846
J . Phys. Chem. 1989,93, 846-854
later species, so they are forced to eventually dehydrogenate.
of cyclopentene at the expense of dehydrogenation. This is interpreted in terms of the blocking of Pt sites required for hydrogen V. Conclusions abstraction by the inert Bi atoms. In the low Bi coverage range Cyclopentene adsorbs molecularly on Pt( 1 1 1) at 110 K. Apwhere dehydrogenation is most strongly poisoned, the influence proximately four Pt surface atoms are required for each chemiof Bi upon the activation energies for desorption and fragment sorbed cyclopentene molecule. These dehydrogenate to c - C ~ H ~ , ~ dehydrogenation are small. Cyclopentene appears to be very by -300 K. At high coverage, cyclopentene also desorbs momobile on the surface at its dehydrogenation temperature (250-300 lecularly, due to the absence of free Pt sites necessary for hydrogen K). abstraction. The activation energies for desorption and dehydrogenation of cyclopentene are -18.3 and 13.3 kcal/mol, reAcknowledgment. We acknowledge the donors of the Petro~,~ further by -570 K to spectively. The c - C ~ Hdehydrogenates leum Research Fund, administered by the American Chemical leave a surface of overall stoichiometry H:C = 2:5, which is Society, for support of this research. interpreted in terms of a mixture of carbon adatoms and C2Ha. Registry No. Pt,7440-06-4; Bi, 7440-69-9;cyclopentene, 142-29-0. Coadsorption with Bi strongly enhances the molecular desorption
Microemulsions as Model Systems To Study the Kinetics and Mechanism of Reactions Occurring in the Extraction of Metal Ions by Lipophilic Extractants: Complexation of Nickel( I I ) by 8-Hydroxyquinoline and Kelex 100 C. Tondre* and M. Boumezioud Laboratoire d’Etude des Solutions Organiques et Colloidales (LESOC), U.A. CNRS No. 406, Universitd de Nancy I , B.P. 239, 54506 Vandoeuvre-l?s-Nancy Cedex, France (Received: February I , 1988: In Final Form: July 14, 1988)
Extractant molecules bearing an alkyl chain substituent have large applications in the field of liquid-liquid extraction of metal species. We have taken advantage of the transparency of microemulsion systems to study the kinetics and mechanisms of reactions that usually take place in highly heterogeneous conditions. The stopped-flow technique has been used here to study the kinetics of complexation of NiZ+with either 8-hydroxyquinoline (HQ) or 7-(4-ethyl-l-methyloctyI)-8-hydroxyquinoline (C,,-HQ) (the latter being the active species of the industrial extractant Kelex loo), in different micellar and microemulsion systems. Equilibrium measurements, using Job’s continuous-variation method, have proven that the stoichiometries of the complexes formed in microemulsions are very different for the two extractants. The kinetics of complexation is considerably slowed by the presence of the alkyl chain substituent: The observed rate constant is decreased by a factor of about 30. A kinetic mechanism based on Berezin’s pseudophases model is proposed, which accounts for the fact that the reaction appears to be essentially a “bulk reaction” with HQ and an “interfacial reaction” with CI1-HQ. The kinetic results reveal the presence of a reaction intermediate in the latter case.
Introduction The extraction of metal species is usually a heterogeneous process taking place at an interface between an organic phase containing a lipophilic extractant and an aqueous phase solubilizing the metal to be extracted.’S2 The hydrophobic character of the extractant molecules ensures a minimum loss of these molecules by transfer in the aqueous phase. A classical extraction process can usually be described as a succession of steps including diffusion processes, adsorption-desorption processes at/from the interface, and complexation reactions. The proper complexation step is often thought to be the rate-limiting step provided that favorable hydrodynamic conditions are set up. It is also now well-known that the rate of extraction can be considerably improved in the presence of tensioactive agent^,^-^
but the mechanisms involved and the location of the different steps are far from being completely elucidated. The rate constants characterizing the reaction of complex formation can in principle be obtained from liquid-liquid extraction experiments in conditions where a “kinetic regime” is established (plateau region in the rate of extraction versus stirring speed), but the values obtained may always be altered by other contributions.2 Microemulsion systems offer a very convenient way to investigate the kinetics of processes taking place at an oil/water interface. Although these systems are microheterogeneous in nature, they permit the use of conventional techniques for homogeneous kinetics. A great deal of work have been done on the use of micellar solutions or microemulsions as reaction media, including studies of their effect on ligand-metal complexation kinetics.*-14 The
(1) Ritcey, G.M.; Ashbrook, A. W. Solvent Extraction Principles and Applications to Process Metallurgy, Part I; Elsevier: Amsterdam 1984. (2)Danesi, P.; Chiarizia, R.CRC Crir. Rev. Anal. Chem. 1980,10, 1-126. (3)Bauer, D.; Komornicki, J. Proc. In?. Solvent Extr. Conf, 1983, 315. (4) Fourre, P.;Bauer, D. C.R . Seances Acad. Sci. Paris, Ser. 2 1981, 292,
1 1986,82, 1515.
1077. (5) Wu,C.-K.; et al. Sci. Sin. (Engl. Ed.) 1980,23(12); Proc. Int. Soluent Exrr. Con$ 1980 Dispersion 80-23.( 6 ) &sea-Asare, K.; Keeney, M. E. Sep. Sci. Technol. 1980, 15, 999. (7)Tondre, C.; Xenakis, A. Faraday Discuss. Chem. SOC.1984, 77, 115.
0022-3654/89/2093-0846$01 SO10
(8) Miyake, Y.; Shigeto, M.; Teramoto, M. J . Chem. Soc.,Faraday Trans.
(9)Fletcher, P. D.I.; Robinson, B. H. J . Chem. Soc., Faraday Trans. 1
1983. -.- -, 79. - , 1959. -.- ..
