J . Phys. Chem. 1990, 94, 6452-6457
6452
HC-solvent contact region. Thus, as the alcohol chain length increases the transfer process becomes increasingly unfavorable. Summary and Conclusion
We have investigated the solubilization tendency of several alcohols in NaLS in D 2 0 , ordinary water, and aqueous urea solutions by eq 1 relating the distribution coefficient to the initial relative lowering ability to depress the cmc and the corresponding initial relative ability to increase the effective degree of micellar ionization. It was found that the coaggregation tendency decreases in going from H 2 0 to D 2 0 and also on going from H 2 0 to 5 M urea. This trend has been confirmed by measuring the coaggregation tendency of the alcohols in these solvent systems by head space gas chromatography. We have also determined by the latter technique the transfer free energy of alcohols from H 2 0 to D 2 0 and from H 2 0 to aqueous urea solutions and then estimated the transfer free energy of an alcohol from micelles in H 2 0 to micelles in D 2 0 and to micelles in urea solutions. Apo(H20-D20) is about zero for ethanol and increases linearly with alcohol chain length. Ag0(H20-u) is positive for ethanol and propanol and
becomes more positive with urea concentration. It is negative for butanol, hexanol, and octanol and tends to become more negative with urea concentration. At a given urea concentration Ap0(H2+u) decreases with increasing alcohol chain length. The transfer free energy of an alcohol from micelles in H 2 0 to micelles in D 2 0 is positive and increases with alcohol chain length. On the other hand, Apom(H20+u) is positive for propanol and increases with urea concentration. For butanol it appears to be favorable at 2 M urea after which Apom(H20-u) becomes positive and increases with urea concentration. For hexanol and to a greater extent octanol Ag0,(H20-u) is negative which becomes more negative with urea concentration, the decrease reaching a maximum at 4 and 3 M urea, respectively. The results have been interpreted in terms of the effect of the "structuredness" of the solvent system on the extent of the HCsolvent region of the micelle and the ability of an alcohol to reduce the extent of this region in the NaLS micelles. Acknowledgment. This research was supported by the University of Kuwait, Project No. S C 0 2 7 and S C 0 4 2 .
Surface Molecular Orientations Determined by Electronic Linear Dichroism in Optical Waveguide Structures D.M. Cropek and P. W. Bohn* Department of Chemistry and Beckman Institute, 1209 W. California St., Urbana, Illinois 61801 (Received: January 17. 1990; I n Final Form: April 16, 1990)
Electronic linear dichroism measurements were performed on monolayers of monophenyldimethylsilanes to determine average orientations of the phenyl rings. The very weak absorption of these silanes at the excitation wavelengths available makes them excellent candidates to test the lower detection limits of this novel approach to orientation studies. By use of thick glass waveguides (150 pm thick), the interaction between the absorbate monolayer and the electric field is maximized, and absorption loss coefficients for the silanes are obtained with high precision. Orientation angles for the long axis of the phenyl groups, measured from normal, are between 58' and 75', in good agreement with an expected tetrahedral angle of 7 0 . 5 O . The exact angle appears to be strongly dependent on the monolayer coating procedure.
