J. PhyS. Chem. 1982,86,3032-3038
3032
This inner-doublet Fe2+may be present as isolated cations on the support or it may be clustered into rafta on the silica near the particles of outer-doublet Fe2+. As the severity of reduction increases, outer-doublet Fe2+is converted into inner-doublet Fe2+,and metallic iron particles (represented by the black region) may be formed during severe reduction treatments such as prolonged exposure to hydrogen at 700 K.
Conclusions The combination of Mossbauer spectroscopy, infrared spectroscopy, and volumetric/gravimetric adsorption measurements suggest that two states of Fe2+exist in reduced Fe/Si02 samples: Fe2+strongly interacting with the support and Fe2+ in small particles of iron oxide. The former cations are of low (e.g., fourfold) coordination, giving rise to the inner doublet in the Mossbauer spectra and to bands at 1910, 1810, and 1750 cm-’ in the IR spectrum of adsorbed NO. The bands a t 1910 and 1810 cm-’ are due to dinitrosyl species, and the band a t 1750 cm-’ is due to mononitrosyl species on these iron cations of low coordination. It is suggested that these iron cations are present as thin “rafts” on the silica support (although this does not exclude the possibility that some of the inner-doublet Fe2+ may exist as isolated cations on the support). All of the cations in these rafts are capable of adsorbing NO; however, for steric reasons, dinitrosyl species are formed at the edges and mononitrosyl species are formed on the faces of these rafts. The Fe2+ in the small particles of iron oxide are of high (e.g., sixfold) co-
ordination, giving rise to the outer doublet in the Mossbauer spectra and the band at 1830 cm-’ in the IR spectrum of adsorbed NO. Cations on the surface of these particles are capable of adsorbing NO, while those cations beneath the surface do not participate in adsorption. These particles are also in intimate contact with the silica support in order to explain their stability against reduction to the metallic state during hydrogen reduction. In competition with this reduction to metallic iron, these iron oxide particles are converted into the Fe2+rafts in strong interaction with the support. Consequently,the dispersion of the iron oxide particles increases, and a greater fraction of the iron cations remaining in high coordination are capable of adsorbing NO. This increased dispersion, coupled with the production of Fe2+of low coordination in strong interaction with the support, causes the NO adsorption uptake of Fe/Si02 to increase with increased reduction severity.
Acknowledgment. We acknowledge stimulating discussions with R. F. Howe (University of Wisconsin-Milwaukee) and B. S. Clausen (Haldor Topsoe Research Laboratories) regarding nitrosyl complexes and Fe/Si02, respectively. In addition, we thank Terrence Udovic (University of Wisconsin-Madison) for assistance in computer subtraction of IR spectra. This work was supported, in part, by the National Science Foundation through grant ENG-7911130, for which we are grateful. Finally, J.A.D. acknowledges the cooperation and hospitality of the Haldor Topsoe Research Laboratories.
Formation and Properties of Large Aggreqates in Concentrated Aqueous Solutions of Ionic Detergents E. Lemner’ and J. Frahm Mex-plsnck-InsMM fuer blqhyskalbche Chemk 03400 Qoefflngen, West Germny (Received: December 15, 198 1;
I n Finel Form: March 24, 1982)
In the first part of this paper kinetic experiments on concentrated aqueous solutions of ionic detergents are presented. At high concentrations the relaxation amplitudes show a distinct maximum, in quantitative contradiction to the theory. It is suggested that the sudden increase of the amplitudes is attributed to the onset of formation of large nonspherical aggregates. This suggestion is supported by comparison with measurements of the concentration dependence of the mean aggregation number m by other authors, which show a sudden increase of m at the same counterion concentration. In the second part of this paper ‘HNMR experiments on the same systems are presented. Again, the line widths show a sudden increase at the same counterion concentration. The TIand T2relaxation time profiles show three distinct regions: the monomer region, the region of ‘spherical” micelles, and that of large aggregates. The analysis of the profiles leads to the conclusion that for spherical micelles any motions other than rapid local conformation changes must be rather restricted. For the large aggregatea the sudden change of the T2profiles indicates that the rapid motions are superimposed by a slower motion, possibly of several segments of the chains or even of whole molecules. I. Introduction In a recently published paper1 we have suggested that ionic micelles may be considered as charged colloidal particles. At lower counterion concentration these particles are stable with respect to coagulation due to the repulsive electrostatic forces. Consequently, they grow by stepwise incorporation of monomers according to the reaction (1) Leeener, E.; Teubner, M.; Kahlweit, M. J. Phys. Chem. 1981,86 1529, 3167. See also: Kahlweit, M. Pure Appl. Chem. 1981,53, 2069.
