J. Phys. Chem. 1991, 95, 7345-7353
7345
Study of DlmethyloctadecybiiyC1HodHied Silica Using Solid-State 18C NMR Line Shapes and Relaxation Techniques Robert C.Zeiglert and Gary E.Maciel* Department of Chemistry, Colorado State University, Fort Collins, Colorado 80523 (Received: July 27, 1990; In Final Form: April 19, 1991 1
The structure and dynamics of dimethyloctadecylsilyl-modifiedsilica gel (Cls-silica)were studied by using "C solid-state NMR line shapes and relaxation measurements. The relative peak intensities observed in the I3C MAS NMR spectra of C18-silicawere found to be dependent on whether cross-polarization (CP) or single-pulse (SP)excitation is used,an observation that is interpreted in terms of the mobilities of the silica-bound silane groups. The I3C MAS spectra in the absence of 'H decoupling show that Cls-silicawith a relatively low surface coverage of silane groups exhibits a marked lack of motion for all segments of the silane groups when no "wetting liquids" are present on the Cls-silicasurface and that the C18-silanechains become more mobile on the addition of common liquids such as water, acetonitrile, or cyclohexane. I3C relaxation studies show that the rate of reorientation of the silane groups is near 50 MHz and that there is a large dispersity of structural environments on the C18-silicasurface.
Introduction Modified silicas are commonly used as stationary h a m in both liquid chromatography and gas chromatography.l! In the past 10 years modified silica surfaces in which the surface-modifying groups are chemically bound to the silica surface have become very popular because of their great stability, particularly at high temperatures. By far the most commonly used type of liquid chromatographic stationary phases are the alkylsilyl silicas.2 Dimethyloctadecylsilyl-modified silica (CI8-silica)is a representative and very widely used member of this class of useful chromatographic stationary phases. A key to the continued development of new and useful chromatographic separations with this type of stationary phase is a fundamental understanding of the surface conformations and dynamics of the alkylsilyl moieties, as well as an understanding of the interactions between C,,,-silica and solvent/solute systems. Many techniques have been used to study these kinds of materials. The most commonly used technique is the correlation of chromatographic performance with changes in the composition of the chromatographic stationary Fluorescence has also been used to study the dynamics in silyl-modified silica
system^.^.' Nuclear magnetic resonance spectroscopy has proven to be a very useful technique for the study of chromatographic stationary phases. I3C NMR spectroscopy has been used to study species adsorbed on silica surfacesel0 and to distinguish different silane structures bound to silica Such studies show that to best resolve the I3C NMR spectrum of surface-bound alkyl silanes, it is necessary to remove the line-broadening influences of chemical shift anisotropy and 'H-I'C heteronuclear dipolar coupling by using magic-angle spinning and high-power proton decoupling. The dynamics of n-alkylsilyl silicas have been studied by using the cross-polarization time constant, TCH,and the spin-lattice relaxation time constant Tl.12+13i1b25 A TCHstudyI6 concluded that the outer ends of the n-alkyl chains are more mobile than the ends that are near the covalent anchor to the silica surface, while the TI and studies suggested that the methyl groups in the CI8 silyl moieties serve as a source of spectral density for relaxation of all of the protons in the C18-silicasystem. Gilpin and Gangoda have used isotopic 13Cenrichment to select specific carbons in CI8-silicasamples and have studied the line shapes as well as the relaxation behavior of C18-silicas,using essentially liquid-state NMR technique^.'^-^^ Bayer and wworken have used high-power 'H decoupling and MAS to narrow the NMR lines in the "C NMR spectra of 'Prcrcnt addreu: HIMONT Remrch & Development Center, 800 Greenbank Road, Wilmington, DE 19808. To whom corrmpondennce should be a d d r d .
OO22-3654/91/2095-7345$02.50/0
alkylsilyl-modified silica^.^^.^^ They have made extensive use of carbon TI measurements to study the dynamics of the system in the presence of different solvents. The various reported spin-lattice relaxation studies have, in general, shown that carbons in "liquid wetted" alkylsilyl silica samples and in general near the unbound ends of the n-alkyl chains exhibit larger TI values. The interpretation of the TI results has been difficult, since no previous studies have conclusively determined whether or not the alkylsilyl silica system is in the motional narrowing regime at any specific frequency, and the TI values often show only small changes (10-20%) for different solvent systems and with different surface silane concentrations.
( I ) Grushka, E.; Kikta, E. J., Jr. Anal. Chem. 1977, 49, 1004A. (2) Bonded Starionary Phases in Chromatography;Grushka. E., Ed.; Ann Arbor Science: Ann Arbor, MI, 1974. (3) Schunk, T. C.; Burke, M. F. Inr. J . Enuiron. Anal. Chem. 1986, 25, 81. (4) Morel, D.; Scrpinet, J. J . Chromatogr. 1982, 248, 231. (5) Lochmulkr, C. H.; Wilder, D. R. J . Chromatogr. Sei. 1979, 17, 574. (6) Carr, J. W.; Harris, J. M. Anal. Chem. 1986, 58, 626. (7) Beaufils.J. P.; Hennion, M. C.; R e t , R. Anal. Chem. 1985,57,2593. (8) Gay, 1. D. J . Phys. Chem. 1974, 78, 38. (9) Kaplan, S.;Rcsing, H. A,; Waugh, J. S. J . Chem. Phys. 1973,59,5681. (IO) Kohler, J.; Chase, D. 8.; Farlee, R. D.; Vega, A. J.; Kirkland, J. J. J . Chromarogr. 1986, 352, 275. (11) Akapo, S.0.;Simpson, C. F. J . Chromarogr. Sci. 1990, 28, 186. (12) Bayer, E.; Albert, K.; Reiners, J.; Nieder, M.; Muller, D. J . Chre malogr. 1983, 264, 197. (13) Bayer, E.; Paulus, A,; Peters, B.; Laupp, G.; Reiners, J.; Albert, K. J . Chromatogr. 1986, 364, 25. (14) Jinno, K. J . Chromatogr. Sci. 1989, 27, 729. (15) Claesscns, H. A.; DeHaan, J. W.; Van De Ven, L. J. M.; De Bruyn, P. C.; Cramers, C. A. J . Chromatogr. 1988, 436, 345. (16) Sindorf, D. W.; Maciel, 0. E. J . Am. Chem. Soc. 1983, 105, 1848. (17) Albert, K.; Even, B.; Bayer, E. J . Magn. Reson. 1985, 62, 428. (18) Gangoda, M.; Gilpin, R. K.;Fung, B. M. 1. Magn. Reson. 1987,74, 134. (19) Gilpin. R. K.; Gangoda, M. E. Talanra 1986, 33, 176. (20) Gilpin, R. K.; Gangoda, M. E. J . Magn. Reson. 1985, 61,408. (21) Gilpin, R. K.; Gangoda, M. E. And. Chem. 1984,56, 1470. (22) Gangoda, M. E.; Gilpin, R. K. J . Magn. Reson. 1983, 53, 140. (23) Gilpin, R. K.; Gangoda, M. E. J . Chromarogr.Sci. 1983, 21, 352. (24) Zcigler, R. C.; Maciel, G. E. Chemicully Modifled Surfaces. Chemically Modifled Surfaces in Science and Indusrry; Leyden, D. E., Collins, W., Eds.;Gordon and Breach Science Publishers: New York, 1988; Vol. 2, pp 319-336. (25) Maciel, G. E.; Zeigler, R. C.; Taft, R. K. Chemically Modfled Surfaces. Silanes, SurjOces and Inrerfaces;Leyden, D. E., Collins, W., as.; Gordon and Breach Science Publishers: New York, 1986; Vol. 1. pp 413-429.
