J . Phys. Chem. 1992, 96, 7506-7509
7506
m
The structural motif reported here for the H--water clusters is in general agreement with the average structures observed for Cl--water clusters in molecular dynamics simulations9 using many-body interaction potentials. Simulation of dilute ionic solutions with these same potentials predicts that the waters in the first hydration shell are, on the average, symmetrically distributed around the ion. In solution, the water molecules of the first solvation shell bond more effectively with the molecules in the second solvation shell than with themselves. In the absence of the second solvation shell, however, the structure of anion-water clusters appears to be very different than that of the anion and its first solvation shell. Calculations are currently underway for the F ( H 2 0 ) , and OH-(H,O), systems. For the n = 1 and 2 fluoride- and hydroxide-water clusters, we observe structural trends similar to the ones reported here for the hydride-water clusters.
A
Top vie\$
Sitlv v i e w
Figure 1. Optimal structures for the H-(H,O), clusters, for n = 1-3, from MP2 calculations with an aug-cc-pVDZ basis set.
eometry has C3symmetry with all H-H- distances equal to 1.601 1 ,all 0-0distances for the hydrogen bonds equal to 3.134 8, and all 0-H-O hydrogen bond angles equal to 165.8'. The calculated geometries of the water molecules in the clusters do not exhibit large variations from the calculated geometry of the monomer, namely, ROH = 0.965 A and 8HOH = 103.8'. The bond angles of the water molecules in the clusters vary from 98.3' to 100.3'. For the 0-H bond lengths we distinguish between those participating in hydrogen bonding between the water molecules, for which the computed range is 0.9674.970 A,and those bonded to the ion for which the range is 1.003-1.035 A. The calculated binding energies (including zero-point energy corrections) for successive addition of water molecules to the anion-water clusters mn.n-1
= En - (En-l + E H 2 0 )
vary little for the first three members of the series. As noted above, the first water molecule is calculated to be bound by 17.4 kcal/mol (AElp). The second water molecule is somewhat less bound, AE = 14.8 kcal/mol. (As noted above, AE2,! has been estimate$; to be 10-15 kcal/mol.) Finally, the binding energy of the third water molecule is calculated to be 13.7 kcal/mol.
Acknowledgment. This work was performed at Pacific Northwest Laboratory under the auspices of the Division of Chemical Sciences, Office of Basic Energy Sciences, U.S. Department of Energy, under Contract DE-AC06-76RLO 1830. Part of the computer resources for this work were provided by the Scientific Computing Staff, Office of Energy Research at the National Energy Research Supercomputer Center (Livermore, CA) and Florida State University (Tallahassee, FL). References and Notes (1) (a) Armbruster, M.; Haberland, H.; Schindler, H. G. Phys. Rev. Lett.
1981, 47, 323. (b) Henchman, M. J.; Paulson, J. F. Radiat. Phys. Chem. 1988,32,417. (c) Hierl, P. M.; Ahrens, A. F.; Henchman, M. J.; Viggiano, A. A.; Paulson, J. F.; Clary, D. C. Faraday Discuss. Chem. SOC.1988,85, 37. (2) (a) Paulson, J. F.; Henchman, M. J. Bull. Am. Phys. SOC.1982, 27,
108. Paulson, J. F.; Henchman, M. J. Ionic Processes in the Gas Phase; Almoster Ferreira, M., Ed.; Reidel: Dordrecht, 1984; p 331. (b) Chalasinski, G.; Kendall, R. A.; Simons, J. J. Chem. Phys. 1987,87, 2965. (3) (a) Griffiths, W. J.; Harris, F. M. Znt. J. Mass. Spectrom. Zon Processes 1987, 77, R7. Griffiths, w. J.; Harris, F. M. Org. Mass Spectrom. 1987, 22, 812. (b) De Lange, W.; Nibbering, N. M. M. Znt. J . Muss. Spectrom. Zon Processes 1987, 80, 201. (4) Dunning, T. H. Jr. J. Chem. Phys. 1989, 90, 1007. (5) Kendall, R. A.; Dunning, T. H. Jr.; Harrison, R. J. J. Chem. Phys. 1992, 96,6796. (6) Frisch, M. J.; Head-Gordon, M.; Trucks, G. W.; Foresman, J. B.; Schlegel, H. B.; Raghavachari, K.; Robb, M.; Binkley, J. S.; Gonzalez, C.; Defrees, D. J.; Fox, D. J.; Whiteside, R. A.; Seeger, R.; Melius, C. F.; Baker, J.; Martin, R. L.; Kahn, L. R.; Stewart, J. J. P.; Topiol, S.; Pople, J. A. GAUSSIAN 90, Revision H,Gaussian Inc.: Pittsburgh, PA, 1990. (7) (a) Reimers, J.; Watts, R.; Klein, M. Chem. Phys. 1982,64, 95. (b) Curtiss, L. A.; Frurip, D. J.; Blander, M. J. Chem. Phys. 1979, 71, 2703. (8) Odutola, J. A.; Dyke, T. R. J. Chem. Phys. 1980, 72, 5062. (9) Dang, L. X.; Rice, J. E.; Caldwell, J.; Kollman, P. A. J. Am. Chem. SOC.1991, 113, 2481.
