428
Ind. Eng. Chem. Process Des. Dev. 1985, 2 4 , 426-429
Santaceserla, E.; Berlendis, D.; CarrB, S. Fluid Phase €qui/. 1979,3, 167. Thomas, E. R.; Newman. B. A.; Nicolaides, G. L.; Eckert, C. A. J. Chem. Eng. Deta 19828,27, 233. Thomas, E. R.; Newman, B. A.; Long, T. C.; Wood, D. A,; Eckert, C. A. J . Chem. Eng. Data 1982b,27, 39Q. Vernier, P.; Raimbauit, C.; Renon, H. J. Chim. Phys. 1989,66,429.
Wardencki, W.: Tameesh, A. H. J. Chem. Tecnol. Biotechnol. 1981.31. 86. Weimer, R. F.; Prausnitz, J. M. Pet. Refiner 1985,44(9),237.
Received for review F e b r u a r y 21, 1984 Accepted June 27, 1984
Separation of Racemic Mlxtures: Optical Resolution of ~~-2-Amino-I-butanol RaJendraN. Samant and Sampatraj B. Chandalla' Department of Chemical Technology, University of Bombay, Matunga, Bombay 400019, India
The L(+)hemitartarates of D(-F and ~(+)-2-amino-l-butanolwere prepared and their solubilities in methanol and water at different temperatures were determined. Data for the ternary phase equilibria involving the diastereomers and the solvent were collected with a view to provide a rational basis for the development of a process for the resolution of DL-2-amino-I-butanol.This approach is likely to be suitable for the design of similar separation processes involving the resolution of racemic mixtures.
Introduction The separation of racemic mixtures is an important operation in the pharmaceutical industry and in the preparation of substances for use in the food industry since only one enantiomer is usually active in the case of many food and drug substances. The problem of optical isomer separation can be tackled in two ways. The substance to be resolved is reacted with an optically active resolving agent and the pair of diastereomers is separated by fractional crystallization. Alternatively, the racemic mixture itself can be separated into two enantiomers by taking advantage of the crystallization kinetics in certain cases. Of the two methods, the former is widely used. Extensive work on the optical resolution by formation of diastereomers has been reported, but the information is in the form of typical recipes. The separation is essentially a problem of crystallization and can be tackled by studying the phase equilibria. The fractional crystallization of inorganic salts and organic compounds has been studied in the past. However, no one has extended it to the separation of diastereomers, and the data for the design and development for the separation of racemic mixtures are missing. In this paper, such an attempt has been made to provide a rational basis for the separation of DL-2-amino-1-butanol. L(+)2-amino-1-butanolis an intermediate for the manufacture of ethambutol, an anti-tubercular drug. The racemic separation of DL-2-amino-1-butanol is generally based on the formation of a pair of diastereomers with L(+)-tartaric acid, followed by fractional crystallization. The solvents employed are water and different alkanols. The information available on the process is, however, very limited (Radke et al., 1954; Pitre and Grabitz, 1969) and is found mostly in the patent literature (Zoja, 1971; Seres and Daroczi, 1970; Ichikawa et al., 1975). Experimental and Analytical Section Materials. ~(+)-2-Amino-l-butanol (Fluka) having a specific rotation of 10' and ~(-)-2-amino-l-butanol(Fluka) having a specific rotation of -10' were used. D L - ~ Amino-1-butanol of commercial grade was distilled and used. L(+)-Tartaric acid (BDH) having a specific rotation of 12.0' (c = 20, H20) was used. 0198-4305/05/1124-0426$01.50/0
Table I. Physical Properties of the Diastereomers obsd
reporte@
salt
(I) anhydrous (I) monohydrate (11) anhydrous (11) monohydrate L(+)-aminobutanol-L(+)-neutral tartarate
141 142 100 102 143
24.4 22.7 11.0 10.3 27.6
142-143
22.7
102-103 142-143
10.3 27.6
P i t r e and Grabitz (1969)