Introduction
Orientational studies on aligned molecular assemblies is an area which is receiving increased attention because of growing interest in the characterization of ordered structures with thicknesses on the order of several monolayers and below. These thin film structures include Langmuir-Blodgett self-assembled monolayer^,^.^ surface-modified electrode^,^,^ and biomolecular assemblies* for various purposes such as monitoring electrode reactions, surface-catalyzed reactions, construction of layers with specific optical and physical properties, chemical sensors, and electronic components. Linear dichroism (LD) techniques allow for the investigation of these aligned samples.9-" LD is the anisotropic response of a partially or completely oriented sample to different polarizations of an excitation source. Information which can be obtained from an LD experiment falls into three main categories: ( 1 ) molecular optical properties, (2) dynamics, and (3) orientation information.I2 LD can be done in numerous experimental modes,I3 including absorption, fluorescence,electron spin resonance, X-ray scattering, and Raman scattering; however, the first two methods are the most prevalent. Traditionally, transmission LD measurements have been used to study either of two kinds of oriented molecular assemblies. The first kind are relatively thick films or even bulk samples which have large concentrations of the oriented structure, for example, guest molecules in liquid crystals14 or stretched polymer films.I5
* 4htFIu
10
\\
hom correqpondence should be addressed
The second kind consists of thin layers of molecules with large absorption cross sections (or fluorescence quantum yields) such as phthalocyanines." These experiments are designed to generate appreciable absorption when passing the excitation radiation perpendicular to the film plane. In the present experiments, we have extended the LD measurements, through the use of integrated optical structures, to the study of submonolayer amounts of even ( I ) Kuhn, H. J . Photochem. 1979, 10, 1 1 1. (2) Swalen, J. D. J . Mol. Electron. 1986, 2, 155. (3) Swalen, J. D.; Allara, D. L.; Andrade, J. D.; Chandross, E. A,; Garoff, S.; Israelachvili, J.; McCarthy, T. J.; Murray, R.; Pease, R. F.; Rabolt, J. F.; Wynne, K. J.; Yu, H. Langmuir 1987, 3, 932. (4) Allara, D. L.; Nuzzo, R. G. Langmuir 1985, 1, 45. ( 5 ) Maoz, R.; Sagiv, J. Langmuir 1987, 3, 1034. (6) Murray, R. W. Annu. Rev. Marer. Sci. 1984, 1 4 , 145. (7) Wrighton, M. S. Science 1986, 231, 32. (8) Blankenburg, R.; Meller, P.; Ringsdorf, H.; Salesse, C. Biochemistry 1989, 28, 8214. (9) Norden, B. Appl. Spectrosc. Rev. 1978, 14, 157. (IO) Schellman, J.; Jensen, H. P. Chem. Reu. 1978, 87, 1359. ( I I ) Palacin, S.; Lesieur, P.; Stefanelli, I.; Barraud, A. Thin Solid Films 1988, 159, 83. (12) Thulstrup, E. W.; Michl, J. In Polarized Spectroscopy of Ordered Systems; Samori, B., Thulstrup, E. W., Eds.; Kluwer Academic Publishers: Boston, 1988; pp 1-24. (13) Kuball, H. G.; Friesenhan, H.; Schonhofer, A. In Polarized Spectroscopy of Ordered Systems; Samori, B., Thulstrup, E W.. Eds.: Kluwer Academic Publishers: Boston, 1988; pp 85-104. (14) Johansson, L. B. A. Chem. Phys. Len. 1985, 118, 516. ( I S ) Tempczyk, A,; Gryczynski, Z . ; Kawski, A.; Grzonka, Z . Z . Natur,forsch. 1988, 43A. 363.
0022-365419012094-6452$02.50/0 C 1990 American Chemical Society
Linear Dichroism Measurements of Silanes
lz wavegi
The Journal of Physical Chemistry, Vol. 94, No. 16, 1990 6453 at each reflection. The magnitude of this evanescent field decays exponentially with distance away from the substrate surface, with a l/e distance on the order of the wavelength of the exciting radiation. This is the electric field which samples the monolayer at the surface. By knowing the incident angle, 0, one can calculate the electric field amplitudes in the three Cartesian directions at the surface using the well-known equations given by HarrickI6 (sin2 p - n212)1/2 cos 0 E , = 2 (1 - n212)I/2[(1+ nzI2)sin2 - n212]1/2
Ey =
Figure 1. Laboratory axes relating the excitation electric field and the absorbate transition moment with the sample geometry.
weak absorbers (e I10 L/(mol cm)). Although analogous experiments have been carried out in the infrared region for many years, they have been plagued by two major problems. First, the optical materials used for common infrared internal reflection elements have large (n > 2.0) refractive indexes. Since the dichroic ratio depends directly on the magnitude of the surface electric field components, which in turn depend on the refractive indexes of the materials comprising the interface, the range of dichroic ratios corresponding to the entire range of molecular orientations (Le., 0-90') is rather small, ca. 0-1.5. Second, infrared experiments are typically performed on internal reflection elements with thicknesses on the order of millimeters, and because the path length related absorption enhancement depends directly on thickness, infrared experiments are limited in sensitivity. Our current LD method exploits the bulk fluorescence of thick optical waveguides to monitor very weak absorbance from monolayers on the surface which considerably extends the detection limits and, thus, the experimental applications of LD measurements. To test the sensitivity and range of this implementation of the LD measurement, we have carefully examined the orientations of phenyl rings on monophenyl- and diphenylsilyl ethers bound to planar SO2 surfaces. These systems were chosen for two reasons. First, since they are used as stationary phases in reversed-phase chromatographic separations, their interactions with solution-borne acenes are of interest for elucidating the molecular mechanisms of retention. Second, the absorption spectrum in these systems is sufficiently blue that the absorption cross section at the wavelengths accessible to our apparatus is small. Therefore, these silanes provide a good test of the sensitivity of the measurements.
