Langmuir 1988, 4, 1346-1352
1346
I
3400
I
I
3100
2800
2500
2200 1900 WAVENUMBERS
1600
1300
I
1000
Figure 14. Experimental external reflection spectrum obtained under the same conditions as Figure 13.
1.20
0.90 OJh
" 0.
0
0
0.00
30.0 60.0 ANGLE OF INCIDENCE
90.0
Figure 15. Angular dependence of electric field intensity at the surface of the substrate for the case of external reflection: (a) (E?) of aimilicon, (b) ( E *) of airsilicon, (c) (E?) of air-PMMA, and (d) (E:) of air-PdMA. When PMMA is used as the substrate, the external reflection spectrum of an LB film shows other features. Figure 13 shows the simulated external reflection spectra of 40 layers of a CdA LB film on a PMMA substrate, obtained by using parallel and perpendicularly polarized lights. In this case, even when parallel polarized light is
used, only the weak negative carboxylate symmetric stretching mode (1430 cm-') is observed at 75O. This fact is due to the lower intensity of the z-component of the electric field on the PMMA substrate compared to that on the silicon substrate (Figure 15). When perpendicularly polarized light is used, the spectrum due to the LB film shows negative absorbances. In these simulated spectra, the denominator of the absorbance unit is the reflectivity of the bare PMMA plate, so that the absorbance of the substrate is theoretically compensated. However, as shown in Figure 13, residual absorbance of the substrate remains as a positive absorbance on the spectrum of the perpendicularly polarized light. The experimental spectrum (Figure 14) shows an spectrum identical with the simulation spectrum. Therefore, it can be concluded that the structure of LB film on PMMA substrate is identical with the structure of LB film on silicon and silver substrates. In the case of PMMA substrate, a t a lower angle than Brewster's angle there is no carboxylate symmetric stretching mode observed because the z-component of the electric field is less intensive than the y-component (Figure 15). Thus, in order to determine the anisotropy of an LB film on PMMA a higher angle of incidence than Brewster's angle is preferred. For the external reflection spectrum on a nonmetallic substrate, the intensity ratio of the carboxylate symmetric and antisymmetric stretching modes, which shows negative and positive direction of peak respectively, can be used as the index of the anisotropy.
Conclusion The use of spectral simulation according to the formula of Drude and Fresnel is quite successful in predicting the actual infrared spectra of LB monolayers on different substrates and under different experimental conditions. By comparing the simulated spectrum to the actual spectrum, the optical effect and the chemical effect on the spectral change can be distinguished. Thus, the structure of uniaxially oriented molecule on any substrate can be determined by using this spectral simulation. Acknowledgment. We gratefully acknowledge the partial financial support of the Bridgestone Corp. We also thank J. M. Tan, Case Western Reserve University, for providing LB films used. Registry No. CdA, 14923-81-0; PMMA, 9011-14-7; Si, 744021-3.
Water Dynamics in Microemulsion Droplets. A Nuclear Spin Relaxation Study Goran Carlstrom and Bertil Halle* Physical Chemistry 1, University of Lund, Chemical Center, P.O. Box 124, S-221 00 Lund, Sweden Received June 13, 1988 The state of water in aqueous microemulsion droplets in the system AOT/D20/isooctane has been investigated by 2Hand *'O NMR. Longitudinal relaxation rates are reported as a function of droplet size, droplet volume fraction, temperature, and resonance frequency. We conclude that the surface-induced perturbation of water rotation is of short range (limited to the primary hydration of the AOT head groups) and of modest magnitude (less than a factor of 10 slower rotation than in bulk water).