31P-113Cdand 31P-2QSiCP/MAS-NMR in Inorganic Semiconductors Deanna Franke, Christopher Hudalla, Robert Maxwell, and Hellmut Eckert* Department of Chemistry, University of California, Santa Barbara, California 931 06 (Received: July 9, 1992) Cross-polarization of insensitive nuclei (29Siand lI3Cd)from abundant 31Pspin reservoirs is demonstrated for the first time. Besides offering remarkable gain in detection sensitivity, these experiments provide unique spectral editing opportunities. By replacing the 90' preparation pulse with a rotor-synchronized DANTE sequence, the magnetization associated with resolved 31Psites can be transferred selectively to directly bonded rare-spin nuclei. Through a series of such experiments, it is possible to assign individual P and Si resonances in silicon phosphide (Sip) and to construct the complete 31P-29Siconnectivity map.
Introduction Much of the power of MAS-NMR has come from the ability of exploiting cross-polarization for facile signal detection and spectral editing for insensitive and rare-spin nuclei.' To date, protons have almost invariably constituted the abundant-spin reservoir for cross-polarization experiments, although, most re0022-3654/92/2096-7506$03.00/0
cently, I9F has found increasing use.24 To our knowledge, there has been only a single CP study using any other nucleus (31P)as the abundant-spin magnetization reservoir, but even that study benefited from the presence of highly abundant lH spins (which were d e t ~ t e d ) .Thus, ~ to date solid-state NMR spectroscopy of inorganic materials that are devoid of protons has been mostly 0 1992 American Chemical Society
The Journal of Physical Chemistry, Vol. 96, No. 19, 1992 15QI
Letters
a
b e
e
e
e
e
e
. *
. e
e
I
0
1
2
3
4
5
Contact time (ms)
1
0
2
4
6
0
10
Contact time (mr)
Figure 1. 31P-29Si and "P-'I3Cd CP/MAS in crystalline CdSiPz. (a) Comparison of single-pulse spectra (bottom trace) and CP/MAS spectra (top trace). All spectra are 16 scans with a 2-min relaxation delay, except for the single-pulse 29Sispectrum (two scans, 15-min relaxation delay). The CP/MAS contact times were 1 and 5.5 ms for the ' W d and 29Sidetection, respectively. (b) Variable contact time 3'PJ9Si (MAS at 3.0 kHz,left) and 31P-113Cd (static sample, right) CP/MAS experiments. The solid curves are fits to an exponential cross-relaxation process, with cross-relaxation respectively. times of 0.9 ms (31P-I13Cd) and 0.4 ms (31P-29Si),
limited to Bloch decay and spin echo studiesa6The spin-lattice relaxation times encountered in such systems are often excessively long, resulting in poor signal-to-noise ratios and long measurement times. Furthermore, the line shapes of disordered inorganic solids are frequently rather broad, poorly resolved, and thus hard to interpret in the absence of spectral editing experiments. We have encountered such problems in our MAS-NMR investigations of crystalline and glassy semiconductor alloys based on a variety of metal phosphides and arsenides.' Recently, various heteronuclear X-Y double-resonance approaches have helped to increase the informational content of static and MAS-NMR spectra of such systems.*-I2 In the present contribution we wish to demonstrate, for the first time, that cross-polarization from 31Pto insensitive nuclei such as Il3Cd and 29Sinuclei is possible in conjunction with MAS, results in significant sensitivity enhancements, and can provide important insights into the structure of inorganic semiconductors.
