4570
J . Phys. Chem. 1994, 98, 4570-4574
Spectroscopic Studies of 5-Coordinate Silicon Compounds Bruno Herreros,' Tery L. Barr>*$Patrick J. Barfie,%and Jacek Klinowski'vt Department of Chemistry, University of Cambridge, Lensfeld Road, Cambridge CB2 IEW, U.K., and Christopher Ingold Laboratories, University College London, 20 Gordon Street, tondon WClH OAJ, U.K. Received: February 1, 1994"
A detailed study of the structural and chemical features of a series of novel compounds in which silicon is in 5-coordination with respect to oxygen has been carried out using XRD, FTIR, solid-state NMR, and ESCA. All compounds contain glycolate groups and different charge-balancing alkali-metal cations. NMR confirms that only 5-coordinate S i is present. There is a unique set of ESCA core level binding energies for these materials. Peak positions may be influenced by (1) the coordination number of Si, (2) the presence of organic moieties, and (3) the kind and concentration of cations. All binding energies in the 5-coordinate silicon compound with Cs+ a r e shifted with respect to those in other compounds to suggest that effects (1) and (3) are dominant.
Introduction In silicates and aluminosilicates, silicon is normally present in 4-coordination with respect to oxygen,' although several minerals with 6-coordinate Si, such as stishovite1q2and thaumasite,lJ are known. In recent years a number of ionic compounds with 5-coordinate silicon have been The reactivity of 5-coordinate organosilicon complexes is enhanced in comparison with 4-coordinate species, especially toward nucleophile^.^*^-^ These materials serve as models of intermediates in nucleophilic substitution reactions at silicon centers. Laine et al.12313 described a direct process in which Si02 reacts with ethylene glycol and an alkali-metal base to produce organosilicon compounds with 5-coordinate Si(M = Li+, Na+, or 2 Si@
+ 2 MOH + 5 (CH20H)2
Hz0
L
2
K+). We shall refer compounds as Li-Siv, N a S i v , and K-Siv, respectively. The corresponding Cs+ compound (Cs-Siv) is a monomer.
A series of these compounds was prepared by recently reported methods.I2J3 All contain glycolate groups and charge-balancing cations. These features may contribute to the X-ray photoelectron spectroscopy (ESCA) shifts found in the core level spectra,14J5 t University of Cambridge.
t Permanent address: Department of Materials and Laboratory for Surface Studies, University of Wisconsin-Milwaukee, Milwaukee., Wisconsin 53201, U.S.A. f University College London. Abstract published in Aduance ACS Abstracts, March 15, 1994. @
0022-365419412098-4570$04.50/0
thus making it difficult to ascribe various patterns simply to changes in coordination, such as Si(1V) * Si(V).
Experimental Section Synthetic Procedures. The raw materials were Cab-0-Si1 M-5 fused silica, group I metal hydroxides (LiOH, NaOH, and CsOH), and ethylene glycol in a molar composition of 1 S i 0 5 1 NaOH:40 ( C H Z O H ) ~ .The 5-coordinate compounds were prepared as follows. Si02 was added to ethylene glycol, and the mixture was stirred until dissolution. It was then placed in a magnetically stirred standard Pyrex distillation apparatus, and MOH was added (where M stands for Li, Na, or Cs). The reaction was carried out a t 195 OC under flowing nitrogen so that the ethylene glycol and the water produced in the course of the reaction were slowly distilled off. The mixture was cooled, and the solid product, was separated out, repeatedly washed with CH3CN, and dried. Two samples of Li-Siv were prepared and dried using different procedures. Upon further heating, L i S i v originally dried at 200 OC for 1 h becomes isostructural with the sample dried at 180 OC for 12 h. This structural change is irreversible. N a S i V was dried at 200 OC for 1 h and Cs-SP at 150 OC for 1 h. Sample Characterization. ESCA Analysis. ESCA analysis was performed in the Surface Analysis Facility at the University of Wisconsin-Milwaukee using a Vacuum Generators ESCALAB system capable of measuring spectra with several anodes (including a monochromator), as well as Auger spectra. All results were acquired using a conventional, direct focus A1 K a anode with optimal resolution for Au(4f7p) of 83.95 eV with a line width of ca. 1.OeV. The samples were examined at a background pressure of ca. 5 X Torr as wafers pressed into an indium foil. All samples showed features characteristic of surface insulators, creating significant charging shifts.'6J7 The shifts, the magnitudes of which are highly system-dependent, varied from ca. 8 to ca. 3 eV. Charging effects were removed, and the binding energy scale was established by fixing the C(1s) peaks for C,H, hydrocarbons at 284.6 eV.I4J7 Detailed justification of the procedure is given el~ewhere.'