MIXED
IONIC SOLVENT SYSTEMS
Mixed Ionic Solvent Systems.
147
111. Mechanism
of the Extraction by James C. Davis and Robert R. Grinstead T h e Dow Chemical Company, Western Division Laboratories, Walnut Creek, California 04608 (Received February 6, 1060)
Mixed ionic solvent systems, consisting of high molecular weight alkylammonium carboxylates in suitable diluents, extract inorganic salts reversibly from aqueous solutions, forming metal carboxylates and alkylammonium salts of the inorganic anion. Analysis of extraction data for MgClz and CaC12leads to the conclusion that the extracted species are monomeric in the organic phase. All species are associated with several unused extractant molecules.
Introduction A preliminary account has been given of a “mixed ionic” solvent extraction system in which both ionic components of an inorganic salt can be extracted reversibly from aqueous solution and subsequently removed by contact with water.’ Because of the considerable potential of such systems in practical applications, a further study has been made of the physical chemistry of some typical mixed ionic systems, with the goal of achieving some understanding of the nature of the extractant system and the mechanism of operation. A typical system involves an alkylammonium carboxylate dissolved in a hydrocarbon diluent. The extraction of magnesium chloride, for example, occurs according to the equation Mg2+
+ 2C1- + 2RdN+R’COO- + Mg(R’C00)Z + 2RiNCJ
(1)
where the bars indicate species in the organic phase. We have determined the distribution of magnesium and calcium chlorides between water and typical mixed ionic systems under various conditions. The data have been examined by the method of slope analysis and lead to a consistent picture of the behavior of these systems.’
Experimental Methods The data used in this study were partially reported previously.’ For the study of this system by the methods of slope analysis we have extended the data down into regions of lower concentration, using the same methods as before. Mixed ionic solvent systems were prepared in most cases simply by mixing the desired amine and acid and diluting with the diluent to the desired concentration. I n the case of quaternary systems, the quaternary ammonium and the acid were and then contacted with a slight excess Of dilute
sodium hydroxide to convert the acid to the corresponding anion. The organic phase was washed with water several times to remove the sodium chloride formed in this conditioning step. Samples were shaken for 1 hr and then separated in a centrifuge. Both organic and aqueous samples were subsequently analyzed. Most experiments were performed with tracer chlorine-36 using counting times adjusted to produce a =k2% maximum standard deviation in the net count. Equivalence of extraction of cation and anion was checked with magnesium chloride in one system by running duplicate experiments, determining magnesium by atomic absorption spectrometry in one case, and determining chloride with chlorine-36 tracer in the other. The distribution of the organic extractants into the aqueous phases was examined in a number of cases in order to make certain that losses into the aqueous phases were negligible. The carboxylate component was determined by examination of the carbonyl infrared absorption after extraction into carbon tetrachloride, while the amine component was determined spectrophotometrically after extraction of the picrate salt into chloroform. Data on aggregate formation were obtained with a Mechrolab osmometer Model No. 301. I n order to conform as closely as possible to the conditions of the extraction studies, water-saturated toluene was used as the reference solvent, and solutions of benzil in that reference solvent were used as standards.
Experimental Data Aggregation and Association of the Extracted Spc&es. One of the common methods for determining the nature of extracting species in liquid-liquid systems is the method of slope analysis. I n this procedure, the effect (1) R.R.Grinstead,J. C.Davis, S. Lynn, and R. K. Charlesworth, I n d . Eng. Chem. Prod. Res. Develop., 8,218 (1969). Volume 74,Number 1 January 8, 1070
JAMES C. DAVISAND ROBERT R. GRINSTEAD
148 of certain variables on the equilibrium is determined, and the desired information is obtained by suitable manipulation of the equilibrium expressions and plotting of data. For the distribution of magnesium chloride, two general situations are possible. If the magnesium carboxylate (denoted by MgAJ and the alkylammonium chloride (denoted by CCl) are associated in the organic phase, the general equation for the distribution of magnesium chloride is -+ +’ 2 -
+ 2C1- + -([CAI,) P
1
-
t
( [il/lgA2* 2CC11t ~vCA) l’ ‘ v/t 2 ([CAI,) -aMgCIa P ~
=
P
+
