Ind. Eng. Chem. Res. 1994,33,3224-3229
3224
Sorption and Extraction of Lactic and Succinic Acids at pH > pKal. 2. Regeneration and Process Considerations Lisa A. Tung?and C. Judson King' Department of Chemical Engineering and Lawrence Berkeley Laboratory, University of California, Berkeley, California 94720
As shown in part 1,carboxylic acids can be recovered from solutions at pH > pKal of the acid by solid sorbents or liquid extractants t h a t are sufficiently basic to maintain substantial capacities even at moderately high values of pH. Appropriate extractants and sorbents can be regenerated by back-extracting the acid into aqueous trimethylamine ( T U ) solution, followed by thermal decomposition of the resulting trimethylammonium (TMAm) carboxylate. Polymeric sorbents and liquid extractants with tertiary amine functionalities are completely regenerable by leaching with aqueous TMA solutions. Fixed-bed breakthrough curves for the sorbent Dowex MWA-1 in both the uptake and regeneration modes are self-sharpening, and the regeneration curve displays a focusing effect. For acids having low-to-moderate solubilities thermal cracking of the resultant TMAm carboxylate yields acid crystals and trimethylamine vapor for recycle. Because lactic acid is more soluble and tends to self-esterify, simple cracking gives only partial regeneration. Another approach is to react TMAm lactate with a n alcohol to form the ester and TMA for recycle. The feasibility of this approach was demonstrated through laboratory-scale experiments with n-butanol.
Introduction Many fermentations producing carboxylic acids operate most effectively at a pH above the PKal of the acid. Some important waste streams also have this characteristic. Therefore the recovery method must function at pH > pKal, if the acid is to be recovered without changing the pH of the feed stream. Aa is demonstrated in Part 1 of this series (Tung and King, 19941, sufficiently basic solid sorbents and liquid extractants sustain capacity for carboxylic acids two and more pH units above the pKal of the acid. Strongly basic extractants and sorbents require correspondingly strong methods for regeneration. One method is back-extraction into an aqueous solution at sufficiently high pH. The aqueous base competes with the less basic sites on the sorbent or extractant, pulling the acid into the aqueous phase. If an ordinary strong base (e.g., aqueous NaOH or Ca(0H)Z)is used, an acid (e.g., HzS04) must be added in order to liberate the carboxylic acid product. This approach consumes chemicals and generates a waste salt stream. An alternative method involves leaching with an aqueous solution of a volatile base. Poole and King (1991,1994) showed that extracts of carboxylic acids in a solvent mixture of Alamine 336 (Henkel Corp.) and methyl isobutyl ketone (MiBK) can be regenerated by leaching with aqueous trimethylamine (TMA). The resulting trimethylammonium (TMAm)carboxylate can be decomposed thermally, as in eq 1, yielding acid product and TMA and water vapor, available for recycle. It is also possible to leach with aqueous ammonia or a primary or secondary amine, but in those cases irreversible amide formation may occur during the thermal decomposition:
RCOO- +HN(CH,),(aq)
+
+
RCOOH N(CH,), t H,O (1) For slightly soluble acids, such as succinic and fumaric t Present address: Rohm & Haas Co., 727 Norristown Road, Spring House, PA 19477.
*To whom correspondence should be addressed at the Department of Chemical Engineering. 0888-5885/94/2633-3224$04.5QlQ
Raffinate Extractant
Acid
Figure 1. Carboxylic acid recovery process.
