Ind. Eng. Chem. Res. 1990,29, 1319-1326
1319
Extraction of Carboxylic Acids with Amine Extractants. 1. Equilibria and Law of Mass Action Modeling Janet A. Tamada,' A. Steven Kertes,t and C. Judson King* Department of Chemical Engineering and Lawrence Berkeley Laboratory, University of California, Berkeley, California 94720
Studies have been made of the extraction of acetic, lactic, succinic, malonic, fumaric, and maleic acids by Alamine 336, an aliphatic, tertiary amine extractant, dissolved in various diluents. The results were interpreted by a "chemical modeling" approach, in which the stoichiometries of acidamine complexes and corresponding equilibrium constants which best represent the experimental results were determined. The acids studied differed in pK, and in the presence or absence of functional groups other than the primary carboxyl group. Diluents were chosen from different chemical classes-electron donating, electron accepting, polar, and nonpolar-so as to examine the effects of diluent-complex interactions. These interactions were found to affect both the stoichiometry of reaction and the magnitudes of the corresponding equilibrium constants. The conclusions reached from these data are compared with those from previous studies. Long-chain, aliphatic amines are effective extractants for separation of carboxylic acids from dilute aqueous solution (Kertes and King, 1986). Generally, the amine extractants are dissolved in a diluent, an organic solvent that dilutes the extractant to the desired concentration and controls the viscosity and density of the solvent phase. Many factors have been found to influence the equilibrium extraction characteristics of these systems. Three important variables are the nature of the acid extracted, the concentration of extractant, and the type of diluent. In this work, the extraction of several carboxylic acids by a tertiary amine extractant in a variety of diluents was examined so as to ascertain and compare the equilibrium behaviors of different systems. Batch extraction experiments were performed with acetic, lactic, malonic, succinic, maleic, and fumaric acids. The extractant used was Alamine 336 (Henkel Corporation), a commercially available tertiary amine, with aliphatic chains that are 8-10 carbons in length. Chloroform, methylene chloride, methyl isobutyl ketone (MIBK), nitrobenzene, and 1-octanol were selected as "active" diluents, i.e., solvents that contain functional groups that interact strongly with the complex. The "inert" diluent studied was n-heptane. Mass Action Law Description of Equilibrium. If "chemical" interactions between the components of the complex are strong compared to the "physical" interactions in the system, the equilibrium behavior can be modeled effectively by postulating the formation of various stoichiometric complexes of acid and amine (Connors, 1987). This mass action law description of a system as a series of stoichiometric reactions will be referred to as chemical modeling. An equilibrium description of the system can be written as a set of reactions of p acid, A, molecules and q amine, B, molecules to form various (p,q) complexes, with corresponding equilibrium constants, /3pq,true:
For practical application, the activities of the organicphase species are assumed to be proportional to the concentrations of the species, with the constants of proportionality (the nonidealities) taken up in the equilibrium constant. The activity of the acid is assumed to be proportional to the concentration of undissociated acid in the equilibrium aqueous phase; thus, differences in properties (e.g., hydrophobicity) among the various acids are incorporated into the equilibrium constant. The apparent equilibrium constant for the overall reaction can be written as Pp,
* To whom correspondence should be addressed.
t Current
address: Department of Chemical Engineering, E25-342, MassachusettsInstitute of Technology, Cambridge, MA 02139. f Affiliation: Hebrew University of Jerusalem; deceased.
(2)
[AIP[BP
where species concentrations are denoted by square brackets and are expressed in molar terms. The loading of the extractant, 2, is defined as the total concentration of acid (allforms) in the organic phase, CLOT, divided by the total concentration of amine (all forms) in the organic phase, CB,tov With appropriate material balances, 2 is determined for a given set of stoichiometries as C'4,org
Z=-=
CPPp,[AlP[Blq
CB,tot
(3)
CB,tot
If diluent interactions with the complex are strong, it may be possible to represent the data by including r diluent, D, molecules specifically into the complex.
In cases where there is a single amine per complex, as in (p,l) stoichiometries, where p = 1, 2, 3, ..., it is convenient to recast the complexation constant for the overall reaction to the constant for the stepwise reaction: A
where the species activities are denoted by braces and the organic-phase species are marked with an overbar.
