Conformation of the Hydrocarbon Chains of Sodium Dodecyl Sulfate

According to Figure 1 the amount of ESR-active Fe varies in the following order: no ... By placing alkane chains on a three-dimensional have been post...
0 downloads 0 Views 724KB Size
J. Phys. Chem. 1989, 93, 2053-2058 agreement with Derouane,I6 that the signals at g = 2.3 and 2.0 can be removed by chemical treatment almost quantitatively. This is not the case for the g = 4.3 signal. However, at the small impurity level of the present samples it is impossible to remove Fe from the zeolite. At large impurity levels (1 750 ppm) Fe is removed by repetitive application of the extraction technique.I6 Extrapolation of Derouane’s results to impurity levels below 500 ppm indicates that extraction of Fe is indeed very difficult if not impossible, in agreement with our results. What happens instead is that Fe is converted to ESR-inactive species. We did not investigate whether this is due to a reduction of Fe3+ to Fez+ or to a change of the ligand field symmetry in the first coordination sphere of Fe3+. The removal of the ESR-active Fe3+ signal at g = 2.0 affects the rate of formation of Ag6+ (Figure 9). This rate is slowest for those samples that are most severely treated such as the NaOAc treatment applied 10 times and the S2O4’-treatment applied 12 times. These samples contain the smallest amount of ESR-active Fe, and Hz is known to be activated on paramagnetic Fe.IS There is however no strong linear correlation. According to Figure 1 the amount of ESR-active Fe varies in the following order: no This is treatment > NaOx > NaOAc > S2O4’-!6) > S2042-.12 not the order of the initial rates of formation of Ag6+ (Figure 9). The reason is that the rate of formation of Ag6+ is governed more by the migration of Ag+, as discussed above, than by the H z activation step. The fact that a dependence on the chemical treatment of Fe removal is observed might indicate that these treatments are not so innocent and attack also the lattice, especially in zeolite A, which is known to have a rather unstable and therefore reactive zeolite lattice. ( 2 5 ) Hidaka, S.; Iino, A.; Mibuchi, T.; Nita, K.; Yamazoe, N. Chem. Lett. 1986, 1213-6.

2053

Conclusions The Ag6+cluster is formed upon H, reduction of AgNa-A and AgK-A zeolites. In the case of Na+ as coexchanged cation the initial rate of formation of the cluster increases with the temperature according to an Arrhenius law with an activation energy of 47-63 kJ mol-I. This indicates that Ag+ migration determines the temperature dependence of the cluster formation. The maximum amount of clusters is formed when on the average 1 Ag+ is present per cubooctahedron. Above that amount Ag6+ is unstable and reacts to metallic Ag. Below 2 Ag+ per unit cell no cluster formation was observed. In the presence of K+ cluster formation is characterized by an induction period, which is due to the fact that the activation energy for K+ migration is larger than that for Ag+ diffusion. In the presence of Cs+ or CaZ+no cluster formation is observed, either due to the small mobility and high activation energy for migration of these cations or due to the presence of acid O H groups, introduced during the ion-exchange process. The zeolite A lattice structure is necessary to stabilize Ag6+. The rate of formation of As6+ cluster is dependent on the chemical treatment for removal of Fe impurities, because H2is activated over the ESR-active Fe3+. Only part of the Fe that is present is active in the activation of HZ.The results suggest also the possibility that the repetitive chemical treatments affect the lattice. Acknowledgment. We thank the National Fund of Scientific Research (Belgium) for support of this research with a grant and with a position as research director for R.A.S. We acknowledge the Bodemkundige Dienst van Belgie for the chemical analysis and Prof. W. Hertogen (K. U. Leuven) for the neutron activation analysis. Fruitful discussions with Prof. P. Grobet are appreciated. Registry No. Ag6+, 73 145-14-9;Fe, 7439-89-6.