acids, Poole and King (1991) found that partial evaporation of the aqueous TMAm carboxylate solution results in precipitation of the product acid. Figure 1presents a flow sheet for a process in which a carboxylic acid is recovered by extraction, with back-extraction into an aqueous solution of TMA, followed by thermal decomposition of the TMAm carboxylate and precipitation of the carboxylic acid in a n evaporative crystallizer. As is demonstrated in this paper, this method of regeneration is strong enough to enable the use of basic sorbents and extractants effective a t a pH above the pKal of the carboxylic acid (King and Tung, 1992; Tung and King, 1994). Because lactic acid is much more soluble in water than are succinic or fumaric acids and tends to form intermolecular esters in concentrated solution, it is more difficult to decompose TMAm lactate into lactic acid and TMA. In a typical experiment, Poole and King (1991) found that only 62% of the TMA present in the initial solution was driven off by prolonged heating of an aqueous solution of TMAm lactate. A viscous solution of water, TMA, lactic acid and lactate polymers was left
0 1994 American Chemical Society
Ind. Eng. Chem. Res., Vol. 33, No. 12, 1994 3225 behind. Addition of MiBK or n-butyl acetate and continued evaporation removes some of the remaining TMA (Tung, 1993). In the current work, esterification by reaction with an alcohol was considered as an improved method for recovering lactic acid from TMAm lactate. Esters can be formed by the reaction shown in eq 2. This reaction
+
RCOO-+HN(CH3)3 R O H RCOOR
& Stirrer
water CO1lcnion Cylinder
+ H,O + N(CH,),
t (2)
is driven to the right by removal of water and TMA by vaporization. The resulting mixture can then be distilled to recover excess alcohol and the lactate ester. The ester can be taken as the ultimate product, if desired, or it can be hydrolyzed to form aqueous lactic acid and alcohol. The alcohol would then be removed by distillation and recycled t o the reactor. A major advantage of this process is that the lactate ester is taken off the top of a distillation column and hence should be recoverable as a pure (e.g., USP grade) product. By contrast, in the thermal cracking process the product is taken as the bottoms. Filachione and Costello (1952) investigated reactions of ammonium lactate with a variety of alcohols to form esters. The reactions were driven by removal of ammonia and water overhead. In general, primary alcohols gave higher conversions than secondary alcohols, and high-boiling alcohols gave higher conversions than low-boiling ones. Boric acid, trihexylamine, basic aluminum acetate, and silica gel were weak catalysts, reducing reaction time by a factor of about one-half. Filachione et al. (1951) also studied the esterification of several alkylammonium carboxylates. The salts were prepared from nonvolatile, long-chain amines, such as tributylamine and decylamine, and hence the amines were not removed overhead. The goals of the current work were 3-fold. First, regeneration of the sorbents and extractants studied in part 1 of this work with aqueous TMA was assessed through batch equilibration experiments. Second, fixedbed studies were carried out for both acid uptake and TMA leaching so as to assess rate effects. Finally, laboratory experiments were used to evaluate the estification approach for conversion of TMAm lactate.
Experimental Section Materials. The carboxylic acids, sorbents and extractants used in this work were described in part 1. Trimethylamine (Aldrich, 25 wt % in water, >99.5% purity), n-butanol (Fisher Scientific, 99.8%),and n-butyl lactate (Aldrich, 99%) were also used. TMA Leaching. Leaching experiments were performed by contacting sorbent, laden with succinic or lactic acid, with aqueous TMA solutions (0.1-0.9 M) in sealed vials placed in a shaker bath a t 25 "C for 48 h. The initial TMA concentration was determined by acidification and back-titration with aqueous NaOH (Tung, 1993), with precautions taken to avoid loss of volatile TMA. The concentration of acid in the final aqueous phase was determined by high-performance liquid chromatography (HPLC), as described in part 1. The final leached sorbent was centrifuged and weighed, and the weight of the wet sorbent was used t o calculate the total solution remaining in the pore volume. The procedure for regenerating acid-laden extractants was entirely analogous to the above procedure.
Figure 2. Apparatus for cracking experiments.
