FUMIKATSU TOKIWA AND KAORU TSUJII
3560 Addition of the nonionic surfactant cetomacrogol 1000 B.P.C. (CieHaa(OCH2CH2)22OH) to solutions of the phenothiazine lowers the pH of the solution. This suggests that the phenothiazine is being taken up in the micelles of the nonionic surfactant and that this results in increased dissociation of the drug. However, the chemical shifts are not what would be expected from this type of association. On increasing the concentration of nonionic a shift of the N-methyl proton signal to lower field occurs (see Table 11). As stated above, such shifts are normally associated with increased protonation. A possible explanation of the observed direction of shift might be that there is association of the nonionic hydroxyls with the head groups of the phenothiazine, i e . , CHZ-0-H. . . : N-(CHa). Hydrogen bonding of this sort would tend to deshield
the N-CHa protons. The addition of more cetomacrogo1 could encourage further dissociation of the head groups by a competing hydrogen-bonding interaction, resulting in a net downfield shift. This would also explain the concurrent drop in pH. Precise analogies have not been detected in the literature. Page and B ~ e s l e rhowever, ,~~ observed a strong downfield shift in the HO-CH, signal as a result of interaction between polyalkylene glycol and pyridine. The exact nature of the interaction between nonionic detergent and the phenothiazine molecules is not clear, but it serves to show that the shifts observed in the first part of this study are primary effects and not simply the result of pH changes in the system. (29) T. E. Page and W. E. Bresler, Anal. Chem., 36, 1981 (1964).
Nuclear Magnetic Resonance Study of Interaction between Anionic and Nonionic Surfactants in Their Mixed Micelles by Fumikatsu Tokiwa* and Kaoru Tsujii Industrial Research, Laboratories, K a o Soap Company, Wakayama-shi 640-91,J a p a n
(Received January 19, 1971)
Publication costs assisted bu the Kao Soap Company
The nmr spectra of aqueous solutions of nonionic surfactants, dodecyl polyoxyethylene ethers (C12POE) with different number of oxyethylene units, have been measured in the presence of anionic surfactants, sodium dodecyl sulfate (NaC12S) and sodium p-octylbenzenesulfonate (NaCsBS), a t different molar ratios of anionic/ nonionic surfactant. When NaCsBS is added to ClzPOE solutions, the peak due to protons of the polyoxyethylene chain of C12POEshifts t o a higher magnetic field, the extent of the upfield shift depending on the polyoxyethylene chain length of ClzPOE and the mixing ratio of NaCsBS/C12POE. However, this peak remains a t the same position and shows no remarkable change on addition of NaC12S. The upfield shift of the polyoxyethylene peak caused by addition of NaCgBS is ascribed to an interaction between the polyoxyethylene chain of CIzPOE and the benzene ring of NaCgBS in their mixed micelle. This interaction has been discussed in relation to the polyoxyethylene chain length of ClzPOE and the mixing ratio of NaCBBS/C12POE. In the mixed micelle, the number of CIzPOE molecules influenced by one NaCsBS molecule was calculated to be 0.6-1.6, depending on the polyoxyethylene chain length and nearly constant for ClzPOE with oxyethylene units larger than about 9. The number of oxyethylene units influenced by the benzene ring of NaCEBS was estimated t o be about 9.
Introduction Mixed surfactant systems are of importance from the fundamental and practical points of view. However, a relatively limited number of papers have been reported on the micellar properties of mixed surfactant solutions, probably because of complication of these systems. In previous works, we have studied the colloidal properties of mixed micelles of two different surfactants by means of electrophoresis, light scatterT h e Journal of Physical Chemistry, Vol. 76, N o . BS, 2071
ing, vapor pressure depression, and other types of measurements.' -6 These studies involved the micellar
(1) F. Tokiwa, J . Colloid Interface Sci., 28, 145 (1968). J a p . , 41, (2) F. Tokiwa, K. Ohki, and I. Kokubo, Bull. Chem. SOC. 2845 (1968). (3) F. Tokiwa and N. Moriyama, J . Colloid Interface Sci., 30, 338 (1969). (4) F. Tokiwa and K. Aigami, Kolloid-Z. Z . Polym., 239,687 (1970).