Conformation of the Hydrocarbon Chains of Sodium Dodecyl Sulfate Molecules in Micelles: An FTIR Study Farida Holler and James B. Callis* Center for Process Analytical Chemistry, Department of Chemistry, BG- 10, University of Washington, Seattle, Washington 98195 (Received: June 28, 1988)

The purpose of this study was to investigate hydrocarbon conformation in micellar systems. FTIR spectroscopy was used to observe methylene wagging modes in the 1300-1400-cm-’ spectral region. The areas of these bands were proportional to the probability of Occurrence of bond sequences in specific gauche/trans conformations along a hydrocarbon chain. The peaks relating to penultimate gauche bonds (bent end-methyl), double-gauchestates, and gauche-trans-gauche (kink) sequences in linear alkane chains were quantitatively compared in neat alkanes and in sodium dodecyl sulfate (SDS) aggregates formed in water and deuteriated methanol. By use of Flory’s rotational isomeric state model, the average numbers of these conformational “defect states” were calculated for a long-chain alkane, heptadecane, and a proportional relationship was used to estimate these values for the shorter chains of SDS and the corresponding liquid alkane, tridecane. Alkane chains of SDS molecules in aqueous micellar solutions exhibited approximately 0.36 bent end-methyl, 0.77 double-gauche, and 0.68 kink states per molecule, compared to 0.68,0.64, and 0.77, respectively, in tridecane. This indicated that the average numbers of gauche configurations were comparable in aqueous micelles and liquid alkanes. Disorder in the aggregates increased as methanol was added to the solvent. The double-gauchestate increased more than the kink state, while the bent end-methyl state remained relatively unaffected by solvent composition.

Introduction have been postulated for the structure of various surfactant aggregates, with particular emphasis on understanding a mean field lattice chain Dill and flory model which they applied to micelles and other chain molecule

* To whom correspondence should be addressed. 0022-3654/89/2093-2053$01.50/0

interphases.’ By placing alkane chains on a three-dimensional lattice, they computed probabilities for segments along the chain occurring at specific sites. These calculations could then be used to predict orientational order within the system. Szleifer, Ben(1) (a) Dill, K. A.; Flory, P.J. Proc. Narl. Acad. Sci. U.S.A. 1981, 78,676. (b) Dill, K. A. J . Phys. Chem. 1982, 86, 1498.