Instruments. The optical measurements were taken on a polarimeter having a least count of 0.01'. All the rotations were taken in a 2-dm tube. Preparation of L(+)-Tartarate Salts. The L(+)hemitartarate salts of L(+) and D(-)-2-amino-l-butanols were prepared by adding 20 g of the required aminobutanol to 33.65 g of I,(+)-tartaric acid dissolved in 75 mL of methanol, followed by cooling to 0 "C. The crystals were dried in a vacuum desiccator. The L(+)-neutral tartarate salts were similarly prepared by adding 40 g of the required aminobutanol, instead of 20 g, to 33.65 g of L(+)-tartaric acid. The physical properties of hemitartarate and neutral tartarate salts are given in Table I. Analysis. The hemitartarate salt, when titrated on a pH meter with standard sodium hydroxide solution, showed two points of inflection, the first one at pH 6.8-7.0 indicating the neutralization of the free carboxyl group and the second one at pH 10.6-10.8 due to the hydrolysis of the salt moiety. Thus, the hemitartarate salt was analyzed for its free carboxyl group by titrating it with 0.1 N sodium hydroxide to pH 6.8 and the purity of the salt expressed as 2-amino-l-butanol-~(+)-hemitartarate was determined. Similarly, the analysis based on the salt moiety was performed by titrating the compound with 0.1 N sodium hydroxide and the amount of sodium hydroxide required to raise the pH from 6.8 to 10.8 was determined. Based on this estimation, the purity of the compound expressed as 2-amino-l-butanol-~(+)-hemitartarate was calculated. If the salt was essentially the hemitartarate with no free tartaric acid and the neutral tartarate salt present, it was 0 1985 American Chemical Society
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 2, 1985 427
expected that the purity determined by both the methods would agree. When subjected t o the same procedure, the neutral tartarate salt showed a single point of inflection at pH 11.0, which indicated the complete hydrolysis of the salt. Hence the salt was titrated with 0.1 N sodium hydroxide to pH 11.0 and the purity expressed as 2-amino-1- butanol-^(+)-neutral tartarate was determined. Determination of the Solubilities of the Hemitartarates. The solubilities of both the hemitartarates were determined in water and methanol at different temperatures. The respective hemitartarates were dissolved in the solvent, and the solution was saturated at higher temperature followed by cooling to the required temperature. The hemitartarate salt, in a preweighed aliquot from the saturated solution, was determined by titrating it with standard sodium hydroxide up to pH 6.8. The solubility was expressed as grams of anhydrous salt present in 100 g of solvent. Ternary Phase Equilibria. The phase equilibria for the ternary system involving the two diastereomers and the solvent was studied and the invariant point at a given temperature (also called drying up point) was determined according to the method given by Mullin (1972). Thus, 16.85 g of L(+)-tartaric acid was dissolved in 60 mL of methanol and 10 g DL-2-amino-1-butanol)was added to it in a semibatch manner. The clear solution was seeded with ~(+)-2-amino-l-butanol-~(+)-hemitartarate (I) and it was equilibrated at 28 "C. The precipitated crystals were separated. The composition of the crystalline residue was determined by taking the specific rotations of about 5% aqueous solution (based on weight of anydrous salt per 100 mL of solution). The specific rotation of a mixture of the diastereomers is generally expected to follow a linear relationship with its composition. This was ascertained by observing the specific rotation of various synthetic mixtures and preparing a calibration curve. Based on the amount and composition of the crystalline material, the compositioli of the mother liquor was determined by material balance. The invariant point in water was determined in a similar fashion by taking 25 mL of water.