0022-3654/82/2086-3032$01.25/0
Ni-1 + N1+ Ni (1.1) With increasing counterion concentration, however, the electric double layer around each particle becomes increasingly compressed, so that the attractive dispersion forces can lead to a reversible coagulation reaction Nk + Ni* Ni k+1=i (1.2) where k and 1 are classes of submicellar aggregates, i.e., of aggregates on the leftrhand side of the minimum of the size distribution, whereas i is a class of the proper micelles, Le., 0 1982 American Chemical Society
The Journal of phvsical Ctmmbtty, Vol. 86, No. 15, 1982 3033
Formation and Properties of Large Aggregates
/--.,-
TABLE I: Fitted Parametem for CsDS Amplitude A,
~
l o 4x
(a In bar-
105x (a 1n ifilap)T, bar-
1.7 1.2 1.0
-4.2 -3.2 -2.2
NTol
temp, "C
cmc, mol m-'
h
03/%
25 35 45
6.4 6.6 7.65
0.27 0.31 0.41
1.6 0.8 0.6
Time Constant 1/7$1(Reaction Path 1.1) temp, "C 25 35 45
M ,s-'
P 0.54 0.59 0.34
68 720 3750
Time Constant 1
4 2.12 2.38 2.03
1 (Reaction ~ ~ ~ Path 1.2) Po(mlo)2:
temp, OC ego, mol m-3 I
I
0
100
200
300
400 L
500
600
700
800
25 35 45
100
110 120
s-'
S
12 94 715
5.2 5.1 5.0
cCsm [mol m-']
1. Relative ampmude A (above) and inverse time constant 1/r2 (below) of the slow relaxation process for H,O-CsDS at different temperatures. Full lines: fit of the theory.
of aggregates on the right-hand side of the minimum. The consequences of this model are in satisfactory agreement with kinetic experiments on aqueous micellar solutions. The question then arises whether or not even proper ionic micelles may coagulate at very high counterion concentrations, either with each other or with submicellar aggregates, to form "giant- micelles which have been visualized by other authors as cylindem2 In order to answer this question we have extended our pressure-jump experiments to very high detergent concentrations. From the analysis of the relaxation amplitudes it follows that the amplitudes first decrease with increasing detergent concentration to become rather small within a limited concentration range but then increase again with further increasing concentration to approach an apparently constant value which is proportional to (a In rii/dp)n where m denotes the mean aggregation number of the proper micelles. Consequently, as long as (dnlldp), is different from zero, the relaxation experiments can be extended to high concentrations. 11. H20-CsDS Systemz4
The first set of experiments was performed on the system H20-CsDS (cesium dodecyl sulfate). The results are shown in Figure 1. Plotted is the relative amplitude A2 (above) and the corresponding inverse time constant 1/r2 (below) of the slow relaxation process vs. the detergent concentration. The full lines represent the fit of the theory' to the experimental results at low and intermediate concentrations. For the mean effective degree of dissociation of the proper micelles, we set a, = 0.3. The mean aggregation number of CsDS micelles a t the cmc at 25 OC was determined as f i = 78.5 Table I shows the fitted parameters. If the fitted parameters for NaDS at 25 "C (Table I in ref 1)are compared, one finds that the parameters are rather (2) Misael, P.J.; Mazer,N. A.; Benedek, G. B.; Young,C. Y. J.Phys. Chem. 1980,84,1044.