INTERACTION BETWEEN ANIONICAND NONIONIC SURFACTANTS size, electrical nature, and solubilization capacity of mixed micelles and also the degree of ionic dissociation of ionic surfactant in mixed micelles. During the course of other work of nuclear magnetic resonance spectroscopy (nmr), the present authors observed an interesting aspect of behavior, as will be described below, with respect to the interaction of benzene with polyethylene glycol. The nmr technique is thought to be one of the powerful tools to investigate an intermolecular interaction of two different surfactants in their mixed micelles, and it will give us some information of this interaction which may not be obtained by means of other techniques. I n the present work, an nmr study has been made of the interaction between anionic and nonionic surfactants, sodium p-octylbenzenesulfonate and dodecyl polyoxyethylene ether, in their mixed micelle. Especially, the interaction between the benzene ring of the anionic surfactant and the polyoxyethylene chain of the nonionic surfactant has been discussed in relation to the polyoxyethylene chain length by paying attention to the chemical shift of the polyoxyethylene proton signal.
Experimental Section Materials. Sodium dodecyl sulfate (abbreviated to NaC12S), C12H250S03Na,was the same sample as that used in a previous work.’ Sodium p-octylbenzenesulfonate (abbreviated to NaCsBS) was synthesized by the following procedure.
I n each step, the reaction product was distilled under reduced pressure or recrystallized from a proper solvent. The product obtained in each step was checked by infrared spectroscopy and element analysis and also by vapor phase chromatography when it was liquid. The final product used for nmr measurement was recrystallized three times from a water-ethanol mixture (20 :SO) and once from ethanol. Dodecyl polyoxyethylene ethers (abbreviated to C12POE), C12HZ~0(CH2CH20),,H,with different p were prepared from dodecyl alcohol of a high purity by addition of ethylene oxide in the presence of sodium hydroxide as a catalyst. Polyethylene glycol, a byproduct of the reaction, was removed by the following method. ClzPOE-2.9 (the number written after C12POE denotes the number of oxyethylene units per molecule): This sample was fractionated from the ethylene oxide condensate of dodecyl alcohol (ethylene oxide, ea. 3
3561
mol) at 150-170” under a reduced pressure of 0.03 mm. Papers and vapor phase chromatographies showed the sample to be free of the unreacted alcohol and polyethylene glycol. ClzPOE-4.0, C12POE-5.9, C12POE9.1, Cl2POE-11, and Cl2POE-14: each sample was purified by the countercurrent solvent extraction method9 using butanol-saturated water and watersaturated butanol. Paper chromatography showed the purified samples to be free of polyethylene glycol. ClzPOE-24: This sample was used without purification because the above two methods could not be applied. The sample contained a small amount of polyethylene glycol (less than 1-2%). The average numbers of oxyethylene units per molecule of these samples were determined from their hydroxyl values. Polyethylene glycol, the molecular weight of which was about 200, was of reagent grade. Benzene was of guaranteed reagent grade. They were used without purification. Nmr Measurements. A sample solution for nmr measurement was prepared by adding a desired amount of anionic surfactant to 5% (by wt/vol) Cl2POE solution containing 0.1% 1,4-di0xane’~as an internal standard. Glass-distilled water was used to make up all sample solutions. The concentrations of sample solutions were sufficiently high as compared to critical micelle concentrations (cmc) and, therefore, most of the surfactant molecules were dissolved in the form of micelles. The cmc values of C12POE, NaC8BS, and to 1 X 3.5 X lO-l, and NaC12S are 1 X 2.3 X lO-l%, respectively, and those of C12POENaC8B8 or C12POE-NaC12Smixtures are probably in an order of at the mixing ratios examined. ClzPOE-2.9 and ClzPOE-4.0 were not soluble (but homogeneously dispersive) in water owing to low affinity for water, but soluble in the presence of anionic surfactant. The chemical shift of the polyoxyethylene proton