0 1989 American Chemical Society

2054

The Journal of Physical Chemistry, Vol. 93, No. 5, 1989

Shaul, and Gelbart calculated molecular and thermodynamic properties for chains in aggregates of various shapes-cylinders, spheres, and planar bilayers.* By minimizing the free energy of a system, they evaluated the equilibrium distribution of chains in aggregates. Such studies indicate that geometric packing constraints lead to a gradient of conformational freedom along the chains, such that there is decreased order at the core of the micelle, which is consistent with data pertaining to N M R order parameters. Spherical and rod-shaped micelles are more disordered than bilayers due to the greater surface curvature of the former. A number of computer simulations have also been carried out to model micellar structure with an intent to understand interactions at the molecular level. Woods, Haile, and OConnell have used molecular dynamics methods to simulate a spherical micelle with dodecyl alkane chain s~rfactants.~ It was determined that about 28% of all bonds were in the gauche configuration. Calculations based on Flory's rotational isomeric state model for chain molecules predict that a normal alkane would have 36% of all bonds in the gauche c~nfiguration.~Thus, the simulation indicated that a micelle is slightly more ordered than an amorphous liquid. The study also clearly demonstrated that hydrophobic portions of the surfactant are not totally shielded from the water molecules as was assumed from early micelle representations. At this point, simulations of micelle structure have produced a far clearer picture of the molecular structure of a micelle than any representations predicted by experimental techniques. Yet many results of the calculations remain untested, and parameters of the models have not been firmly established. The models can only be tested and refined as experimental data become available. Experimental studies so far have failed to produce a detailed picture of the interior of a micelle. Techniques that rely on measurement of bulk properties of the system, such as conductivity or surface tension, do not provide definitive information on conformations and interactions at the molecular level. As Menger has pointed out, such data are often interpretable in different ways.S Direct evidence of microscopic structural characteristics, such as chain conformation, surface roughness, viscosity, and shape of individual micelles, is difficult to obtain. Vibrational spectroscopies such as infrared and Raman techniques can be used to measure molecular configurational states and thus produce direct information about interactions of individual chains. Some qualitative studies of surfactant systems have been carried out by these techniques. Okabayashi et al. found that the Raman spectra for a series of short-chain alkyl sulfates and carboxylates contained bands due to the presence of gauche isomers, and the intensities of these bands increased with chain length? A Raman study by Kalyanasundaram and Thomas showed that aqueous micellar solutions of some surfactants, including sodium dodecyl sulfate (SDS), existed in a liquidlike state and that peaks due to gauche conformations were present in spherical and rod-shaped micelles but decreased in intensity for the larger aggregates.' Picquart also used Raman spectroscopy to study temperaturedependent behavior of concentrated SDS solutions.* FTJR spectroscopy could potentially provide further quantitative information on conformational order of hydrocarbon systems. The basis for this conformational study comes from pioneering work by Snyder, in which he has calculated and tabulated the frequencies for an extensive number of vibrations in alkane systems, including some that are specific to localized "bent" structures containing gauche bonds.9 His normal-mode calculations give (2) (a) Ben-Shaul, A.; Szleifer, I.; Gelbart, W. M. J. Chem. Phys. 1985, 83,3597. (b) Szleifer, I.; Ben-Shaul, A.; Gelbart, W. M. J. Chem. Phys. 1985, 83, 3612. (3) (a) Haile, J. M.; OConnell, J. P. J. Phys. Chem. 1984,88,6363. (b) Woods, M. C.; Haile, J. M.; OConnell, J. P. J . Phys. Chem. 1986,90, 1875. (4) Flory, P. J. Statistical Mechanics of Chain Molecules; Wiley: New York, 1969. (5) Menger, F. M. Acc. Chem. Res. 1979, Z2, 111. (6) (a) Okabayashi, H.; Okuyama, M.; Kitagawa, T.; Miyazawa, T. Bull. Chem. SOC.Jpn. 1974, 47, 1075. (b) Okabayashi, H.; Okuyama, M.; Kitagawa, T. Bull. Chem. SOC.Jpn. 1975, 48, 2264. (7) Kalyanasundaram, K.; Thomas, J. K. J . Phys. Chem. 1976,80, 1462. ( 8 ) Picquart, M. J. Phys. Chem. 1986, 90, 243.

Holler and Callis All-Trans

End-Gauche La

fJp

bmi

Gauche-Gauche ~

W

Figure 1. Alkyl chain conformations: (A) all-trans, (B) end-gauche (1341 cm-I), (C) gauche-gauche (1354 cm-I), (D) kink (1367 cm-I). From ref 1 1.

a very detailed assignment of various C-H stretching, scissoring, rocking, and wagging modes. The intensities of these bands can be used to obtain quantitative information on various "defect" structures within the chains that contribute to the overall disorder of the system. The three bands chosen for this study occur at 1341, 1354, and 1367 cm-I and are the carbon-hydrogen bending modes for segments of the chain containing specific trans-gauche sequences. The first peak indicates a penultimate bond oriented such that the terminal methyl group is in a gauche configuration relative to the methylene group three carbons away (bent endmethyl), the second is due to two adjacent gauche bonds along the carbon backbone (double-gauche), and the third arises from a gauche-trans-gauche sequence (kink) (see Figure 1). These wagging vibrations are localized; i.e., they are relatively unaffected by chain length or by conformational changes elsewhere in the chain. Since these bands are directly proportional to the number of segments in a particular conformation, they can be quantitatively related to the equilibrium distribution of gauche defects and hence the degree of departure from an all-trans state in the molecule. The bent end-methyl state provides the least perturbation to chain order since it occurs only at the terminal methyl ends of the chain. The kink state provides some disruption in the order of the interior segments of the chain, while the double-gauche state produces a severe departure from the alignment of chains in the all-trans conformation. The Occurrence of these conformation-related peaks has been monitored to observe the phase behavior of solid n-alkanes.IO They have also been used for the determination of conformational order in reverse-phase liquid chromatography packing materials." The presence of hydrocarbon chains in molecules of the alkyl surfactant sodium dodecyl sulfate (SDS) allows the investigation of the degree of order present in micelles formed by aggregation of these molecules. The effect of solvents on the conformational states of these molecules can also be investigated by this method. It had been assumed, until recently, that very little methanol could actually penetrate into the oily interior of the micelle. However, a paper on N M R self-diffusion measurements by StilbsIz indicates that the partition coefficient for methanol between the SDS micelle and water is about 0.16. The presence of methanol would un(9) (a) Snyder, R. G. J . Chem. Phys. 1967, 47, 1316. (b) Snyder, R. G.; Schachtchneider, J. H. Spectrochim. Acta 1963, 19, 85. (10) Maroncelli, M.; Qi, S. P.; Strauss, H. L.; Snyder, R. G. J. Am. Chem. SOC.1982, 104, 6237. (1 1) Sander, L. C.; Callis, J. B. Field, L. R. Anal. Chem. 1983, 55, 1068. (12) Stilbs, P. J . Colloid Interface Sci. 1981, 80, 601.