Fixed-Bed Studies. Fixed-bed rate studies were carried out with a 1.0 cm inside diameter x 30 cm long glass column packed with Dowex MWA-1 and equipped with a Masterflex (Cole-ParmerCo.) pump. Precautions were taken to minimize effects due t o mixing in extraneous volumes (Tung, 1993). Samples were collected from the outlet of the column using a fraction collector. During TMA leaching, the solution reservoir was kept tightly sealed to avoid loss of TMA, and the outlet samples were capped as soon as they were taken. The volume and pH of each outlet sample were then measured, and the concentrations of acid or other species were determined as in previous experiments. Simple Thermal Cracking. The apparatus for the TMAm lactate cracking experiments is depicted in Figure 2. An aqueous solution, typically containing 0.076 mol each of lactic acid and TMA and 3.5 mol of water, was placed in a three-neck, lOO-mL, roundbottom flask and heated by a mantle attached to a voltage regulator. The experiments were performed at 460-480 mmHg absolute pressure, and nitrogen was sparged into the solution at a flow rate of approximately 75 mumin to prevent possible oxidation of TMA. The temperature of the liquid was monitored continuously. As TMA and water vapor were driven off, the temperature of the solution increased gradually as the boiling point increased. Late in the run, when most of the TMA and water vapor had been driven off, there was generally a sharp rise in temperature. At this point, the voltage regulator was adjusted to keep the temperature below 120 "C. The rates of evolution of water and TMA were monitored by absorption into HzS04 containing an indicator. The amount of TMA remaining in the concentrated solution was determined by elemental analysis of the nitrogen present. The evolution of water was monitored by the level in the water collection cylinder. The water from the collection cylinder and the final absorber liquid were titrated with H z S O ~and NaOH, respectively. Mass balances for TMA generally closed within 5%. Esterification. Esterification was carried out with n-butanol, which produces commercially valuable nbutyl lactate and allows for easy removal of water as the butanol-water azeotrope. Concentrated TMAm lactate solution was reacted with n-butanol in the apparatus shown in Figure 3. The reaction mixture was heated in a three-neck, 500-mL, round-bottom flask, equipped with heating mantle and voltage regulator. A nitrogen sparger and a thermometer passed through the side necks of the flask. Vapor from the reacting mixture passed through a silver-jacketed column packed with glass helices and was condensed over a modified DeanStark trap. The top, n-butanol-rich phase was automatically returned to the reaction flask through the column; the bottom, aqueous phase was periodically withdrawn from the trap using the stopcock. As in the
3226 Ind. Eng. Chem. Res., Vol. 33, No. 12, 1994
:v1
0.3
To Vacuum
0 4
Injection of sulhuic Acid
-
Nitrogen
for Sampling
0
Results and Discussion Leaching with Aqueous "MA, Sorbents. Figure 4 shows results for leaching lactic acid from Dowex MWA-1 with varying amounts of aqueous T U . The solid curve is calculated from a theoretical model (Tung, 1993) based solely upon the pKa of TMA and the
6
mols W m o l s acid
Stirrer
cracking experiments, the rates of evolution of TMA and water were monitored over time. The temperature in the reaction flask increased as the boiling point of the mixture increased. Typically, all evolution of water and TMA ceased by the time the reaction mixture reached 142 "C. The amounts of n-butanol and n-butyl lactate in the final reaction mixture were determined by gas chromatography. In some cases, the nitrogen content was determined via elemental analysis. The overall mass balance closed within 5-10%. Some of the reaction mixtures were distilled a t 420 mmHg abs, using the silver-jacketed column packed with glass helices. Reaction Rates. Experiments to determine rates of esterification were carried out a t the (changing) boiling point of the mixture, and the heating rate was kept approximately constant from one run to the next by maintaining a constant setting on the voltage regulator. Small (0.1-0.2 mL) samples were withdrawn periodically through a septum and analyzed by gas chromatography to monitor the rate of n-butyl lactate production. The rates of evolution of water and TMA were monitored as in the simple thermal cracking experiments. The experiments involving potential catalysts (concentrated sulfuric acid, Amberlyst 15 cation exchange resin, silica gel, or alumina powder) were identical except that the catalyst was added to the reaction flask at the beginning of the experiment. Hydrolysis of n-Butyl Lactate. In the hydrolysis experiments, 1mol of n-butyl lactate was combined with 3 mol of water and 1 mL of concentrated sulfuric acid catalyst in a three-neck, 500-mL, round-bottom flask. The solution was heated under total reflux at atmospheric pressure, and samples were periodically withdrawn through a septum seal placed in one of the flask necks. Additional details on the apparatus and procedures for the various experiments are given elsewhere (Tung, 1993).
4
Figure 4. Leaching of lactic acid from Dowex MWA-1 into aqueous solutions of varying TMA concentration (initial loading 0.36 g/g, phase ratio 0.015 Ug).
8 Figure 3. Apparatus for esterification experiments.
2
1.2 1'5
5
o
r 0
2
6
mols W m o l s acid
Figure 5. Leaching of succinic acid from Dowex MWA-1 into aqueous solutions of varying TMA concentration (initial loading 0.52 g/g, phase ratio 0.016 Ug).