(3) Klar, G., unpubliahed measurements.
similar, except for M and Po, which are 2 orders of magnitude smaller than the figures for NaDS. We recall that both parameters include quantities which characterize the rate constants and the concentrations of the aggregates near the minimum of the size distribution. We further note that the flocculation value cgois lower for CsDS than for NaDS, which agrees with the order for flocculation values of dispersion colloids. Finally, we again find that the temperature dependence of ego. is higher than that of the slope s, which agrees qualitatively with the predictions of the DLVO theory. We thus find that at low and intermediate concentrations the kinetic theory fib the experimentalresults quite well. At high concentrations, however, Figure 1shows a deviation: the amplitudes rise, pass through a maximum, and decrease again. Since the level of the plateau at intermediate concentrations is a measure for (e In nl/dp)T, this result indicates that at these detergent concentrations new species appear, the properties of which differ from those of the proper micelles at lower concentrations. As one can further see, the maximum of A2 is shifted toward higher concentrations and decreases as one raises the temperature. The concentration dependence of the inverse time constant, on the other hand, shows no comparable deviation from the expected behavior. We recall, that the concentration dependence of the amplitudes is independent of the reaction path, along which the micelles are formed, depending only on the change of their equilibrium properties with pressure or temperature, respectively. The concentration dependence of the time constant, on the other hand, is determined by the reaction path. The decrease of 1 / at~ low ~ detergent concentrations, i.e., low counterion concentrations, is a consequence of reaction 1.1. At the minimum of the l / r 2curve, reaction path 1.2 takes over. With further increasing concentration, the coagulation reaction rate becomes increasingly faster, finally approaching that of rapid coagulation. From then on 1 / r 2 remains practically constant. Since the size of submicellar aggregates and that of proper (spherical)micelles should differ only little, the rate constants for rapid coagulation of submicellar aggregates should not differ significantly from those for the rapid coagulation of proper micelles. One would, therefore, ex-
3034
The Journal of Fhysicai Chemisby, Vol. 88, No. 15, 1982 50
Lessner and Frahm 25
I
I
I
20
t
1
I
10 iN
01
L
2000
1
0
200
0
50
600
800
1000
150
200
250
I 25
,e
1000
IE
500
t ::: 50
'
50
- c
J
100
200
NaDS + cNaC,
500
1000
[mot m 9
Figuo 2. Reletlve amplitude A (above) and mean aggregatbn number f i (below) for H20-NaDS-NaCI at 25 OC: (0)ref 4; (0)ref 5.
pect that the appearance of large micelles-with different equilibrium properties-manifests itself mainly in the concentration dependence of the amplitudes, whereas the corresponding time constants should show no significant change. In view of recent studies by other authors4i5who have found a sudden increase of the mean aggregation number m of NaDS micelles after adding a sufficient amount of NaCl to a diluted NaDS solution, this raised the question of whether or not the increase of the relaxation amplitudes is correlated to the onset of formation of large aggregates. Since corresponding experiments on CsDS are not reported in the literature, we tried to answer the question for the system H20-NaDS-NaCl. 111. H,O-NaDS-NaCl System While Mazer and co-workers4 performed their experiments with 69 mol m-3 NaDS, Ikeda and co-workers5 performed their experiments at the cmc, both adding NaCl. If plotted vs. the sum of detergent and salt concentration (c + cJ, both sets of experiments show that the apparent molecular weight increases rather weakly at low and intermediate concentrations, to rise steeply a t about c + c, = 300 mol m-3. To obtain the same counterion concentration with NaDS alone, one would have to apply about lo00 mol m-3 NaDS, which is difficult because of its limited solubility. We, therefore, repeated the experiments of Mazer and co-workers. Instead of measuring the molecular weight of the aggregates, however, we performed relaxation experiments using the same NaDS concentration (69 mol m-3) and then added NaC1. At c = 69 mol m-3 and c, = 0 the amplitude of NaDS has almost reached the plateau (see Figure 1 in ref 1). Since our experiments, reported in ref 1,have shown that the addition of counterions leads to a shift of the curves toward lower detergent concentrations, we accordingly expected that the addition of a sufficient amount of NaCl would lead to a shift of the maximum of (4).Mazer, N. A.; Carey, M. C.; Benedek, G . B. In "Mizellization, Solubihtion and Microemulsions";Mittal, K. L., Ed.; Plenum Press: New York, 1977; Vol. I; p 359. (5) Hayashi, S.;Ikeda, S.J . Phys. Chem. 1980,84, 744.