In brief, the 300-MHz output is mixed with that of a local (PTS-250) oscillator, low-pass filtered, amplified (Doty Scientific amplifier) to ca.300 W, and passed through a narrow band-pass filter prior to entering the probe. On the low-frequency side the amplifier output from the regular multinuclear spectrometer channel is routed through a 0-87-MHz low-pass filter (MR Resources). The 90" pulse lengths employed were 5-8 I.CS (all nuclei) in CdSiP2and CdGeP2single-resonance experiments and 11-14 ps (all nuclei) in the CP/MAS experiments. Spinning speeds employed were 0-4.5 kHz. Additional high-field (1 1.7 T) MAS-NMR spectra were obtained on a General Electric GN-500 instrument at frequencies of 202.415 and 99.36 MHz for and 29Si,respectively. These measurements employed a 5-mm probe from Doty Scientific at spinning speeds of 9 KHz, 90" pulse lengths of 4.0 and 4.5 ps, and delays of 1 and 90 min for and 29Si, respectively. For all the measurements, chemical shift references are 85% H3P04("P), tetramethylsilane (29Si),and dimethylcadmium (Il3Cd).
Experimental Section Sample Preparation and Characterization. The samples under
Results and Discussion
study, crystalline CdSiP2,CdGeP,, and Sip, were prepared from the elements (Aldrich; Cd, 99.5%; Ge, 99.99%; P, 99.999%; Si, 99.9999%) in evacuated (lo-) Torr) silica glass ampules. Sip was prepared from a solution of the elements in molten tin at 1150 OC,as previously described.I3 Identities and purity of the samples were ascertained by X-ray powder diffraction, using a Scintag diffractometer, and magic-angle spinning NMR. Solid-state NMR Studies. All CP/MAS-NMR spectra were obtained on a General Electric GN-300 instrument, at approximate frequencies of 121.65,66.70, and 59.71 MHz for 31P," F d , and 29Si,respectively. The measurements employed a 7-mm double-broad-band tuned probe from Doty Scientific, The spectrometer modification for generating the second multinuclear frequency (31Pin the present case) has been described previously.'o
Figure l a compares single-pulse Il3Cd and 29SiMAS-NMR spectra obtained in crystalline CdSiPz with the corresponding 31P-113Cdand 3'P-29Si CPMAS NMR spectra, illustrating the expected sensitivity advantage of CPMAS NMR. This advantage is especially critical for the detection of the 29si nuclei which have excessively long spin-lattice relaxation times. Table I illustrates that the 31P-31P dipolar coupling constant is of the same order of magnitude as the spinning frequency. Under such conditions, interference and modulation effects due to the averaging of the dipoledipole couplings by magic-angle spinning have been predicted and observed.Icl6 Due to this situation, the Hartmann-Hahn condition is more difficult to locate than in the familiar IH-l3C case. The effect of magic-angle spinning on the width and shape of the Hartmann-Hahn condition
Letters
7508 The Journal of Physical Chemistry, Vol. 96, No. 19, 1992 3' P
TABLE I: Dipolar Coupling Constants (in Wz) and Experimentally Measured Cross-Relaxation Rates in tbe Samples under Study
P
D-
compd ("P-"P) CdSiP2 0.56 CdGeP2 0.53 SIP 0.60
D-
(31P-29Si) ("P-"'Cd) 0.63 0.70 0.67 0.65
TP-Si
a
TP-Cd
(ms) (ms) 0.4 f 0.14 0.9 f O.lb
0.5 f 0.1' 3.0 f 1.w
"MAS at 3.0 kHz. bStatic. EMASat 3.5 kHz. dMAS at 4.2 kHz; measured with DANTE selection of the "P resonance at -143.6 ppm.