[APB,I -
=
+ A,,B
2
ApB
[A71 Kpl = [ A ] [ A y B ]
(5)
The association constant for the stepwise reaction, K is the ratio of the association constants for the overafieactions, Ppl/Ppl,l, and Kll is equal to P11. Characteristicsof Calculated Loading Curves. The loading curve is a plot of 2 vs log [A]. Overloading, i.e., loading greater than unity, indicates that complexes with more than one acid per amine have been formed. For
0888-5885/90f 2629-1319$O2.50f 0 0 - 1990 American Chemical Societv
1320 Ind. Eng. Chem. Res., Vol. 29, No. 7, 1990
systems with only one amine per complex, there is no effect of total amine concentration on the loading. If there is more than one amine per complex, loading increases with increasing amine concentration at low acid concentrations. Systems that include the diluent specifically in the complex stoichiometry show decreasing loading with increasing amine concentration. Systems that exhibit aggregation, formation of complexes with large numbers of acid and amine molecules, exhibit an abrupt increase in loading at a critical acid concentration. Limitations of the Mass Action Law Approach. The use of the chemical modeling approach to describe these systems has limitations. By ignoring physical interactions between components, the model may be less realistic than models that account for aqueous- and organic-phase nonidealities in other ways. Minimizing the error between the extraction data and a hypothetical model may imply that complexes of a specific stoichiometry are formed. Without independent evidence, such as spectroscopic data, the existence of these complexes must be viewed with caution, since they may be artifacts of the nature of the model and/or the fitting procedure. Despite the limitations of the chemical model, if complexation is strong, chemical modeling is a useful tool to describe data quantitatively and in deepening understanding of the mechanism of extraction. In part 2 of this work, the results from law of mass action modeling are interpreted in terms of spectroscopic evidence. Experimental Section Alamine 336 (Henkel Corp.), a C8-Cl0 straight-chain tertiary amine mixture, was used as received. Alamine 336 is a pale yellow liquid, with a specific gravity of 0.81 (25 "C/25 "C), a viscosity of 23 CPat 40 "C, and a surface tension of 53 dyn/cm saturated solution at 25 "C (Henkel Corp., 1984). From organic-phase potentiometric titrations of the amine dissolved in l-propanol (Baker reagent grade), using approximately 0.1 mol/L HCl dissolved in 1propanol (Baker, reagent grade) as the titrant, the effective molecular weight was calculated to be 406 g/mol. All solutions were prepared by dilution of the appropriate acid into distilled water, which had been further purified by passage through a Milli-Q water filtration system (Millipore Corporation). Lactic acid (Mallinkrodt, analytical reagent grade, 85%) was diluted to approximately 15% (w/w) and refluxed for 10 h to hydrolyze any dimers or multimers present in the concentrated solution (Holten, 1971). High-performance liquid chromatography (HPLC) of the lactic acid before and after heating confirmed the success of the transformation into a monomeric solution. Malonic (Aldrich Chemical Company, 99%), succinic (Mallinkrodt, analytical reagent grade), acetic (Fisher Scientific, reagent grade), fumaric (Aldrich), and maleic (Aldrich) acids were also used as received. Methyl isobutyl ketone (Mallinkrodt, 99% 1, n-heptane (Baker, reagent grade), methylene chloride (Fisher), and nitrobenzene (Aldrich) were used as received. In the initial experiments, chloroform (Mallinkrodt, analytical reagent grade) was washed three times with water immediately prior to use to remove the ethanol stabilizer. Experiments with washed and unwashed chloroform showed that the presence of ethanol stabilizer had no noticeable effect on the equilibrium data, and later experiments were done without prewashing of the chloroform. Known volumes of aqueous and organic solutions of known concentration were added to Erlenmeyer flasks and equilibrated in a temperature-controlled shaker bath at 25 "C for at least 12 h, which preliminary tests demonstrated to be a sufficient time for equilibration. The
aqueous and organic phases were separated, and each phase was centrifuged (Damon, IEC)at 3000 rpm for at least 5 min. Aqueous-phase pH measurements were performed with an Orion Model 601A pH meter equipped with an Orion Ross pH electrode. Aqueous-phase acid concentrations were determined by colorimetric titration with aqueous NaOH solution, using phenolphthalein as an indicator. In most cases, organic-phase acid concentrations were determined by two-phase titration with aqueous NaOH, using phenolphthalein as an indicator. However, potentiometric organic-phase titrations were used in cases where the indicator change was not visible (methylene chloride or 1octanol diluent). Potentiometric organic-phase titrations were performed by dissolving the organic phase in methylene chloride and titrating with KOH (Mallinkrodt) in 1-propanol (Baker, reagent grade). For both aqueous- and organic-phase measurements, titrant concentration and sample volume were varied in order to give approximately the same relative precision for low-concentration measurements as for high-concentration measurements. Material balances closed within 590, and generally within 2%. Tabulations of the experimental results are given by Tamada and King (1989). Numerical Treatment of Data A computer program was developed to determine parameters in the law of mass action model so as to minimize the error between the experimental data points and points calculated from the model (eqs 2-4). Various sets of stoichiometries of the complex(es) formed were postulated, and the minimization routine was performed for each set of stoichiometries to determine the corresponding best-fit equilibrium constants and the error. The minimization parameter used was the sum of the absolute values of (CA,org,pred - CA,og,expS/ CA,org,expt, where the subscript %red" denotes the value predicted by the mass action law equations and the subscript "expt" denotes the experimentally determined value. This parameter was used because the relative (percentage) error in the organic-phase acid concentration was approximately constant for the experimental measurements within each data set. Determination of [A]. The experimentally determined pH and the pK, values of the acids (Dean, 1985) were used to determine [A], the concentration of undissociated acid in the aqueous phase. The experimentally determined pH values were up to one pH unit higher than the values that would be expected based on the aqueous acid concentration alone, with the deviation increasing with increasing total amine concentration and decreasing aqueous acid concentration. This type of behavior is indicative of a small amount of a basic impurity, such as a low molecular weight amine, being soluble in the aqueous phase. This is consistent with the product literature (Henkel Corp., 1984),which reports small amounts ( 2 4 % ) of impurities in Alamine 336. The amount of these impurities is relatively small, as evidenced by the fact that loading curves do develop plateaus at 2 = 1. A chromatographic analysis of Alamine 336 is reported by Ricker (1978). Correction for Extraction by Diluent. Physical extraction by the diluent alone was taken into account by subtracting a correction factor from the experimentally determined value of CA,org. The correction factor at each value of [A] was taken as the concentration of acid extracted by the diluent alone (obtained by separate experiments) multiplied by the volume percent diluent in the solvent mixture. A summary of experimental data for physical extraction for the acid-solvent systems used in this study is given in Table I. The amount of acid ex-
Ind. Eng. Chem. Res., Vol. 29, No. 7, 1990 1321 2 5
2 5 Oiluent -
/
Dlluent
t 2-Ethyl-1-Hexanol (Rlcker. 1978) 2
-
-+- Nltrobenzene + Chloroform (Spala. 1980)
Chloroform -+-MlBK '-m- Xylene (Sato et al
.