Q 1991 American Chemical Society
7346 The Journal of Physical Chemistry, Vol. 95, No. 19, 1991
Another NMR approach that has been used to study the dynamics of alkylsilyl-modified silica is to observe the effects of motion on NMR line shapes. The general principle behind NMR line-shape analysis is that molecular motion will change the shape of and, in general, narrow an NMR resonance that is otherwise broadened by an interaction such as chemical shift effects, the heteronuclear dipolar interaction or a quadrupolar interaction. Line-shape analysis of broad 13CNMR resonances has been used in the study of motions in materials such as solid and and bonded to29*30 surfaces. small molecules adsorbed Line-shape analysis of I3C NMR resonances broadened by chemical shift effects has shown that carbons near the unbound ends of silica-bound n-alkylsilyl chains exhibit narrower lines than carbons near the surface-bound end of the silane group, indicating that the free ends of the n-alkyl chains are either more mobile or are structurally more homogeneous than the bound ends of the chains.21*22 A further study using similar techniques showed that the terminal methyl resonances of n-alkylsilanes bound to silica become narrower with increasing concentrations of dioxane in D20.20,23 We have chosen Cta-silica as a model system to study the surface structure and dynamics of these kinds of materials. Preliminary studies of the system showed that differences exist in the Ct8-silicasystem as a result of differences in surface coverage and the presence of solvent^.^^^^^ A wide-line ,H NMR study was run on C18-silicasamples similar to those described in this paper.3132 These 2H studies showed that the cta chains on Ct8-silica samples move at frequencies between 10 and 50 MHz and that a dispersity of surface environments and C t 8chain motions exists on the C18-silicasurface. 0
3
1
~
~
9
~
Zeigler and Maciel TABLE I: Solvent Louling Levels on C,,-silicn Sampleso silane loading level (silane/silica), mg/g 64 64 64 207 207 207
g of solvt added/g
solvt
HZO
CD3CN C6D12
H20 CD,CN C6D12
of silica 0.076 0.181 0.112 0.033 0.016 0.068
~
Experimental Section Synthesis of C18-silicasand Solvent-Modified C18-silicas. All solvents that are specified to be “dry” were commercially available reagent grade materials that were distilled from a sodiumbenzophenone mixture under N,. (a) silica Siylation. The alkylsilyl silica samples were prepared by adding a solution of dimethyloctadecylchlorosilane (DMODCS, from Petrarch Systems, Inc., Bristol, PA, used as received) in dry toluene to silica gel (Fisher S-679, evacuated at 0.020 Torr for 16 h at 180 OC prior to reaction) and then heating the mixture to 1 IO OC for 48 h. Each reaction mixture was then cooled and rinsed three times with dry toluene, and then the excess toluene was removed by heating the sample at 100 OC and 0.020 Torr for 16 h. The synthesis and all subsequent handling of the modified silica samples were carried out under a dry N2 atmosphere to minimize contamination of the sample by atmospheric water or oxygen. The high loading level (207 mg of silane/g of silica) sample was prepared by adding 88 g of an 80% (w/w) solution of dimethyloctadecylchlorosilanein dry toluene to 36.65 g of silica. The low loading level sample (64 mg of silane/g of silica) was prepared from the reaction of 28 g of a 5% (w/w) solution of dimethyloctadecylchlorosilanein dry toluene and 10.297 g of silica. The alkylsilyl loading levels were determined gravimetrically. (b) Solvent Addition to Cla-silicaSamples. Solvent vapor was added to the alkylsilyl silicas by placing the modified silica samples for 72 h in a chamber filled with nitrogen saturated with the solvent vapor. The gravimetrically determined solvent loading-level for each C18-silica-solventsystem is listed in Table I. NMR Instrumental Conditions. The 50.3-MHz 13C NMR spectra were run on a Nicolet NT-200 NMR spectrometer that had been severely modified to run solid samples. All of the NMR spectra were taken of samples sealed by torch under vacuum in (26) Bovey, F. A.; Jelinski, L. J . Phys. Chem. 1985,89, 571. (27) Majors, P. D.; Ellis, P. D. J . Am. Chem. Soc. 1987, 109, 1648. (28) Marshall, D. B.; McKenna, W. P. Anal. Chem. 1984, 56, 2090. (29) Kelusky, E. C.; Fyfe, C. A. J . Am. Chem. Soc. 1986, 108, 1746. (30) Boddenburg, B.; Grosse, R.; Breuninger. U. Sur/. Sci. 1986, 173, L655. (31) Zeigler, R. C. Ph.D. Dissertation, Colorado State University, 1989. (32) Zeigler, R.C.;Maciel, G. E. J . Am. Chem. SOC.,in press.
Solvent and silane loading levels measured gravimetrically. Estimated standard deviation f 0 . 0 0 2 g of solvent.