100
ccs+ [ m o l m - 3 ~
Flgure 3. Relative amplitude A , for H,O-CsDS-CsCI at 35 OC: (above) vs. CsDS concentration with CsCl concentration as a param eter, (below) vs. counterion concentration; CsCl = (+) 0, (0) 50, and (A)100 mol m-'.
the amplitude from 1000 to 69 mol m-3 NaDS. The experiment confirmed this prediction; Figure 2 shows the result. In the upper part we have plotted A2 vs. c + c, on a double logarithmic scale; the lower part shows the corresponding plot for m.476 As one can see, the amplitudes increase rather weakly at intermediate concentrations, to rise steeply at about c + c, = 300 mol m-3, in evident correlation with the mean aggregation number. We thus suggest that the increase of the amplitudes at high counterion concentration is attributed to the formation of large aggregates. If this interpretation is correct, the measurement of the amplitudes could be used as a rather simple and sensitive detection method for the onset of the formation of these aggregates. The reason for the fact that the amplitudes pass a maximum to decrease again, however, cannot be clarified by kinetic experiments.
IV. H20-CsDS-CsCl System In order to study the influence of counterion concentration on the formation of large aggregates, we have measured the amplitudes as a function of CsDS concentration at 35 "C, adding different amounts of CsCl. The upper part of Figure 3 shows the result; as expected, the maximum of the amplitudes is shifted toward lower detergent concentrations with increasing CsCl concentration. This suggested that we should plot the curves again vs. the total counterion concentration cg instead of detergent concentration (lower part of Figure 3). As one can see, the three curves coincide. This again indicates that the equilibrium and kinetic properties of ionic micellar solutions appear to be determined by the total counterion concentration rather than by the concentration of the detergent. For CsDS solutions we thus find that the formation of large aggregates starts at cg i= 100 mol m-3, compared to cg = 300 mol m-3 for NaDS solutions (Figure 2). We note that in both systems these concentrations are close to the corresponding flocculation values, i.e., close to the onset of rapid coagulation of the proper micelles.
Formation and Properties of Large Aggregates
V. NMR Measurements2s In order to decide whether the coagulated micelles undergo a conformational change or remain as flocculates, we have determined the mobility of the different groups of the detergent molecules by NMR measurements. For this purpose we have studied the widths of the different lines as well as their spin-lattice relaxation times Tl and their spin-spin relaxation times TP This was done by lH NMR, which permits one to obtain high S/N ratios in rather short measuring times and, furthermore, suffices for a qualitative discussion of relative changes of the mobilities. Its disadvantage, compared with 'Y!NMR,- lies in the fact that it only permits the resolution of the resonances of four groups, namely, the a-CH2,8-CH2,(CH,),, and w C H ~groups. The 'H NMR measurements were performed at 270 MHz on a Bruker WH 270 pulse spectrometer in the Fourier transform mode using a deuterium lock. The line ) taken from the spectra as full width widths ( A V ~ , ~were at half-maximum intensity. The spin-lattice relaxation times T1 of individual lines were measured by means of the inversion-recovery sequence 180"-~-90"acquisition using nonselective pulses. The spin-spin relaxation times T2were measured by means of Carr-Purcell-Meiboomacquisition with Gill (CPMG) sequences 90"-(~-180"-~), a phase cycling of the 180" pulse. Fourier transformation of the second half of the nth echo then yields a normal high-resolution spectrum, the peak intensities of which are partially relaxed as exp(-2n~/T2). The pulse spacing was in the range of 1-4 ms. This prevented J modulation of the echo train due to spin-coupling effects. Moreover, a nonspinning sample was used to avoid modulation by sample rotation. Under these experimental conditions a dependence of Tzon the pulse spacing was not observed, indicating the absence of contributions to spin-spin relaxation from either diffusion or kinetic^.^ In most cases the calculated T2values were longer by a factor of 2-10 than the value T2* = (7rAv1/2)-1 derived from the line width. Only for very short relaxation times of single resonance peaks did both values tend to become equal. In general, strictly exponentially decaying peak heights were found in all Tl and T2measurements. In both types of experiments a delay of at least 6T1 was chosen before accumulating further spectra ?'ypically, five to eight T values were measured to establish a single Tl or T2value. The accuracy of the relaxation times was about 5%;the temperature was constant within 0.5 K. Deuterium oxide (99.75% Uvasol) was purchased from Merck. VI. D20-NaDS-NaCl System In the first set of experiments we determined the line width as well as the T1 and T2 profiles for the system studied by Mazer et aL4composed of 69 mol m-3 NaDS and NaC1. Figure 4 shows the line widths of the a-CHz and w-CH3groups vs. counterion concentration cNa+. In evident correlation with Figure 2 the line widths start to increase a t about c + c, = 300 mol m-3. The line widths of the &CH2 and (CH,), groups show the same features. Figure 5 shows the T1 and T2 profiles for different concentrations of NaCl. Using the individual relaxation times for the four groups, we constructed approximate molecule profiles. Apparently, one can distinguish between (6) Williams, E.; Sears, B.; AUerhand, A.; Cordes, E. H. J.Am. Chem. SOC.1973,95,4871. (7) Henriksson, U.; Odberg, L. Colloid Polym. Sci. 1976, 254, 35. (8)WennerstrBm, H.; Lindman, B.; SMerman, 0.; Drakenberg, T.; Rosenholm, J. B. J. Am. Chem. SOC.1979,101,6860. (9) Frahm, J. J. Magn. Reson. 1982, 47, 209.