Theory Most LD experiments are done on uniaxial samples, that is, samples for which two of the three experimental axes are equivalent. This also holds for the experiments described here. There is no external force which causes the monolayer under investigation to orient preferentially in either of the in-plane directions during either the coating or measurement process. Therefore, these two axes are equivalent, simplifying some of the mathematical treatment. Figure 1 shows the experimental axes in relation to the cross section of the substrate and the plane of incidence. The x and y axes are parallel to the surface of the substrate and the z axis is perpendicular to the surface, or, equivalently, the x and z axes are in the plane of incidence defined by the incident and reflected light rays of the laser beam (vide infra) while the y axis is perpendicular to this plane. Transverse electric (TE) polarized radiation is oriented with its electric vector in the y direction. Only those absorbers which have a component of their absorption transition moment in the y direction will absorb this radiation. Transverse magnetic (TM) polarization is comprised of electric vector components in both the x and z directions. Only absorbers with absorption transition moments having components in either the x or z direction will absorb T M radiation. This anisotropic response is used to determine the orientation of absorbers on the surface of the substrate. In the ray-optics approximation, the electric field extends out past the surface of the substrate into the external environment
(la)
2 cos p (1 - n212)1/2
where nzl is the ratio of the refractive index of the rarer medium to that of the denser medium, and the dielectric properties of the surface adlayer have been approximated as being equal to the overlayer in calculating the surface field amplitudes. The absorbance measured is given by the following equation where (ml and (k( are the states involved in the transit@, cI is a constant, ji is the absorption transition moment, and E is the electric field vector. This equation can be expressed in terms of the three laboratory axes as A = k l ( ~ , E ,+ pYEy+ p,Eh2
(3)
where k is a constant and 1 is an effective path length. The absorption due to T E and T M radiation can thus be calculated separately. Figure 1 also shows the relation between experimental axes and the absorption transition moment. \k is the angle between the electric field vector of interest and the transition moment vector, and it can be expressed in terms of the other two angles, 4 and 0, where 4 is the angle the transition moment vector projection into the xy plane makes with the x axis and 0 is the angle between the transition moment vector and the z axis. Simple trigonometric relations give the following three equations for the angle between the transition moment vector and electric field vectors lying along each of the three axes cos P = sin 0 cos 4
(4a)
Ey: cos \k = sin 0 sin 4
(4b)
E,:
E,: cos \k = cos 0 (4c) When T E radiation is used, E, = E, = 0 and only Ey is nonzero. Absorption of T E radiation results in eq 3 being simplified to ATE = kl(Ey121p12sin2 0 sin2 4
(5)
after substitution of eq 4b. Because this is a uniaxial experiment, all values of 4 are equivalent and averaging from 0 to 27r gives ATE
= y2kllEy121pL(2sin2 0
(6)
An expression for T M absorption can be obtained in a similar manner by separating the absorption in the x direction from that in the z direction. This separation, including substitution of eqs 4a and 4c and averaging over 4, results in the following two equations = !/2k/lE,121p12sin2 0
(7a)
ATM,, = y2k/lE,121p12cos2 0
(7b)
ATM,,
It is seen that absorption in the x direction is identical with absorption in the y direction due to the uniaxial nature of the sample. (16) Harrick, N. J. lnrernal Reflection Spectroscopy;Interscience Publishers: New York, 1967.
6454 ..A,
Cropek and Bohn
The Journal oJPhysico1 Chemistry. Vol. 94, No. 16. 1990
bra.