(10) Fletcher, P. D. I.; Robinson, B. H. J . Chem. SOC.,Faraday Trans. 2417. (11) Mackay, R. A. Adu. Colloid Interface Sci. 1981, 15, 131. (12)Letts, K.; Mackay, R. A. Inorg. Chern. 1975, 14, 2990; 1975, 14, 2993. 1 1984, 80,
0 1989 American Chemical Society
Microemulsions in Extraction Kinetics
The Journal of Physical Chemistry, Vol. 93, No. 2, 1989 847
originality of the present work lies in the use of microemulsions as model systems to probe the kinetics and mechanism of liquid-liquid extraction of metal ions. In a previous work, we used the stopped-flow technique to investigate the effect of an alkyl chain substituent on the kinetics and thermodynamics of complexation of 8-hydroxyquinolines with Ni2+ and Co2+in methanolic solution^.'^ The alkylated form of hydroxyquinoline (HQ) was the 7-(4-ethyl- 1-methyloctyl)-8hydroxyquinoline (which will be abbreviated CII-HQ in the following), which constitutes the active species of the industrial extractant Kelex In the present paper, we will compare the complexation of Ni2+ with the two preceding extractants in micellar solutions and in microemulsions. A few equilibrium data obtained in these media will be reported. The kinetic results will be shown to be consistent with a general mechanistic scheme taking into account the existence of pseudophases. Experimental Part Chemicals. 8-Hydroxyquinoline (puriss.) was obtained from Fluka and used without further purification. Kelex 100 was from Shering (F.R.G.). It was thoroughly purified, by a previously described chromatographic method,I5 to get the active component 7-(4-ethyl- 1-methyloctyl)-8-hydroxyquinoline. Recrystallized 99% sodium dodecyl sulfate (SDS) was obtained from Roth (F.R.G.); 1-pentanol (puriss.) and dodecane (purum.) were from Fluka. Triethanolamine Merck (pro analysi) was used with perchloric acid to buffer the solutions. Techniques. Equilibrium measurements were performed with a Varian DMS-100 UV-visible spectrometer on-line with a DS-15 data station. pH measurements in micellar and microemulsion systems have been performed with a combined glass electrode Metrohm following the recommendations given in previous works.25 The stopped-flow technique with optical detection was used for the kinetic measurements. The apparatus was a Durrum D-1 10 on-line with a data acquisition system (Biomation 805 transient recorder interfaced with a N M 4/30 Computer-Automation set up (Yrel, France)). For all the experiments, the overall metal ion concentration was a t least 10 times larger than the ligand concentration. The microemulsions mixed were of the same composition except that one contained the metal ion and the other the ligand. All the concentrations given refer to the overall stoichiometry after rapid mixing. The observed rate constants (or reciprocal relaxation times) were obtained from a nonlinear least-squares analysis of the change of optical density recorded at 400 nm, on the absorption band of the complex. When the standard deviation was too high with a monoexponential fitting, the data were fitted with a biexponential function. The analysis performed on a sampling (13) Robinson, B. H.; Steytler, D. C.; Tack, R. D. J. Chem. Soc., Faraday Trans. 1 1979, 481. (14) Holt, S . L. Inorganic Reactions in Organized Media, ACS Symposium Series 177; American Chemical Society: Washington, DC, 1982. (15) Boumezioud, M.; Lagrange, P.; Tondre, C. Polyhedron 1988,7,513. (16) Hartlage, J. A. Trans. SOC.Min. Eng. AIME 1969. (17) Ashbrook, A. W. Coord. Chem. Rev. 1975, 16, 285. (18) Leveque, A.; Helgorsky, J. Proc. Int. Solvent Extr. Con5 1977, CIM Special Volume 21, 439. (19) Demopoulos, G.P.; Distin, P. A. Hydrometallurgy 1983, 11, 389. (20) Cote, G.;Bauer, D. Hydrometallurgy 1980, 5, 149. Bauer, D.; Fourre, P.; Lemerle, J. C. R . Seances Acad. Sci. Ser. 2 1981, 292, 1019. Fourre, P.;Bauer, D.; Lemerle, J. Anal. Chem. 1983,55, 662. Marchon, B.; Cote, G.;Bauer, D. J. Inorg. Nucl. Chem. 1979, 41, 1353. Guesnest, P.; Sabot, J. L.; Bauer, D. J. Inorg. Nucl. Chem. 1980, 42, 1459. (21) Lakshmanan, V. I.; Lawson, G.J. J. Inorg. Nucl. Chem. 1973, 35, 4285. (22) Flett, D. S.; Lox, M.; Heels, J. D. J. Inorg. Nucl. Chem. 1975, 37, 2197. Flett, D. S.;Hartlage, J. A.; Spink, D. R.; Okuhara, D. N. J. Inorg. Nucl. Chem. 1975, 37, 1967. (23) Ritcey, G. M.; Lucas, B. H. CIM Bull. 1974,64,87; 1975,68, 105. (24) Haraguchi, K.; Freiser, H. Inorg. Chem. 1983,22, 1187. Bag, S.T.; Freiser, H. Anal. Chim. Acta 1982, 135, 319. Zhu, L.; Freiser, H. Anal. Chim. Acta 1983, 146, 237. (25) Berthod, A.; Saliba, C. Analusis 1985, 13, 437; 1986, 14,44. Bahri, H.; Letellier, P. J. Chim. Phys. 1985, 82, 1009.
SDS Pentanol- 2/3
H20
Dodecan e
Figure 1. Pseudo ternary phase diagram at 25 "C giving the compositions of the microemulsions used as reaction medium. Z(S)
xX
X
x
x
X X X X
Zshort
mo
o o o o
0
0
0
1
0
% dodecane ( W / W ) (
Figure 2. Relaxation times characterizing the CII-HQ/NiZ+ reaction vs the oil content of the microemulsion: [Cll-HQ] = 4.5 X l e M, [Ni2+IT = 8.3 X lo-) M; no buffer; T = 25 "C.
of 100 points involved the optimization of three adjustable parameters in the first case (relaxation amplitude, relaxation time, and asymptote) and of five adjustable parameters in the second case. The experiments were carried out at 25 OC.
Results and Discussion The microemulsion systems that we have chosen for these experiments were made of water/dodecane/sodium dodecyl sulfate/pentanol- 1.26 The compositions of the microemulsions used as reaction medium are indicated in the pseudo ternary phase diagram represented i n Figure 1. This diagram, in which the ratio between 1-pentanol and SDS has been kept equal to 2, has been the subject of numerous investigations. The letters D, B, and I refer, respectively, to direct, intermediary, and inverse microemulsions. Kinetic Measurements in Unbuffered Media. A first series of kinetic measurements have been performed with no buffer present in the microemulsions, because we wanted to avoid as much as (26) Tondre, C.; Zana, R. J. Dispersion Sci. Technol. 1980, I , 179.