signal, to which attention was paid in the present work, of these two samples in the absence of anionic surfactant was assumed to be the same as the chemical shift for other samples which coincided within the error of measurement. As is easily understood from the chemical structure of (312(5) F. Tokiwa and N. Moriyama, Nippon Kagaku Zasshi, 91, 903 (1970). (6) F. Tokiwa, N. Moriyama, and H. Sugihara, {bid., 90,449 (1969); 90, 454 (1969); 90, 673 (1969). (7) F. Tokiwa, J. Phys. Chem., 72, 1214 (1968). (8) K. Shinoda, T. Nakagawa, B. Tamamushi, and T. Isemura, “Colloidal Surfactants,” Academic Press, New York, N. Y., 1963, p 165. (9) K. Nagase and K. Sakaguchi, Koggo Kagaku Zasshi, 64, 635 (1961). (10) Dioxane was also used as an internal standard in aqueous solutions of surfactants in the following papers: e.g., H. Inoue and T. Nakagxwa, J . Phys. Chem., 70, 1108 (1966); T. Nakagawa and K. Tori, Kolloid-Z. 2. Polym., 194, 143 (1964); B.R. Donaldson and J. C. P. Schwarts, J . Chem. SOC.B , 395 (1968). It is not possible a t this time to apply bulk susceptibility corrections to the observed chemical shifts since the magnetic susceptibilities of the surfactants used have not yet been measured.
The Journal o j Physical Chemistry, Vol. 76, No. $3,1971
3562 POE, all protons of the polyoxyethylene chain are equivalent in nmr spectroscopy except the proton of terminal -OH. This is independent of the polyoxyethylenene chain length. 'Therefore, it may be assumed that the polyoxyethylene chain of C12POE2.9 and CnPOE-4.0 surfactants have the same chemical shift as that of their higher homologs. The nmr spectra were obtained with a Japan Electron Optics JNNI-3H-60 spectrometer (60 Mcps) operating at about 22'. Chemical shifts were measured in cycles per second from the dioxane The precision of measurements of the distance between the dioxane peak and a peak due to protons of the polyoxyethylene chain of ClzPOE was *0.5 cps.
Results and Discussion When NaClzS is added to ClzPOE solutions, the peak due to protons of the polyoxyethylene chain of ClzPOE shows no change and remains at the same position. On the other hand, this peak shifts to a higher magnetic field and becomes broader on addition of NaCeBS. Broadening of the polyoxyethylene signal in this case is not owing to an increase in viscosity because of the absence of broadening of other resonance signals of protons. I n other signals of CIZPOE than the polyoxyethylene signal, no appreciable change in chemical shift was observed. The extent of the upfield shift of the polyoxyethylene signal, caused by addition of NaC8BS, depends on the chain length of polyoxyethylene and the mixing ratio of NaCaBS/ C12POE. This upfield shift may be ascribed to an interaction of the polyoxyethylene chain of ClzPOE with the benzene ring of NaCsBS in their mixed micelle, since the only difference between the two anionic surfactants is whether they have a benzene ring in the molecule or not. Incidentally, on mixing ClzPOE Kith NaCsBS, the peak due to phenyl protons of NaCsBS near to the alkyl chain shifts t o a lower field to the extent of 3-4 cps. An evidence of this interaction may be provided by changes in chemical shift of polyethylene glycol by addition of benzene. When polyethylene glycol is mixed with benzene, the polyoxyethylene proton signal shifts to a higher field and the benzene proton signal to a lower field.12 This is indicative of an interaction of polyethylene glycol and benzene molecules and corresponds to the result for the mixtures of ClzPOE and NaC8BS. It also suggests an interaction of the polyoxyethylene chain and the benzene ring in the mixed micelle of C12POE and NaC8BS. To discuss quantitatively the interaction in the mixed micelle, it is necessary to evaluate the average chemical shift of the polyoxyethylene proton signal broadened by addition of NaC8BS. This was done on the nmr chart by calculating the quantity fvI(v)dv/ ,fl(v)dv, where v is the chemical shift in cycles per second and I(v) is the absorption intensity at v. The Journal of Physical Chemistry, Vol. Yb, No. 93, 1971
FUMIKATSU TOKIWA AND KAORU TSUJII
10
vi
a
u ;s' q
5
0
I 2 C (NoC,BS/CI,POE), mol.