log ( m y ) ’ M g C l r K2’ (7) Equivalence of magnesium and chloride in the organic phase requires that q([Mg&Ig*zCA) = (MgC12) (8)
r/2([CClI,*@A)
log (MgC12) =
([RilgA2*2CCl]t.~CA) (2)
where v is the number of unreacted extractant molecules associated with the extracted species and t is the degree of aggregation of that species. The equilibrium con-’ stant for the reaction is
K1
+
1 2 - log ([IC~~A~],*ZCA) ;log ([CCl],*yCA) =
where (MgCl2) is again the analytical magnesium chloride concentration in the organic phase. Using this equation in (7) we get
V
Mg2+
Taking logarithms we get
+
(3)
34r log ( m * y ) ~ p ~fi alog K,” 2q
+r
(9)
Thus by plotting the log of the analytical concentration of magnesium chloride in the organic phase against the log of may for magnesium chloride, the slope of the r. The minimum slope plot has the value of 3pr/2q which can occur, when no aggregation occurs, Le., r = q = 1, is 1.0. I n Figure 1,the data for magnesium chloride distribution with four different solvent systems are plotted
+
If the magnesium chloride concentration in the organic phase is kept low, the CA concentration is essentially a constant and, taking the logarithms of eq 3, we get 10-1
The actual concentration of the species following the log term on the left side of (4) is simply equal to l / t times the analytical Concentration of magnesium chloride in the organic phase, represented by (MgC12). Making this substitution, and noting that a Y g C 1 2 = 4(mrI3 log (MgCl2) = 3t log ( m *Y ) M g C h
+ log Kll’
Mg2+
+ 2c1- +
( [CAI,)
+
-
_
10-0
10-4
10-6 10-8 ,
2 1 -( [M~A~],.zCA) -( [CCl],*yCA) ( 6 ) r 4 The Journal of Physical Chemistry
,
(5)
where m is the aqueous molality of magnesium chloride. Thus eq 5 is in the slope-intercept form and if (MgC12) is plotted against may on a log-log plot, the resulting line will have a slope of 3t. The minimum possible slope will be 3.0 if no aggregation occurs in the organic phase. If the extracted magnesium and chloride species are not associated with each other in the organic phase, the general equation for magnesium chloride distribution will be
A
10-2
10-2
10-1
my for aqueous MgCL.
Figure 1. Magnesium chloride distribution between water and alkylammonium 2-ethylundecanoates in toluene. Slopes of straight-line portions given on curves. Open points, chlorine-36 tracer data; solid points, atomic absorption data.
M I X E D IONIC
against the quantity rn. y for magnesium chloride. Activity coefficient values were obtained from Robinson and Stokes.2 The four systems consisted of 0.5 M 2-ethylundecanoic acid plus 0.5 M amine in toluene.
species (magnesium or chloride) into the aqueous phase to a greater extent than the other. If the difference in distribution is significant compared to the actual magnesium or chloride level in the organic phase, eq 8 does not hold. For all the svstems studied here. these differences were negligible. Another source of failure of eq 8 is through hydrolysis of one of the extracted species. For example, hydrolysis of the magnesium carboxylate would form magnesium hydroxide; hydrolysis of the alkylammonium chloride would give HC1. Formation of either of these inorganic hydrolysis products would reduce the organic phase concentration of magnesium or chloride ion, respectively, and lead to an excess of the other. pH measurements on the aqueous phases in the primary amine systems were always less than 8.3, which M hydroxide ion. This is equivalent to 2 X represents a negligible hydrolysis. KO measurements were made on the other amines, but since aliphatic amines are of comparable base strengths in aqueous solution, the extent of their hydrolysis should be comparable to the primary amine. The interesting feature of Figure 1 is that all four systems studied have the same slope of the straight-line portion of each curve, and the slope is approximately 1.0. Equation 5 , which suggests a minimum slope of 3.0, obviously does not hold for this system, and we conclude that the magnesium and chloride are present as separate species. Assuming then that eq 9 is the valid equation for the distribution of magnesium chloride, the resulting slope (1.0) must equal 3qr/ 2q r . Values of q = 1 and r = 1 are the only integral solutions to this equation. That is, both the magnesium carboxylate and the alkylammonium chloride exist as unaggregated monomers. This situation holds in the straight-line region, below an organic phase concentration of about 0.02 M magnesium chloride for the quaternary and primary amine, and below about 0.03 M for the secondary and tertiary amines. Similar results were obtained for calcium chloride extraction by the quaternary and primary ammonium systems, and the data are shown in Figure 2. The
+
149
SOLVENT SYSTEMS 1’
I
I
10-8 10-2
10-8
my for
10-1
1
aqueous MgClz.