a
'
t
0
100 200
300 Loo
500 600 700 800 900 low cNa+[mol m?I
--c
(e)for D20-
Flgure 4. 'H NMR line wktths of aCH, (U) and A H , NaDS-NaCI at 25 "C. 5-
r
t :
a-CH, p-CH, -lCH2)n- w-CH,
0.1 c
-
005 1
001
.
0 005
1
i 55NaDS
t 0001
1
monomers
69NaDS
69NaDS UONaCI micelles ~
69NaDS 69NaDS 560NaCl 910NaCl large aggregates
_____~_
~~~
f1
-
Flgure 5. 'H NMR T i and T 2 relaxation time profiles for D,O-NaDSNaCl at 25 "C. Concentrations in mol m4.
three regions. As one increases the detergent concentration from below (5.5 mol m-3) to above the cmc (69 mol mS), the mobilities of all groups decrease by a factor of about 2. If one then adds NaC1, the level of the Tl values remains essentially unchanged, except for the fact that the mobilities of the a-CH2and o-CH3 groups decrease. The T2 values, on the other hand, show a second distinct decrease, as one proceeds from 300 to 500 mol m-3 NaC1, in correlation with the increase of the line widths in Figure 4. Before discussing these resulta in more detail, we shall first present the corresponding results for CsDS. VII. D20-CsDS System Figure 6 shows the line widths of the a-CHz and o-CH3 groups vs. detergent concentration at 25 and 65 "C. The curves show the same features as the corresponding curves for NaDS (Figure 4). Again the line widths (at 25 "C) start to increase a t about the same concentration as that at which the relaxation amplitudes do (Figure l ) , and again the increase is shifted toward higher concentrations with rising temperature. At high concentrations, one observes a Gaussian line shape due to direct dipolar couplings not averaged to zero (apparent line width 1 1 kHz), which indicates the formation of liquid crystals in the system.1° If compared with the concentration dependence of the amplitudes (Figure 11, this suggests that the decrease of the amplitudes at high concentrations is caused by the appearance of patches of liquid crystals in the solution, until at very high concen(10) Wenneretr6m, H. Chem. Phys. Lett. 1973, 18, 41.
3036 The Journal of Physical Chemistry, Vol. 88, No. 15, 1982
Lessner and Frahm 100
200
-
100 :
-~ ----
-B-
2’ /
s I
50
,.--
Y
/
,:20
,
4
A*’
/ 0
i
0-
8’
-
0
100
200 300 bo0
500 600 700 800 900 1000
. )
1 -
0
100
200 300 LOO
500 600 700 800 900 1000
Flgwe 6. ’H NMR line widths of aCH, (squares) and w-CH, (circles) for D,O-CsDS at 25 O C (full points) and 65 OC (empty points). 5-
2
c-
i
I
a-CH,
01
-p-CH, -(CH2b- w-CH,
0 05
-A
J
0 01
r 25.C
___->
0 005 _1
000,
LCsDS monomers
110 C s D S 23OCsOS micelles
160CsDS
580CsDS
I
---------large -aggregates -----
FIgm 7. ‘HNMR TI (at 25,45, and 65 OC) and T , (at 25 OC) profiles for D,O-CsDS. Concentrations In mol m-,.