3
Figure 1. The upfield shift, AP, of the polyoxyethylene proton signal plotted against the concentration, C,of added NaCsBS for C12POE-p with different p .
I n the following discussion, the upfield shift Av of the polyoxyethylene proton peak, on addition of NaCsBS, from the original peak will be expressed by this average value. Relation between Av and p. The curves of upfield shift Av vs. concentration C for ClzPOE with different p are shownin Figure 1where Cis the concentration of NaCsBS expressed by the molar ratio of NaCeBS/ClZPOE, the concentration of ClzPOE being kept constant. When one compares the value of Av a t a constant C, it decreases with increasing number of p . This suggests that, in the mixed micelle, the influence of the benzene ring of NaCsBS on the polyoxyethylene protons cannot extend to the whole polyoxyethylene chain when the chain is sufficiently long. Here, if the number of oxyethylene units influenced by the benzene ring is assumed to be 1 and the dependence of Av on C is common to all samples of ClZPOE, then Av will be proportional to l / p (where I < p) at a definite C. Further, if the extent of influence of the benzene ring on E oxyethylene units is denoted as AVI, Av is also proportional to Avd/p at a definite C. Then, the following equation may be written 1
Av = Av,.-f(C), whenp
P
>1
(1)
where f(C) is a function related to the concentration, C. Here, we have t o explain the physical meaning of 1 in more detail. The ClzPOE and NaCsBS molecules form mixed micelles in the form of so-called palisade(11) H . Inoue and T . Nakagawa, J. Phys. Chem., 70, 1108 (1960). (12) I n this case, tetramethylsilane was used as an external standard and, therefore, the difference in magnetic susceptibility between polyethylene glycol and polyethylene glycol-benzene mixture should be taken into account. The magnetic susceptibility correction was made to the chemical shifts observed. However, the extent of correction is insignificantly small.
INTERACTION BETWEEN ANIONICAND NONIONIC SURFACTANTS
3563
water phase
'
~
:\\hydrocarbon core
I
the region influenced ', by the benzene rings
--
/
~
--
, polyoxyethylene
, ,
shell
/
V 0
--#'
I
3
Figure 3. The p . Av vs. C plots for ClzPOE-p with different p .
type micelles, orientating their hydrocarbon part in side and the ionic head of NaCsBS and the poly oxyet'hylene part of CuPOE out'side. Figure 2 is a schematic representation of this mixed micelle. The benzene rings of t'he NaCSBS molecules are probably located in the region between the hydrocarbon core and the polyoxyethylene shell of the micelle. If this is the case, the influence of the benzene rings would be stronger on the oxyethylene units near the hydrocarbon part and weaker on the oxyethylene units remote from the hydrocarbon part. This aspect seems to be reflected on the pattern of the polyoxyethylene signal: namely, the width of the signal becomes broader in the presence of P\TaC8BS. In the present' model we are assuming that, when p > I, the benzene rings in the micelle exert an influence of Avl on tho oxyethylene units from 1 to I and no influence on the oxyethylene units ( I 1) to p . Then, there would appear two imaginary peaks; one is a t the original position and the other at a position corresponding to Avl. Tho relative intensity of the two peaks would depend on the ratio of l / p . The average of the chemical shifts of the two peaks corresponds to the shift Av calculated from the spectrum observed. If the above assumption is reasonable, cq 1 is rewritten in the form
+
>1
I
2
C.
Figure 2. A schematic representation of a part, of a mixed micelle of ClzPOE and NaCsBS ( p > 1).