8-mm glass NMR tubes and spun at the magic angle at 2.2 kHz, with a spinning system based on a modification of a design by Gay.33 The basic Gay spinning system was extrapolated to spin 0.5 g of sample in 8-mm glass NMR tubes by machining the cylindrical bearing surface of a Gay-type stator so that it is 0.003 in. larger in diameter than the 8-mm NMR tube. This bearing surface in the stator allows the NMR tube itself to serve as the bearing of the spinner. Torque to spin the tube is provided by a small plastic cap that fits around the top end of the glass NMR tube. With 24 grooves cut into the plastic cap, this NMR tube assembly may be spun at 2.2 kHz/s with 30 psi of air pressure. The power levels of the carbon and proton channels were set so that the Hartmann-Hahn match was achieved at 55 kHz. The proton TI values were measured by using a modified inversion-recovery experiment34in which the proton polarization is inverted and then allowed to relax for a time T . At the end of this relaxation time the proton magnetization is rotated by 90° and cross polarization (CP) is used to transfer polarization from the protons to the I3C nuclei. After CP, the I3C NMR spectrum was taken. The relaxation behavior of the protons could therefore be observed indirectly through the carbon spectrum and, in addition, the TI for the protons associated with each resolved carbon resonance could be observed. The I3C TI values were determined by using the inversion-recovery techniquegsor the Freeman-Hill modification to the inversion-recovery e~periment.’~The carbon polarization was generated by using either singlepulse (90° pulse) excitation (SP) or using CP.37 The T,; values were determined by observing the decay of spin-locked polarization that had been generated by using single-pulse excitation. The TCH and TIPH values were determined from the double-exponentialcurve-fitting analysis of variable contact-time data.38 The 25-MHz I3C TI measurements were made using the NMR TI experiments described above on a home-built NMR spectrometer based on a Nicolet 1180 computer, a Nicolet 2938 pulse programmer, and a wide-bore Nalorac magnet. The sealed-sample spinning system was also used for these experiments. A 5-s repetition delay was employed in all NMR experiments. With the exception of some methyl I3C resonances in this study, this delay is at least 3 times longer than the relevant TI values (see Tables 11-IV). Results and Discussion 13CNMR Spectra of C18-silica. (a) I3C MAS Spectra witb IH Decoupling. To ensure that the C18-silicasamples were not contaminated by oxygen or water, it was necessary to run all of the samples in sealed glass tubes. The modified Gay-typeg3 torch-sealed glass spinner system described in the Experimental Section was used for this purpose. Figure 1 shows the singlepulse (SP) and cross-polarization (CP) MAS NMR spectra of a high-loading C18-silica (207 mg of si(33) (a) Gay, 1. D. J . Magn. Reson. 1984, 58, 413. (b) Shoemaker, R.
K.;Apple, T.M.J . Magn. Reson. 1986,67, 367.
(34) Sullivan, M.; Maciel, G. E. Anal. Chem. 1982, 51, 1606. (35) Lyerla, J. R.,Jr.; Levy, G. C. Top. Carbon43 N M R Spectrarc. 1974,
I . ,79. _
(36) Freeman, R.;Hill, H. D. W . J . Chem. Phys. 1971,51, 3367. (37) Torchia, D. J . Magn. Reson. 1978. 30, 613. (38) Demo, D. E.;Tegenfeldt, J.; Waugh, J. S.Phys. Reo. B 1974,11, 4133.
The Journal of Physical Chemistry, Vol. 95, NO. 19, 1991 7347
Dimethyloctadecylsilyl-ModifiedSilica A
I
CH3
I' I
Z 3
4-15
I6
17
18
C4-CI5
Cross Polrizrtlon
Slnplr Pulse
b 40 40
20
0
PPM
40
20
0
PPM
Figure 1. 50.3-MHz 13CMAS spectra of high-loading CIB-silicawith no adsorbed solvents (dry) and high-loading CIB-silicawith adsorbed liquids as indicated. The spectra on the left side were measured using a single-pulse (SP) technique (5-s repetition delay), and the spectra on the right were measured by using a cross-polarization(CP) technique (5-s
CP
I
D 20
(39) Pines, A.; Gibby, M.G.; Waugh, J. S. J . Chem. Phys. 1973,59,569.
W90 PPU
c
20 I
iiom SP
I
"
= ' 0 ' I
PPM
CP
.
Ill,
repetition delay). Asterisks indicate solvent resonances. lane/g of silica), in both the presence and absence of "wetting liquids", along with the peak assignments. The peak assignments were made from well-established additivity rules and have been reported previously.16 Two of the seven observed peaks result from overlapping signals from two or more carbons in the c18 alkyl chain. The S P and CP I3C spectra in Figure 1 exhibit different relative peak intensities for corresponding carbons in the various segments of the alkylsilane groups. The resonances of those carbons near the unbound ends of the chains, C16 and C18, appear more intense, relative to the C4-CI5 peak, in the S P spectrum, while the resonances for Cl', C1, and C3, those carbons near the anchored ends of the CISchains, are enhanced in the CP spectrum. A previous p ~ b l i c a t i o showed n ~ ~ that the same kinds of intensity differences are observed in the SP and CP I3C MAS spectra of both the high-loading C18-silicaand the low-loading C18-silica(64 mg of alkylsilane/g of silica) saturated with the vapor from a variety of liquids. The intensity differences observed between the SP and CP 13C MAS spectra of Figure 1 arise because each of the two excitation techniques is most efficient for a different state of molecular motion. The SP pulse sequence generates magnetization from spin-lattice relaxation, which is dependent upon the presence of motional components near the Larmor frequency, 50 MHz in this case. This excitation technique is therefore most efficient when vigorous molecular motions are present. In contrast to the SP pulse sequence, the CP excitation requires that a static component of the IH-l3C dipolar interaction be pre~ent.3~This static component can result from a lack of motion, unlikely in the case of the C,,-silica samples at room temperature, or nonrandom motion, that is, motion that is highly anisotropic and does not give effective motional averaging in all three orthogonal spatial dimensions. In the C P MAS spectra the resonances of those carbons near the silica-bound ends of the CISchains are more intense than they are in the SP MAS spectra. The higher intensities for those carbons near the anchored ends of the alkylsilyl groups in the CP spectra suggest that C P polarization transfer from the protons to these carbons is more efficient than it is for those carbons near the unbound ends of the CI8chains. Since CP polarization transfer is most efficient in the presence of static interactions, it appears that carbons CI-C3 are less mobile than those carbons near the unbound ends of the chains. This lack of mobility for the C1, C2, and C3 carbons could indicate either a correlation time, T,, of molecular motion that is larger than the inverse of the frequency that characterizes the C-H heteronuclear dipolar interaction (approximately 23 kHz) or a nonrandom type of motion that will
0
Figure 2. 50.3-MHz 13CMAS spectra, obtained with no IH dccoupling, of low-loading CIB-silicawith no adsorbed solvents (dry) and high-loading CIB-silicawith adsorbed liquids as indicated. The spectra on the left side were measured by using a single-pulse technique (5-s repetition delay), and the spectra on the right were measured by using a cross-polarization technique (5-s repetition delay). Asterisks indicate solvent resonances.