trations the entire system becomes anisotropic. Figure 7 shows the Tl and Tzprofiles. Also shown is the temperature dependence of the spin-lattice relaxation time TIfor the micellar systems. As one can see, the profiles show essentially the same features as in the D20-NaDSNaCl system. VIII. D20-CsDS-CsCl System The experiments were performed at 56 mol m-3 CsDS, CsCl being added. Figure 8 shows the line widths of the a-CHz and w-CH, groups vs. counterion concentration at two different temperatures. The curves are qualitatively similar to those for the pure CsDS system but differ insofar as the plateau is more pronounced than in the system without salt (Figure 6), ita level, however, being only about half as high. Furthermore, the spectra show no indication of the formation of liquid crystals a t high counterion concentration. The T2profiles (not shown), as well, show qualitatively the same features as in the D20-CsDS system, with the exception of the T2 value of the a-CH2 group at high counterion concentrations, which is about twice as high as that for the system without salt, corresponding to the lower plateau level in Figure 8. IX. Discussion The increase of the line widths with increasing counterion concentration has been interpreted as being caused
ccs+ [mol m-’]
Flgwe 8. ’H NMR line widths of aCH, (squares) and w-CH, (circles) for D,O-CsDS-CsCI at 25 OC (full points) and 65 OC (empty points).
by the formation of large aggregates.”-13 There seems, however, to be a difference between the CsDS systems with and without CsCl (compare Figures 6 and 8): in the system without salt the aggregates appear to continue to grow with increasing detergent concentration, until patches of liquid crystals are formed and the system, finally, becomes anisotropic (-750 mol m-3 at 25 “C). In the CsDS system with constant detergent concentration, on the other hand, the plateau which is reached at high concentrations of the added salt can be explained in different ways: e.g., Porte et al.14 found a similar dependence of the NMR line width for cetylpyridinium bromide upon addition of N&r, whereas light-scattering measurements still show the micelles to increase in size. Provided that the motions relevant for NMR T2relaxation in these systems (for a more detailed discussion see below) can be related to the size or length of the aggregates, e.g., as by overall tumbling, then this finding indicates the existence of a critical or persistence length for the elongated cylindrical aggregates. Its value can be defined as the minimum length separating two subunits of a micelle with negligible correlation of motion. Thus, the appearance of a plateau region parallels the onset of a finite “resistance of bending” of the elongated micelles. Another explanation, however, could be that the aggregates do not continue to grow with increasing counterion concentration because of the limited detergent concentration but instead reach a rather stable configuration. In fact, such a behavior has been reported by Porte et al.,14 when the bromide counterions were exchanged for chloride counterions. The authors deduced from QLS measurements that the mean hydrodynamic radius (RH= 30 A) of CTACl remains unchanged when NaCl is added in the concentration range from 100 to 2000 mol m-3. Unfortunately, corresponding light-scattering data for the CsDSCsCl system are not available, so that a clear decision cannot be drawn from the line width measurements alone. Both effects would also explain the dependence of the apparent activation energies of the line width on salt concentration (Figure 8): at low counterion concentrations, i.e., in the range of spherical micelles, one finds for the a-CH2group an activation energy of about 10 kJ mol-’, in the transition region (ccB+= 130 mol mS) one of about 95 kJ mol-’, and at high counterion concentration (ccs+ = 600 mol m-3) again only about 20 kJ mol-’. A similar high value (85 kJ m o l 9 is found for the relaxation amplitudes in the transition region. We attribute the low activation energies (11) Ulmius, J.; Wennerstrom,H. J. Magn. Reson. 1977, 28, 309. (12) Staples, E. J.; Tiddy, G. J. T.J . Chem. Soc., Faraday Trans. 1 1978, 2530. (13) Nery, H.; Marchal, J. P.; Canet, D.; Cases, J. M. J. Colloid Znterface Sci. 1980, 77, 174. (14) Porte, G.; Appell, J.; Poggl, Y. J. Phys. Chem. 1980, 84, 3105.
Formation and Properties of Large Aggregates
to intramolecular motions in the micelles, being relevant for the T zrelaxation, but independent of overall properties, while the high activation energies in the transition region seem to directly reflect the formation of large aggregates. The Tl relaxation times increase with rising temperature (Figure 7). From this it follows that the motions, responsible for the spin-lattice relaxation processes, must have correlation times in the "extreme narrowing" regime (7,2wO2 1 relevant for T2relaxation), and (iii) an overall tumbling process of the aggregates which only partially averages the static dipole-dipole couplings to be responsible for the band shape in the associated liquid crystalline phase. It is interesting to note that their fitted T2values for large CTAB micelles are rather close to the spin-spin relaxation times reported in this paper. We would like to emphasize again that our relaxation times were obtained (20) Frahm, J. , in preparation. (21) Quantitative evaluations of the T2data are complicated by the fact that the characteristicproperties of the motions will greatly influence the calculations. The assumption of axial symmetric or isotropic motions
would yield T~ == lo-* 8.