Results and Discussion Preparation of Hemitartarates. When 2-amino-lbutanol is reacted with L(+)-tartaric acid, two different salts are formed depending on the amount of amine taken. Thus, with equimolar amounts,a hemitartarate is formed, while a neutral tartarate is formed if two moles of amine are taken for one mole of tartaric acid. Though equimolar amounts give rise to a hemitartarate, the formation of a neutral tartarate cannot be ruled out if the reactions are not consecutive. Therefore, the salt, obtained after taking equimolar amounts of acid and amine, was analyzed to determine the purity with respect to free carboxyl group as well as the salt moiety. I t was observed that the purity with respect to acid and salt was 99.4% and 99.6%, respectively. This indicated that the salt formed was a hemitartarate since any contamination with the neutral tartarate would have given low purity with respect to acid and high purity with respect to salt. The exclusive formation of a hemitartarate was further confirmed when the same salt was obtained even after reversing the mode of addition, i.e., by adding tartaric acid to a solution of aminobutanol in a semibatch manner. When the hemitartarate salt was crystallized from water, it showed 92.7% purity with respect to acid and 92.8% purity with respect to salt, when expressed as hemitartarate. This suggested that the salt may be a monohydrate since a purity of 93% is expected in the case of
'or
20
U
lP
10.
0.
J
I
-tot
0'2
0.1 MASS FRACIION OF HEMITARTARATE SALTS
Figure 1. Solubility curves for hemitartarate salts in methanol: (-0-) solubility of I; (-A-) solubility of 11.
-I01 0
0.1
0.2
0.3
0.4
0 . 5 0.55
MASS FRACTION OF HEMITARIARATES
Figure 2. Solubility curves for hemitartarate salts in water: (-O-) solubility of I; (-A-) solubility of 11.
monohydrate. The observation that the monohydrate is formed, when the hemitartarate is crystallized from water, is in agreement with the published literature (Pitre and Grabitz, 1969). Preparation of Neutral Tartarates. It was observed that ~(+)-2-amino-l-butanol-~(+)-neutral tartarate easily separated from the methanolic solution, but ~ ( - ) - 2 amino-1-butanol-L(+)-neutral tartarate did not crystallize from methanolic solution. We could not obtain the letter in crystalline form even after changing the solvent and temperature of crystallization. Therefore, the study of the binary and ternary equilibria of neutral tartarates could not be carried out. The neutral tartarate of ~ ( + ) - 2 amino-1-butanolshowed 99.5% purity. Since it was found that the hemitartarates were more conveniently separated than the neutral tartarates, no attempts were made to pursue the work based on neutral tartarates. Binary Phase Equilibria. After the preparation and characterization of the hemitartarates, their solubilities were determined in methanol and water at different temperatures (Figure 1 and 2). Some of the data were replicated and the difference was found to be not more than 3% of the mentioned value. The study showed that L(+)-2-amino-l-butanol-~(+)-hemitartarate (I) was less soluble in methanol and more soluble in water than D-
428
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 2, 1985
Table 11. Ternary Equilibrium Data for L( )-2-Amino-l-butanol-~( +)-hemitartarate (1)~(-)-2-Amino1-butanoh( )-hemitartarate (11)-Solvent compn of saturated s o h at invariant point mf of solvent temo. "C mf of I" mf of 11" solvent" 0.744 28 0.06 0.196 methanol 0.213 0.527 water 28 0.26 0.572 7 0.3 0.128 water 0.216 0.682 aqueous methanol 7 0.102
+
+
"mf = mass fraction
(-)-2-amino-l-butanol-~(+)-hemitartarate (11). This observation is in agreement with the trend previously reported by Radke et al. (1954) and by many patentees. In the case of methanol, it was observed that the differences in the solubilities of the two diastereomers went on increasing with temperature. Hence, better resolution could be achieved at higher temperatures. However, the yield of the crystals would be low at higher temperatures. Therefore, an intermediate temperature of 28 O C was selected for the resolution with methanol. The use of water as a solvent was considered at two typical temperatures, viz., 28 O C and 7 "C.The difference in the solubilities of the diastereomers at 28 "C was not significantly different from that at 7 "C (Figure 2). In view of this, a lower temperature of 7 "C was preferred because it is expected to give a better yield of crystals without affecting the purity of I1 in the crystalline residue. Since the solubility behavior of the two diastereomers is opposite in methanol and water, an alternate crystallization scheme using methanol followed by water can be employed. Thus the desired diastereomer (I) can be crystallized from methanol. The mother liquor containing an excess of the undesired diastereomer (11) can then be crystallized from water after removal of methanol. This step would give I1 as the crystalline residue. If the yields of the crystalline material in the two steps are properly adjusted, the mother liquor from the second crystallization would contain nearly equal amounts of the two diastereomers, and these could be recycled after removing water. If a scheme based on a single solvent was envisaged, it would have needed several stages for the separation of various mixtures of different compositions encountered in the process. Ternary Equilibria. The ternary equilibrium between the two diastereomers and the solvent was studied and the invariant point at a given temperature was determined. The invariant point refers to the composition of the solution which is saturated with respect to both the solutes. This is determined by starting with a system which on crystallization gives a solid fraction that contains both solutes (Mullin, 1972). The composition of the solution pertaining to the invariant point D (Figure 3) for a methanolic system at 28 OC was determined (Table 11).