848
The Journal of Physical Chemistry, Vol. 93, No. 2, 1989
I
/
P
, pH=8.5
I
2
Tondre and Boumezioud
4
6
,
,
8
Figure 3. kob vs total Ni2' concentration for HQ/Ni2' reaction in microemulsion D1 at 25 OC: [HQ] = lo4 M; TEA buffer, 0.1 M.
possible the introduction of additional substances. As the complexation of Ni2+ with hydroxyquinolines involves the departure of the hydroxylic proton, the pH of the solution is slightly more acidic when the final equilibrium is attained. The kinetic curves for complexation of Ni2+ with Cll-HQ are perfectly represented by a sum of two exponential functions. The reciprocal relaxation times, characteristic of the exponential decreases, both vary linearly with the total Ni2+concentration. We have shown in Figure 2 how these relaxation times vary at fixed ligand and metal concentrations, when the composition of the microemulsion is changed along the paths indicated in Figure 1. The shorter relaxation time is almost insensitive to the microemulsion composition, whereas the longer one is slightly decreasing with the dodecane content. Such a small effect of the microemulsion composition had previously been observed for the PADA/Niz+ r e a ~ t i o n .In ~ the following, most of the experiments will be performed in the microemulsion system represented by D, in Figure 1, with for composition 5.9% SDS, 11.8% 1-pentanol, 6.8% dodecane, and 75.5% water (w/w). Contrary to the preceding case, the reaction of Ni2+ with H Q in system D1 is strictly monoexponential, as is also true in waterz7 where Cl1-HQ is not soluble. On the other hand, the reaction in a pure micellar solution of SDS, with the same amount of SDS as in microemulsion D1, gives rise to two relaxation processes with opposite directions. Everything goes on as if the complex formation was followed by its partial destruction. This effect is less pronounced in mixed SDS/pentanol micellar solutions. A possible explanation of this observation is that the complex, after it is formed, is preferentially located close to the micellar surface where the H+ concentration can be very high. Desextraction is indeed known to occur in acidic medium.2s The decrease of the electrostatic field when going from SDS micelles to mixed SDS/ pentanol micelles and finally to microemulsions would then explain the progressive disappearance of this second relaxation process. We will see below that pure exponential signals are obtained for the HQ/NiZ+ reaction in buffered micellar solutions. Kinetic Measurements in Buffered Media. The choice of a buffer was a crucial problem because it has been previously demonstrated that many buffers can interact with micelles or microemulsions, thus introducing a ApH comparatively to pure aqueous buffersSz9 The interaction between the charged particles and the buffer components can be electrostatic or hydrophobic: A buffer component can be adsorbed at the surface or repelled, and it can also be solubilized. In addition, many buffers are susceptible to form complexes with Ni2+ions. Owing to the range
Figure 4. kobs(or ~ ~ ~ vs~ total ; l ) Ni2' concentration for CIl-HQ/Ni2+ reaction in microemulsion D, at 25 'C: [Cl,-HQ] = IO4 M; TEA buffer, 0.1 M.
t 0
0.061
0.04
pH-7.86
t-
0 X X
0
h
A
h
0.02 -
X
A
h
A
pH.8.5
X
2
X
1O3x[Ni1~(MII) 0
2
4
6
8
Figure 5. kob for the slow relaxation process (or Ts]oa-l) vs total Ni2' concentration for CIl-HQ/Ni2' reaction in microemulsion D,. Same conditions as in Figure 4.
of pH we wanted to investigate, our choice fell on triethanolamine/HC104, which was previously used by Hague and Eigen30 in an investigation of the complexation kinetics of divalent metal ions with HQ. It was expected that triethanolamine would have very weak complexing power due to steric hindrances for approaching the nitrogen atom.31 On the other hand, both forms of this buffer (TEA and TEAH') can probably interact with micellar particles. The kinetics of complexation of Ni2+ with H Q and CII-HQ has been studied in the microemulsion system D,, adjusting the macroscopic pH with the preceding buffer. A purely monoexponential absorption change was always obtained with HQ, whereas biexponential signals characterized the second extractant in the range pH 7.2-8.5. Outside this range, the amplitude of the second relaxation process vanishes and monoexponential signals are obtained. We will always compare the faster relaxation time, which accounts for the larger part of the absorption change, with
~
(27) Boumezioud, M.; Tondre, C. J . Chim. Phys. 1988, 85, 719. (28) Stary, J. Anal. Chim. Acta 1963, 28, 132. (29) Mackay, R. A.; Jacobson, K.; Tourian, J. J Colloid Znterfuce Sci. 1980, 76, 5 1 5
(30) Hague, D. N.; Eigen, M. Trans. Faraday Sor. 1966, 62, 1236. (31) Mulla, S. T.; Jose, C. I. J . Chem. Soc., Farday Trans. I 1986, 82, 681. Antelo, J. M.; Arce, F.; Rey, F.; Varela, A. Polyhedron 1987, 6, 1279.