3038
The \owns1of phvsical Chemistty, Vol. 86,No. 15. 1982
T
cmc I
micelles I I
t Water
Detergent Salt
Water
Detergent
Flgure 9. Schematic phase dlagram of the binary system H,O-ionic detergent (above) and of the ternary system H,O-ionic detergentelectrolyte (below).
independently from resonance peak heights by means of multiple-pulse sequences. Therefore, the T2values represent well-defined relaxation times reflecting intramolecular correlated motions, in contrast to values derived from line shapes. However, except for small effects close to the base line, we were unable to detect deviations of the signals from Lorentzian line shapes (at 270 MHz). Since there are a variety of further reasons for such deviations, as, e.g., chemical shift dispersion of the (CH,), resonance peak, overlap of broad lines with neighboring resonances (P-CH,), unresolved spin-spin splittings (a-CH2,w-CH2), or, finally, polydispersity of the structures, it is obvious that anisotropic contributions cannot be detected unambigiously in this system. Small contributions, however, cannot be ruled out. As one can see (Figures 5 and 7), the mobility of the a-CH2 and &CH2 groups appears to approach the same level as the one for the central groups, whereas the w-CH3 tail groups appear to retain their mobility.22 This would fit with the model of a rod-shaped aggregate, in which the packing density of the tail groups should remain about the same as in spherical micelles, whereas the mean area per (22) From the differences in the TIprofiles for the large aggregates, one may conclude that spin diffusion to a heat sink is not very important in these systems. For the liquid crystalline D,O-CsDS system (0.7 M, 298 K) identical spin-lattice relaxation times are obtained for all resonances, the values of which are still very close to those of the large aggregates.
Lessner and Frahm
head group should decrease. We note that the T I profiles of the large aggregates show a strong similarity to the profiles of the order parameters23 and the spin-lattice relaxation timed8 in phospholipid bilayers and vesicles. As, with rising temperature, these large aggregates are destroyed, one again observes the tube-shaped profiles typical for spherical micelles. The T2profiles show a similar change as one proceeds from spherical micelles to large aggregates: at low counterion concentrations, the mobilities of the head groups are similar to those of the central groups. As the large aggregates are formed, the mobility of these head groups decreases markedly, leading to a steep mobility gradient from the head groups to the CH3 tail groups. As already mentioned above, there seems to be a difference between pure D20-detergent solutions and D20detergent-salt solutions. As one can see from Figures 5 and 7, the ratio T2(o-CH3)/T2(a-CH2)is about 10 for 500 mol m-3 CsDS (without CsCl), but only about 4.5 for 69 mol m-3 NaDS plus 800 mol m-3 NaCl. This difference is apparently due to the fact that the T2 values, i.e., the mobilities of the head group, are about twice as high in the detergent-salt systems as in the systems with pure detergent, in accordance with the behavior of the linewidth (see Figures 6 and 8).
IX. Conclusion In kinetic experiments in ionic micellar solutions with very high counterion concentrations, one finds a sudden increase of the relaxation amplitudes, which indicates the formation of species the properties of which differ from those of spherical micelles at low and intermediate counterion concentrations. By comparison with aggregation number measurements by other authors and with NMR measurements presented in this paper, we suggest that these species are large compact aggregates, possibly rod shaped. A rise in temperature leads to a shift of the critical concentration for the formation of these large aggregates toward higher detergent concentrations, whereas the addition of a salt with the same counterions as the detergent molecules leads to a shift toward lower detergent concentrations. For the binary system H20-ionic detergent and the temary system H20-ionic detergent-salt we, therefore, suggest phase diagrams as they are schematically shown in Figure 9. The NMR measurements further indicate that in pure D20-detergent systems the aggregates increase in size with increasing detergent concentration, until they form patches with ordered structures (liquid crystals).
Acknowledgment. We are indebted to Professor M. Kahlweit for suggesting the problem and supporting the work. (23) Seelig, A.; Seelig, J. Biochemistry 1974, 13, 4839. (24) The experimentspresented in sections 11-IV were performed by
E.L.
(25)
J.F.
The experiments presented in sections V-IX were performed by