sorption equilibria reported in part 1. With Reillex 425, Duolite A7, and Dowex MWA-1, essentially complete recovery of lactic acid was achieved when 1mol of TMA was present for every mole of acid (Tung, 1993). These results indicate that the ionized, aqueous TMA is a much stronger base than any of these three sorbents. For Amberlite IRA-35 laden with lactic acid, somewhat more than a stoichiometric amount of TMA was required for full regeneration. About 87% recovery was achieved when the molar ratio of TMA to acid was equal to 1.0, and essentially 100%recovery was achieved at a ratio of 1.45 (Tung, 1993). Regeneration of IRA-35 is more difficult because of its greater basicity. Only about 27% recovery of lactic acid was achieved from Amberlite IRA-910, even with 2.5 mol of T W m o l of acid. This recovery compares with the 19%recovery achieved by leaching with pure water. These results reflect the high basicity of the quaternary ammonium resin. The results of TMA leaching studies conducted with succinic acid and Reillex 425, Duolite A7, Dowex MWA1,and Amberlite IRA-35 were similar to those obtained with lactic acid, except that 2 mol of TMA are needed per mole acid. Results for leaching of succinic acid from Dowex MWA-1 are shown in Figure 5. As for lactic acid, a somewhat greater than stoichiometric amount of TMA was required to regenerate the more highly basic IRA35 (Tung, 1993). Bio-Rad AG3-X4 laden with succinic acid was fully regenerable at a molar ratio of TMA to acid slightly greater than 2.0, except for about 8% of the acid, which was presumably held by the strong-base sites. Amberlite IRA-910 (Figure 6) was about 62% regenerable when approximately 1.5 mol of TMA was present for every mole of acid. This much higher recovery than that achieved with lactic acid most likely
Ind. Eng. Chem. Res., Vol. 33, No. 12, 1994 3227
1.2
I 2
80
0
6
2
malr ThWmolr acid
0
TMA Fted
20
2
6
8
Time (hours)
Figure 6. Leaching of succinic acid h m Amberlite IRA-910into aqueous solutions of varying TMA concentration (initial loading 0.55 g/g, phase ratio 0.017 Ug).
Acid Frod
0
40
Nvmkr of Bed Volvmss
Figure 7. Breakthrough and leaching curves for succinic acid on Dowex MWA-1, F = 1.5 mumin, L = 18 em. 5.25 wt % acid feed, 5.25 wt % TMA feed.
reflects conversion of the resin from the bisuccinate to succinate form. Extractants. Amberlite LA-2 (0.3 M) in n-octanol was fully loaded with lactic acid and then back-extracted with aqueous solutions of TMA of varying concentrations. Full recovery of the acid was achieved when at least one mole of TMA was present per mole of acid. Poole and King (1991)have demonstrated that Alamine 336/MiBK extracts of lactic and succinic acids are fully regenerable by back-extraction with aqueous solutions of TMA. Fixed-Bed Studies. Figure 7 presents the results of a cycling, fixed-bed experiment involving breakthrough of 5.25 w t % aqueous succinic acid on Dowex MWA-1, regeneration by leaching with 5.25 wt % aqueous TMA, and then a second breakthrough step. There are several important results. First, the uptake breakthrough curves are virtually identical, and the amount of acid leached, as calculated from the area under the leaching curve, is equal to the amounts sorbed, as calculated from the areas to the left of the uptake curves. This demonstrates that TMA leaching regenerates the resin fully, without diminishing the acid-uptake capacity. Because of the highly favorable isotherms, the breakthrough curves for both uptake and leaching are self-sharpening, and a constant pattern is quickly achieved (Shenvood, et al., 1975). The ratelimiting step most likely occurs in the particle phase, and a pore-diffusion model fits the data reasonably well (Tung, 1993). Another interesting feature is that the concentration of acid a t the peak of the leaching curve is 1.4 times higher than that in the feed stream. This result can be explained in terms of a focusing effect (Busbice and Wankat, 1975; Wankat, 1990). Focusing, or trapping,
Figure 8. Simple thermal cracking of trimethylammonium lactate 0.076 mol ofladie acid, 0.077 mol ofTMA, 3.5 mol of water initially present. P = 470 mmHg absolute.