The Journal of Physical Chemistry, Vol. 93, No. 2, 1989 a49
Microemulsions in Extraction Kinetics
Figure 6. Light transmission (7') vs time during the Cl1-HQ/NiZ+reaction in microemulsion D,;effect of adding HC1 to the solution: X = 400 nm: T = 25 OC. TABLE I: Apparent Rate Constants for Extractants/Ni2+Reactions in Different Media at 25 OC HQ/Ni2+ Cll-HQ/Ni2+
reaction medium microemulsion D,: 5.9% SDS; 11.8% C,OH; 6.8% 75.5% H20
water
methanol
pH 9.2 8.5 7.9 1.4 7.2 8.5 8.05 7.3 6.9 10.0 9.4 9.0 8.5
kppp,
kdaPP,
ktpp,
M-I s-I 4750 1660
s-I
M-I s-l 140 55 34.6
1.6 0.8
660
0.2
21750 7830 2050 1200 1558 583 391 258
10 2.5 1 .o 0.7 0.1 0.1 0.1 0.1
32.2
388 137 78 45
kdaPP, S-1
0.12 0.06 0.02 0.015
-0 0 0 0
the sole relaxation time measured with HQ. Figure 3 represents the variation of the observed rate constant versus the total Ni2+ concentration at different pHs for the HQ/Ni2+ reaction. Figure 4 shows the corresponding plots for the C11-HQ/Ni2+ reaction, whereas the variations of the observed rate constant for the slow process are indicated in Figure 5. The disappearance of this slow process, when the pH becomes lower than 7.2, is illustrated in Figure 6 where HC1 has been added drop by drop to the solutions, the kinetic curve being recorded after each addition. If the acidification is pursued further more, the complexation is no longer possible, as expected from the variation of the yield of extraction when the pH is varied.28 We will come later on to the possible significance of the slow process observed with Cll-HQ, but we will first focus our attention on the comparison of the complexation kinetics measured for the two ligands. The linear behaviors observed in Figures 3 and 4 are consistent with a classical complexation reaction of the type A + B kiC C whose reciprocal relaxation time (or observed rate constant) is given by32 1/ 7 =
kobs
= kf[Al tot
+ kd
Figure 7. Plot of k p p p vs pH for reaction HQ/Ni" in water and in microemulsion D1,respectively. The horizontal lines indicate that superposition can be obtained from a simple translation.
(2)
when [A], >> [B],. In the present case, an apparent formation rate constant kf"ppcan thus be deduced from the slope of the linear plots, the intercept giving the apparent dissociation rate constant kdaPP.
The values of these apparent rate constants are given in Table I, in which some results obtained in ordinary solution^'^*^^ are also reproduced for the sake of comparison. The value of kaPP is about 30 times larger for H Q than for Cll-HQ in the microemulsion DI,whereas a factor of only about 4-5 was obtained in methanol.Is One can also notice that the values of kfaPP for Cll-HQ have the same order of magnitude in microemulsion D1and in methanol, (32) Bernasconi, C. F.Relaxation Kinetics; Academic: New York, 1976.
1
0
1
10 20 30 Figure 8. Plot of k p p p vs the oil content of the microemulsion for HQ/NiZ+(X) and ClI-HQ/Ni2+ (0),respectively: [extractant] = lo4 M; [Ni2+]= 3.2 X 10-3 M;pH 9.4.
while for H Q the values in the microemulsion is intermediate between those obtained in water and in methanol, respectively. This is a first indication that the reaction takes place in a different environment for the two ligands. We have compared in Figure 7 the pH dependence, in microemulsion D1and in water, respectively, for the HQ/Ni2+ reaction, according to the values of kfpp given in Table I. It is noteworthy that the two curves can be deduced from one another by a simple translation of f 1.3 pH units along the x-axis. The effect of changing the microemulsion composition is shown in Figure 8. We have plotted in this figure the variation of the observed rate constant at fixed concentrations of metal ion and extractants, when the microemulsion composition is changed following paths D and B in Figure 1. For both extractants kobs begins to decrease when the amount of dodecane in the microemulsion is increased, but when the amount of active mixture is maintained constant (path B), kobscontinues to decrease for CII-HQ whereas it remains constant for HQ. This difference suggests that HQ partitions preferentially in the amphiphilic membrane and CII-HQ in the dodecane core. Such an explanation is in agreement with comparative solubility measurements that we have reported elsewhere33in order to precisely determine the preferential site of solubilization of these two ligands. We have also determined the kinetics of complexation in pure micellar solutions of SDS (some of the experimental data obtained (33) Boumezioud, M.; Derouiche, A,; Tondre, C. J. Colloid Interface Sci., in press.
850 The Journal of Physical Chemistry, Vol. 93, No. 2, 1989
Tondre and Boumezioud
TABLE II: Reciprocal Relaxation Times (or Observed Rate Constants) for Extractant/Nizt Reactions at 25 OC ([Ni2+] = 6.4 X M, lHQl = lo4 M, ICll-HQ1 5 lo4 M, pH 8.5) C ,I-HQ/Ni2+ qa,,-l, T , ~ ~ ~HQ/Ni2+: - ~ , reaction medium s-l s-I T-l
micellar SDS solution, 5.88% (w/w) 2.64 mixed micellar solution 1.76 5.88% SDS 5.88% CSOH microemulsion D, 0.41
0.43
30
0.16
26.5
0.04
11.7
with the system HQ/NiZ+can be seen in Figure 11 which will be discussed below). The kinetics are greatly enhanced, for both extractants, at low concentration of SDS, due to a classical micellar catalysis effect.34 This is related to the increased concentration of reagents a t the micelle surface, but this effect progressively disappears for larger SDS concentrations because the reagents are redistributed on a greater number of micelles. Moreover, when the SDS concentration is around 0.2 M (i.e., comparable to its concentration in microemulsion DJ,the addition of pentanol and then of dodecane has no dramatic effect (see Table 11): The relaxation time is increased about 6-fold for CII-HQ and only 3-fold for HQ. The slow relaxation time measured with CII-HQ was found to be practically independent of the SDS concentration when all the other parameters are kept constant. This indicates that the slow process is involving an intramicellar reaction. Equilibrium Measurements. The interpretation of the preceding results requires a more precise knowledge of the location of the two ligands in the microemulsions and of the geometrical constraints susceptible of governing the stoichiometry of the complexes. The first information was obtained from solubility measurements with a saturating concentration of extractant33and the stoichiometry of the complexes were determined by Job's continuous-variation method. The solubility measurements have been discussed in great details elsewhere33because they had important fundamental implications and potential applications. Their original object was to determine the association constants of H Q and CI1-HQ to micelles and microemulsion droplets, respectively. In fact only the association constant of HQ to the micellized surfactant could be evaluated, defining the equilibrium constant K H L for the partitioning of the neutral form of the extractant between the two pseudophases bylo (3) where C is the concentration of surfactant contributing to the micellar pseudophase and and [HLIB are the concentrations in the micellar and bulk pseudophases, respectively (concentrations referring to,the total solution volume). Provided that the concentration of the solubilized species is low compared to the total surfactant concentration, the preceding expression can be approximated by (4)
Introducing the solubilities of HQ, at neutral pH, in the micellar solution (Stat) and in water (So)
The plot of (St,, - &)/So versus (Ctot- cmc) is represented in Figure 9. The slope of the straight line obtained has led us to a value of 70 M-' for K H L . This value should be considered as a lower limit because the approximation made above does not appear to be totally justified here. (34) James, A. D.; Robinson, B. H. J . Chem. Soc.,Faraday Trans. 1 1978, 74, 10. Diekmann, S.; Frahm, J. . I Chem. . Soc., Faraday Trans. I 1979, 75, 2199.