CI'
YO
20
0
PPU
YO
20
0
PPM
Figure 3. 50.3-MHz I3C MAS spectra, obtained with no IH dccoupling, of high-loading CIB-silicawith no adsorbed solvents (dry) and highloading Clrsilica with adsorbed liquids as indicated. The spectra on the left side were measured by using a single-pulsetechnique (5-s repetition delay), and the spectra on the right were measured by using a cross-po-
larization technique ( 5 4 repetition delay). Asterisks indicate solvent resonances.
result in a larger static component in the heteronuclear dipolar interaction. This nonrandom motion would not average the heteronuclear dipolar interaction as effectively as the more random motion experienced by the carbons near the unbound ends of the CI8chains. Either of these effects may be responsible for the more efficient cross polarization that is observed for those carbons near the silica surface, and it is impassible to determine which of these two effects is responsible on the basis of just the relative intensities observed in the CP and SP I3C MAS spectra. (b) I3CMAS Spectra with No Decoupling. The NMR spectra in Figure 1 were all obtained with high-power 'H decoupling, SO that sharp, well-resolved peaks could be observed; but by applying this high-power decoupling, one precludes the possibility of obtaining potentially useful information about the IH-W dipolar interaction. Like the relative peak intensities in the IH-decoupled C P MAS spectra, the observed line widths in a proton-coupled I3C MAS spectrum are dependent on the presence of a static component of the 13C-'H dipolar coupling. While a totally static C-H pair will exhibit a Pake pattern that is 23 kHz wide, the rapidly tumbling molecules in a liquid sample show essentially no line broadening associated with the heteronuclear dipolar interaction because of complete motional averaging. A requirement for complete motional averaging of the IH-l3C dipolar interaction is that 7;' for the random motion must be greater than about 23 kHz. Hence, any sharp resonances in the normal IH-decoupled I3C MAS N M R spectra of C18-silicas that remain sharp in a i H-coupled spectrum must represent CH, moieties that are moving rapidly, that is, with 5;' 1 23 kHz, and have rapid motion in all three spatial dimensions. The I3C SP-MAS NMR spectra in the absence of ' H decoupling have been published p r e v i o u ~ l y . ~ ~ The fact that the C P pulse sequence preferentially excites a slower moving component of the CI8silyl groups is shown by the
7348 The Journal of Physical Chemistry, Vol. 95, No. 19, 1991 loading level (silane/silica), mg/g 64 207
CI' 0.82 0.51
CI 0.87 0.52
c3 0.68 0.47
Zeigler and Maciel
C4-C 15 0.75 0.48
C16 0.83 0.51
C2
+ C17 0.78 0.49
C18 1.1 0.43
"Detected by I3Cvia cross polarization. Taken from ref 24. Estimated standard deviation *IO%, proton-coupled CP-MAS NMR spectra shown in Figures 2 and 3. The sharp, motionally averaged resonances observed in the S P spectra of these modified silica samples are absent or nearly absent in all of the CP-MAS spectra. This result is in agreement with the other CP/SP comparisons of spectra shown in Figure 1 and further demonstrates that the CP technique preferentially excites the slower and less randomly moving component of the CI8-silicasystem, while the SP technique excites the faster moving component of the.C18-silicasystem better than the CP method does. Relaxation Studies. (a) 'Hand I3CSpin-Lattice Relaxation ( TIHand TIC). Several relaxation studies were carried out to evaluate the sensitivity of the relaxation parameters to details of sample preparation and to evaluate the utility of relaxation studies for elucidating the configuration and motion in the CI8silicas. T I is the time constant that describes the rate at which a nonequilibrium component of the nuclear magnetization that is parallel to H,,returns to equilibrium. No detectable ESR resonances could be found in the high-loading C18silica with no wetting liquids, suggesting a negligible concentration of paramagnetic species in the silica used in this study. Therefore, we assume in the present work that the only relaxation mechanism that significantly contributes to the observed T,values in C18-silicais the dipolar relaxation mechanism, in which TI relaxation is due to the fluctuating magnetic fields that occur due to the reorientation of relevant 'HJH or IH-W internuclear vectors. To calculate exact theoretical values of T I or any other relaxation time for a specific chemical system, it is necessary to know in detail the rates of motion and the modes of motion of the system. The details of a system's motion are often unknown, or it may be impractical to calculate the theoretical TI values for a system in which a complex variety of motions exist. In this case the significance of any observed trends in relaxation times can often be determined qualitatively by comparing observed relaxation times to the predictions of a simple motional model. A commonly used motional model is that of isotropic, or totally random, m o t i ~ n . ~ ~ *For ' Q ~this ' model, the value of TI for a proton (TIH)that is relaxed by the reorientation of one IH-IH vector is described by the following equation:
The T I of a carbon (TIc) that is relaxed by the reorientation of one 'H-I3C vector is described by the equation l/TIC =
In these equations y is the magnetogyric ratio, r is the internuclear distance, and J(w) is the spectral density function at the frequency u.@ For the case of isotropic motion J(w)is given by the equation
J(w) = 2Tc/(1
+ u2~f2)
(3)
In cq 3, T , is the correlation time describing the molecular reorientation.a TICexhibits a minimum at a T~ value near l/wL, where wL is the Larmor frequency in r a d d . The presence of this makes it desirable to know whether a chemical minimum in TIC system mwcs with a value that is greater or less than the 13C (40) Farrar, T.C. Introduction to Pulse NMR Spectroscopy; Farragut Press: Chicago, IL, 1989. (41) Redfield, A. G. IBM J . 1957, I , 19.
a. 10.0
I
1 .o T:
0.1
I
10"
I
10. l o
I
I
loe
10'
I
I -7
10
106
T~ (set)
b.