of a solute wave can occur when there exists some thermodynamic variable (in this case, pH) which, when changed, causes a large change in the amount of solute sorbed. Focusing causes a high concentration of solute to become trapped a t the pH wave front. Both the selfsharpening and focusing effects are beneficial for process efficiency. Thermal Cracking. In the TMAm lactate cracking experiments, essentially 100%of the total water and an average of 65% of the total TMA present initially were recovered overhead. The final solution was viscous and contained, on average, 0.27 mol of T W m o l of acid, as determined by elemental analysis, indicating that some TMA may have been lost. Figure 8 presents the results of a typical thermal cracking experiment performed a t 470 mmHg absolute under a nitrogen atmosphere. The rate of evolution of TMA is slow until most of the water has been driven off, a t which point the temperature of the solution rises, and TMA is evolved a t a higher rate. It appears that the cracking reaction occurs a t a significant rate only a t elevated temperatures. However, it is also possible that the presence of water helps to stabilize the TMAm lactate ion pair, making cracking less favorable thermodynamically. There are several possible reasons for the incomplete removal of TMA during thermal cracking. The cause could be thermodynamic; i.e., the ratio of lactic acid t o TMA increases throughout the experiment, which may reduce the volatility of TMA. It is also possible that byproducts are formed irreversibly, although there is no other evidence of that. On the other hand, kinetic and/or transport limitations may prevent complete removal of TMA. At lower concentrations of TMA, the driving force for the reaction may become so low that the rate diminishes greatly. Alternatively, the high viscosity solution may introduce transport limitations that inhibit removal of TMA. Color changes accompanied heating to temperatures above approximately 115 "C. These are discussed and interpreted elsewhere (Tung, 1993). Esterification. For an initial molar ratio of nbutanol to lactate equal to 2.5, the average percent conversion of TMAm lactate to n-butyl lactate, calculated from measurements of the n-butyl lactate in the final solution and lactic acid originally added to the system for 10 experiments was 68 f 4%, a t the 95% confidence level. This value is close to that achieved by Filachione and Costello (1952) with ammonium lactate. The conversion was unaffected by the presence of silica gel, alumina, sulfuric acid, or Amberlyst 15 (Rohm
3228 Ind. Eng. Chem. Res., Vol. 33, No. 12,1994
and Haas Co.) a strong cation exchange resin. The duration of the preconcentration step also had little effect on the conversion. Unconcentrated (1.5M) TMAm lactate reacted with n-butanol gave a conversion of 70%. An initial molar ratio of n-butanol to lactate equal t o 10.0provided 64% conversion, not much different from the 68% achieved using the molar ratio of 2.5. Lactic acid with no TMA was reacted with n-butanol at a molar ratio of 2.5. The conversion to n-butyl lactate was 75%, somewhat higher than the 68% for TMAm lactate. The result from this lactic acid experiment may be somewhat high, since Filachione and Costello (1952) performed the same experiment and achieved a conversion of only 62%. Thus, the evidence is not conclusive that the reaction of TMAm lactate with n-butanol gives a lower conversion than that of lactic acid with nbutanol. Comparison of the gas chromatograms for the reactions of n-butanol with TMAm lactate and with lactic acid revealed no major differences in the composition of the final products, aside from the presence of TMA in the former case. Chromatograms of the final solutions showed a small, wide peak at a high retention time, perhaps corresponding t o a long-chain ester or esters. If such multimers or polymers are formed, they can be fed back to the esterification reaction, where they will eventually be converted to n-butyl lactate under more dilute conditions (Kirk-Othmer, 1979;Nielsen and Veibel, 1967;Filachione and Costello, 1952). On the basis of results from elemental analyses, the final reaction mixture from the esterification experiments contained approximately 10% of the nitrogen originally present in the initial solution. This nitrogen may be in the form of molecular TMA or TMAm salts that are “trapped” in solution by kinetic or thermodynamic limitations. Most of the remaining TMA is removed overhead in the subsequent distillation step at 420 mmHg absolute, as part of the fractions removed at 94 “C or below. Therefore, it appears that most or all of the TMA remaining in the final esterification mixture is not in the form of irreversibly formed byproducts. The retention of TMA may be related to the fact that it is readily absorbed by alcohols (Windholz, 1983). In the original esterification experiments, any TMA vapor released from the reaction flask passed over the Dean-Stark trap, which contained condensed nbutanol and water. TMA absorbed into the n-butanol phase passes back into the reactor. Rates. Rate experiments were conducted using initial solutions containing 0.2 mol of lactic acid, 0.5 mol of n-butanol, and amounts of water and TMA that varied with the degree of preconcentration. Under favorable conditions the esterification reaction was complete in 2-3 h (Tung, 1993). The esterification process appears t o consist of an induction step, during which water is driven off and very little ester formation occurs, and then a second step, in which most of the ester formation occurs. At the lowest water content, there is virtually no induction period. During the induction period, the temperature of the solution remains approximately constant, as water is evaporated. The rate of evolution of TMA is smaller during the induction period. Presumably, esterification of TMAm lactate takes place in two steps, the first being cracking of TMAm lactate into lactic acid and TMA, and the second being reaction of lactic acid with the alcohol to form the ester and water. High concentrations of water drive both reactions toward the left.