A v . p = A v l . l . f ( C ) , when p
I
I
consisting of Q ClzPOE molecules. Divide the micelle surface into small sections, each containing q ClzPOE molecules (Q >> q). Assume that if one NaC8BS molecule enters into a particular small section, all of the q ClzPOE molecules in this section are influenced to cause the upfield shift of the polyoxyethylene proton signal, and also assume that further entering of NaCgBS molecules (more than one molecule) into this section exerts no influence upon the upfield shift of the signal. Under this assumption, an increase in upfield shift, d ( A v , , , , ) (the subscript mic refers to one micelle), when dn of NaCsBS molecules enter into the micelle, will be proportional to the probability, P, of finding the small sections not occupied with even one NaC8BS molecule in the micelle. Then, we have the following equation d( AVrnic) = kP(n) (3) dn where k is a constant. The P(n) in eq 3 can be calculated as follows. The probability of finding a particular NaCgBS molecule in a particular small section is q / Q . When n NaCsBS molecules exist in a micelle being watched, the probability of finding m particular NaCsBS molecules in the small section and other (n - m ) molecules outside the small section is written by
(2)
According to eq 2, the plots of A v a p vs. C will give us a single curve irrespective of p if f(C) is common to all numbers of p , because I and Avl are constants. Figure 3 shows the A v . p vs. C plots for ClzPOE with different p . As expected, these plots can be reduced to a single curve except the plots for ClzPOE with p of 2.9, 4.0,and 5.9. The reason for deviation from this curve may be easily understood if I is assumed to be larger than 5.9. When p < I, eq 1 or 2 has no physical meaning. Relation between Av and C. In the next place, let us determine the function f(C). Consider a micelle
As there are n!/m!(n - m)!ways of selecting m out of n, the probability of finding m arbitrary NaCsBS molecules in the small section is given by
('>"(1
n! m!(n - m)! Q
-
a)"-"
This is equivalent in form to the Poisson's equation for " f l u ~ t u a t i o n . " ~The ~ probability of finding no (13) L. Landau and E. Lifshits, "Statistical Physics" (Japanese ed), Vol. 11, Iwanami Books Corp., Tokyo, Japan, 1958, p 124.
The Journal of Physical Chemistry, Vol. 76, N o . 83, 1971
FUMIKATSU TOKIWA AND KAORU TSUJII
3564 NaCgBS molecule in the small section being watched is then { I - (q/Q)In, which can be obtained by putting m = 0. This probability, 11 - (q/Q)]", also holds for other arbitrary small sections. Thus, the probability P(n) of finding small sections containing no NaCsBS molecule in the whole of the micelle (Q molecules) is given by
AV = Avs{l - e ~ p ( - ~ y C ) }
By definition, e-k' = 1 - (q/Q), ie., -k' = In { 1 ( q / Q ) ) . Further, k' = p/Q since In { 1 - ( q / Q ) ) N -q/Q (q/Q is sufficiently small as compared to unity). Then, eq 11 becomes
/
Av = Av8{l
(4)
(5)
In
(12)
Av = Avs(l - e-,')
Integration of eq 5 gives
- tydn
- exp( - g C ) }
The value of L M N is equal to Q N , both values being the number of ClzPOE molecules in unit volume of solution. Then, ( ~ L M N I Q Nin ) eq 12 is equal to q and eq 1 2 is simplified to
Substituting eq 4 for eq 3, we obtain
AVmio = kJ (1
(11)
+ constant = (1 -
- (dQ>j
i)n+
Comparing eq 1 with eq 1 3 Av,
constant
(6)
Here, the boundary conditions are that A m o = 0 at n = 0, and Avrnio 3 Av, as n where Av, is the saturation value of the upfield shift. Equation 6 then becomes
-
(8)
The variable n in eq 8 is the number of NaCsBS molecules per micelle, as defined above, and can be convertible to C (C = M A I M N , where MA and M N are the molar concentrations of NaCsBS and CEPOE, respectively). Then n can be related to M A by the equation
I
Av1.P
(14)
1 - e-@' = f (C)
(15)
1 AV = Avl-(l - e-*c), 1 < p P
(16)