Figure 2. Distribution of magnesium and calcium chlorides between water and alkylammonium 2-ethylundecanotates in toluene: 0.48 M Aliquat 336S:EUD: A, CaClz; U, MgC12; 0.50 M Primene JM-T:EUD: A, CaClt; H, MgCh. Slopes of straight-line portions are all
corresponding magnesium chloride data are included here €or comparison. In the low concentration region the calcium chloride curves also exhibit a slope of 1, indicating that the extraction mechanism is the same as for magnesium chloride. It is interesting to note that the distribution of these two salts in the quaternary system is roughly the same, while in the primary system, the calcium exhibits a significantly greater distribution toward the organic phase than the magnesium. Association of the Extracted species with Unused Extractant. The preceding data have all been taken a t a constant extractant concentration, and at sufficiently low loadings that the extractant concentration is essentially constant throughout a given experiment. I n order to obtain information about the participation of the extractant in the extraction mechanism, additional experiments were done in which the extractant concentration was varied. These data are shown in Figure 3 for magnesium chloride in the quaternary ammonium system. Now, if in eq 6 the organic concentration of magnesium chloride is held constant, we get, upon taking logarithms of the equilibrium expression log([CAl,) = -2y 3p -+-+2 r
log (m* 7)MgClz f log K,”’
q
(10) Corresponding values of my and extractant concentration can be obtained at any given organic magnesium (2) R. A. Robinson, and R. H. Stokes, “Electrolyte Solutions,” 2nd ed., Academic Press, Inc., London, 1969.
Volume 74, Number 1 January 8, 1070
JAMES C. DAVISAND ROBERTR. GRINSTEAD
150
Table I : Aggregation of Organic Extractants“ Aggregation no., p = nominal molality/ Nominal conon, m
0.10 0.39 0.90 0.10
0.39 0.90
observed Molarity
T,OC
molality
0.09 0.28 0.46 0.09 0.28 0.46
25
3.2 3.2 3.0 3.2 3.2 3.4
25 25
37 37 37
a Trioctylmethylammonium 2-ethylundecanoate in toluene system equilibrated with water. I
10-
,
I
,
I
1
10-3 10-1 my for aqueous MgClz.
Figure 3. Effect of extractant concentration on distribution of magnesium chloride between water and Aliquat 3368: EUD. Extractant concentrations: 0,., 0.48 M (filled points, chlorine-36 data; open points, atomic absorption data); A, 0.29 M ; 0,0.10 M . Slopes of straight-line portions of lines are 1.0.
p = 1, r = 1, and p = 3 for the extraction of magnesium
chloride, the slope of - 1.0 gives, from eq 10
Explicit values of y and x are not obtained from eq 11, but even so, these calculations suggest that the association of unreacted extractant with the extracted species is quite significant.
Discussion
0.001
0.005
0.002 my for
0.01
0.02
aqueous MgCIz.
Figure 4. Relation between extractant (CA) concentration and aqueous magnesium chloride activity for constant organic composition. Extractant is Aliquat 336s-EUD in toluene. Slopes of lines are both -1.0; 0 , (MgCle) = 5.0 X M; 1 ,(MgC12) = 3.0 X 10- M .
chloride concentration from the data in Figure 3. A log-log plot of the extractant concentration vs. the quantity my will then have slopes which contain the variables y, x , Y, p , and p as shown in eq 10. Such a plot is given in Figure 4 for two concentrations of magnesium M and 3.0 X chloride in the organic phase, 5.0 X 10-3 M . The slopes of both plots are about - 1.0. Table I contains the data obtained by vapor pressure osmometry to determine the degree of aggregation of the extractant molecules. Data are given for the quaternary amine system at two temperatures, 37” and 26’. The aggregation does not seem to vary greatly with temperature, being equivalent to about a trimer for both temperatures studied. If we use the values The Journal o j Physical Chemistry
The extraction of metal cations into organic solvent systems of low polarity normally involves replacing the water in the coordination sphere of the cation by donor groups of a hydrophobic nature, such as alcohols, amines, etc. I n the case of the magnesium and calcium cations, the stoichiometric requirement of two carboxylate ions per cation could account for up to four of the coordination positions of these cations. Further coordination may be due to additional carboxylate ions of the “unused” extractant, or possibly even to water which remains in the coordination shell. Since the carboxylates involved are branched, the steric hindrance might be sufficient t o limit the coordination to two of these species per cation. I n a previous paper,a it was shown that a t high concentrations of amine hydrochlorides in toluene, the amine salts were invariably aggregated. Yet, in the mixed solvent systems studied here, the same amine hydrochlorides are not noticeably aggregated at similar concentrations. This result is presumably due t o the presence of unused alkylammonium carboxylate which is available for association with the inorganic-organic ion pairs, n’either the cation carboxylate nor the ammonium salt can be involved in an association with the other since the mechanism indicates no association of the two inorganic extracted ions. The “solvent” (3) R. R. Grinstead and J. C. Davis, J. Phys. Chem., 7 2 , 1630 (1968).