Top View
1.i
Figure 3. The slide assembly configuration consisting of a high refractive index prism mounted on a glass substrate. The lower region of the slide
Side View
indicates the coated region
' " I C "
Figure 2. Top and side view of the experimental apparatus. See text for
details. The ratio of T E absorption lo T M absorption is the dichroic ratio. Using eqs 6 and 7, one can relate the dichroic ratio, p. to the orientation angle, 8, by ATE
p=-=
ATM
lE,f IEx12+ 2lEZl2cot2 0
(8)
The orientation angle, 0, is the angle between the transition moment vector and the z axis. If the position of the transition moment is known with respect to the molecular geometry of the absorber, then the molecular orientation of the absorber with respect to the z axis can be obtained. It is important not to underestimate the importance of being able to make this connection. Experimental Section Apparatus. The experimental configuration is similar to experiments described previously" and is shown in Figure 2. A Coherent lnnova 90-4, 4-W, Ar' laser with UV capabilities permits the use of the following useful wavelengths: 351, 363, 457. 488, and 514.5 nm. Appropriate extracavity prisms and apertures were used to separate the desired wavelengths from competing plasma radiation. Selection of polarization states was done using a Babinet-Solei1 compensator (Optics for Research Model RC-IO) and a Clan-Taylor polarizer (Melles Griot, Model 03PTA101). Interchangeable mirrors were either UV reflective coated mirrors for use at 351 or 363 nm or visible reflective mirrors for all other wavelengths. An fJ20 500-mm focal length planoconvex lens focused the laser beam onto the slide assembly. The side view of Figure 2 shows the geometrical constraint forced by the slide assembly (vide infra) in which the beam must travel down to reach the substrate. Mirror M 2 is mounted on an Ealing 35-2500 rotational stage for precise adjustment of angle y (see Figure 2). A I-mm Hewlett-Packard HFBR-3500 polymer fiber collected the emission from the slide assembly and directed it onto a JJ3.3 aspheric antireflection coated glass lens, which then wupled emission into a Jobin-Yvon U-1000 I-m focal length fJ8 double monochromator equipped with an RCA C31034A photomultiplier tube in a thermoelectrically cooled housing (Products for Research, Inc.). Standard photon counting electronics (EG&G Ortec) were used. A thin beam splitter was used to split a small portion of the beam to a Hamamatsu GI I16 GaAsP photocell to monitor laser power fluctuation. The rotational stage, photon counting (17) Stephens. D. A.; Bohn, P. W.A n d Chem. 1989.61. 386.
electronics, photocell. and the monochromator position were all under computer control. The slide assembly is the heart of the experiment. Figure 3 shows the integral parts to this assembly. The laser beam was focused so that the minimum beam waist was incident on the coupling edge of an LaSF 9 prism (Precision Optical) which was held by pressure on the substrate. The substrates used were 24 mm X 60 mm BK 7 wver slips (Corning), ca. 150 pm thick. The laser beam was coupled into this thick waveguide and propagated down the length of the cover slip. The fiber optic was moved up and down this length to collect the signal at 20-25 different positions. This experiment was done in air, and in experiments which will be described in a future paper, the slide was dipped into a solvent bath. Coating Procedure. The preparation of coated slide is essentially a two-step process involving cleaning of the substrates and the coating application. Initially, the substrates were degreased by washing with Alwnox detergent followed by rinsing with doubly deionized water. The slides were then successively dipped into three different sulfuric acid baths at 180 "C for 3 min each, rinsed with copious amounts of doubly deionized water, dipped into a 4 1 N H , 0 H / H 2 0 2 etching solution for 3 min, and finally rinsed again with doubly deionized water. A filtered nitrogen stream was used to drv the slides. and thev were stored under nitroeen until used. Chlorodimethvlvhenvlsilane (Aldrich) was the coating material .. . of merest in these experiments.' The coating procedurehescribed in ref 18 wah cloiely followed. The silane nas vacuum-distilled over CaH, before use. Coating solutions consisted of distilled toluene (Baker ScientificJand the distilled d i n e in a 1Oo:l volume ratio. The clcm whirate;. mere immeried in this solution to coat ihr. bottom half only (see Figure 3). The solution w3s heated to between 70 and 80 OC. and the coating reaction proceeded under a dr) nitrogen atmosphere for 1 h. The slides uere remwed irom the reaction vessel. rinsed w i t h cnpi(ius amounts of chloroform. toluene. and chloroform again. and then stored under a nitrogen atmosphere u n t i l needed Severdl modifiwtions to thiscoAng p r m d u r e were introduced in attempts tu incredsc the surface coverage by the silane The firrt involved the addition of 3 b3se catalyst. trieth)lamine (AldrichJ. TEA. used as receired. to the coating suIution in the cancentration 0.5 ml. of TEA to 100 mL or solution. A second alteration involved dr)ing the glass cover slips at I20 OC prior to watinp to eliniinate the ph)sirurbed water Idyer at the surface. Finall). curing the slides at 100 OC after coating and before rinsing nas also tried. Lorr Measurrmmrs. The excitation beam uas polariicd first in either the TE 0r.r.U orienwtiun. and the fiber optic uas initially positioned near the coupling edge 01 the prism so that the fiber
-
(18) K3llur). K. M. R.. hrull. U. J , 60. 109
Thompson. M
Anal Chem. 1988.