Figure 9. Plot of (Smt- So)/Sovs (CsDs - cmc) from solubility measurements of HQ at pH 7.5.
Figure 10. Absorption vs mole fraction of NiZ+,Yr,= [Ni*']/([extractant] + [Ni2+])at 400 nm (Job's plots). No buffer. (a) In methanol (mixing of solutions of concentrations lo-' M): (+) HQ/l\ji2'; (0) C, I-HQ/Ni2+. (b) In microemulsions (mixing of solutions of identical concentrations as indicated): left scale, (+) HQ/Ni2+ (lo-' M); right scale, (0)Cll-HQ/Ni2+(2 X lo-' M), (0) CII-HQ/Ni2+(6 X lo4 M). All in microemulsion D,,except for the dashed line (- - -), which is in microemulsion I , .
The low solubility of CI1-HQ in pure micellar solutions has prevented us from carrying out a similar determination with this extractant. The association constants to microemulsion droplets were impossible to measure, although we tried to extend the preceding method to this situation. This was due to the fact that the microemulsion droplets cannot be considered as well-defined objects during the solubilization process: Their composition is affected by a competition for solubilization between the extractant molecules themselves and either the cosurfactant or the Nevertheless, the solubility measurements have shown that the association of HQ should not be very much different from that observed in pure SDS micelles, whereas for Cll-HQ a true microemulsification process of the extractant occurs, as soon as a cosurfactant is added to the system. The stoichiometry of the complexes formed in microemulsions when the extractants are mixed with Ni2+ appears to be very different for HQ and CI1-HQ, respectively. It is also very different from the situation encountered in homogeneous solution (results in methanol). This is illustrated in parts a and b of Figures 10, in which the stoichiometry can be deduced from the position of the absorption maximum. The results in methanol (Figure loa) have been given for the sake of comparison. They show a maximum at 0.33 for both extractants, corresponding to the formation of a 1:2 (metal to ligand) complex. In microemulsion systems (Figure lob), a maximum of 0.25 is obtained for HQ, indicating the formation of a 1:3 complex. Complexes with three ligands cannot be excluded;28they could be favored in the neighborhood of a microemulsion droplet, either by the more acidic environment or by a concentration effect. This observation suggests that HQ is solubilized near the surface of the particles rather than deep inside it. The situation is more complex for CI1-HQ for which Job's curve can be bimodal as well for a direct microemulsion (D,) as for a reverse one (I,). The first maximum is at around 0.3 and the second one at around 0.6. The interpretation is not as straightforward as in homogeneous solutions, but this probably indicates that 1:2 and 1:l complexes can coexist, although the
The Journal of Physical Chemistry, Vol. 93, No. 2, 1989 851
Microemulsions in Extraction Kinetics SCHEME I H ,(I
the kinetics ~bserved,~' all these equilibria can be considered to be always established. For this reason, the above kinetic model is characterized by only one observable relaxation time. We have applied the treatment of Berezin et al.36to obtain the overall rate for complex formation u as a sum of individual rates in the pseudophases, weighted by the volume fraction 4 of the pseudophase considered
+ ( L - ) ~+
with
maxima in this case should be obtained at 0.33 and 0.5, respectively. The existence of a 1:l complex is perfectly conceivable in this case because the extractant is located inside the particle and the steric hindrances render more difficult the formation of a 1:2 complex. As expected in this case, Figure 10b shows a more pronounced second maximum when the ligand and metal concentrations are lower, because the probability of having a 1:l complex is increased. It must be emphasized that only 1:l complexes have to be considered in the kinetic experiments, in which the extractant concentration was always very low compared to the metal concentration. Knowledge of the preceding stoichiometries gives nevertheless some useful indications concerning the location of the extractant molecules. Interpretation of the Kinetic Results. On the basis of the kinetic mechanisms existing in homogeneous solution^^^^^^ and of the data for the partition of the extractants between the pseudophases, Scheme I can be proposed to interpret the results. This mechanism can be applied to HQ as well as to Cll-HQ, in pure micellar solutions or in microemulsions. Considering the pH range investigated in the present study, the diprotonated form of the extractant (HzL+) has not been introduced in this scheme. In any case, this species has been previously shown to have very little r e a c t i ~ i t y . ~Nevertheless, ~ it will be introduced in the mass balance when justified by the experimental conditions. On the other hand, metal ion hydrolysis is assumed to be negligible because Ni2+ ions are expected to be located mainly at the micellar surface where the pH is more acid than in the bulk. The above mechanism is based on the pseudophase model first introduced by Berezin and c o - w ~ r k e r s . ~The ~ reaction is assumed to take place simultaneously in the bulk aqueous pseudophase (subscript B) and in the micellar pseudophase (subscript M). It involves the neutral form (HL) and the anionic form (L-) of the extractant and the partition of the reagents between the pseudophases. In fact, the constants KM, and KHL have been treated as association constants of the considered species to the micelles (see Equilibrium Measurements):
All the concentrations refer to the total volume of the solution, and kfaPP is the apparent rate constant for complex formation. Neglecting the activity coefficients, additional relations are provided by the ionization constants of the extractants in the pseudophases:
An expression for kaPP can be obtained by solving eq 6-18 in the form kfaPP = kiL
mM[(l+
+
kb
+E
KZM'([H+l M/$M)
KZMr([H+l 1 M/4M) + KlM'([Hi1M/4M))
+
-
The micellar particles have not been introduced in the reaction scheme for the sake of simplicity. KHL' defines the partition constant of H L between the micellar pseudophase KHL' =
([HLlM/[HL]C)('#)C/'#'M)
with
(7)
(or "membranen pseudophase in the case of microemulsions) and the oil core (subscript C) of the droplets. Owing to the fact that the proton-transfer reaction between H L and L- is very fast and that the partition reactions of the reagents between the pseudophases are expected to be much faster than (35) Johnson, W. A.; Wilkins, R. G. Inorg. Chem. 1970, 9, 1917. (36) Berezin, I. V.;Martinek, K.; Yatsimirskii, A. K. Russ. Chem. Rev. (Engl. Transl.) 1973, 42, 787.