-1 I
I
10"
\
Single Tc
I
10- l o
I
I
to9
I
I
l0*
106
7 , (set)
Figure 4. I3C spin-lattice
relaxation time constant ( T I c )as a function of the correlation time (io) for C-H pairs that are under oing isotropic motion: (a) demonstrates the effect of determining the TIt values at two different magnetic field strengths, based on eq 2; (b) effect of a distribased on plots in ref 45. bution of correlation times ( T J on TIC, Larmor frequency in order to relate the observed trends in TIC to relative rates of molecular motion in a particular chemical system. One method of determining whether a given system moves with a T ~ - I value that is greater than or less than the Larmor fr uency can be seen in Figure 4a, a plot (based on eqs 1-3) of T vs 7, for two different Larmor frequencies. One sees that T It! varies as a function of wL for T, values near l/uL or higher, but if T , is much lower than l/wL,then TICis independent of ut. If a chemical system is moving fast enough so that T, > l/oL the I3C{IH)NOE is constant at a value of 1.0. For systems with values of T , that are near 1/wL the value of the NOE is intermediate between 1 .O and 2.998, and in this motional regime the NOE is dependent on H,,.The effect of a distribution of T, values on the I3C{'H)HOE is to broaden the range of average T~ values over which the I3C{'HJNOE values are intermediate between 1.0 and 2.998.42 The experimental NOE values measured at 50.3 MHz for the dry high-loading C18-silicasamples are shown in Table V. The NOE values for the low-loading C18-silicasamples exhibit more
Zeigler and Maciel TABLE V: 'C{'H) Nuclear Overhauser Enhancements (NOE) for Drv Cwdlica SamDles" loading level (silane/silica), V k CI' CI C3 C 4 4 1 5 C16 C 2 + C 1 7 C18 64 mg/g 50.3MHz 25.3 MHz 207 mg/g 50.3MHz 25.3 MHz
2.0 2.5 3.4 1.7 2.0 1.5
2.3 2.0
2.1 2.2
2.4 2.0
2.5 1.5
2.3 2.2 2.2 1.6 1.9 2.0
2.3 2.0
2.1 2.0
2.4 2.0
2.1
"Single-pulse excitation used. Estimated precision f0.2. frequency.
1.5
Larmor
scatter than those for the high-loading C18-silicasamples, because of the lower signal-to-noise inherent in the I3C spectra of these samples. The NOE values reported in Table V show that the C18-silicacarbons move with an average rclvalue near 50 MHz. However, these NOE values cannot be interpreted quantitatively in terms of the predictions of eq 4, which was constructed from a theory based on a model of isotropic motion. Apparently, the lower NOE values exhibited by each C18-silicasample a t 25.3 MHz cannot be accounted for in terms of theories that do not account for anisotropic molecular motion explicitly. (c) Rotating-Frame Spin-Lattice Relaxation ( TIPH and TIPC). The rotating-frame spin-lattice relaxation time (TIP) is a measure of molecular motion at the precessional frequency in the magnetic field induced by a radio-frequency spin-locking field,"4*4s55 kHz in the present case. The value of TI, that is expected if the rf field ( H I ) is greater than the local dipolar field Hd and if T~ is much less than the rigid-lattice T2 is given by the e q u a t i ~ n ~ , ~ ~ The (1 - q ) factor in eq 5 is a function of the geometry of the system's motion and is independent of the rate of motion. The minimum value of TI, will be observed when 7,is near l / q , where wIis the precession rate of the nuclei in the spin-locking rf field. As long as 7c is less than ulrthe larger relative values of T I ,can be interpreted as indicating greater rates of motion in the system under study. The value of w1 is 55 kHz for both the 13Cand 'H nuclei in the present study. TIPC values for the high-loading and low-loading C18-silicaswere measured by using SP MAS techniques and are published elsewhere;24they are listed in Table VI for the convenience of the reader. TIPHdata were measured via I3C, using the CP-MAS variable-contact-time technique and are listed in Table VII. The patterns of comparison between the TIPCvalues for the dry/ high-loading and dry/low-loading C18-silicasare similar to those between the SP-based TICvalues in Table 111. For the low-loading sample, C1 carbons at the anchored ends of the CISgroups exhibit larger TIPC values and the carbons near the unbound ends of the c 1 8 chains exhibit smaller TIPCvalues. For the high-loading C18-silicasample, the measured TIPH values are larger for protons near the unbound ends of the c18 chains than for the protons near the anchored ends of the c18 chains. A serious limitation in TIPHfor the present study is apparent in Table VII. Due to the high rf power levels that are required to maintain a spin-lock in solid samples, it is technically demanding on the rf hardware to keep the nuclei spin-locked for longer than about 20 ms. All of the solvent-loaded CI8-silicasamples exhibit TIPH values that are much longer than this 20-ms limit, so it is values. A relevant impossible to determine accurately these values ISthe likely possibility factor in the interpretation of the TIPH of flip-flop spin exchange between the protons during the spinlocking period of the T I , experiment. Rotating-frame spin exvalues, thus decreasing change leads to an averaging of the TIPH the ability of the TIPHtechnique to resolve differences in motion for different regions of the C18-silicasystem. Nevertheless, a substantial degree of diversity is seen in the TIPHvalues for a (44) Douglass, D. C.; Jones, G. P.J. Chem. Phys. 1966, IS, 956. (45) VanderHart, D. L.; Garroway, A. N. J. Chem. Phys. 1979,71,2713.
The Journal of Physical Chemistry, Vol. 95, NO. 19, 1991 7351
Dimethyloctadecylsilyl-Modified Silica TI: Values (9) for CI1-dlier Simples' loading level (silane/silica), CI mg/g CI'
TABLE VI:
0.10 0.055
64 207
0.37 0.022
C4415 0.034 0.041
c3 0.025 0.022
" Measured by single-pulse experiments. Estimated standard deviation *IO%.
