Meeting: Washington, DC, 1988. Helfferich, F. Ion Exchange; McGraw-Hill: New York, 1962; Chapter 5, pp 134-145. Meister, A. Biochemistry of the Amino Acids, 2nd ed.; Academic Press: New York, 1965; Vol. I, p 28. Myers, A. L.; Byington, S. Thermodynamics of Ion Exchange: Prediction of Multicomponent Equilibria from Binary Data. In Zon Exchange Science and Technology; Rodrigues, A. E., Ed.; NATO AS1 Series E., No. 107; Nijhoff Dordrecht, 1986; pp 119-145. Nozaki, Y.; Tanford, C. The Solubility of Amino Acids and Two Glycine Peptides in Aqueous Ethanol and Dioxane Solutions. J . Biol. Chem. 1971,246, 2211-2217. Pieroni, L. J.; Dranoff, J. S. Ion Exchange Equilibria in a Ternary System. AIChE J . 1963, 9, 42-45. Saunders, M. S.; Vierow, J. B.; Carta, G. Uptake of Phenylalanine and Tyrosine by a Strong Acid Cation Exchanger. AIChE J. 1989, 35, 53-68. Smith, N. D. Multicomponent Cation Exchange in Aqueous Systems. Ph.D. Dissertation, Illinois Institute of Technology, Chicago, 1960. Thien, M. P.; Hatton, T . A.; Wang, D. I. C. Separation and Concentration of Amino Acids using Liquid Emulsion Membranes. Biotechnol. Bioeng. 1988, 32, 604-615. Yu, Q.; Wang, N.-H. L. Multicomponent Interference Phenomena in Ion Exchange Columns. Sep. Purif. Methods 1986,15,127-158. Yu, Q.; Yang, J.; Wang, N.-H. L. Multicomponent Ion Exchange Chromatography for Separating Amino Acid Mixtures. React. Polym. 1987, 6, 33-44. Receiued for reuiew October 5, 1989 Accepted January 30, 1990
Selective Transport of Aldehydes across an Anion-Exchange Membrane via the Formation of Bisulfite Adducts Manabu Igawa,* Yasuko Fukushi, and Takashi Hayashita Faculty of Engineering, Kanagawa University, Rokkakubashi, Kanagawa-ku, Yokohama 221, J a p a n
Michael R. Hoffmann California Institute of Technology, W . M . Keck Laboratories 138-78, Pasadena, California 91125
Organic nonelectrolytes can be selectively transported through a n ion-exchange membrane if they are specifically converted to electrolytes in the membrane. Aldehydes react with bisulfite to form hydroxyalkanesulfonates (HASA), which are the conjugate bases of strong acids. Aldehydes are shown to be transported efficiently across an anion-exchange membrane via the coupled countertransport of HMSA ion with hydroxide ion and the relay of an aldehyde from one membrane-bound bisulfite ion to another. Permeation rates are in the order of formaldehyde > acetaldehyde > acetone; this relationship parallels the relative order of the stability constants for formation of the respective adducts with bisulfite ion. Aldehydes can be separated readily from other organic solutes by this method. The selective transport of organic solutes plays a very important role in biological membranes (Kotyk et al., 1988). However, in synthetic membranes, the controlled transport of organic solutes is very difficult to achieve. The selective transport system for aldehydes, which is described in this paper, may be useful as a commercial separation process. Similar systems can be developed to separate other organic solutes. There have been many reports on the selective and facilitated transport of electrolytes under a concentration
* To whom correspondence should be addressed.
gradient through polymer or liquid membranes. Macrocyclic compounds, such as monensin (Choy et al., 1974) and dibenzo-18-crown-6 (Reusch and Cussler, 1973), are very effective carriers for the selective transport of metal ions. Electrolytes can be separated from each other with various membranes by their ion sizes, ion valences, hydrophilicities, and chelate formation properties. On the other hand, only a few papers have been reported on the selective transport of organic nonelectrolytes. Nonelectrolytes have been separated by their sizes, hydrophobicities, and their degree of dissociation. The permselectivities of sugars were regulated by pH in polyvinyl-poly-
08S8-5885/90/2629-085~~02.50/0 0 1990 American Chemical Society