Linear Dichroism Measurements of Silanes
The Journal of Physical Chemistry, Vol. 94, No. 16, 1990 6455
Glass Coverslip + Polymer Fiber
(a)
-0 Yrm
m
m
1Ym
Imn
Wavenumber (cm-1) (b)
I
adeorbent 1
&
h?t Figure 4. (a) Fluorescence spectrum of Ti:Zn glass, using 351- and 514.5-nmexcitation. (b) A simulated situation indicating ideal positioning of both the excitation and detection wavelengths.
aperture was centered on the width of the propagating beam. From a side view, the end of the fiber was very close to the substrate without physical contact, on the order of 0.5 mm. The fluorescence intensity at this position was measured by photon counting for 20-30 10-s integrations and then averaged to obtain a single data point. The detection fiber was then moved, and the process was repeated to obtain the fluorescence intensity at 20-25 different positions down the length of the slide. The excitation beam was then polarized in the second orientation, and the experiment was repeated. These experiments resulted in two sets of data: the fluorescence intensity versus distance from the prism edge for T E and T M polarizations. Since the boundary between coated and uncoated regions was known beforehand, loss measurements were separately determined for the two regions.
Results and Discussion Loss Measurements. The excitation wavelength was chosen to be on both the red edge of the glass absorption band and an absorption band of the coating material. In these measurements, we rely on fluorescence from the glass substrate to act as a reporter for the guided power density; thus, there must be some absorption of incident radiation due to minor components of the glass matrix. This is typically of the order 0.02 mm-l (including scattering losses). Of course, the surface bound molecule of interest must also absorb. The detection wavelength is on a fluorescence peak of the glass. Figure 4a shows the fluorescence spectra for BK-7 glass with 35 1 - and 5 14.5-nm excitation, respectively. The sharp peaks are due to Raman scattering from the polymer fiber optic. If one uses 35 1-nm excitation, maximum fluorescence from the fiber at its peak maximum at 2 0 0 0 0 - ~ m -is~ nearly a factor of 10 smaller than the intensity of the Raman band at 25 500 cm-'. With 5 14.5-nm excitation, there is negligible fluorescence from the fiber at 15 000 cm-' and none to the red side of 14 000 cm-l. Fiber fluorescence, therefore, does not interfere with these measurements and can be ignored. The selection of the detection wavelength is dictated by the fluorescent properties of the substrate and adsorbate as shown schematically in Figure 4b. Ideally, the detected wavelength discriminates against unwanted fluorescence
from the adsorbate and is positioned on a fluorescence maximum of the substrate. Fortuitously, fluorescence from this particular silane does not overlap the glass fluorescence and therefore, using 351-nm excitation, the large band at ca. 18 000 c-l can be monitored. Obviously, use of this more intense fluorescence band gives a better signal-to-noise ratio. The underlying principle of this experiment is that, by monitoring the decrease in fluorescence of the glass with path length, the absorption due to any component such as the substrate, the monolayer, or the solvent can be separately determined. Since the slide is divided into coated and uncoated regions, the scattering and absorption losses due to the substrate and solvent can be measured separately and eliminated from the total loss in the coated region, allowing the absorptive loss of the adsorbate to be measured directly. It is interesting to note that this is the surface analogue of a dual-beam absorption measurement. There are two distinct advantages to the coating geometry shown in Figure 3. One is the ease of restricting the area of the coating to a specific region. The second is to allow signal and background to be obtained in the same experiment as stated above. The slope obtained in the uncoated region is the background. The absorbance loss coefficient measured here is due to glass absorption and solvent absorption. The absorbance measured in the coated region is due to all of the possible absorption processes in the uncoated region plus the additional absorption by the coating monolayer. Subtracting the uncoated slope from the coated slope gives the absorption loss coefficient due to the monolayer alone. Performing the experiment with both T E and T M polarizations obtains the dichroic ratio, p, used to calculate the orientation angle, B (see Figure 1). Statistical consideration^.'^ The t distribution was frequently used to test the null hypothesis, which stated that the coated slope is not statistically different from the uncoated slope; that is, there is no increased absorbance due to the monolayer. The t value can be calculated by lC =
[
+
mc - mu ( M - l)EV,
( N - l)EV, N+M-2
(9)
]k i) +
where m, and mu are the coated and uncoated loss coefficients respectively, N a n d M a r e the number of points used to ascertain the coated and uncoated losses, and EV, and EV, are the error variances for the coated and uncoated losses. The error variances are calculated from
where Ri is the residual for the ith data point. These experiments gave values of t, from 1.O to 6.4, corresponding to confidence levels for rejecting the null hypothesis, Le., for accepting the reality of adsorbate absorption, of 90-99.9%. Throughout the experiments, however, great importance was placed upon balancing large t, values, which can be obtained from a large number of measurements, and minimizing the experimental data acquisition time. Selection of a reasonable number of collection points narrowed the experimental time to less than an hour for each polarization. The uncertainty in the orientation angle, AB, can be calculated by using the uncertainty in the dichroic ratio, Ap, from the following equation AB = (dO/dp)Ap (1 1) where dB/dp is given by the expression
/ (19) Box, G.; Hunter, W. G.; Hunter, J. S.Statistics for Experimenters; Wiley: New York, 1978.