-A-
1985)
p 1 5 -
eP
._
0
0
1
-
2 0
1
1 0
-
05
-
-3
8
J
1 0 -
0 5
-2
-1
0 0
0
-5
Figure 1. Extraction of acetic acid by Alamine 336 in 2-ethyl-lhexanol (Ricker, 19781, Alamine 336 in nitrobenzene (present work), trioctylamine in chloroform (Spala, 1980), and Alamine 336 in 15% (v/v) chloroform in n-heptane (present work).
0
-1
1
log [AI (mol/L)
Figure 2. Extraction of lactic acid by Alamine 336 in chloroform, Alamine 336 in methyl isobutyl ketone (present work), and trilaurylamine in xylene (Sato et al., 1985).
tracted by most diluents is negligible compared to the amount extracted by amine-diluent systems, although alcohols and ketones show significant extraction in some cases.
Results Experimental results and model predictions are presented as loading curves. The data points for different amine concentrations are not differentiated unless amine concentration affected the loading curve. More detailed plots of the data are given by Tamada and King (1989). The solid curves show the calculated results of the "best" model, i.e., the simplest model that gave low error between the experimental and predicted results. If the error was not reduced significantly by inclusion of a particular complex, that complex was not considered to be part of the best model. An additional consideration was that the model should represent physically reasonable complex formation. Preference was given to complexes whose existence has been inferred from spectroscopic experiments (see part 2 of this work). Values of the equilibrium constants used in the model are shown in Table 11. Further details of this procedure are given by Tamada and King (1989). Systems Showing ( p,1) Stoichiometry. Figure 1 shows the results for the extraction of acetic acid, a monocarboxylic acid, by 0.29 mol/L Alamine 336 in nitrobenzene and by 0.097,0.29, and 0.58 mol/L Alamine 336 in a mixed diluent, in which chloroform comprised 15% of the total diluent volume and n-heptane made up the remaining volume. Also shown are data from previous workers for the extraction of acetic acid by 0.02 and 0.10 mol/L Alamine 336 in 2-ethyl-1-hexanol diluent (Ricker, 1978; Ricker et al., 1979) and 0.68 mol/L trioctylamine in chloroform (Spala, 1980). In all of the diluents, loading is independent of initial amine concentration, confirming (p,l) stoichiometries, and overloading is apparent. Extraction in the presence of nitrobenzene, a polar solvent, and chloroform and 2-ethyl-1-hexanol, proton-donating solvents, is greater than in the presence of the relatively noninteracting heptane-chloroform mixture. Lactic acid, a hydroxymonocarboxylic acid, shows behavior similar to that of acetic acid. Loading is independent of CB,tot,and overloading of the amine is evident. Figure 2 shows the results for extraction of lactic acid by 0.29 mol/L Alamine 336 in chloroform and in MIBK. Also shown are data from Sato et al. (1985) for extraction of lactic acid by 0.1, 0.2, 0.5, and 1.0 mol/L trilaurylamine in xylene. The data indicate formation of both (2,l) and (1,l) lactic acid-amine complexes in chloroform. For lactic
-2
-3
-4
log [AI (mol/L)
2 0
1 5
.-P
= 8
l o
_1
05
_ _ -8
no
-
-7
-6
-5
-4
-3
-2
-1
0
1
log [AI (mol/L)
Figure 3. Extraction of maleic, malonic, and succinic acids by Alamine 336 in methyl isobutyl ketone.
" "
-8
-6
-4
-2
0
log [AI (mol/L)
Figure 4. Extraction of maleic and succinic acids by Alamine 336 in chloroform.
acid in MIBK and in xylene, as for acetic acid in chloroform-heptane, a set of complexes with one, two, and three acids per amine must be hypothesized to account for the full extent of overloading. Similar (p,l) stoichiometries were found by previous workers for the extraction of acetic acid by tridecylamine in carbon tetrachloride and benzene (Chaikhorskii et al., 1966),trilaurylamine in o-xylene (Hogfeldt and Fredlund, 1967), Amberlite LA-2 (a secondary amine) and trioctylamine in hexane and benzene (Kawano et al., 1983a,b),and Adogen 283 (a secondary amine) in 2-heptanone (Ricker et al., 1979). The extraction of trichloroacetic acid by trilaurylamine in o-xylene (KuEa and Hogfeldt, 1967) and propionic acid by Amberlite LA-2 and trioctylamine in benzene and hexane (Kawano et al., 1982, 1983a) also showed (p,l) stoichiometries. The results for the extraction of dicarboxylic acids are shown in Figures 3-5. Figure 3 shows the results for the
1322 Ind. Eng. Chem. Res., Vol. 29, No. 7 . 1990 Table I. Partition and Dimerization Coefficients for Carboxylic Acids Extracted from Water into Organic Solvents at 25 O C (Kertes and King, 1986) aC*,org
OI&!
.-.A,Chloro'orm
= P[AI + &"[AI2
acid acetic
solvent nitrobenzene 15% ( v / v ) chloroform in n-heptane lactic m-cresol MIBKb chloroform malonic MIBK succinic 1-octanol MIBK methylene chloride maleic MIBK chloroform fumaric MIBK
P 0.524 0.0075 0.314 0.110 0.010' 0.166 0.264 0.192 0.026 0.354 0.042 2.P
Diluent .