Hydrocarbon Conformation in Micellar Systems doubtedly affect chain packing and order within the micelle. A formidable barrier to the study of aqueous systems by IR spectroscopy is presented by the large absorptions of the solvent. This problem necessitates the use of extremely short path lengths through the sample. An internal reflectance sampling accessory such as the "Circle Cell", made by Spectra Tech, provides a fairly reproducible sampling depth for solutions of specific refractive index and a particular wavelength of radiation. The total effective path length at 1400 cm-' is on the order of 10 pm. In addition, the wide dynamic range of modern FTIR spectroscopy allows for the subtraction of solvent spectra from those of solutions in order to obtain spectra of the solutes with excellent signal-to-noise ratios. Thus, by quantitative evaluation of peak areas, information can be obtained on the equilibrium distribution of conformers of alkane chains of aggregated surfactant molecules in aqueous solutions and in other solvents. Experimental Section FTIR spectra were obtained of neat alkanes and solutions containing 10 wt % SDS in order to compare the degree of conformational order in the two systems. A high concentration of surfactant was required in order to observe the bands related to gauche defects with adequate signal-to-noise ratios for quantitative comparisons. For the concentration used, the micelles would be expected to be in the form of large rod-shaped structures. Spectra were also obtained of 10% SDS in deuteriated methanol and in waterlmethanol-d4 mixtures. Light-scattering measurements indicated that, at this high concentration of SDS, large aggregates were present for all solvent compositions including 100% methanol. The normal alkanes, such as tridecane and heptadecane, were obtained from Sigma at 99% purity, and electrophoresis grade sodium dodecyl sulfate was obtained from Mallinckrodt. Deuteriated methanol used was from Aldrich and ICN Biomedicals Inc. (d4,99.5%). Chemicals were used without further purification. The spectra were obtained on a Perkin-Elmer 1800 FTIR spectrometer, with a wide-band MCT detector. A Spectra Tech "Circle Cell" internal reflectance attachment (with ZnSe crystal) was used for sampling. Samples were kept at 30 f 3 'C. Measurements were made at 2-cm-I resolution with an interval of 0.2 cm-' between data points. Approximaely 12000 scans were averaged (data collection time of about 2 h) for each single-beam spectrum of solvent and solution. The instrument was kept under a dry nitrogen purge for several hours in order to remove residual water vapor which interfered in the spectral region of interest. Data Analysis In the case of surfactant systems, subtraction of the bulk solvent spectrum from that of the solution gave the spectrum of the solute, SDS. Since the critical micelle concentration of SDS is 0.2 wt %, the fraction of the surfactant molecules in micelles would be about 98% in a 10% (w/w) solution. Thus, the observed spectrum would represent primarily the molecules in micellar aggregates. In order to obtain the spectrum of SDS, the fraction of solvent subtracted could be adjusted in order to compensate for differences in the amount of solvent present in each solution. However, since the highly charged solute, SDS, causes the peaks in the H20 spectrum to undergo slight spectral shifts and changes in intensity, the spectral subtraction produces derivative-shaped features that result in a complicated base line profile. The fraction of solvent subtracted was therefore adjusted slightly to reduce the curvature in the base line, but was kept close to 1. Since the peaks of interest were overlapped, peak height was not a reliable measure for quantitative analysis. A nonlinear least-squares curve fitting method was used for quantitative estimation of the areas of individual peaks. A FORTRAN subroutine by Bevingt~n'~ utilizing the Marquardt algorithm was used to optimize parameters for a fitted function composed of the sum of Lorentzian peaks and a simple base line function. In addition, a procedure was developed to fit the second derivatives of the (13) Bevington, P.Data Reduction and Error Analysis for rhe Physical Sciences; McGraw-Hill: New York, 1969.