(37) Almgren, M.;Grieser, F.; Thomas, J. K. J . Am. Chem. SOC.1979, 101, 279.
852 The Journal of Physical Chemistry, Vol. 93, No. 2, 1989
Tondre and Boumezioud be the same in the two pseudophases, and the values previously obtained in water2' have been entered in the calculations:
kgL = kBL = 707 M-' s-' k p = kf = 5
X
lo5 M-' s-'
Such an approximation is not unusual in micellar kinetics, where the changes of reaction rate are often due to simple concentration effects, the rate constants remaining unchanged. The same kinds of approximation have been made for the values of the ionization constants KIM' = KIBr = 1/Kl = 109.9 K2Mr = K ~ B=' 1 / K2 = lo5,'
0
0.2
0.4
0.6
Figure 11. Plots of k,, vs micellized concentration of SDS for the HQ/Ni2+ reaction at different pHs. The symbols refer to experimental results at pH 7.08 (+), 8.25 ( O ) , and 9.0 (0). The lines have been calculated as indicated in the text from eq 19 (-) or eq 21 (---). Conditions: [HQ] = IO4 M; [Ni2+]= 4.25 X M; TEA buffer, 0.1 M.
We have previously indicated that kaPP should be experimentally obtained from the slope of the straight lines representing the variation of kabsvs the total metal ion concentration since
kobs= kpP[Me2+]
+ kdaPP
(20)
The same kind of treatment as just developed for k;pp could be applied to the reverse reaction in order to obtain an expression for kdaPP. Owing to the low values obtained experimentally for this constant (see Table I and Figures 3 and 4), we will neglect it in the following and we will consider that the first term alone in eq 20 provides a good enough approximation for the calculation of the value of kobr. We will first check the validity of eq 19 by using it to calculate the theoretical values of kobs for the HQ/Ni2+ reaction in pure micellar solutions of SDS. In this case, most of the constants needed can be measured or estimated. We will then try to examine whether the model can predict the values of kobsmeasured in mixed SDS/pentanol micelles and in microemulsions. Case of Pure Micellar Solutions. All the (K,C)-' terms in eq 19 are expected to be very small when the association constants of the reagents with the micelles are large, so that eq 19 can be approximated by
kaL +
kh KZM'(
fH+l M / h ) \
in which the (KHLr)-'term has also been discarded because there is no oil core in this case. The theoretical simulations obtained both with the complete expression (eq 19) and with the simplified one (eq 21) have been compared in Figure 11 with the experimental points obtained at different pHs and different SDS concentrations. The agreement is very satisfactory for pH 7.08 and 8.25, but at pH 9, the experimental points are significantly different from the theoretical prediction, particularly at low SDS concentration. However, when the surfactant concentration is large, the simplified equation can be considered to furnish a good enough prediction even at pH 9. In order to obtain the preceding theoretical curves, and owing to the large numer of parameters involved in eq 19, several assumptions had to be made. The rate constants were assumed to
where pKl and pK2 are the ionization pK's in water.35 This is a very rough approximation, which neglects the possible influence of the micellar particle. The association constants Kx to the micelles have been taken such that KNi= l o 3 M-' lo and K H L = 70 M-' (from the solubility measurements described above). The volume fraction of the micellar pseudophase 4Mis given by
4 M = CY
(22)
4B = 1-4M
(23)
where C has already been defined as the concentration of micellized SDS and Vis the volume contributed per mole of micellized surfactant to the micellar pseudophase.I0 This volume should not be confused with the molar volume of SDS since it includes all the space in which the reaction rate is affected by the presence of the particle. The quantities [ H + ] M / ~ and M [H+IB/4B represent the local proton concentrations in the pseudophases. They can be treated as constants when the pH is fixed by a buffer, and PHB will be taken as the measured pH. As a first approximation, the following relation was used, with concentration in place of activity: [ H + ] B / ~=B
(24)
We are left with only two adjustable parameters: [H+]M/$Mand V (see eq 22). The best simulations, which are reproduced in Figure 1 1 , have with pHM such that pHM been obtained with [ H + ] M / ~ = M = pH - 1.25 and V = 1.5 dm3/mol. The value of pHM indicates a difference of 1.25 pH units between the bulk and micellar pseudophases. The lower value of the pH close to the negatively charged surface of the SDS micelles is consistent with previous observation^.^^ A difference of up to 2 pH units can exist in unbuffered systems, which should be reasonably decreased when a buffer is present. Note also that Figure 7 was suggesting that a shift of 1.3 pH units, quite comparable to the value obtained in the simulations, could exist in the microemulsion D,. On the other hand, the value obtained for V, the second adjustable parameter, appears to be much larger than the values usually found in micellar catalysis studies. The volume of the micellar pseudophase is often found to be of the order of the volume of the Stern layer, Le., around 0.2-0.4 dm3/mo1.9J0~38,39 The unusually high value of Vwould indicate that the pseudophase volume in which the reaction occurs include, in addition to the Stern layer, a crown about 12 A thick. A distance of 10 A corresponds to the Debye screening length in 0.1 M of a 1:l electrolyte, which is equal to the buffer concentration. The value obtained thus roughly defines the volume in which most of the Ni2+ ions are concentrated. It can be concluded that the values obtained from the theoretical simulations are not unreasonable. We have intentionally limited the number of adjustable parameters. It is obvious that we could have adjusted as well other parameters, for instance the rate (38) Biresan, G.; Bunton, C. A. J . Phys. Chem. 1986, 90, 5849. (39) Bunton, C. A,; Romsted, L. S.; Savelli, G. J. Am. Chem. SOC.1979, 101, 1253.