C16 0.038 0.05 1
C2 + C17 0.049 0.053
C18 0.59 0.20
+ C17
C18
Taken from ref 24.
TABLE VII: TIAH Values (SI for CI.-silica Samples' loading level
(silane/silica), added solvt
CI' 0.04 0.082
>o. 1 >o. 1
0.022 0.020
>o. 1
0.017
" Measured by "C
CI 0.033
>o. 1 >o. 1 >o. 1
0.020 0.021 >o. I 0.020
c3 0.016 0.062 >0.1 >0.1 0.024 0.019
>o. 1
0.021
C4-C 1 5
C16
0.040
>o. 1 >0.1 SO.1
0.05
>0.1 >o. 1 >o. 1
>0.1 >0.1 >o. 1 >o. 1
>0.1
>0.1
0.032
0.030
>O.l
>o. 1 >o. 1
0.054 0.035
>o. 1
C2
0.095 0.024
0.036
0.026
>o. 1 >o. 1
>o. 1
>O.l >O.l
>o. 1
via variable contact-time cross-polarization experiments. Estimated standard deviation &IO%.
typical sample reported in Table VII. Though the computer fits of the variable contact-time data are in general excellent for these samples, one might question the origin and nature of this specific kind of diversity and the absence of another kind of diversity-i.e., or TcH component is detected the fact that not more than one TIPH for a given I3C peak. This degree of skepticism has its basis in the fact that both TIPHand TCHwere derived from variable contact-time data, from which it would be difficult to detect different TIPHand TcH components for a particular peak. While the concerns described immediately above limit our ability to use TIOH as a definitive probe of the C18-silicasystem, some provocative trends are nevertheless observed in both the TI: and the TIOH results. Both the CP-based TI," values and the SP-based TIPC values exhibit identical trends: an increase in surface silane loading level leads to an increase in TI, for most carbons and protons, except for those on Cl', C1, and C3, carbons near the silica surface. The TICand NOE data show that the motional I;T values of the CISchains in C18-silicaare near 50 MHz; therefore, the c18 chain motion certainly occurs with a I;T greater rate than 55 kHz. Under this circumstance the C18-silica system is certainly in the motional narrowing limit for T!,Cor TIPH, and it may be stated that an increasing T I Pvalue indicates a higher frequency of molecular motion. According to this interpretation, the TIPC and TIPH data indicate that an increase in surface silane loading increases the motional rate for those carbons near the unbound ends of the C I 8chains, while it appears that the same increase in surface silane loading may decrease the motional rate of those carbons near the silica surface. This resuli is consistent with the results of the TI studies reported above. It is worth stressing that these changes in TI, with loading level do not involve more than a factor of 2, so these conjectures conceming the effects of surface silane loading-level on the c18 group mobility are not overwhelmingly persuasive. The effect of the addition of solvents to the low-loading C18-silica sample is, in contrast to the change in loading level, quite dramatic. When solvent was added to the low-loading C18-silica,the TIPH values increased to the point that accurate measurements could not be made by using the available hardware. This dramatic shows clearly that rC of the c18 chains in lowincrease in TIOH loading C18-silicadecreases with the addition of any of the liquids data to draw studied here, but it is not possible to use these TIPH any conclusions concerning the relative effects of different solvent systems on the C18-silicasystem. In the case of the high-loading C18-silica,little liquid-wetting effect was observed in the TIPHvalues, except in the case of deuteroacetonitrile, where the T I , values increased beyond the range of measurement. This apparent decrease in the motional
TABLE VIII: TCHValues (ms) for C18-silicaSamples with and without Added Solvent9
loading level (silane/silica), addedsolvt
CI'
CI
C3
C4-CI5 C16 C 2 + C 1 7
C18
0.22 0.09 0.17 0.27 0.17 0.23 b 0.18 0.22 0.190.120.15
0.17 0.23 0.27 0.35
0.25 0.42 0.35 0.66
0.28 0.36 0.34 0.28
1.31 2.81 1.67 2.74
0.22 0.28 b 0.20
0.20 0.24 0.25 0.15
0.25 0.32 0.44 0.20
0.22 0.28 0.42 0.27
1.18 1.59 3.37 1.62
0.14 0.19 0.18 0.13
0.14 0.22 0.19 0.13
"All values run at 50.3 MHz unless noted otherwise. Estimated standard deviation & 10%. Measured via variable-contact-time crosspolarization experiments. bNot determined due to solvent peak interference. T~ of the C18groups in C18-silicais very interesting in light of the fact that the acetonitrilewater system is one of the most effective mobile phases for use with these C18-silicastationary phases in chromatography. It is possible that acetonitrile interacts particularly well with the C18-silicasystem because it can interact with both the polar silica surface and the nonpolar CI8chains. Under this circumstance, the acetonitrile would increase the mobility of the c18 chains and thus make the stationary phase more accessible to chromatographic solutes than either water or cyclohexane can. (d) Cross-Polarization Relaxation Time ( TcH). Values of the cross-polarization time constant, TCH,were obtained for the C18-silica samples via variable-contact-time experiment^.'^ The results of this study are shown in Table VIII. TCHis a measure of static I3C-lH dipolar interactions, and a smaller TCHvalue is associated with a less mobile and/or shorter C-H internuclear vector. An expression for the CP time constant, TCH,was obtained by Demco et al.38 If both the IH spin-locking irradiation and the I3C spin-locking irradiation are on resonance and the spin-locking fields generate magnetic fields that are much stronger than the local dipolar fields in the sample, then TCHfor a single IH-I'C pair is described by the equation
where Awemis the difference between the precessional rates of the 13Cnuclei and the IH nuclei spin-locked in their respective effective fields. Jx(Awem) describes the spectral density of IH-I3C dipolar
7352 The Journal of Physical Chemistry, Vol. 95, No. 19. 1991
TABLE IX: TIc Values a d Line Widths for C,,-silica h m p l e s effective obsd natural no. of sample line width: line width: contributing Hz Hz resonancesd resonance T 2 c s~ Low-Loading C18-silicawith No Added Solvents 27 0.053 162 6.0 CI' 6.2 0.031 64 10.3 CI 10.3 0.032 103 10.0 c3 5.8 58 10.0 C4-CI5 0.032 44 8.8 5.0 0.036 C16 40 6.5 6.2 0.049 C2 + C17 48 67 1.4 C18 0.22
0.14
43
2.3
I
19
Determined by using SP excitation and Carr-Purcell experiments. Estimated standard deviation *lo%. bMeasured from SP MAS spectra. CNatural line width = I / ( r T Z c ) .dEffective number of contributing resonances = (observed line width)/(natural line width).