Several potential catalysts were tested for their effects on the rate of conversion of TMAm lactate to n-butyl lactate. The addition of sulfuric acid to the reaction mix reduced the reaction time by about half (Tung, 1993). Since this amount of acid is sufficient to neutralize only about 3.5% of the TMA present, the mineral acid appears to act as a true catalyst, rather than simply displacing lactic acid from TMAm lactate. Silica gel and Amberlyst 15 also showed some catalytic effect, but basic and acidic alumina did not (Tung, 1993). Distillation. Batch distillation of the final mixture from the esterification reaction resulted in three fractions, the first containing n-butanol and some residual water, the second containing 95% n-butanol, and the third containing 89% n-butyl lactate. The residue from distillation was viscous and dark brown, with a caramel odor. This is a common occurrence (Schopmeyer,1954) and one that is difficult to avoid except possibly under very high vacuum. Even at pressures as low as 90 mmHg absolute the residue from distillation of the hydrolysis products was brown and viscous. Hydrolysis. Very little hydrolysis occurred in 4 h without the H2S04 catalyst. With HzS04, 50%conversion was achieved within 30 min, and no further conversion was observed after 2 h. For three runs at the indicated ratio of water t o n-butyl lactate the average conversion was 50 f 4% at the 95% confidence level. The average equilibrium constant calculated from these data for the forward esterification reaction was 4.9 f 1.1. This value is within the normal range for esterification reactions (Kirk-Othmer, 1979). An experiment was also performed using 18 mol of water, 1 mol of n-butyl lactate, and 1 mL of concentrated H2SO1, yielding an apparent equilibrium constant of 5.4. With the 3:l ratio, the reaction was complete in 16 min or less, and with the 18:l ratio the reaction was complete in 29-39 min. Process Considerations. n-Butanol was used in this work because it allows for easy removal of water from the system. Methanol would have been more dificult to use because it is more volatile than, and fully miscible with, water. Thus, there is no convenient method of driving water out of solution without also removing methanol. In an industrial process, however, it may be desirable to use some other alcohol. Methanol, for example, has been used to purify lactic acid industrially. The alcohol is continuously recovered overhead and recycled back t o the process. There are several potential advantages to a process using methanol. Because methanol is more volatile than water, the final step, in which residual alcohol is removed from the hydrolyzate, is simplified. Methanol has a lower latent heat of vaporization than does n-butanol. On the other hand, if methanol is used, the system will probably have t o be pressurized t o produce a higher temperature for rate purposes. Problems encountered in a batch esterification process are that concentrated lactic acid is formed in the distillation pot, and at higher temperatures these solutions polymerize and discolor. One approach to this problem is to design the process such that the reaction and distillation steps occur simultaneously. The alcohol can be recycled continuously to the esterification reaction vessel (Schopmeyer, 1944;Wenker, 1943;Schopmeyer and Arnold, 1944;Weisberg and Stimpson, 1942; Keyes, 1932). In this manner, concentrated lactic acid solution never accumulates in the bottom of a distillation column. Likewise, the hydrolysis step and the
Ind. Eng. Chem. Res., Vol. 33, No. 12, 1994 3229 second distillation can occur simultaneously. Discoloration and polymerization are minimized because an aqueous environment is always maintained in the hydrolysis tank. Relatively long residence times may be required if the rates of esterification are slow. Another probably very desirable alternative is to use combined reaction-distillation columns for both the esterification and hydrolysis steps. In this configuration, the products of the reactions are continuously removed, thus driving the reaction toward completion. Simultaneous reaction-distillation should work well for the hydrolysis reaction, since the reaction kinetics are relatively fast, and the relative volatilities of the various components are sufficiently different. Simultaneous distillation-reaction may not work as well for the esterification process. Because this reaction proceeds more slowly, large column holdups might be necessary to obtain complete conversion. If complete conversion is not achieved, lactic acid will be present in the bottoms of the first distillation column.