....*...M W
-7
-4
-3
-2
-1
9ctaw
0.007 ?
0.014 ? -4
3
-2
log [ A I Imol/L)
,
'I
--C Nitrobenzene
4791 Octansl 53% Chloroform
+
K,"
a [A] = concentration of undissociated acid (moi/L). *MIBK. methyl isobutyl ketone. CHolten (1971). "Starr (1988).
--eMethylene Chloride
+
0
log IAl (mOl/LJ
Figure 5. Extraction of succinic acid by Alamine 336 in nitrobenzene, methylene chloride, and methyl isobutyl ketone.
extraction of succinic, malonic, and maleic acids by 0.05-0.58 mol/L Alamine 336 in MIBK. Figure 4 shows the results for the extraction of succinic and maleic acids by 0.05-0.58 mol/L Alamine 336 in chloroform. Figure 5 shows the results for the extraction of succinic acid by 0.10-0.58 mol/L Alamine 336 in methylene chloride and nitrobenzene. The MIBK data from Figure 3 are repeated in Figure 5 for comparison. For succinic acid, the chloroform, methylene chloride, and nitrobenzene systems show higher values of Kll than does the MIBK system, indicating that these solvents have greater abilities to solvate the complex than does MIBK. Formation of strong (1,l) complexes and minor, if any, (2,l) complexes is consistent with the chloroform and methylene chloride data, whereas significant (2,l) as well as (1,l) complex formation is required to match the MIBK and nitrobenzene data. Overloading is most apparent for MIBK. The magnitudes of the (1,l)equilibrium constants are comparable to those for lactic acid in the same diluents. But the (2,l) complex formation is quite different, with succinic acid showing virtually no formation of a (2,l) complex in chloroform but showing greater formation of a (2,l) complex in MIBK when compared with lactic acid. Malonic and maleic acids extracted by Alamine 336 in MIBK diluent behave similarly to succinic acid in the MIBK system; formation of strong (2,l) and (1,l) complexes is necessary to fit the data. The comparison of maleic (pK,, = 1.911, malonic (pKa1= 2-85),and succinic (pK,, = 4.21) acids shows that K,, increases with increasing ionizing acidity (lower pK,), but KS1is much less strongly affected. Thus, a plateau in the loading curve appears at 2 = 1 as solute pK,, decreases.
Figure 6. Extraction of succinic acid by 0.29 mol/L Alamine 336 in chloroform, 47% (v/v) 1-octanol and 53% (v/v) chloroform, and 1-octanol.
In chloroform diluent, maleic acid behaves similarly to succinic acid, except that the overloading region is more apparent for maleic acid because its greater solubility in water allows for higher experimental values of [A]. For both maleic and succinic acids, overloading is much less pronounced in chloroform than in MIBK diluent. Previous workers (Manenok et al., 1979; Vieux et al., 1974; Vieux and Rutagengua, 1977) concluded that dicarboxylic acids form only (1,l)complexes with tertiary amines. The results from the current work indicate that dicarboxylic acids are capable of overloading the amine at sufficiently high acid concentrations, but the ratio of stepwise equilibrium constants for (1,l) to those for (2,l) complex formation is small. For all of these systems, the loading is independent of the initial amine concentration, indicating that any complexes formed have only a single amine molecule per complex. The dibasic character of the acids thus does not seem to engender complexes with multiple amines in these active-diluent systems; Le., there is no significant formation of a (1,2) complex. Systems Showing ( p,2) Stoichiometries. In contrast to the (p,l) stoichiometries observed for succinic acidAlamine 336 complexation in the solvents reported above, multiamine complexes are inferred from the results using 1-octanol as the diluent. Experiments for the extraction of succinic acid in 0.15 and 0.29 mol/L Alamine 336 in octanol (Tamada and King, 1989) indicate that loading increases with increasing amine concentration. In Figure 6, the results for the extraction of succinic acid by 0.29 mol/L Alamine 336 in diluents composed of 1-octanol, a mixture of 47% (v/v) chloroform and 53% 1-octanol, and chloroform are compared. The loading curve for the octanol diluent has an unusual shape, being elevated but flat at low loadings and showing no obvious plateau a t a loading of unity. The loading curve for the mixed chloroform-octanol diluent system lies between the chloroform system and the octanol system. (1,2) and/or (2,2) complex formation is required to fit the results for the octanol and octanol-chloroform diluents. Puttemans et al. (1985) investigated mixed diluents of pentanol and chloroform for the extraction of some carboxylic acid dyes. They concluded that the alcohol modified the chloroform to produce increased extraction by combining the solvating powers of each diluent. No similar effect was found in this work. Figure 7 shows the results for the extraction of fumaric acid by 0.05-0.29 mol/L Alamine 336 in chloroform. In chloroform, evidence of (1,2) and (2,2) complex formation is seen through the increase in loading with increasing amine concentration and the convergence of the curves as
Ind. Eng. Chem. Res., Vol. 29, No. 7, 1990 1323 1 .o
-
0 - 0 DUI md/L A 0 146 mdlL
10
F
E
'8
P
s
8 -I
0.5
05
0.0 00
-7
-5
-6
-4
-2
-3
-1
log [AI (mol/L)
Figure 7. Extraction of fumaric acid by Alamine 336 in chloroform.
I,
-5
-4
0
1
-2
-3
-1
0
log [AI (mol/L)
Figure 9. Extraction of succinic acid by Alamine 336 in a mixed diluent of methylene chloride and n-heptane. Percentages show the volume percent methylene chloride in the diluent. The solid curve at 70% methylene chloride represents a calculated model with KI1 = 1.9. The solid curve a t 20% methylene chloride represents a calculated model with Kll = 0.3 for comparison with the dashed curves, which indicate the experimental loadings.