The Journal of Physical Chemistry, Vol. 93, No. 5, 1989 2055

1284

:4

1

I

+

t

1

0.324

0.024

25W

3wo

15W

2 m

WAVENUMBER 17oi1

"

i

'

'

"

'

"

'

'

"

"

'

"

'

'

"

'

'

[ 't

\

1 0356

'

"

.

\ I

3000

25iM

15W

2000

1wO

WAVENUMBER 0.135

c'

"

"

"

"

"

"

"

"

"

'

"

(

I1

'

"

'

r

WAVENUMBER

Figure 2. Spectra of (a) tridecane, (b) 10%SDS micellar solution, and (c) SDS after subtraction of bulk water.

spectra in order to reduce background pr0b1ems.l~ Results and Discussion Since the SDS molecule contains a 12-carbon chain with 10 methylene groups and 1 terminal methyl group, it is appropriate to compare it to tridecane, the linear alkane with an equivalent number of methylene groups and a methyl group at each end. The spectrum of tridecane is shown in Figure 2a. The spectrum of the aqueous micellar solution of SDS (Figure 2b) is dominated by the absorption due to water. After subtraction of the spectrum of bulk water, the spectrum of SDS emerges, with absorption peaks corresponding to the alkane chain as well as the sulfate head group (Figure 2c). The negative absorptions are due to displacement of water by SDS. (14) Holler, F.; Burns, D. H.; Callis, J. B., manuscript in preparation.

The Journal of Physical Chemistry, Vol. 93, No. 5, I989

2056

a

Holler and Callis

c

w

! . --

5m

-

0

-

-

5v1 2 '

m 4

-

-

-

-

8: 1 . 3 3 1 . 3 4 1.35 1.36 1.37 1.38 1.39 2 . 4 0 WAVENUMBER x 10

I

, , 1 , , , , l , , , , I , , , , 1 , , , , 1 , ,

, I l , / l l , , l l l l j

1.37 1.38 1.39 WAVENUMBER x 10

1 . 3 3 1.34 1.35

1.36

2.4"

Figure 3. (a) Spectrum of tridecane (1325-1405 cm-') with superim-

posed best fit (note increasing wavelength scale). (b) Residuals between best fit and spectrum.

Figure 4. (a) Spectrum of 10% SDS in water (1300-1410 c d ) with

superimposed best fit using an exponential base line (note increasing wavelength scale). (b) Residuals between best fit and spectrum.

TABLE I: Comparison of Peak Areas for Tridecane, 10%SDS in Water, and 10%SDS in Methanol-d, normalized areas bent doubleend-methyl gauche kink umbrella tridecane 0.15 0.24 0.35 1.0

10% SDS in water peak + exponential fit peak + quadratic fit second-derivative fit 10% SDS in methanol-d, peak + exponential fit peak + quadratic fit second-derivative fit