Conclusions With the exception of quaternized resins, all of the sorbents studied are fully and efficiently regenerated by leaching with an aqueous solution of TMA. For most sorbents, 100% recovery of the acid is achieved when one equivalent of TMA is present for each equivalent of carboxylic acid. Amberlite IRA-35, which is a more strongly basic resin, requires somewhat more than a stoichiometric amount of TMA for complete regeneration. Amberlite LA-2 and Alamine 336 extracts of carboxylic acids are also fully regenerable by this method. The resulting TMAm carboxylate can be cracked thermally, yielding acid product ant TMA vapor. With low-to-moderate solubility acids, such as succinic and fumaric, acid crystals are formed. With lactic acid, which is much more soluble and tends to self-esterify, simple thermal cracking performed in the batch mode results in a viscous mass still containing significant amounts of TMA. Cracking with solvent addition removes some additional TMA but does not result in a pure product. Esterification of TMAm lactate may be preferable to thermal cracking, with or without solvent addition, because a much purer final product can be achieved. Preferred implementations for the esterification and hydrolysis involve simultaneous reaction and distillation, with recycle of reactants t o the esterification/ hydrolysis reaction vessels. Thermal decomposition of lactic acid in the bottoms of distillation columns is thereby avoided.
Acknowledgment This research has been supported by the Biological and Chemical Technology Research F’rogram, Advanced Industrial Concepts Division, Office of Industrial Technologies, U.S.Department of Energy, and by a National Science Foundation Graduate Fellowship. Literature Cited Busbice, M. E.; Wankat, P. C. pH Cycling Zone Separation of Sugars. J . Chromatogr. 1975,114,369. Filachione, E. M.; Costello, E. J. Lactic Esters by Reaction of Ammonium Lactate with Alcohols. Ind. Eng. Chem. 1952,44, 2189. Filachione, E. M.; Costello, E. J.; Fisher, C. H. Preparation of Esters by Rection of Ammonium Salts with Alcohols. J . Am. Chem. SOC.1951,73,5265. Keyes, D.B. Esterification Processes and Equipment. Ind. Eng. Chem. 1932,24,1096. King, C. J.; Tung, L. A. Sorption of Carboxylic Acid from Carboxylic Salt Solutions at pHs Close to or Above the pKa of the Acid, with Regeneration with an Aqueous Solution of Ammonia or Low-Molecular-Weight Alkylamine. U S . Patent No. 5,132,456, July 21, 1992. Kirk-Othmer Encyclopedia of Chemical Technology, 3rd ed.; John Wiley & Sons: New York, 1979;Vol. 9. Nielsen, J . I.; Veibel, S. The Reactivity of Lactic Acid and Some of Its Simple Derivatives: A Review. Acta Polytechnica Scandinavica, Copenhagen, 1967;Chapter 63,pp 25-57. Poole, L. J.; King, C. J. Regeneration of Carboxylic Acid-Amine Extracts by Back-Extraction with an Aqueous Solution of a Volatile Amine. Ind. Eng. Chem. Res., 1991,30,923. Poole, L. J.; King, C. J. US.Patent pending, 1994. Schopmeyer, H. H.; Arnold, C. R. Lactic Acid Purification. U.S. Patent 2,350,370, June 6, 1944. Schopmeyer, H. H. Lactic Acid. In Industrial Fermentations; Underkofler, L. A., Hickey, R. J., Eds.; Chemical Publishing Co.: New York, 1954;pp 91-419. Sherwood, T. K.;Pigford, R. L.; Wilke, C. R. Mass Transfer; McGraw-Hill: New York, 1975. Tung, L. A. Recovery of Carboxylic Acid at pH Greater than pKa, Report No. LBL-34669, Lawrence Berkeley Laboratory, Berkeley, CA, Aug 1993. Tung, L. A.; King, C. J. Part 1 of this series. Ind. Eng. Chem. Res. 1994,33,3217. Wankat, P. C. Rate-Controlled Separations; Elsevier Applied Science: New York, 1990;pp 257-258, 338-340. Weisberg, S.M.; Stimpson, E. G. Preparation of Lactic Acid, U S . Patent 2,290,926,July 28,1942. Wenker, H. Purifymg Hydroxy Aliphatic Acids. U.S. Patent 2,334,524,Nov 16,1943. Windholz, M., Ed. The Merck Index, 10th ed.; Merck & Co., Inc.: Rahway, NJ, 1983. Received for review January 27, 1994 Revised manuscript received August 12, 1994 Accepted August 29, 1994@ ~
Abstract published in Advance ACS Abstracts, November 1, 1994. @