The Journal of Physical Chemistry, Vol. 93, No. 2, 1989 853
Microemulsions in Extraction Kinetics constants, but it does not appear to be justified. On the other hand, if the pk"s of the extractant were modified in the micellar pseudophase, this would mean that the change attributed to [H+]M/4M (compared to [ H + ] B / ~ Bshould ) have been distributed between KXM' and [H+]M/4M,which always appear as a product in eq 19 or 21. The bad agreement between theory and experiment at pH 9 can be attributed to a charge separation effect, which has not been taken into account: At this pH there will be more extractant molecules in their anionic form, which is repelled by the negatively charged micelles whereas the NiZ+ions are attracted. This is expected to reduce the value of kobs,as observed in Figure 1 1 . Extension to the Cases of Mixed Micelles and Microemulsions. Although the association constant K H L could not be measured in these cases, solubility measurement^^^ show that it is always at least of the same order as in pure micellar solutions of SDS, so that the (KXC)-' terms can always be neglected in eq 19. In the case of mixed pentanol/SDS micelles, we have seen (see Table 11) that pentanol is responsible for a decrease of l / ~ This effect can be easily explained by the increase of +M in eq 21. Indeed, the value of 4Mmust include in this case the molar contribution VA of the micellized alcohol to the micellar pseudophase:
(=Fobs).
This simple change can reasonably account for the small effect observed for the HQ/Ni2+ reaction. A further decrease of 1 / ~ , by a factor slightly larger than 2, occurs when going from the mixed micelles to the microemulsions. As mentioned before, we must in this case keep the (KHL')-' term in the expression of k f p , to take into account the partition of the extractant into the oil core. A simple estimation of the different terms in the denominator of eq 21 shows that the missing (KHL')-'&/4M term does not need to be much higher than 1 to take quantitatively into account the effect of adding dodecane for the HQ/Ni2+ reaction. In the case of the C11-HQ/Ni2+ reaction, the partition of the extractant into the oil core is of course much more important, leading to a larger contribution of the (KHL')-' term. This will only partly explain the factor of about 30 between the rates of reaction of the two extractants in microemulsion D,. Indeed, the data in Table I1 show that a factor of 11-15 already exists in SDS and in mixed SDS/pentanol micelles. Remember also that the alkyl chain has been found to be responsible for a 4- to 6-fold decrease of kfpp in methanol (Table I). For all these reasons, it is obvious that the slowing down of the reaction kinetics is also related to a decrease of the rate constants compared to HQ. The degree of freedom of the neutral form of the alkylated extractant solubilized in a micellar particle is considerably reduced, and it has to move toward the interface to react with a NiZ+ion. On the other hand, the anionic form has a nonnegligible solubility in the bulk water, but the accessibility of the complexation site is decreased due to the steric hindrances created by the alkyl chain.ls The effect of the reduction of dimensionality of the system may play an important part in the preceding observations. Both the activation energies and the collision probabilities are expected to be modified when a reaction takes place at an interface instead of in the bulk. Contradictory opinions have been developed in the literature regarding this effect. Astumian and Schellym have applied the collision, transition-state, and diffusion control theories to demonstrate that the geometric effects of the reduction of dimensionality are a significant decrease of the specific rates of bimolecular elementary reactions. On the contrary, the results of Gratzel et al."1*42support an opposite conclusion, showing that the reduction of dimensionality dramatically increases the probability of reaction. (40) Astumian, R. D.; Schelly, Z. A. J . Am. Chem. Soc. 1984,106,304. (41) Frank, A. J.; Gratzel, M.; Kozak, J. J. J . Am. Chem. SOC.1976,98, 3317. (42) Hatlee, M. D.; Kozak, J. J.; Rothenberger, G.; Infelta, P. P.; Gratzel, M. J . Phys. Chem. 1980,84, 1508.
Another very interesting result has been obtained by Szyman ~ w s k i who , ~ ~ has measured the extraction rate of copper by hydroxy oximes having alkyl substituents with lengths varying from C,to CI2. The contribution of the volume reaction is shown to be less than 10% for oximes containing 1 1 carbon atoms or more in the alkyl chain and higher than 90% for oximes whose alkyl chain has no more than 3 carbon atoms. It may be entirely coincidental, but the extraction rate when the bulk reaction predominates was found to be about 30 times larger than when the reaction is mainly interfacial. This is exactly the factor that we have found in the microemulsion D, between the H Q and Cl1-HQ reactions with Ni2+. Both results would thus support the idea that the reduction of dimensionality has the effect of reducing the reaction rate. Origin of the Second Relaxation Time in the Case of CII-HQ. We have mentioned above that the second relaxation process observed with C,,-HQ was not specifically related to the presence of an oil core, since it also exists in pure micellar solutions of SDS. A first explanation that has occurred to us was that the slow process could be related to a slow exit rate of the extractant out of the micellar particle. This would mean that the equilibrium (HL)M F! (HL), (see general reaction mechanism) is no longer rapidly established with the alkylated ligand, resulting in a splitting into two relaxation processes. This interpretation is not satisfactory because it would contradict the fact that the exit rate for a variety of neutral solutes, even when they are very hydr~phobic,~' is quite fast. The exit time is seldom longer than a few milliseconds, whereas the relaxation time of the slow process is in the order of several tens of seconds. One can also recall that the characteristic time for the exit of an SDS molecule out of a micelle is in the order of microseconds.44 A more likely explanation is that an intermediate species forms during the reaction, which contrary to the case of HQ'O would have a measurable lifetime:
O\H
Due to the relative positions of the alkyl chain R and of the hydroxyl group, the dissociable proton will be buried in an hydrocarbon environment, which will make difficult its removal. This will be particularly true if the intermediate species is stabilized through Coulombic attractions with the negatively charged SDS molecules. According to classical examples in relaxation kinetics,32if the second step in reaction 26 is much slower than the first one, and in our experimental conditions (pH buffered, [Ni2+], >> [HL],), the slow relaxation time will be characterized by the following expression:
In this equation, k , must be substituted by the complicated expression of kfaPP and k2 by an expression for kdaPPnot developed in this work. The expressions for k3 and k4 will depend on the mechanism of proton removal. One can think of three different possible reactions, implying the proton transfer on a water molecule, on a hydroxylic ion, or on a buffer molecule:45 (43) Szymanowski, J. Polyhedron 1985, 4 , 269. (44) Aniansson, A. G.; Wall, S. N.; Almgren, M.; Hoffmann, H.; Kielmann, J.; Ulbricht, W.; Zana, R.; Lang, J.; Tondre, C. J. Phys. Chem. 1976, 80, 905. (45) Barth, D.; Tondre, C.; Delpuech, J. J. Int. J. Chem. Kinet. 1986, 18, 445.