I P
P 3 -5
-4
-3
-2
1
0
log [AI (mol/L)
Figure 8. Extraction of fumaric acid by Alamine 336 in methyl isobutyl ketone.
loading approaches unity. Unfortunately, the limited solubility of fumaric acid in water at 25 "C prevents exploration of possible overloading tendencies. The results obtained by Starr (19881, shown in Figure 8, indicate that fumaric acid extracted by Alamine 336 in MIBK exhibits significant (2,1), as well as (1,2) and (2,2) complex formation. The results for fumaric acid are quite different than those for its cis isomer, maleic acid, and the other dicarboxylic acids. Reasons for the differences are discussed in part 2. Systems Showing Aggregation. The systems presented above are well described by simple stoichiometric complexes. This "ideal" behavior may be interpreted as a result of the ability of the diluent to solvate the complex effectively; the diluent prevents the complexes from interacting strongly with one another so that the activity coefficients of the organic-phase species do not change much with respect to composition. However, if the diluent is a poor solvating medium for the species formed, the polar complexes tend to cluster together, away from the low-polarity bulk solvent. In extreme cases, the polar acid-amine complexes may form a separate phase, a coacervate, or a precipitate. The results for the extraction of succinic acid by Alamine 336 in n-heptane (Tamada and King, 1989) show highly nonideal behavior. In this system there is formation of a viscous third phase between the bulk organic phase and the aqueous phase. Organic-phase analysis revealed that the third phase contains most of the extracted acid. The distribution of acid into the combined organic phases was low. The loading increased with increasing amine concentration throughout the entire amine concentration range, signifying aggregation. Because of the complications of third-phase formation, no modeling analysis was made of this system. Figure 9 illustrates the effect of amine concentration for the extraction of succinic acid by Alamine 336 in a mixed
-4
-3
-2
1
0
log [AI (mol/L)
Figure 10. Extraction of succinic acid by Alamine 336 in methyl isobutyl ketone for the full amine concentration range.
diluent containing heptane with various proportions of methylene chloride (volume percent). At 100% and 70% active diluent (methylene chloride), there is no discernible dependence of loading on amine concentration. But at low active diluent content, 20%, loading increases as extractant concentration increases. Experimental loading curves in 20% methylene chloride diluent are steeper (dashed curves) than what would be predicted by (1,l) complex formation (solid curve). Similar results were found for the extraction of succinic acid in a mixed diluent containing 10% (v/v) chloroform in heptane (Tamada and King, 1989). The behavior of these systems indicates substantial aggregate formation in systems with low active diluent content. Vaiiura and KuEa (1976) examined the extraction of citric acid, a tricarboxylic acid, by trilaurylamine in toluene. As for the extraction of succinic acid in inert diluents, loading increased with increasing extractant concentration. The authors quantitatively modeled the results with (1,2), (4,5), and (5,6) stoichiometries. The large stoichiometric coefficients are suggestive of aggregation of the citric acid complexes. Data of Sato et al. (1985) indicate that the extraction of succinic, tartaric, and citric acids by trilaurylamine in xylene also results in increased loading with increasing extractant concentration. The behavior of the di- and tricarboxylic acid systems in inert diluents is in contrast to results for monocarboxylic acids in inert diluents, in which the loading data for different amine concentrations lie on the same curve (see Figures 1and 2), implying that aggregation does not occur. High Amine Concentration. At high amine concentrations, systems containing active diluents showed a decrease in loading with increasing amine concentration, a reverse of the behavior in inert diluents. Figures 10 and