TABLE II: Nearest-Neighbor Bonding Geometries in Silicon Monopimphide, SIP site direct bonds
Si(1) Si(2) Si(3) Si(4) Si(5) Si(6) P(1) P(2) P(3) P(4) P(5) P(6)
Si(2), P(2), 2P(5) Si(l), P(l), 2P(4) Si(4), 2P(1), P(3) Si(3), P(4), 2P(6) Si(6), 2P(2), P(6) Si(5), 2P(3), P(5) Si(2), ZSi(3) Si( l), 2Si(5) Si(3), 2Si(6) 2Si(2), Si(4) 2Si(l), Si(6) 2Si(4), Si(5)
d IA) a (den)" 2.2949 109.44 2.2949 109.40 2.2834 109.14 2.2823 109.42 2.2835 109.23 2.2801 109.45 2.2775 97.45 97.44 2.2760 2.2619 97.21 95.91 2.2706 95.89 2.2702 97.07 2.2604
B
( d e d b A (den)' 100.51 15.05 100.54 15.10 24.11 102.23 101.52 29.08 102.15 23.98 101.62 29.34 108.05 13.49 108.02 13.63 111.25 20.33 107.37 16.37 107.35 16.45 111.26 20.93
" Average bond angle including nearest neighbors. bAverage bond angle including next-nearest neighbors. Range of next-nearest neighbor bond angles. is currently the subject of more detailed investigations. Notwithstanding this situation, preliminary variable contact time experiments (see Figure 1b) reveal an approximately exponential buildup of rare-spin magnetization and no dramatic effects of MAS. The cross-relaxation times TRsiand Tpcd are on the order of 1 ms, substantially longer than typical 'H-13C values. This is expected because both the heteronuclear 31P-1'3Cdand 31P-29Si as well as the homonuclear 31P-31Pdipolar coupling constants (square root of the second moments) in CdSiP2 are significantly smaller their 'H-W and 'H-IH analogs in a typical organic solid. An important aspect of CPMAS experiments is the potential for spectral editing due to the distance dependence of the crossrelaxation rates. We will discuss this point for the compound silicon monophosphide, Sip, which possesses six crystallographically inequivalent silicon (four-coordinated) and phosphorus (three-coordinated) sites. Table 11 summarizes important crystallographic inf~rmation'~ about the detailed nearest-neighbor coordination environments for the individual phosphorus and silicon sites. It is evident that the silicon sites and the phosphorus sites can be categorized into three distinct pairs, each with almost identical geometries. The 11.7-T single-resonance MAS-NMR spectra of this compound (see Figure 2) show five resolved resonances at -143.6,-152.6,-232.8,-236.1 (double intensity), and -239.6 ppm and five resolved %i resonances at -6.0 (double intensity), -13.2,-16.5,-31.2, and -34.5 ppm, respectively. While the MAS-NMR spectra are generally consistent with this information, they neither offer peak assignments nor reveal the P s i connectivities. This question can be addressed by selective cross-polarizationexperiments. An important difference between cross-polarization from 'H and cross-polarization from 3*Pdiscussed here is that, in the present case, the abundant-spin species (j'P) shows well-resolved chemid shift spectra. The homonuclear dipoledipole coupling among inequivalent sites is generally much smaller than the chemical shift difference between them, resulting in a quenching of homonuclear zeroquantum transitions. Consequently, spin diffusion among inequivalent sites is slow, making it possible to selectively spin-lock the magnetization associated with a selected 31Presonance. The most elegant way of accomplishing this is to replace the 90° preparation pulse by a rotor~ynchronized'~J* DANTEI9 sequence; see Figure 3. The crystal structure of Sip reveals that, of the three Si atoms bonded directly to each of the individual P atoms, two are crystallographically equivalent. If we thus cross-polarize selectively
2g~i
- r - r r - - - v
r l - ~ r - T r - l - ' T T r r
-2
-10
U
c
-3 0
-40
PPM
Figure 2.
Single-pulse 11.7-T 3'P and 29SiMAS-NMR spectra of silicon phosphide with relaxation delays of 60 s (16 scans) and 90 min (12 scans), respectively. Spinning sidebands are indicated by asterisks.