I
0.4
0.2
-
----,-----I---
MASS FRACTION
0.6
OF (11)
0.e
18
Figure 3. Ternary phase diagram for the system I-11-methanol at 28
OC.
When some experiments were performed for determining the tie lines by taking different concentrations of the salt, an approximately straight line connecting the points of solubilities of .the pure diastereomers and the invariant point was obtained (Figure 3). Depending on whether the composition of the starting mixture lies in triangle AED or triangle BFD, the crystalline material would be essentially pure I and 11, respectively. The weight of solids obtained from a starting composition, e.g., P I , can be predicted from this phase diagram with the help of the lever-arm principle wt of crystalline solids PICl =wt of starting solution AC1 If the starting composition falls in triangle ADB, the composition of the solution would correspond to the invariant point and the amount of solids precipitated can be determined from the material balance. A few experiments were carried out with different starting compositions, and the yield as well as composition of the crystals were determined. The results so obtained and those predicted from phase equilibria are compared in Table 111. The experiments showed that the predicted and observed values were almost in agreement when the two components of the crystalline reside differed greatly in amounts, though there was wide deviance when the components were more nearly equal in amounts, such as in trial no. 4 (Table 111). The invariant point for the aqueous system was determined at two different temperatures, viz., 28 "C and 7 "C. The compositions of the saturated solutions pertaining to the invariant points are given in Table 11. The tie lines were drawn assuming straight lines between the points of solubilities of the individual diastereomers and the invariant point (Figure 4). In order to crystallize pure I1
Table 111. Typical Results of the Separation of the Diastereomers with Methanol and Aqueous Methanol compn of cryst residue wt of cryst residue, g pred obsd trial no. solvent, .g pred obsd I, % 11, % I, % 11, % 10 20 3" 40
5*
78.7 63.0 47.2 39.0 15.6
8.45 9.23 9.6 13.3 10.7
8.46 9.4 9.6 11.0 9.3
100 100 91 77
9 23 100
98.5 97.0 91.0 89.0 6.0
1.5 3.0 9.0 11.0 94.0
"Starting composition was 13.425 g of I and 13.425 g of I1 in methanol as a solvent. *Starting composition was 4.3 g of I and 13.14 g of I1 in water containing 36% w/w methanol.
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 2, 1985
429
k
\
0
\I4 0.2
0.L
06
M A S S F R b C l l O N OF
0 8
(111
-.. Q!
Figure 4. Ternary phase diagram for the system I-11-water a t 7 O C .