Introduction Water-in-oil microemulsions are thermodynamically stable systems which, under certain conditions, consist of 0743-7463/88/2404-1346$01.50/0
discrete aqueous droplets dispersed in a continuous oil medium. The phase behavior, molecular organization, and dynamics of such systems have been examined by a variety 0 1988 American Chemical Society
Water Dynamics in Microemulsions of experimental technique^.'-^ In the present work we focus on the water dynamics in the aqueous microemulsion droplets, in particular, on the reorientational motion of the individual water molecules. This property is of interest in connection with the use of microemulsions as media for chemical reactions. In addition, microemulsions constitute well-defined model systems suitable for studying the generic properties of interfacial water. Water dynamics in microemulsions has been studied mainly by three experimental methods: fluorescence dep ~ l a r i z a t i o n ,quasi-elastic ~~ neutron ~ c a t t e r i n g , ~and -~ nuclear spin relaxation.'O-13 The fluorescence technique monitors the reorientational motion of a large (perturbing) probe molecule solubilized within the aqueous droplet. Information about the state of water in the droplet is then deduced, somewhat indirectly, by introducing the concept of a local viscosity. However, since viscous dissipation is a highly nonlocal phenomenon, this concept does not seem to be a useful one. The other two methods are potentially more informative, as they directly probe single water molecule motion. Neutron-scattering data (interpreted in terms of an unrestricted isotropic translational diffusion model) have so far been reported only for very small microemulsion droplets (up to 16 molecules of water per surfactant). Consequently, it has not been possible to ascertain the range of the surface-induced perturbation of water dynamics. In nuclear spin relaxation studies of water one has a choice of three stable isotopes: the proton (lH), the deuteron (2H),and the oxygen-17 isotope (170).The 'H isotope is relaxed by intra- and intermolecular magnetic dipole-dipole couplings, while the other two isotopes are relaxed by, essentially intramolecular, electric quadrupole couplings. Consequently, the 2H and 170relaxation rates reflect the motion of single water molecules, whereas the 'H rate is also affected by relative water-water and water-surfactant motions. In the early 'H studies,'l this complication was not recognized. More recently,12 isotopic substitution experiments were used to separate out the water-surfactant contribution to the 'H rate. On account of the fast proton exchange in water, this approach cannot be used to separate the intra- and intermolecular contributions to the water lH relaxation rate. It is therefore not possible to extract the water reorientational correlation time unless simplifying assumptions are made about the water pair distribution and relative motion in the interfacial region. Clearly, the most direct means of probing water reorientational dynamics is provided by the quadrupolar nuclei 2H and 170.The relaxation of these nuclei has been used (1) Micellization, Solubilization and Microemulsions; Mittal, K. L., Ed.; Plenum: New York, 1977. (2) Surfactants in Solution: Mittal, K. L., Lindman, B., Eds.: Plenum: New York, 1984; Vol. 3. (3) Surfactants in Solution; Mittal, K. L., Bothorel, P., Eds.; Plenum: New York, 1986; Vol. 6. (4) Wong, M.; Thomas,J. K.; Gratzel, M. J.Am. Chem. SOC.1976,98, 2391. (5) Zinsli, P. E. J. Phys. Chem. 1979, 83, 3223. (6) Bardez, E.;Monnier, E.; Valeur, B. J. Colloid Interface Sci. 1986, 112,200.
(7) Tabony, J.; Llor, A.; Drifford, M. Colloid Polym. Sci. 1983, 261, 938. (8) Tabony, J. Chem. Phys. Lett. 1986, 113, 75. (9) Fletcher, P.D. I.; Robinson, B. H.; Tabony, J. J. Chem. SOC., Faraday Trans. 1 1986,82, 2311. (10)Hansen, J. R. J. Phys. Chem. 1974, 78, 256. (11)Wong, M.; Thomas, J. K.; Nowak, T. J. Am. Chem. SOC.1977,99, 4730. (12) Llor, A.;Rigny, P. J. Am. Chem. SOC.1986, 108, 7533. (13) Quist, P.-0.; Halle, B. J. Chem. SOC.,Faraday Trans. 1 1988,84, 1033.
Langmuir, Vol. 4, No. 6, 1988 1347 extensively in recent years to study water dynamics and ordering in a wide variety of heterogeneous aqueous solution systems. However, in the microemulsion field, very little advantage has so far been taken of the 2H and 170 isotopes in studies of water dynamics. In fact, we know of only two 2H studies. One of theselo was concerned with a microemulsion containing hexanol as cosurfactant. Because of the possibility of fast deuteron exchange between water and alcohol, information about water dynamics cannot be deduced directly from these data. In the other 2H study13this problem did not arise, but the focus there was on water dynamics and aggregate structure in the supercooled regime, where only a small amount of water (two to four water molecules per surfactant) remains unfrozen. The aim of the present work is to quantify the magnitude and, in particular, the range of the surface-induced perturbation of water reorientational dynamics in large microemulsion droplets (with radii of the order 100 A). We have thus measured the longitudinal spin relaxation rates of the 2H and 170nuclei in DzOin the three-component system AOT (sodium bis(2-ethylhexyl)sulfosuccinate)/ D,O/isooctane, which forms an oil-continuous microemulsion phase over wide ranges of concentration and temperature. The 2H and 170relaxation data, obtained under conditions of variable composition (droplet size and volume fraction), temperature, and resonance frequency, allow us to determine the magnitude and range of the perturbation. These results do not depend in any essential way on model assumptions that have not been independently checked.