1324 Ind. Eng. Chem. Res., Vol. 29, No. 7 , 1990 Table 11. Summary of Mass Action Law Modeling Results for (p,l)Carboxylic Acid-Amine Systems at 25 "C" log Kp1 extractant (concn, mol/L) diluent (1J) (2J) (3J) (4,l) source (method) Acetic Acid, pK, = 4.76 -0.07 2-heptanone 2.09 0.50 Adogen 283 (0.429) 2-ethyl- 1-hexanol 1.83 -0.01 Alamine 336 (0.02-0.10) 1.25 0.65 chloroform Alamine 336 (0.68) 0.86 0.96 [-0.17] nitrobenzene Alamine 336 (0.29) -0.53' 0.25' diisobutyl ketone Alamine 336 (0.39, 0.69) 0.5 0.73 -0.09 benzene Amberlite LA-2 (?) 0.13 0.07 0.65 benzene tridecylamine (0.186-0.745) benzene [0.21 [0.21 triisooctylamine (0.10) w.31 xylene [0.11 [O. 11 triisooctylamine (0.10) [-0.11 0.11 -0.13 -0.23 15% CHCI, in n-heptane Alamine 336 (0.1-0.58) 0.1 0.25' carbon tetrachloride tridecylamine (0.372-1.49) o-xylene -0.6 0.44 -0.57 trilaurylamine (0.08-0.32) -1.30 -0.2 0.44 hexane Amberlite LA-2 (0.44-0.510) -0.68' -3.0' n-heptane trilaurylamine (0.08-0.32) -0.85 -0.35 -1.6 hexane trioctylamine (0.05-0.20) Trichloroacetic Acid, pK, = 0.52 6.93@ 2.51 @
trilaurylamine (0.0154-0.161)
o-xylene
Amberlite LA-2 (0.054-0.20) Amberlite LA-2 (0.044-0.21) trioctylamine (0.05-0.20)
benzene hexane hexane
Alamine 336 (0.29) Alamine 336 (0.29) Alamine 336 (0.1-1.0)
chloroform MIBK xylene
trilaurylamine (0.05) triisooctylamine (0.05) triisooctylamine (0.05)
Oxalic Acid, pKa1 = chloroform 1,2-dichloroethane o-dichlorobenzene
Alamine 336 (0.048-0.58) triisooctylamine (0.10) triisooctylamine (0.10)
Malonic Acid, pKal = MIBK o-dichlorobenzene benzene
Alamine 336 (0.29, 0.58) Alamine 336 (0.048-0.58) Alamine 336 (0.10-0.29) triisooctylamine (0.05) triisooctylamine (0.05, 0.10) Alamine 336 (0.048-0.29) triisooctylamine (0.05, 0.10) triisooctylamine (0.05) trilaurylamine (0.05)
Succinic Acid, pKal = 4.207, pKa2 = 5.635 2.54 methylene chloride chloroform 2.44 [-0.95] nitrobenzene 2.43 -0.27 chloroform 2.35 1,2-dichloroethane P.01 MIBK 1.39 0.52 o-dichlorobenzene l0.81 benzene l0.11 benzene l0.01
triisooctylamine (0.05, 0.10) triisooctylamine (0.05, 0.10) triisooctylamine (0.05)
Glutaric Acid, pKn1 = 4.34, pKa2 = 5.41 chloroform U.61 o-dichlorobenzene w.51 [0.01 benzene
Alamine 336 (0.29, 0.58) Alamine 336 (0.15, 0.29)
chloroform MIBK
Propionic Acid, pK, = 4.874 0.87 0.18 -0.13
1.48 1.14 0.10
0.70 -0.02 0.17
Lactic Acid, pK, = 3.858 2.57 1.31 -0.11
-0.45 0.21 -0.02
-0.35 -1.29
1.271, pKa2 = 4.272 [3.01 [3.01 ~ 4 1 2.826, pKa2 = 5.696 3.10 0.14 ~.41 [1.71
3 (C*) 3 (C*) 11 (C*)
12 (GI 13 (G*) 13 (G*) 3 (C*) 13 (G*) 13 (G*)
14 (G*) 14 (G*) 14 (G*)
Maleic Acid, pKn1 = 1.91, pKn2= 6.33 6.00 -0.28 5.69 0.60
Phthalic Acid, pKal = 2.95, pK,2 = 5.408 chloroform i3.71 benzene [3.21 12.61 n-heptane Kpl in molar units. pK, values from Lunge's Handbook of Chemistry (Dean, 1975). [ ] indicate large uncertainty in the value. + indicates values were calculated as j3,, rather than Kp, because fitting with stepwise model did not result in good agreement with the data. @ indicates the activity rather than the concentration used as the drying force. C indicates a computer fitting routine was used to determined the model parameters. G indicates a graphical approach was used to determine the model parameters. * indicates the fit was performed in this work; otherwise the fit was performed by following authors: 1 = Ricker (1978), Ricker et al. (1979); 2 = Spala (1980); 3 = this work; 4 Kawano et al. (1983a); 5 = Chaikhorskii et al. (1966); 6 = Vieux, (1969); 7 = Hogfeldt and Fredlund (1967); 8 = Kawano et al. (1982); 9 = Kawano et al. (1983b); 10 = KuEa and Hogfeldt (1967); 11 = Sat0 et al. (1985); 12 = Manenok et al. (1979); 13 = Vieux and Rutagengua (1977); 14 = Vieux et al. (1974). trilaurylamine (0.01) trilaurylamine (0.01) trilaurylamine (0.01)
11show data for the extraction of succinic acid by Alamine 336 in MIBK and in chloroform, respectively, over the full range of amine concentrations. At 100%Alamine 336 (1.98 mol/L) a third phase was formed, *nd experimental
loading is shown as the total moles of acid in both organic phases divided by the total moles of amine. For amine concentrations less than 0.29 mol/L, increasing the amine concentration has little effect upon loading, but as the
Ind. Eng. Chem. Res., Vol. 29, NO. 7, 1990 1325 Table 111. Summary of Mass Action Law Modeling Rssults for Systems Not Showing (p,l) Stoichiometrp
16
extractant (concn, mol/L)
7F
10
3 05
00 -4
-3
-2
1
0
log [AI (mol/L)
Figure 11. Extraction of succinic acid by Alamine 336 in chloroform for the full amine concentration range.