6456 The Journal of Physical Chemistry, Vol. 94, No. 16, 1990
Cropek and Bohn
TABLE I: Loss Coefficients from Three Samples Coated in the Same Batch"
TE
TM
slope
slope
sample
uncoated
coated
IC
uncoated
coated
1,
1 2 3
-0.086 I -0.0752 -0.0794
-0.121 -0.0951 -0.1 17
2.263 0.953 2.124
-0.08 16 -0.071 1 -0.0895
-0.129 -0.105 -0.133
3.499 1.176 1.635
6
A6
61.0 58.2 63.0
1.9 4.2
2.5
av 60.7 f 2.4
'Slopes are given in units of mm-' and angles in deg.
Ap is calculated knowing the standard errors of the loss measurements Ap = [AaTE'
+ PaTM2]'/*
(13)
2.0
-
where I % -
A a = s/N112
(14)
where s is the standard deviation of the loss measurement and N is the total number of points used to calculate the slopes in the coated and uncoated regions. The small value of dO/dp in eq 12 results in an uncertainty in the orientation angle that is exceedingly small. For example, using typical data where t , for the T E polarization is 2.2 and t, for the T M polarization is 3.5, we obtained an orientation angle of 61.0' with an uncertainty of only 1.9'. Silylated Surfaces. Experiments were first done on uncoated substrates to verify agreement with the expected behavior. Highly linear behavior was observed for both T E and T M polarizations, thus validating the use of these glass cover slips as homogeneous substrates. The coating material was chosen with a dual purpose. First, it was desired to examine a material often used as a reversed-phase liquid chromatographic (RPLC) stationary phase, for later experimentation. Second, the absorbate must have some appreciable absorbance in the wavelength region accessible with the available instrumentation. The long alkyl chain RPLC stationary phases, while easily the most common stationary phases, have absorbance bands in the vacuum-UV region which make them unusable. A phenyl phase was selected because the K-a* absorption band was much closer to the 35 1 -nm short wavelength limit of our experiment. Diphenylsilane adsorbates were initially selected but were abandoned for two reasons. The steric hindrance of two phenyl rings can significantly decrease the surface coverage, and the bifunctional nature of this silane starting material allowed polymerization off of the substrate surface. In addition, interpreting the results of an experiment using this silane could be clouded by different positionings of the two absorbing phenyl groups on the same molecule. Thus, a monofunctional, single-phenyl silane was preferable. Chlorodimethylphenylsilane became the coating of interest. Its structure when bound to the surface and its absorption spectrum are shown in Figure 5. The phenyl ring is drawn with a finite thickness for ease of viewing. The structure is drawn in accordance with previous N M R experiments done on similar silanes adsorbed on silica gel.20 By analyzing the chemical shift anisotropy in carbon- 13 NMR, Slotfeldt-Ellingsen and Resing determined three main geometrical parameters for these types of silanes: (1) Sirurfsce-O bonds are oriented approximately 35' from the surface plane, (2) 0-Si bonds are perpendicular to the surface, around which there is free rotation at room temperature, and (3) the torsional angle, 6, between the 0-Si