fluctuations at the frequency Aw,~. In eq 6 MzCHis the second moment of the I3C resonance due to the 13C-'H dipolar interaction,M which can be described for a given "C-IH interaction by the equation4'
(7) where the ( 1 - 3 cos2 0 ) term is the time average of the trigonometric factor in the angle between the C-H internuclear vector and Ho.The degree of averaging is dependent on the mode of motion as well as the frequency of the motion; isotropic motion at a rate much greater than 23 kHz averages MzCHto zero (the case of nonviscous liquids). Inspection of Table VI11 reveals that the TCH values for the C18-silicasystem do not change significantly with changes in surface silane loading level, although a complex array of changes occurs in the observed C18-silicaTcH values upon the addition of solvents. In the case of both the low-loading and high-loading C18-silicas,the addition of solvents generally increases the TCH values for the carbons in the CI8-silica,indicating a general increased motional averaging in bound C18groups. It is interesting to note that added cyclohexane has a smaller effect on the TCH values for carbons (21'423, those carbons near the silica surface, than more polar solvents do. The TCH values, which are smaller for those carbons relative to those of C4-Cl8, indicate less motion for those alkylsilyl carbons near the silica surface, and it appears that cyclohexane does not affect those carbons near the silica surface as strongly as more polar solvents do. This lack of effect could be interpreted in terms of a relative lack of cyclohexane molecules this close to the polar silica surface, perhaps indicating that the cyclohexane preferentially interacts with the nonpolar C I Schains of the surface silane groups. The pattern of smaller TCHvalues for C18-silicasamples with cyclohexane than with more polar solvents is less clear for the carbons from C4 to C18, compared to the C1-C3 segment. The TCH values for C4-Cl5 and C16 are actually smaller in the case of the high-loading silica with cyclohexane than those for the dry high-loading C18-silicasample. This reduction in TcH may indicate that the cyclohexane is present among the C18chains and causes steric restrictions to the motion for the center sections of these long hydrocarbon chains. (e) Spin-Spin Relaxation ( T2).Tzc values, as measured by using spin-echo SP MAS experiments, for the high-loading and (46) Abragam, A. The Principles of Nuclear Magnetism;Clarendon Press:
Oxford. 1961. -----
low-loading C18-silicasamples with no solvents present, can be found in Table I X . TZ-l is the rate constant that describes the dephasing of the macroscopic spin magnetization component that is perpendicular to H,,:For a I3C nucleus that is relaxed entirely by dipoledipole relaxation involving a single proton and undergoes isotropic motion the value of the I3C T2 may be calculated by using the equation3s
~ J I ( w H+) ~ J & H + WC) ( 8 )
High-Loading C18-silicawith No Added Solvents 70 7.2 9.7 CI' 0.044 0.017 52 18.7 2.8 CI 43 14.5 3.0 c3 0.022 34 8.0 4.2 C4-CI5 0.040 0.043 21 7.4 2.8 C16 33 4.8 6.9 C2+ C17 0.066 C18
Zeigler and Maciel
(47) Mehring, M . High Resohtion N M R in Solids; Springer-Verlag: Berlin, 1983.
The J(0) term in eq 8 shows that T2 is dependent on static interactions in the system as well as the high-frequency motions that dictate the observed value of T I . The minimum value of the full width at half-maximum (fwhmh) that can be observed in a NMR experiment, the "natural line width", is l / ( r T z ) .If the line width observed in the NMR spectrum is greater than the natural line width, then an exchange broadening or an inhomogeneous broadening (e.g., chemical shift dispersion) must be present." The measured T2cvalues are nearly all within 20% of the T I / values reported in Table VI. This near equality of the T2cand TIPcvalues, considering the € and I,,T, dependences of TIc,places the C18-silicaroughly in a "limited motion" regime, in which T;I in C18-silicais in the region that is nominally between the Larmor frequency of 13C (50 MHz in this case) and the precession frequency of the spin-locked 13C in the TI/ experiments (55 kHz in this case).@ The upper boundary of the limited motion regime is not precisely at 50 MHz; thus, the above observation that C18-silicais in the limited motion regime is consistent with the interpretation above that the T;I value characterizing motion of C18-silicais near or slightly above 50 MHz. The natural line width of an NMR resonance is equal to 1/ ( ~ 7 ' ~ The ) . natural line widths for the dry low-loading and high-loading C18-silicasamples were calculated from the measured Tzc values, and these natural line widths are presented along with the observed line widths for the C18-silicasamples in Table IX. The actual line widths observed for the C18-silicaresonances all e x 4 their natural line widths by a factor of 2.8 or more, so an inhomogeneous line-broadening interaction is present in the C18-silicasystem. The most likely line-broadening interaction in this case is chemical-shift dispersion, which would indicate that the carbon nuclei in the C18chains of C18-silicaare present in a variety of surface environments. This variety of surface environments indicates that the surface topography of silica probably forces the CI8chains into a variety of conformations and interactions with other components of the C18-silicasystem.