Experimental Section Materials. AOT (sodium bis(2-ethylhexyl)sulfosuccinate) of 98% purity was obtained from Fluka. It was further purified as described by Kunieda and Shinoda,14except that active charcoal, rather than ether, was used to extract impurities from a solution of AOT in methanol. After evaporation the residual AOT was vacuum dried at 40 "C for 12 h. Isooctane (2,2,4trimethylpentane) from Ega-Chimie (99%) was used as supplied. The water used was a mixture of DzO (>99% 2H) from Norsk Hydro and 170enriched D 2 0 (22% 170, 62% l80)from Ventron. Microemulsionsamples were made by weighing the components into NMR tubes, which were then shaken until an isotropic, transparent solution was obtained. The droplet size was varied by successive additions of D260 to the samples. As a consequence, the oxygen isotope composition of the water in the samples varied somewhat (1-10% 170). The effect of this variation on the relaxation rates is insignificant?2 In order to eliminate contributions from scalar relaxation to the transverse relaxation rates (reported elsewhere), the pD of the added DzO was adjusted to 4.1. The volume fraction of droplets (AOT + water) was calculated by using the weighed sample composition, the molar masses (20.0, 114.2, 444.6 g mol-'), and the bulk densities (1.105, 0.692, 1.138 g ~ m - for ~ ) D20, isooctane, and AOT, respectively. As a reference for the relaxation measurements, we used a sample of acidified D 2 0 (1.5% 170) with pD 2.4. NMR Relaxation Measurements. Water 2H and 170relaxation rates were measured at different resonance frequencies by using the following instruments: a Bruker MSL 100 (%frequency 15.37 MHz, 170frequency 13.57 MHz), a home-built 6-T spectrometer (39.14 and 34.57 MHz), a Nicolet NM 360 spectrometer (55.54 and 49.05 MHz), and a Bruker CXP 100 equipped with a flux stabilizer HS 90 var (170frequencies in the range 2.0-12.0 MHz). The longitudinal relaxation rates (R,) were determined by the conventional inversion recovery method. Each R1value is the result of a least-squares fit of the magnetization plotted against delay time ( T ) for a t least 20 different T values. The frequency dependence of the longitudinal relaxation rates of 2H and 170was studied by using all the spectrometers listed (14) Kunieda, H.; Shinoda, K. J.Colloid Interface Sci. 1979, 70, 577.
1348 Langmuir, Vol. 4,No. 6, 1988
Carlstrom and Halle Table 1. Experimental Conditions UO; MHz X ~
series A B C D E
variable X X Y
T YO
T, "C 20 20 20 15.1-21.7 20
15.37/13.57 15.3711357 15.37113.57 15.37/13.57 2.0-49.OC
hb
Y
7.5-51.1 14.9-51.9 25.3 25.3, 50.8, 51.1 28.2
4.75 35.0 2.60-48.9 48.8, 35.0, 4.75 4.75
0.40-0.63 0.10-0.19 0.66-0.095 0.095, 0.18, 0.63 0.53
OResonance frequencies are given as U ~ ( * H ) / V ~bVolume ( ~ ~ O ) .fraction droplets (AOT + D20).cOnly 170data. above. The uncertainties in R1is estimated to 10% for the Bruker CXP 100 and to 5% for the Nicolet NM 360 and the home-built 6-T spectrometer. These experiments were performed at 20.0 f 0.5 "C, controlled by the passage of thermostated air. Measurements were carried out within a week from sample preparation on the Nicolet and the 6-T spectrometer but 4 months later on the Bruker CXP. Remeasurementa on the Nicolet showed no significant time dependence. Having established the frequency independence of R1 (6. Table I11 below), we used the Bruker MSL 100 to study the temperature and composition dependence. From the reproducibility of the measured relaxation rates for the reference water sample and for the microemulsion samples, using two different samples of the same composition, we estimate the uncertainty in R1 to be less than 3% for these measurements. The experimentswere carried out at temperatures ranging from 18.9 to 21.3 "C, measured with a calibrated thermocouple. All relaxation rata were corrected to 20.0 "C, using the temperature dependence data in Figure 4. These corrections were always less than 4%. During each experiment the probe temperature was kept constant to within 10.2 "C by the passage of thermostated air. The experiments were done within 2 days from sample preparation. For 170the longitudinal relaxation is strictly exponential only under conditions of extreme narrowing (Le., when R1 = R&. The relaxation is, however, effectively exponential as long as R2/Rl is of order 1, and approximate analytical expressions can then be used for the measured rates.16 The error in such a treatment can be evaluatedl6and is negligible in the present study.