amine concentration is increased further, loading decreases. The amount of acid extracted from a solution of fixed initial aqueous acid concentration is maximized at an intermediate amine concentration and decreases at high amine concentrations. Similar results were obtained for the extraction of acetic acid by Alamine 336 in 2-ethyl-1-hexanol, Adogen 383 in 2-heptanone (Ricker, 1978), and trioctylamine in chloroform (Spala, 1980), as well as for the extraction of malonic acid by Alamine 337 in MIBK (Tamada and King, 1989). At high amine concentrations, loading decreased with increased amine concentration; the amount of acid extracted also decreased at high amine concentrations. When two organic phases were formed, it was in most cases not possible to separate them so as to enable analyses and volume measurements of the two phases individually. Such an analysis was made, however, for the case of extraction of malonic acid by Alamine 336 with no diluent. I t indicated that the middle phase contained from 1.4 to 2.4 mol/L acid, about 2 mol/L water, and about 1.8 mol/L amine, while the top phase contained from 0.001 to 0.004 mol/L acid, 0.02 to 0.06 mol/L water, and approximately 2 mol/L amine (Tamada and King, 1989). In Figure 10, extraction of succinic acid by Alamine 336 in MIBK is modeled by (2,1,1) and (1,1,2) complexes. In Figure 11, extraction with chloroform diluent is modeled by a (1,1,2) complex. Inclusion of the diluent specifically in the complex qualitatively describes the decrease in loading with increased amine concentration. However, at the higher amine concentrations (1.5 and 2.0 mol/L), the slopes of the experimental loading curves show an abrupt rise, indicating aggregate formation, which has not been accounted for by the simple model presented. The simple acid-amine-diluent complexes shown overpredict the effect of increasing amine concentration at low amine concentrations. The model also does not account for acid-base complex solvated by the amine itself or by nonstoichiometric assemblages of diluent molecules, which may be the dominant effect. Discussion a n d Conclusions A summary of the results for systems studied in this work and by previous workers that follow (p,l) stoichiometries is given in Table 11. The equilibrium constants are presented as the logarithm (base 10) of the stepwise formation constants in molar units. The source of data and the method (C, computer minimization program; G, graphical analysis) used to calculate the equilibrium constants are given in the last column. An asterisk indicates that determination of the complex stoichiometries and equilibrium constants was done in this work; otherwise, the values were taken from the published work. Other error-reduction functions were used for computer fitting
(P,d or (P,(7,4 diluent
log
a,
or 1% a,,
Succinic Acid, pK,, = 4.21, pKa2= 5.64 Alamine 336 (0.15, 0.29) 1-octanol (12) (22) [3.50] [4.91] Alamine 336 (0.29) 40% 1-octanol in (1,2) (1,l) (2,l) chloroform 3.28 2.27 2.42 Alamine 336 (0.05-2.0) chloroform (1,1,2) 0.54 ( 2 J A (1,1,2) Alamine 336 (0.05-2.0) MIBK 1.34 -0.35 Fumaric Acid, pK,, = 3.1, pK,, = 4.6 Alamine 336 (0.05-0.29) chloroform (12 5.54 Alamine 336 (0.02-0.29) MIBK (12) 4.33
(2.2) 7.74 (22) 1.49
(2,U 4.81
a&,, in molar units. Brackets [ ] indicate a large uncertainty in the value of 8.
by previous authors. Table I11 reports overall equilibrium constants and stoichiometries used to fit the curves shown in Figures 7-11, where (p,l) stoichiometry was not followed. A comprehensive review and analysis of carboxylic acid-amine extraction systems is given by Tamada and King (1989). The findings are summarized below. Stoichiometry. The formation of complexes with more than one acid per amine is common behavior, especially for monocarboxylicacids. The ratio of (1,l)to (2,l) complex formation is diluent dependent. Halogenated hydrocarbons and alcohols inhibit overloading; the ketone enhances overloading. In active diluents, most of the systems showed no detectable formation of (1,2) acid-amine complexes, as might be expected for dicarboxylic acids. No evidence of (1,2) complex formation with Alamine 336 was seen for succinic acid in chloroform, nitrobenzene, MIBK, or methylene chloride, for malonic acid in MIBK, or for maleic acid in MIBK or chloroform. Formation of (1,2) complexes was manifested for succinic acid in 1-octanol and for fumaric acid in chloroform and MIBK. Nonideal Behavior. When active diluents are employed, at a given aqueous acid activity, loading decreases with increasing amine concentration at higher amine concentrations. For nonaggregating systems in inert diluents, for example, monocarboxylic acids in alkane or aromatic diluents, there was no effect of amine concentration on loading. For aggregating systems, for example, dicarboxylic acids in inert diluents, loading increases with increasing amine concentration. A few nonaggregating systems-fumaric acid in chloroform or MIBK, and succinic acid in 1-octanol-showed increasing loading with increased amine concentration, apparently because two amines were bound to each complex. Amine complexes of di- and tricarboxylic acids show a greater tendency to aggregate than do those of monocarboxylic acids. Aggregation is more pronounced in solvents with a low concentration of active diluent (e.g., if there is a high concentration of amine), and in the extreme is manifested as third-phase formation. Degree of Extraction. Generally, the greater the ionizing acidity of the acid, as measured by pK,, the more it is extracted. For most of the acids studied, the strength of solvation of the complex by the diluent decreases in the order alcohol (e.g., 2-ethyl-1-hexanol) L nitrobenzene L
1326 Ind. Eng. Chem. Res., Vol. 29, No. 7, 1990
proton-donating halogenated hydrocarbon (e.g., methylene chloride, chloroform, 1,2-dichloroethane)> ketone (e.g., MIBK, diisobutyl ketone (DIBK), 2-heptanone) > halogenated aromatic (e.g., dichlorobenzene, chlorobenzene) > benzene > alkyl aromatic (e.g., toluene, xylene) > aliphatic hydrocarbon (e.g., hexane, heptane, octane). Analysis of the chemical interactions that lead to these behaviors is the subject of part 2 of this work.
Acknowledgment We thank Vista Soroush, Jane Tong, James McKinley, and Pamela Leong, who all made significant contributions to the experimental portion of this work. This work was supported by a National Science Foundation Graduate Fellowship and by the Assistant Secretary for Conservation and Renewable Energy, Office of Energy Systems Research, Energy Conversion and Utilization Technologies (ECUT) Division, US. Department of Energy, under Contract DE-AC03-76SF00098. Registry No. Acetic acid, 64-19-7; lactic acid, 50-21-5; malonic acid, 141-82-2; succinic acid, 110-15-6; maleic acid, 110-16-7; fumaric acid, 110-17-8.