0.08

0.04 0.03

0.29 0.21 0.17

0.31 0.26 0.23

0.5 0.5 0.5 1.34

0.06 0.03 0.03

0.35 0.36 0.28

0.36 0.31 0.28

1.35

1.36 1.37 WAVENUMBER

0.5 0.5 0.5

The three defect bands in the tridecane spectrum occur in the 130C-l400-cm-~ region, along with the methyl symmetric bending (umbrella) mode at 1378 cm-l. This region of the spectrum was fitted with parameters for four Lorentzian peaks and a base line offset parameter (Figure 3). Since the intensity of the umbrella mode should be conformation independent in the liquid state, the areas of the other three peaks were normalized by ratioing to this band. The residuals between fitted spectrum and the data (Figure 3b) contain some structure due to the inability to describe the peak shapes by a simple Lorentzian function. In addition, the absorptions from residual water vapor in the instrument are always superimposed on the spectra and contribute to the residuals. However, the magnitude of oscillations in the residual spectrum are approximately fl mAU-less than 5% of the intensity of the smallest peak (AU = absorbance unit). The spectrum of SDS exhibits the methyl umbrella mode, as well as the three conformation-related peaks. The intensity of the smallest peak is now about 6 X IO4 AU. In addition, the spectrum is complicated by a peak at 1390 cm-I, probably caused by interactions between solvent and SDS. This peak is not present in the spectrum of solid SDS or in the spectra of the solvents used. It was fitted by an additional Lorentzian profile in all the curve-fitting procedures. The major problem with the base line

1.39

1.38

3.40

x 10

il

n

I'

L

E

\I I

I

I

I

I

I

1.34

I

I

I

1.35

I

I

I

I

I

1.36

I

I

I

I

1.37 WPVENUMBER

I

I

I

I

I

1.38

1

'1

I

1 1.39 3 . 4 0 x 10 I

I

I

1

Figure 5, (a) Spectrum of 10% SDS in water (1330-1402 cm-I) with superimposed best fit using a quadratic base line (note increasing wavelength scale). (b) Residuals between best fit and spectrum.

is caused by the intense S - 0 stretching absorption at 1200 cm-I due to the sulfate group. Different functional forms were used to approximate the base line in the region of interest. Figure 4a illustrates the best approximation using an exponential base line for the 1300-1410-~m-~range. A quadratic functional form for the base line produced a poor fit over the same wavenumber range, but the fit improved when a smaller wavelength range was used (Figure 5 ) . Second-derivative fit also improved over the smaller

The Journal of Physical Chemistry, Vol. 93, No. 5, 1989 2057

Hydrocarbon Conformation in Micellar Systems

A

n

\ I

L

1,

I

I

I

I

I

I

1.34

I

I

I

I

I

1.35

I

I

I

I

I

1.36

>

I

I

t

I

1.37

,VI

I

I

1.38

WAVENUMBER

I

,

I

I

I

I

I

I

1 . 3 3 1.34 1.35 1.36 1.37 1.38 1.39 2 . 4 0 WAVENUMBER xA 0

I

1.39 2 . 4 0 x 10

b

b h

(Ut

1

1

1.34

1.35

1.36

1.37

WAVENUMBER

1.38

1.39 x 10

9.40

Figure 6. (a) Second derivative of spectrum of 10% SDS in water (1330-1402 cm-I) with superimposed best fit; 4.6-cm-' smoothing win-

dow (note increasing wavelength scale). (b) Residuals between best fit and second derivative of spectrum. wavelength range (Figure 6). The final normalized peak parameters are shown in Table 1. There is significant variation in the areas and peak widths among the three methods, particularly for the 1341-cm-' (bent end-methyl) peak which is most affected by the tail of the interfering peak. It appears that the exponential base line should be the most reasonable approximation in that region, since it provides the best fit over a wide spectral region. Since the SDS molecule has the same number of methylene groups as tridecane, but only half as many methyl groups per molecule, any vibration related to the terminal methyl group, such as the umbrella mode, should have half the band intensity in SDS relative to tridecane. The normalized peak areas for SDS were divided by 2 for comparison to those obtained for tridecane. The comparison of normalized areas indicates that the aqueous SDS micelle solution has approximately half the number of bent end-methyl states as tridecane when considering the exponential fit. The other two fits indicate lower values, but these areas are probably underestimated because of the shape of the base line. Since the SDS molecules have only half the number of end-methyl groups relative to tridecane, they would appear to be in a similar state of disorder. The double-gauche and kink states are also similar to the alkane within about 30%, although the differences between the three methods indicate significant uncertainty in the actual areas. However, one can conclude that the average proportions of conformational defects in a micelle are not very different from that in a liquid alkane at the same temperature. The actual distribution of conformational states can be calculated for a long-chain n-alkane by using Flory's rotational isomeric state modeL4 Since simplifying assumptions of infinite chain length were used, the calculated values should be most accurate for long chains. Therefore, the calculated numbers of defect states per molecule were assumed to be proportional to peak areas for heptadecane, the longest chain in the liquid state at 30 "C. Estimates were made of the average numbers of defect states in different alkane systems by scaling normalized peak areas for these compounds to those of heptadecane. The average numbers of bent end-methyl, double-gauche, and kink states per molecule of tridecane were found to be approximately 0.68, 0.64, and 0.77, respectively. Using peak areas for the fitted SDS spectrum with