Results The ternary phase diagram of the system AOT/H20/ isooctane has been partially determined16 a t 25 "C. In terms of the composition variables X = moles of H20/mole of AOT and Y = moles of isooctane/mole of AOT, the isotropic microemulsion phase (usually denoted L2) extends from the isooctane corner down to Y = 1-2 and from the isooctane-AOT axis up to X = 40-140 (depending on Y). The composition of our samples is given in the schematic phase diagram of Figure 1. (The indicated boundary of the stability region of the L2 phase is not precisely known; it depends sensitively on temperature and trace amounts of impurities. However, all our samples were transparent and optically isotropic and showed no sign of phase separation. In addition, for all samples, only one Lorentzian peak was obtained in the NMR spectrum, indicating a single phase.) Relaxation data from five experimental series will be reported (Table I). In series A and B we vary the size of the aqueous droplets, while in series C the droplet volume fraction is varied a t constant size. (The droplet size is determined essentially by X.19) Series D is a temperature dependence, and series E refers to measurements at variable resonance frequency. Only longitudinal spin relaxation rates will be reported here. We have also measured the transverse rates, which are somewhat larger. The interpretation of the transverse rates is more involved and requires consideration of several other dynamic processes besides water reorientation. The transverse relaxation data will therefore be discussed in a separate rep0rt.l' (15) Halle, B.; Wennerstrom, H. J. Magn. Reson. 1981, 44, 89.
(16) Tamamushi, B.; Watanabe, N. ColloidPolym. Sci. 1980,258, 174.
isooctane
Figure 1. Schematic extension of the isotropic microemulsion phase (L,) in the system AOT/D20/isooctane at 20 O C . Concentrations are expressed as percentage by weight. The compositions of the studied samples (seriesA, B, and C) are indicated (0).
I
0
1
2
3
4
5
I
6
1OOlX
Figure 2. Water 170longitudinal relaxation rate at 20 O C plotted against X-' = (moles of D20/moleof AOT)-' for series A (0) and The solid line is a linear least-squares fit of eq 11 series B (0). to the data with X > 25. The dashed curves are the predictions of e_q9 obtained by using XB _= 15 and the following: (a) Ala-= 2, RIB = 575 s-'; (b) X/a = 5, RIB = 500 s-'; and (c) X/(Y = 30, RIB = 380 5-l. For each choice of X/a, RIB was optimized to give the best fit for X > 40. Note that eq 9 presupposes that X 2 XB = 15. The insert shows the relaxation rate over a wider range of X-l values. The central experimental results are the variations of the 2Hand 1 7 0 longitudinal relaxation rates with X, as shown in Figures 2 and 3. For both nuclei, R1 is an essentially linear function of 1/X for the larger droplets, while for X 5 15 there is a clear deviation from linearity. (The significance of the curves in Figures 2 and 3 is explained in the following section.) Further, it is seen that R1 is determined by the composition variable X the results of the A and B series fall on the same curve. The inde(17) Carlstrom, G.; Halle, B., submitted for publication in J. Phys. Chem.
Langmuir, Vol. 4, No. 6, 1988 1349
6-
6.- 4-1'70
- 1.9
*H
, 5 r l -
- 1.7
6.2
- 1.5
I
9 6.0 a
-c
I
-
-
- 1.3
t
- 1.1
5.23
5.6
1
0
2
1
3
4
I
Figure 3. Water 2H longitudinal relaxation rate at 20 O C plotted and against X-I = (moles of D20/moleof AOT)-' for series A (0) The solid line is a linear least-squares fit of eq 11 series B (0). to the data with X > 25. The insert shows the relaxation rate over a wider range of X-I values. Table 11. Longitudinal Relaxation Rates at Variable Droplet Volume Fraction' R1(170),s-l 4D R1(2H),s-l Y
OX
0.66 0.42 0.26 0.16 0.095
4.83 4.83 4.83 4.82 4.91
475 475 475 480 480
= 25.3,T = 20 "C,uo = 15.37/13.57MHz for 2H/170.