Literature Cited Chaikhorskii, A. A.; Nikol’skii, B. P.; Mikhailov, B. A. Complex Formation in Nonaqueous Solutions X. Interaction of Tridecylamine with Acetic Acid. Sou. Radiochem. (Engl. Transl.) 1966, 8, 152-158; Radiokhimiya 1966,8, 163-171. Connors, K. A. Binding Constants. In Binding Constants. A Measurement of Molecular Complex Stability; John Wiley & Sons: New York, 1987; Chapter 2, pp 21-102. Dean, J. A., Ed. Lange’s Handbook of Chemistry, 13th ed.; McGraw-Hill: New York, 1985; pp 5-62-5-67. Henkel Corp. Alamine 336. Red Line Technical Bulletin; Henkel Technical Center: Minneapolis, MN, 1984. Hogfeldt, E.; Fredlund, F. Aggregation of Long-chain Amine Salts Studied by Two-Phase EMF Titrations. Some Results for the Extraction of Acetic Acid by Trilaurylamine Dissolved in Heptane and o-Xylene. Solvent Extraction Chemistry, Proc. Int. S o h . Ext. Conf. ’66, Gothenburg, Sweden; North Holland Publishing: Amsterdam, 1967. Holten, C. H. Physical Properties. In Lactic AcidlProperties and Chemistry of Lactic Acid and Derivatives; Verlag Chemie, Copenhagen, 1971; Chapter 4, pp 20-58. Kawano, Y.; Kusano, K.; Takahashi, T.; Kondo, K.; Nakashio, F. Extraction Equilibria of Lower Carboxylic Acids with Long-chain Alkylamine. Kagaku Kogaku Ronbunshu 1982,8 (4), 404-409. Kawano, Y.; Kusano, K.; Kusano, K.; Inoue, K.; Nakashio, F. Extraction Equilibria of Aqueous Solution of Lower Carboxylic Acid (sic) with Benzene Solution of Long-chain Alkylamine, Amberlite
LA-2. Kagaku Kogaku Ronbunshu 1983a, 9 (4), 473-475. Kawano, K.; Kazushito, K.; Nakashio, F. Extraction and Interfacial Adsorption Equilibria of Aqueous Acetic Acid and Propionic Acid with Tri-n-Octylamine in Hexane. Kagaku Kogaku Ronbunshu 198313, 9 (2), 211-213. Kertes, A. S.; King, C. J. Extraction Chemistry of Fermentation Product Carboxylic Acids. Biotechnol. Bioeng. 1986,28,269-282. KuEa, L.; Hogfeldt, E. On Extraction with Long Chain Tertiary Amines VIII. Extraction of Trichloroacetic Acid and Water by Trilaurylamine Dissolved in o-Xylene. Acta Chem. Scand. 1967, 21, 1017-1027. Manenok, G. S.; Korobanova, V. I.; Yudina, T. N.; Soldatov, V. S. Influence of the Nature of the Solvent on Extraction of Certain Mono- and Dicarboxylic Acids by Amines. Russ. J . Appl. Chem. (Engl. Transl.) 1979, 52, 156-160 Zh. Prikl. Khim. 1979, 52, 173-183. Puttemans, M.; Dryon, L.; Massart, D. L. Extraction of Organic Acids by Ion-Pair Formation with Tri-n-Octylamine, Part 4. Influence of Organic Phase Composition. Anal. Chim. Acta 1985, 178, 189-195. Ricker, N. L. Recovery of Carboxylic Acids and Related Organic Chemicals from Wastewaters by Solvent Extraction. Ph.D. Dissertation, Department of Chemical Engineering, University of California, Berkeley, 1978. Ricker, N. L.; Michaels, J. N.; King, C. J. Solvent Properties of Organic Bases for Extraction of Acetic Acid from Water. J . Sep. Proc. Technol. 1979, I (l),36-41. Sato, T.; Watanabe, H.; Nakamura, H. Extraction of Lactic, Tartaric, Succinic, and Citric Acids by Trioctylamine. Bunseki Kagaku 1985, 34, 559-563. Spala, E. E. A Thermodynamic Model for Solvating Solutions with Physical Interactions. M.S. Thesis, Department of Chemical Engineering, University of Washington, Seattle, 1980. Starr, J. N. Personal communication, University of California, Berkeley, 1988. Tamada,, J. A. Ph.D. Dissertation, Department of Chemical Engineering, University of California, Berkeley, 1989. Tamada, J. A.; King, C. J. Extraction of Carboxylic Acids by Amine Extractants. Report LBL-25571; Lawrence Berkeley Laboratory: Berkeley, CA, Jan 1989. Vafiura, P.; KuCa, L. Extraction of Citric Acid by the Toluene Solutions of Trilaurylamine. Collect. Czech. Chem. Commun. 1976, 41, 2857-2877. Vieux, A. S. No. 575-Sur l’extraction de l’acetate d’uranyle par la tri-iso-octylamine en solution dans des solvants organiques divers. (Note preliminaire). Bull. SOC.Chim. Fr. 1969, 9, 3364-3367. Vieux, A. S.; Rutagengua, N. Extraction of Glutaric and Succinic Acids by Tri-Isooctylamine. Anal. Chim. Acta 1977,91, 359-363. Vieux, A. S.; Rutagengwa, N.; Rulinda, J. B.; Balikunger, A. Extraction of Some Dicarboxylic Acids by Tri-Isooctylamine. Anal. Chim. Acta 1974,68, 415-424.
Received for review August 30, 1989 Revised manuscript received February 12, 1990 Accepted February 21, 1990