1

1

1 33

1

#

1

1

1

(

#

1

1

1

, I I I , # I I I I I

,

1 . 3 4 1.35 1.36 1.37 WAVENUMBER

, , , , I , , , , I 1.38 1.39 x 10

3.40

Figure 7. (a) Spectrum of 10%SDS in 70%methanol-d4/30%water. (b) Second derivative of spectrum (4.6-cm-I smoothing window).

exponential base line, we estimated the corresponding numbers of these states per molecule to be 0.36,0.77, and 0.68, respectively. These numbers indicate similar proportions of these states given that the tridecane has an equivalent number of methylene groups and twice the number of methyl groups as SDS. The peak areas for SDS in methanol-d, are also displayed in Table I. Although there are still significant differences between the areas obtained from the three curve-fitting methods, each method shows a similar trend in going from the spectrum of SDS in water to the spectrum of SDS in methanol. The bent endmethyl states do not show a significant difference for the two solvents, within each fitting procedure. The number of doublegauche states shows a significant increase of about 40% for each of the three methods. This indicates that these states, which produce a severe disruption of chain order, are much more likely to occur in aggregates formed in methanol than in aqueous aggregates. The number of kink states is also increased in methanol, but only by about 20%. Thus, the methanol solvent produces a significant increase in disorder of the SDS chains relative to an aqueous solvent. The spectra of SDS in mixed solvents had severely sloping backgrounds which could not be adjusted by solvent subtraction. This was caused by exchange of deuterium from the hydroxyl group on the methanol-d, and the hydrogen ions from water. The solvent spectrum was no longer flat in the region of interest, and an effective subtraction of the solvent from the solution could not be achieved. The resulting base lines for the spectra of SDS were inadequately fit with any simple functional form, making quantitative comparisons impossible. (See Figure 7 for a spectrum of 10% SDS in 70% methanol/30% water.) The second derivative of the spectra of SDS in mixed solvents was effective in removing a large part of the sloping background so that qualitative comparisons could be made. These comparisons indicated that the gauche defect peaks increased gradually over the whole range from 0 to 100% methanol. Another method that has frequently been used to monitor the environment of the alkane chains is to observe shifts in peak position^.'^ The wavenumbers of the methylene antisymmetric stretch absorption vs solvent composition are plotted in Figure (15) Mendelsohn,

H.H.Biochemistry

R.; Dluhy, R.; Taraschi, T.; Cameron, D. G . ;Mantsch, 1981, 20, 6699.

J . Phys. Chem. 1989, 93, 2058-2065

2058

2 928

1

2.9275

4

2 925

n

4

?

~

~

2 9245

--I

0

20

40

60

00

100

parcent d-4 mdhanol

Figure 8. Peak position of methylene antisymmetric stretch vs percent methanol-d, in solvent.

8. They also indicate a trend of gradually changing environment of the alkane chains. Previously, experiments on conductivity and light scattering of SDS solutions at lower concentrations indicated that aggregates were not present in solvents with methanol content above 30%.16 This study demonstrates that surfactant molecules can form aggregates in methanol-rich solutions.