3'P
-7-
- -7-
-7-
c
-
-T-
-T-
Decoupler on
contact time
-7-
I
contact time
Figure 3. Pulse sequence for DANTE-selected cross-polarization, m a g ic-angle spinning NMR.
?I
A
i
i \
'I
IT\ f"
'
-239.7
!
4 I
2c
"
"
'
"
1
6
/
\/bi '
I
-*c
--
(29~i;'(ppm)
Figure 4. DANTE-selected CP/MAS of silicon phosphide, obtained with 3-6-ms contact times and a 3-min relaxation delay. The chemical shifts of the selected I'P resonances are indicated.
from &chosen phosphorus site, we expect the corresponding 29Si MAS-NMR spectrum to be dominated by the two resonances of the silicon atoms (present in a 2:l ratio) directly bonded to this
Letters
The Journal of Physical Chemistry, Vol. 96, No. 19, I992 7509
-143 6 opm h
En
-1526
W
I
7
OW
I
00
- 2 3 2 B p p m P(2orll - 2 3 1 5 8 - 2 3 6 . 6 opmp-
-i
8
- 2 3 9 7 ppm P(4or51
Figure 5. Heteronuclear correlation map constructed from the D A N T E selected CP/MAS-NMR results shown in Figure 4. The final assignments of the 3'P and %i resonances are indicated in the figure. Larger circles indicate the presence of two bonds, whereas smaller circles indicate the presence of one bond.
site. Thus, from a series of such experiments, the complete heteronuclear correlation map can be derived. Figure 4 shows the experimental verification of this idea, using DANTE pulse trains of eight 11.25O pulses separated by the inverse of the sample rotation rate. All of the spectra shown in Figure 4 are obtained with a 3-6-ms contact time, and excellent reproducibility was observed in triplicate runs. Note that, as expected, the individual 31P-29SiCPMAS spectra are distinctly different, showing generally two dominant lines each. The more intense one is assigned to the two identical Si atoms directly bonded to the P atom under consideration, whereas the second-most intense line is assigned to the remaining silicon neighbor. In some cases, the area ratio differs somewhat from the expected 2:l value, suggesting that there may be slight differences in the individual "P29si cross-relaxation rata. (Variable contact time studies over the range 1-10 ms show that the contact time used in these experiments influences the relative peak ratios generally very little, however.) The results reveal further that there is a distinct difference between the " P s i CP/MAS-NMR spectra obtained by spin-locking at -235.5 and at -236.6 ppm, respectively. This is taken as evidence that the above chemical shifts are to be assigned to the two unresolved resonances centered at -236.1 ppm. We note that most of the spectra also show weaker peaks due to partial cross-polarization of more remote Si atoms and/or incomplete excitation selectivity due to partial overlap of the 31P resonances. These interference peaks are most intense in the CP spectra obtained from the spins in the crowded -235 to -237 ppm region and have been identified by careful studies of CP peak ratios as a function of 31P irradiation frequency. A final assignment constraint arises from the fact that each 29Sisite can be cmss-polarizedonly from two of the six phosphorus sites present. The P-Si connectivity map constructed from this information is shown in Figure 5. For arriving at final peak assignments, it is important to note that the local environments of two of the six P sites, namely P(3) and P(6), differ dramatically from the others in both the average P-Si bond lengths and the average P-Si-X bond angles (see Table 11). On the basis of this structural feature, we assign the two resonances at -143.6 and -152.6 ppm, which are shifted distinctly downfield from the others, to P(3) and P(6),
respectively, or vice versa. From this initial starting point, all of the other assignments can be derived, and they are listed in Figure 5 . Note that this assignment independently arranges each two silicon sites with similar geometries into pairs with very similar chemical shifts. This result adds further codidence to the validity of our conclusions. A final ambiguity, originating from the ambiguity of the P(3) vs P(6) peak assignments, remains to be resolved. To address this question, ?3-?3i COSY or INADEQT experiments on 29Si-enrichedmaterial will be required. Past NMR studies of silicon oxidez0and phosphorus oxide2' systems have sucessfully correlated 29Siand 31PNMR chemical shifts with geometrical parameters such as bond lengths, angles, and valence sums. In conjunction with our peak assignments, Table I indicates that no simple linear correlations appear possible in Sip. This conclusion confirms our general result, that in covalently bonded systems, chemical shift data are not straightforwardly interpretable in terms of local bonding environments.22 In conclusion, heteronuclear X-Y cross-polarization (with or without spin-locking of selected resonances) is a powerful approach to enhance both the detection sensitivity and the informational content of MAS-NMR spectra in materials devoid of hydrogen. Applications to a wide range of inorganic ceramics, semiconductors, and glasses are currently underway in our laboratory. Acknowledgment. This research was supported by NSF Grant DMR-8913738. We thank Mr. David Schmidt for technical assistance.