Determination of Values of Av,, Avl, q, and 1. As expressed in eq 13, Av is a function of C. If h is taken as an arbitrary constant, then
(7) In the above calculation, only one micelle has been considered. However, this result, i.e., eq 7 , holds for all micelles in unit volume because the extent of the upfield shift is irrespective of the number of micelles, although the intensity of the signal is proportional to the number of micelles. Thus, AVmia in eq 7 is equal to the upfield shift observed, Av. Since 0 < { l - (q/Q)) < 1, the term ( 1 - (a/&)) in eq 7 can be expressed as 1 - (q/Q) = e-k' (IC' > 0) for convenience. Then, eq 7 is written in the form
=
Finally, we have
-f
Av = Avs(l
(13)
Av(C) = Av,(l - e-"') Av(C
+ h) = Av,(l
(17)
-
(18)
Eliminating e--Pc from eq 1 7 and 18, we obtain
+ h) = e-4n-Av(C) + (1 - e-gh)Avs (19) According to eq 19, the plot of Av(C + h) VS. A v ( c ) Av(C
should give us a straight line. The values of Au, and q will be then evaluated from the slope and intercept of this straight line. Figure 4 shows the Av(C -Ih) vs. Av(C) plots which were obtained from the Av vs. C curves shown in Figure 1 by taking h = 0.1. The values of Av, and q calculated from the straight lines in Figure 4 are given in Table I. The values of q and Av, are essentially independent of the extent of h since h is arbitrarily chosen so as to obtain an appropri~~
~
Table I : The Values of A v ~and p 9,
where N is the number of micelles in unit volume of solution and L is the Avogadro's number. Since C = M A / M N , the relation between n and C is then
Substituting eq 10 for eq 8, we obtain The Journal of Physical Chem&ry, Vol. 76, No. 23, 1971
molecules
Samples
ClzPOE-2.9 CIzPOE-4.0 ClrPOE-5.9 CizPOE-9.1 ClaPOE-11 ClzPOE-14 ClzPOE-24
15.9 14.3 13.5 12.0 10.9 8.5 4.6
1.6 1.2 1.1 0.62 0.58 0.54 0.65
3565
INTERACTION BETWEEN ANIONICAND NONIONIC SURFACTANTS
IO
I
I
I
5
0
I
I
10
AWCI, cps!
+
Figure 4. The A v ( C h ) us. A v ( C ) plots for C12POE-pwith p of 4.0 and 11: h = 0.1. Straight lines were also obtained for other CIzPOE with different p although they are not shown here.
+
ate plot of Av(C h) us. Av(C). (In the present case, h is taken as 0.1. Even if h is taken as 0.2 or others, the result is the same.) The value of Avs, i.e., the saturation value of the upfield shift, decreases with increasing p , and the value of q, i e . , the number of ClzPOE molecules influenced by one rjaCsBS molecule, also decreases with increasing p and becomes nearly constant for p larger than 9.1. The number, I , of the oxyethylene units on which the benzene rings of NaCsBS exert an influence may be estimated from the plot of Avs vs. l/p. This plot should be linear when 1 < p , according t o eq 14, and will deviate from the linear relation when 1 > p . Figure 5 shows the plot of Av, us. l/p. The value of 1 estimated
(l/p) x IO. Figure 5. The plot of Ava vs. l/p.
from the inflection of this plot is about 9, which seems somewhat larger than an expected value. However, this could be interpreted as showing that the polyoxyethylene chains in the mixed micelle of CuPOE and NaCsBS are being in motion, not always stretching straight into the water phase, and even the oxyethylene units remote from the hydrocarbon part of the micelle can approach to the benzene rings of NaCeBS molecules t o jnteract with each other. From the inflection of the Av, us. l / p plot, one can also estimate the value of A Y I ,L e . , the upfield shift of the signal for the polyoxyethylene chain with 1 oxyethylene units. The value estimated is about 12-13 cps.
Acknowledgments. The authors express their thanks to Dr. H. Rita, the Director of the Research Laboratories, for his encouragement and permission to publish this paper, and to Dr. Y. Inamoto for his help in preparing the samples.
The Journal of Physical Chemistry, Vol. 76, N o . $9,1971