Conclusions The data indicate that the alkane chains of SDS have approximately the same proportions of bent end-methyl, double(16) (a) Ward, A. F. H. Proc. R. SOC.London, A 1940, 176, 412. (b) Parfitt, G. D.; Wood, J. A. Kolloid Z . 2.Polym. 1968, 229, 5 5 .

gauche, and kink states as tridecane. Although these results support the models of disordered micellar systems, it is not possible to predict long-range order by using these experimental methods. Since this FTIR study provides an average estimate of conformational states, information about the gradient of conformational order along the chain (such as that provided by N M R order parameters) cannot be confirmed. Increasing methanol content of the solvent affects the environment of the hydrocarbon chains. The chains are gradually being exposed to a more polar environment, but there is no "critical point" beyond which the aggregates are no longer present. If the aggregation of surfactant molecules was completely disrupted, the chains would see a fairly constant polar environment, and further addition of methanol to the solvent should not significantly affect their environment or their conformational state. Technical advances in FTIR spectroscopy in recent years, along with better data-handling capabilities, make this an appropriate tool for the acquisition of important experimental data in this field. The study of conformations of alkane systems could help to elucidate the internal structure of a micelle with respect to chain organization and order. It has also been demonstrated that the technique could be used to study solvent interactions with micellar systems.

Acknowledgment. The authors thank Perkin-Elmer for donating the PE Model 1800 FTIR for this research. Thanks are due to Dr. David Burns for assistance with the computer programming and data analysis. F.H. thanks the Center for Process Analytical Chemistry at the University of Washington for the graduate research assistantship. Registry No. SDS, I5 1-21-3;methanol, 67-56-1;heptadecane, 62978-7; tridecane, 629-50-5.

Effects of Preparation Variables on Particle Size and Morphology for Carbon- and Alumina-Supported Metallic Iron Samples Scott A. Stevenson,? Scott A. Goddard, Masuhiko Arai,t and J. A. Dumesic* Department of Chemical Engineering, University of Wisconsin, Madison. Wisconsin 53706 (Received: June 28, 1988)

Triiron dodecacarbonyl, Fe3(C0)12,was used to prepare highly dispersed metallic iron on carbon black, graphite, and dehydroxylated alumina. Magnetic susceptibility measurements were used to estimate particle size distributions;these experiments indicate that after reduction at 673 K much of the iron is present in particles containing only a few iron atoms. The high dispersions obtained for the cluster-derivedFe/A1203and Fe/carbon black catalysts were resistant to sintering during treatment in hydrogen at up to 773 K. Carbon monoxide chemisorption measurements are in agreement with the dispersions estimated from the magnetic susceptibility data for the alumina- and graphite-supported catalysts. Mossbauer spectroscopy suggests that the cluster-derived, carbon black supported iron particles may be more spherical and possibly more widely separated than conventionally prepared iron particles on the same support.

Introduction promising for organometallic cluster oneof the compounds in the field of heterogeneous catalysis is for the preparation of highly dispersed zerovalent metal particles. of catalyst genesis, which involve the deConventional position of metal salts, require reduction of the metal cations to the zerovalent state. Highly dispersed cations of elements such as iron, however, are difficult to reduce completely. In contrast,

* Author

to whom correspondence should be addressed. Present address: Institut fiir Physikalische Chemie der Universitat

Miinchen, Sophienstrasse 11, 8000 Miinchen 2, West Germany. 'Present address: Chemical Institute of Non-Aqueous Solutions, Tohoku University, Sendai 980, Japan. 0022-3654/89/2093-2058$01.50/0

the metallic atoms in many organometallic compounds are already in the zerovalent state, so that reduction is not required. Early work d e " m a t e d the value of this concept by the production of highly dispersed zerovalent an element that is difficult to reduce once supported in an oxidized state. Because of the difficulty of reducing isolated ferrous cations,3d organo(1) Brenner, A,; Burwell, R. L., Jr. J . Cafal. 1978, 52, 353. (2) Brenner, A.; Burwell, R. L., Jr. J . Cafal. 1978, 52, 364. (3) Garten, R. L.; Ollis, D. F. J . Cafal. 1974, 35, 232. (4) b u d a r t , M.; Delbouille, A,; Dumesic, J . A.; Khammouma, S.;Topsw, H.J . Coral. 1975, 37, 486. (5) Raupp, G. B.; Delgass, W. N. J . Cafal. 1979, 58, 337. (6) Yuen, S.; Chen, Y . ;Kubsh, J. E.; Dumesic, J. A,; Topsae, N.; Topsae, H . J . Phys. Chem. 1982, 86, 3022.

0 1989 American Chemical Society