Table 111. Longitudinal l ' 0 Relaxation Rate at Variable Resonance Frequency' YO, MHz R p O ) ,s-l YO, MHz R1(170), s-l 2.03 6.18 12.2 OX
442 447 450
13.57 34.57 49.05
441 473 454
= 28.2, Y = 4.75,T = 20 "C.
pendence on Y over an even larger range is documented in Table 11, which shows the relaxation data from the C series. (In contrast, the transverse relaxation rates depend on Y as well as on X . 9 Since the droplet size is determined by X,1-3we may thus conclude that R1 depends on droplet size but not on the droplet volume fraction. The temperature dependence of the relaxation rates in the range 15-22 OC is shown in Figure 4 for three compositions. The Arrhenius plots yield an apparent activation energy of 25 kJ mol-' for all samples (within the experimental uncertainty). For comparison, the activation energy for the 2H and 1 7 0 relaxation rates in bulk D20 within this temperature range 21 kJ mol-l. The observed temperature dependence is consistent with a fast-exchange situation, where the observed relaxation rates are spatial averages over all local environments experienced by the D20 molecules in the microemulsion (e.g., interfacial region, droplet core, and droplets of different size). This conclusion is also consistent with the observation that all line shapes are Lorentzian. Within the experimental uncertainty, the longitudinal relaxation rates are independent of the resonance frequency in the range 2-49 MHz (Table 111). (This is also true for the transverse rates.17) Consequently, the longi(18)Hindman, J. C.; Zielen, A. J.; Svirmickes, A.; Wood, M. J. Chem. Phys. 1971,54,621. (19) Lang, E. W.; Liidemann, H.-D. Ber. Bonsenges. Phys. Chem. 1981, 85, 603.
3.40
3.45
6
5
roo/x
2.60 7.22 14.4 27.7 48.9
13.35
3.35 IO~K/T
3.40
3.45
Figure 4. Temperature dependence of the I7O (13.57MHz) and the 2H (15.37 MHz) longitudinal relaxation rates in microemulsions of different compositions: X = 25.3,Y = 48.9 (@; X = 51.1, Y = 4.75 (0); and X = 50.8, Y = 35.0 (0). tudinal relaxation is not affected by any dynamic processes with correlation times of the order lo4 s or longer. Since the local reorientational motion has a correlation time of about 10-l' s (cf. next section) and averages out all but 1-2% of the quadrupolar interaction,17it followsmthat the longitudinal relaxation is affected only by the local water reorientation (since (10-2)2X lo4 X/a over the entire size distribution. We now consider the experimental R,(X) data in Figures 2 and 3. The onset of nonlinearity (see inserts) suggests a value for XB of about 15. The stronger than linear 1/X dependence for X 5 15 indicates, not unexpectedly, that the local relaxation rate varies within the interfacial region (or that the interfacial environment varies with droplet size for x 5 15). From the high degree of linearity in the range
I
60
Figure 5. (R, - R1")-'for 170plotted against the composition and series B (0). Rlm was variable X at 20 " C for series A (0) taken from the intercept of the solid line in Figure 2.
with a = 1-2 Combining eq 5-8 and performing the integration, we obtain
Rl(x) = R1F
50
(21) Woessner, D. E.
J. Magn. Reson. 1980, 39, 297.
Water Dynamics in Microemulsions
Langmuir, Vol. 4, No. 6,1988 1351
has been reported22thus cannot be detected in our experiments. However, it has also been suggested23that a fraction ( p M ) of the AOT is present in the form of small droplets (reversed micelles), which presumably contain a small amount of water (xM moles of D20/mole AOT). Since the reversed micelles do not contain any bulklike water (xM xF. However, the nearby presence of ionic solutes tends to decrease the qcc's and may nearly cancel the effect of reduced hydrogen b ~ n d i n g . ' ~Further, the subpicosecond averaging of the qcc's will also be affected to some extent by the altered molecular environment in the interfacial region. Although there is some uncertainty as regards xB, the available evidence suggests that the R~B/R~ ratio F is determined mainly by the correlation times. In any case, we can safely conclude that water rotation in the interfacial region is less than one order of magnitude slower than in the bulk liquid. This conclusion applies to water droplets with x > xB; in small reversed micelles with x