References and Notes (1) For a review, see for instance: Duncan, T. M.; Dybwski, C. R. Surf. Sci. Rep. 1981,1, 157. (2)Sebald, A,; Merwin, L. H.; Schaller, T.; Knoller, W. J . Maan. Reson. 1992,96, 159. (3) Klein-Douwel. C. H.: Maas. W. E. J. R.: Veeman. W. S.: Werumeus Buning, G. H.; Vankan, J. M. J. Macromolecules 1990,'23, 4d6. (4)Kohn, S.C.; Dupree, R.; Mortuza, M. G.; Henderson, C. M. B. Am. Mineral. 1991,76,309. (5) Crosby, R.C.; Reese, R. L.; Haw, J. F. J . Am. Chem. Soc. 1988,JIO, 8550. (6)For a recent review, see: Eckert, H. Eer. Bunsen-Ges. Phys. Chem. 1990,94,1062. (7) Franke, D. R.; Maxwell, R.; Lathrop, D.; Eckert, H. J . Am. Chem. Soc. 1991,113,4822.Franke, D. R.; Banks, K.; Maxwell, R.; Eckert, H. J . Phys. Chem. 1992,96, 1906. (8)Makowka, C. D.; Slichter, C. P.; Sinfelt, J. H. Phys. Rev. 8 1985,31, 5663;Phys. Rev. Lett. 1982,49,319. (9)Boyce, J. B.; Ready, S.E. Phys. Rev. 1988,838,11008. Van Eck, E. H. R.; Veeman, W. S.Solid State NMR 1992,1, 1. (10) Maxwell, R.; Franke, D.; Lathrop, D.; Eckert, H. Angew. Chem., Int. Ed. Engl. 1990,79,884. (11)Van Eck, E. R. H.; Janssen, R.; Maas, W. E. J. R.; Veeman, W. S. Chem. Phys. Lett. 1990,174,428. (12) Franke, D.; Hudalla, C.; Eckert, H. SolidState NMR 1992,1, 33. (13) Wadsten, T.Chem. Scr. 1974,8, 63. (14)Stejskal, E. 0.; Schaefer, J.; Waugh, J. S.J . Magn. Reson. 1977,28, 105. (15)Stejskal, E. 0.; Schaefer, J.; McKay, R. A. J . Magn. Reson. 1984, 57, 471. (16)Sardashti, M.; Maciel, G. E. J . Magn. Reson. 1987,72,467. (17)Caravatti, P.; Levitt, M. H.; Ernst, R. R. J. Magn. Reson. 1986,68, 323. (18)Moran, L. B.; Berkowitz, J. K.; Yesinowski, J. P. Phys. Rev. 1992, 45,5341. (19) Bodenhausen, G.; Freeman, R.; Morris, G. A. J . Magn. Reson. 1976, 23, 171. (20)Grimmer, A. R.; Radeglia, R. Chem. Phys. Lett. 1984, 106, 262. Pettifer, R. F.; Dupree, R.; Farnan, I.; Sternberg, U. J. Noncryst. Solids 1988, 106,408. (21)Grimmer, A. R. Z . Chem. 1984,88,1518. Cheetham, A. K.;Clayden, N. J.; Dobson, C. M.; Jakeman, R., J. B. J . Chem. Soc., Chem. Commun. 1986,195. (22) Eckert, H. To be published.