J . Phys. Chem. 1989, 93, 854-858
854
head,46 and it cannot be excluded that TEA behaves like a cosurfactant, which would facilitate its penetration. Reactions 28 and 30 appear thus to be quite plausible. Equation 27 predicts that at low Ni2+concentration (k, [Ni2+] > k,) T , ~ should ~ ~ - tend ~ toward a constant value. This value should increase when the proton concentration increases. Although a plateau has only been attained at pH 8.5, there is no major disagreement between this prediction and the experimental results.
The expressions for k3 and k4 would be, respectively, in this case k3 = kH20
+ kOH[OH-lM + kTEA[TEAlM
k4 = k - ~ , o [ H + l+~ k-oH
+ ~-TEA[TEAH+]M
(3l) (32)
Proton removal by the hydroxylic ion (reaction 29) is very unlikely because of the electrostatic repulsions with the negatively charged surface of the micelles. It is admitted that water molecules can penetrate the micelles until the first CH2 group after the polar
Conclusions We have shown in this work that microemulsions can be employed to improve our understanding of the complicated reactions taking place in liquid-liquid extraction processes. Due to their transparency, they allow the use of methods generally limited to homogeneous kinetics, although providing information about heterogeneous kinetics. As the complexation step is often rate limiting in the extraction of metal ions, it is of major interest to have at our disposal this very convenient way of studying the influence of the nature of the components of the extraction medium, as well as the influence of potential additives. Registry No. Ni, 7440-02-0; HQ, 148-24-3; 8-quinolinol, 291 71-27-5. (46) Zana, R. J. Chim. Phys. 1986,83, 603.
Nucleophllic Reactions In Zwltterlonic Micelles of Amine Oxide or Betaine Sulfonate Surfactants Clifford A. Bunton,* Marutirao M. Mhala, and John R. Moffatt Department of Chemistry, University of California, Santa Barbara, California 93106 (Received: February 8, 1988; In Final Form: July 12, 1988)
Reactions of CI- and Br- with methyl naphthalene-2-sulfonate (MeONs) are speeded by micelles of N-hexadecyl-N,Ndimethyl-3-ammonio-1-propanesulfonate(SB3-16), but reactions of OH- and SO$ are inhibited. Reactions of o-iodosobenzoate ion and its 5-octyloxy derivative and of F with p-nitrophenyl diphenyl phosphate (pNPDPP) are speeded by micelles of SB3-16 and less strongly by micelles of dodecyldimethylamine oxide. Comparison of these micellar effects with those of cationic micelles provides a qualitative estimate of the concentrations of the nucleophilic anions at the surface of micelles of SB3-16. For dephosphorylation by iodosobenzoate ions reactivity at the micellar surfaces is very similar in water and at the surfaces of cationic and zwitterionic micelles.
Ionic micelles in water concentrate counterions at their surface and repel co-ions due to Coulombic interactions. But non-Coulombic interactions are also important because hydrophilic, high charge density counterions are bound less strongly than polarizable, low charge density counterions. The concentration of reactants at micellar surfaces is the major source of enhancements of the rates of bimolecular reactions involving counterions, and examination of reaction rates and equilibria provides information on micelle-ion The specificity of counterion binding to micelles can be described in terms of a competition between ions. In one widely (1) Fendler, J. H. Membrane Mimetic Chemistry, Wiley-Interscience; New York, 1982. (2) (a) Romsted, L.S.In Micellization, Solubilization and Microemulsions; Mittal, K. L., Lindman, B., Eds.; Plenum: New York, 1977; p 509. (b) Romsted, L.S.In Surfactants in Solution; Mittal, K.L., Lindman, B., Eds.; Plenum: New York, 1984; Vol. 2, p 1015. (c) Romsted, L. S. J. Phys. Chem. 1985,89, 5107, 5113.
(3) Quina, F. H.; Chaimovich, H. J. Phys. Chem. 1979, 83, 1844. (4) Sudholter, E. J. R.; van de Langkruis, G. B.; Engberts, J. B. F. N. Recl.: J . R . Neth. Chem. Soc. 1980, 99, 73. ( 5 ) (a) Bunton, C. A. Coral. Rev. Sci. Eng. 1979,20, 1. (b) Bunton, C. A,; Savelli, G. Prog. Phys. Org. Chem. 1986, 22, 213.
0022-3654/89/2093-0854$01 S O / O
used treatment one for one competition is assumed and ion-exchange parameters are calculated as for binding to ion-exchange resin^.^-^ An alternative treatment uses a Langmuir equation without stipulating one for one e~change.~,'Other treatments involve calculation of Coulombic effects by solving the PoissonBoltzmann equation, if necessary with inclusion of separate, non-Coulombic, interaction^.^^^ There is kinetic and equilibrium evidence for the binding of both hydrophobic and hydrophilic co-ions to ionic micelles. 1-5~10 Many hydrophobic organic cations bind readily to cationic micelles because dispersive and hydrophobic attraction overcome the Coulombic repulsions. Hydrophilic anions, e.g., OH-, are not completely excluded from anionic micelles if the solution is concentrated in electrolyte." This partial binding is explained by (6) Bunton, C. A.; Gan, L.-H.; Moffatt, J. R.; Romsted, L. S.;Savelli, G. J . Phys. Chem. 1981, 85, 4118. (7) Rodenas, E.; Vera, S. J. Phys. Chem. 1985, 89, 513. (8) (a) Bunton, C. A,; Moffatt, J. R. J. Phys. Chem. 1986,90, 538. (b) Bunton, C. A.; Moffatt, J. R. Ibid. 1988, 92, 2896. (c) Bunton, C. A.; Moffatt, J. R. Ann. Chim. 1987, 77, 117. (9) Rcdenas, E.; Ortega, F. J. Phys. Chem. 1987, 91, 837. (10) Cordes, E. H.; Gitler, C. Prog. Bioorg. Chem. 1973, 2, 1.
0 1989 American Chemical Society