Conclusions Comparison of the observed line widths of the C18-silicaresonances to the natural line widths calculated from measured T2 values indicates that the CI8chains in C18-silicaexist in a variety of environments. The irregular surface topography of silica a p parently forces the C18chains of C18-silicainto a variety of surface configurations. An obstacle to interpreting I3C-based relaxation data for the C18-silicasystem quantitatively as a function of the surface silyl loading and solvent addition is that the observed variation for any specific relaxation parameter when sample preparation parameters are varied systematically is generally small, from 0 to 5096, and in some cases smaller than the estimated variance in the relaxation data. It is nevertheless possible to draw some qualitative conclusions concerning the motion of the CI8chains in the C18-silica systems. The TIcand NOE results show that the carbons in the C18-silica system move with an average r;l near 50 MHz. It follows that the TIPHand TIPCmeasurements are in the motional narrowing and TIPC values are proportional limit, so that the measured T I P H to the average T;I of the CI8chains. Since in general the T I , values measured for different portions of the CISchains do not differ by more than a factor of 5 for any given C18-silicasample,
J. Phys. Chem. 1991, 95,7353-7357 it may be concluded that TCI values for all of these carbons do not differ by more than this factor of 5. The only exception to this rule is C18,the terminal methyl carbon. The very large T I , TIP,and TcH values observed for CI8result from the combination of the reorientations of the CISchains and the rapid local rotation of these methyl groups at room temperature. In general, the measured quantities T I ,TI,,or TCHshow only small high-loading vs low-loading differences in the motional behavior of the C18chains, as reflected in T ; ~ . The TICresults suggest that the high-loading sample exhibits a lower T;I than the low-loading sample. No consistent trend is observed for the T I ,and T C H values measured for dry C18-silicas. The relaxation measurements indicate that any changes in motion associated with changing the loading level are small, probably involving changes in T;] of less than a factor of 2. The addition of wetting liquids introduces regular changes in the relaxation parameters. In the case of the low-loading C18-silica, both TI,and TCH increase upon the addition of either polar or nonpolar wetting liquids, indicating that motion in the C18-silica chains increases in l;7 and/or amplitude. The addition of acetonitrile to the high-loading C18-silicaincreases TI, and TcH,suggesting an increase in T ; ~ for alkylsilyl chain motion, as in the low-loading case. The decrease in T I P H upon the addition of H20suggests that H 2 0 causes the motion
7353
of the clg chains to be attenuated, this suggests that water induces the hydrophobic CISchains to clump together in a "separate phase", perhaps in a "haystack" configuration.'s The general pattern that emerges from line-shape and relaxation studies of the C18-silicasystem is that the C18-silicasurface is complex and heterogeneous and that there are no dramatic changes in the motional frequency spectrum for the various carbons in the C18-silicasystem as the sample is varied. This lack of dramatic changes in measured relaxation times as sample characteristics are varied indicates that the substantial differences observed between the proton-coupled I3C MAS spectrum of the dry lowloading C18-silicaand the protoncoupled 13CMAS spectra of the other C18-siIicasamples may be related to the mode of motion rather than the frequency spectrum of CI8-chainmotion. Such changes in the mode of motion are not accounted for in the equations cited above, which are based on an isotropic model of motion. Acknowledgment. We gratefully acknowledge partial support of this work by NSF Grant CHE-8610151 and use of the Colorado State University Regional NMR Center, funded by National Science Foundation Grant No. CHE-86 16437. (48) Karch, K.; Sebcstian, I.; Halasz, 1. J . Chromurogr. 1976, 122, 3.
Energy Transfer and Migration in Langmuir-Blodgett Films: Monte Carlo Simulations S. Blonskit Center for Materials Characterization, University of North Texas, P.O.Box 5308, Denton, Texas 76203-5308
and K. Sienicki**t DCpartment de Chimie, UniversitC de Montreal, C.P.6128, Succursale A, MontrCal. Quebec, Canada H3C3J7 (Received: July 30, 1990)
Monte Carlo simulations were used to investigate energy migration and transfer in twedimensional Langmuir-Blodgett films. A general procedure of Monte Carlo simulations of excitation transport is presented and discussed. The results obtained
for quantum yields of donor fluorescence, fluorescence depolarization,and donor fluorescence decays reveal two important characteristics. The first is that it is not necessary to include an orientational factor in the description of energy migration and transfer. Second, all of the mentioned photophysical observables are dependent not only on the concentration of molecules but also on the ratio of Fbrster radii. A comparison with the available theoretical models of excitation transport in two-dimensional structures shows that their agreement with Monte Carlo simulations is satisfactory only at low concentrations. An increase of concentration leads to the situation where the configurations of not only donor molecules but also acceptors which come into play are significant.
1. Introduction
Within the past two decades there seems to have been a renaissance in the studies of Langmuir-Blodgett films.'+ Monomolecular films or multilayers transferred onto solid substrates have unique physical properties that can be controlled during an assembling procedure. Although the extent of the desired controllability of the structure of Langmuir-Blodgett films is still limited, one can foresee further progress. The search for supramolecular devices based on Langmuir-Blodgett films has just started.I0 In spite of the important technological implications of these studies, basic scientific questions in regard to the structure and properties of Langmuir-Blodgett films have to be addressed. Since Kuhn's first work," fluorescence spectroscopy was extensively used in studies of Langmuir-Blodgett films. Experimental studies of transient and stationary photophysical observables of chromophores embedded into Langmuir-Blodgett films. 'On leave from the Institute of Technical Physics and Applied Mathematics, Technical University of Gdansk, 80-952 Gdansk, Poland.
Experimental studies of transient and stationary photophysical observables of chromophores embedded into Langmuir-Blodgett films have yielded a number of important results in regard to the structure and properties of these films.'+ In recent years, it became more and more evident that chromophoresembedded into (1) Kuhn, H.; Mabius, D.; Bbher, T. In Techniques ofChemisrry; Weiaberger, A., k i t e r , B. W., Eds.; Wiley: New York, 1972; Vol. 1, Part 3B, pp 577-702. (2) Blumberg. W. E. In Physical Methods on Biological Membranes and Their Model Systems; Conti, F., Blumberg, W. E., de Gier, I., Pocchiari, F., Eds.; NATO AS1 Scrim; Plenum: New York, 1985. (3) Roberts, G. 0. Contemp. Phys. 1984, 25, 109. (4)Roberts, G. G. Adu. Phys. 1985, 34,475. ( 5 ) Proceedings published in Thin Solid F i l m 1983,99,and 1989, 179. (6) Blinov. L. M. Sou. Phys. Usp. 1988, 31, 623. (7) Blinov, L. M. Russ. Chem. Reo. 1983,52, 713. (8) Kuhn, H.J . Phorochem. 1979, 10, 1 1 1 . (9)Mabius, D.Acc. Chem. Res. 1981, 14, 63. (IO) Barraud, A. J . Chim. Phys. Phys.-Chim. Biol. 1988,85, 1121. ( 1 1) A summary is given by: Kuhn, H. Narunvissenscha/ten 1967, 51, 429. See also: Zwick, M. M.; Kuhn, H. Z . Nururjorsch. 1962, 17A, 411.
0022-365419112095-7353502.50/0 0 1991 American Chemical Society