MIXEDIONIC SOLVENT SYSTEMS then is not just a hydrocarbon but a mixture of "solvating" species within a toluene diluent. The exact nature of this interaction can only be speculated upon, but must involve a close association of the alkylammonium carboxylate with either the metal ion carboxylate or the alkylammonium chloride or both. It is not possible to determine from our data the extent of association for the individual species, but, if we assume for example, equal association with the magnesium and chloride species, then x and y in eq 11 will each be approximately 2. A value of p = 4 would increase the association numbers to about 3 each. Thus, on the average, each magnesium and chloride ion is associated with 2 or 3 extractant molecules in addition to the stoichiometric requirement. It should be pointed out here that the above arguments for association apply only to quaternary amine systems, for no data were obtained for other amines. It should also be clear that the discussion applies only to systems in which the organic phase contains only small concentrations of inorganic species. Since it is possible to reach rather high organic loading levels of inorganic salts, the amount of unused extractant available for association cannot in these cases conform to the figures given above. Similarly, although association of metal ion and halide ion species does not exist at low organic loadings, it may at higher levels corresponding to the curved portions of the log-log plots. One of the most important features of the distribution data is the wide variation in extraction among the amine types. There are two possible explanations for this behavior, both of which may be valid. In the first place, recalling the basic reaction for extraction by the mixed ionic solvent (eq 1) we see that for a particular aqueous concentration, the position of equilibrium will be influenced by the relative free energies of the two amine salts, the carboxylate and the chloride. If we replace a given alkylammonium component by another for which the chloride salt is relatively more stable with respect to the carboxylate, the equilibrium will shift to the right; Le., the extraction will be more complete. Barrow and Yerger4-6have shown that the extent of reaction of an amine and acetic acid to form the salt is relatively independent of the type of amine. I n comparison, amine hydrochloride formation is very dependent on the amine type.3 Thus the amine itself might be expected to have a considerable effect on the overall distribution by shifting the equilibrium to the
151 right through a relative increase in the hydrochloride stability as compared to the carboxylate stability. Admittedly, the cause of the great variation in amine hydrochloride stability in toluene is due to aggregation while in the mixed ionic systems the hydrochlorides are monomers. However, the ease of aggregation in toluene is also a reflection of the need for solvation, and those amine hydrochlorides which were found to aggregate most readily in toluene can be expected to most readily accept and be stabilized by the unused extractant in the mixed ionic solvent systems. The unreacted solvent then could stabilize some amine hydrochlorides more than others, essentially by stabilizing the least sterically hindered in preference to the more sterically hindered amines. The fact that the stability of amine hydrochlorides in toluene is in the order 1" > 2" > 3" and that the order of ease of salt extraction in the mixed solvent is the same suggests that such an argument is not unreasonable. The quaternary ammonium compound provides a special case. The argument involving completeness of reaction with acids is inapplicable, since quaternary salts are completely ionized; i e . , salt formation is complete. A more fruitful explanation is to be found in the fact that quaternary ammonium salts have considerably larger dipole moments than any of the other three types.' This is a consequence of the larger separation of charge which must exist, and also of the fact that no dissipation of the dipole can occur through hydrogen bonding of the ammonium ion to the anion. The quaternary salt is thus in greater need of solvation, which it can obtain by extraction of a metal salt. The quaternary ammonium chloride which is formed may not provide a great deal of reduction of the dipole moment. However, it should be sterically easier for the unused extractant to combine with these new species than to combine with itself; Le., formation of the inorganic salts allows for greater stabilization by association with the unused extractant.
Acknowledgment. This research was supported by the Office of Saline Water, U. S. Department of the Interior, undei Contract No. 14-01-0001-1134. (4) G. M. Barrow and E. A. Yerger, J . Amer. Chem. SOC.,76, 5211 (1954). (5) E. A. Yerger and G. M. Barrow, ibid.,77,4474 (1955). (6) E.A.Yerger and G. M.Barrow, ibid., 77,6206 (1955). (7) K.Bauge and J. W. Smith, J. Chem. Soc., A , 616 (1966).
Volume 74,Number 1
January 8, 1970