from the aqueous solution, the starting composition should lie within the triangle MNQ (Figure 4), N being the invariant point at 7 "C. A similar system had been used by Pitre and Grabits (1969),but it was not possible to compare their results with those observed by us because they had not mentioned the crystallization temperature. From the weight of crystalline residue they had obtained, it appears that the temperature of crystalliition might have been around 7-10 "C,because in this temperature range comparable results were obtained. Typical Process Version. From the phase equilibria in both the solvents, it was found that the resolution worked better in methanol. A product of 97% purity could be obtained when the first crop was restricted to 35% of the t ~ t asalts. l A second crop, either by removal of solvent or cooling, would give crystals of low purity of the desired diastereomer. Therefore, it was thought appropriate to distill out methanol and then separate out I1 from aqueous system equal to the amount of I already removed during earlier stages. Thus, nearly 5050 mixture of I and 11would be obtained and after removal of the water, the salt mixture may be conveniently recycled along with the fresh racemic mixture to the first step. In actual practice, it would be difficult to completely remove methanol and then add water, because this would necessitate the handling of a highly viscous or pasty mass. It would therefore be desirable if the second step were carried out in a typical aqueous solution where some methanol is allowed to remain from the previous step. In view of this, water containing 36% w/w methanol was chosen as a solvent, and the equilibrium data were collected at 7 "C (Figure 5). The composition of the saturated solution corresponding to the invariant points was determined (Table 11). Typical Separation Scheme. L(+)-Tartaric acid (16.85 g) was dissolved in 80 mL of methanol and 10 g D L - ~ amino-1-butanol was added in a semibatch manner. After the addition, the hot supersaturated solution was cooled to 28 "C and seeded with I. The precipitated salts were filtered. The yield of the crystals was 9.4 g and the composition was 97% of I and 3% of II. The filtrate containing 4.3 g of I and 13.14 g of 11 was distilled at 75 "C under atmospheric pressure to give a thick syrup. The weight of the syrup was 23.04 g, which contained 17.45 g of the salts. Ten milliliters of water was added to the syrup so that the composition of the solvent became approximately 36% w/w methanol in water. The syrup was dissolved in
MASS FRACTION O F
Ill1
Figure 5. Ternary phase diagram for the system I-11-aqueous methanol at 7 O C .
the solvent by heating to about 45 "C and it was then cooled to 7 "C. The crystallized solids containing the monohydrate of hemitartarates were filtered. Weight of the solid fraction was 10 g, which was equivalent to 9.3 g when expressed on the basis of anydrous salts. The composition of the solid fraction was determined and it was found to contain 94% of I1 and 6% of I. The comparison between the results predicted from the phase diagram and those experimentally observed is given in Table 111. Thus 4.4 g of I1 and 3.75 g of I remained in the filtrate, giving salt compositions of 54% I1 and 46% I. It should be possible to suitably alter the crystallization conditions whereby the solution after the second step would contain a 5050 mixture of the two diastereomers. The desired hemitartarate (I) was then dissolved in water and hydrolyzed with lime. The calcium tartarate formed, was filtered, and the filtrate was subjected to fractional distillation. The aminobutanol so obtained had a specific rotation of +lO.Oo, which is generally acceptable for the manufacture of ethambutol. The undesired hemitartarate (11)was similarly treated to recover I,(+)-tartaric acid in the form of calcium tartarate. The latter can be reacted with sulfuric acid to obtain L(+)-tartaric acid for reuse. Conclusions For the separation of racemic mixtures involving preparation of the diastereomers followed by fractionation based on the difference in solubilities of the diastereomers in various Solvents, it is possible to provide a rational basis for the development and design of the process by studying the relevant phase equilibria. The possibility of using such an approach has been ascertained for the separation of DL-2-amino-1-butanolusing L(+)-tartaric acid as the resolving agent. Registry No. I, 94944-55-5; 11, 26293-36-7; DL-2-amino-1-butanol, 13054-87-0; methanol, 67-56-1.
Literature Cited Ichlkawa, Y.; Hirata. H.; Sawaki, T.; Honda, Y.; Ishimaru, K. Japan, Kokai 75 11 1 005, 1975. Mullin, J. W. "Crystallization", 2nd ed.; Butterworths; London, 1972; pp 114-121. Pitre. D.;Grabitz, E. 6 . Chimia ISSO, 23, 399. Radke, F. H.; Fearing, R. B.; Fox, S. W. J . Am. Chem. SOC. 1954, 76, 2801. Seres, J.; Daroczi, I . Hungarian Patent 157225, 1970. Zoja, G. U S . Patent 3579586, 1971.
Receiued for reuiew October 14, 1983 Accepted June 28, 1984