~J~J17 In the present case, the carbonaceous species were the adventitious carbon and the ethylene glycol moiety. X-ray Diffraction. X-ray diffraction (XRD) patterns were recorded in Cambridge using a Philips 17 10powder diffractometer with Cu K a radiation (40 kV, 40 mA), a 0.025O step size, and a 1-s step time. InfraredSpectra. Fourier transform infrared (FTIR) spectra, were recorded in Cambridge with a Nicolet 205 spectrometer using a KBr wafer disk. Solid-state NMR. lH-29Si magic-angle-spinning (MAS) N M R spectra with cross-polarization (CP) were recorded in 0 1994 American Chemical Society
The Journal of Physical Chemistry, Vol. 98, No. 17, 1994 4571
Spectroscopic Studies of 5-Coordinate Si Compounds
u-siv 200oc
v
.-2c. c
Na-Si"
v)
a
c
S
3
13
23
33
43
53
5 10
20
30
50
40
60
2 8 (degrees) Figure 1. XRD patterns for 5-coordinatesilicon compounds. London at 7.05 T, with a single-contact pulse sequence with a proton 90° flip back-pulse after acquisition, 8-ms contact time, 4.8-ps 1H 90° pulse, and 10-s recycle delay. The HartmannHahn condition was established using a sample of QaM8. Chemical shifts of 29Si are given in ppm from external tetramethylsilane (TMS).
Results X-ray Diffraction. XRD patterns of all 5-coordinate compounds (Figure 1) show good crystallinity. The space groups, unit cell parameters, and atomic positions have not been determined so far. FIIR. Since the glycolate group is part of the structure, FTIR spectra (see Figure 2) have been empirically assigned by comparison with the spectra of ethylene glycol (EG). The main EG absorption bands appear at 880 cm-l ( Y C X ) , 1040 and 1100 cm-1 ( Y C ~ ) 1350 , to 1450 cm-1 (SCH), and 2900 and 2950 cm-* (YCH). In addition, a comparison was made with silica and the purely siliceous form of sodalite.la Band shifts, the disappearance of EG bands, and the appearence of new bands are expected to accompany the creation of C - o S i bonds. The bands at 3400 and 1650 cm-l are due to HzO, and the band at 2400 cm-l is due to atmospheric COZ. We note the following changes in the spectrum of Na-Siv, consistent with the expected structure, in comparison with the spectrum of ethylene glycol: (a) The fine structure of YCH, ~ C H , and v c bands ~ appears at 2800-3000,1350-1450, and 850-950 cm-1, respectively. This indicates that the glycolate group cannot move freely in the structure. (b) The EG band at 1040 cm-l, corresponding to v c a , disappears, which indicates that H-O-C bonds have been replaced by Si-O-C bonds. (c) The vcx band at 880 cm-1 splits into two components and shifts to 900 and 930 cm-1, due to the presence of Si near the C-C bonds. (d) The creation of Si-0-C bonds is shown by the presence of absorption bands at 1100 cm-l and from 600 to 800 cm-l, assigned by comparison with the spectra of purely siliceous sodalite18 to v,,(Si-O-C), v,(Si-O-C), and S(O-Si-0). (e) Absorption bands corresponding to free ethylene glycol disappear, which indicates that the glycol is incorporated into the structure. The same changes are observed for L i S i v . Here, the shift of the Y C X band is even more pronounced (to 930 and 950 cm-I). We note that even after several washes with CH&N there is still residual glycol (bands at 880 and 1040 cm-1). The latter bands might also be due to the presence of monomeric species with C-O-H bonds, but such species are not seen in the NMR spectra (see below).
Na-Si"
Transmittance
Li-Siv
~ O O O C
Li-siv 2 0 0 ~ ~
cs-SI'
1000 3200
2400
1800
1400
1000
600
Wavenumbers (cm- I )
Figure 2. FTIR spectra of 5-coordinate silicon compounds.
Cs-Siv is a monomer containing C-Os1 and C-O-H bonds. Accordingly, it FTIR spectrum (Figure 2) shows the changes observed for Na-SiV and the characteristic EG bands at 880 and 1040 cm-1. The band at 1040 cm-l is too intense to be due to residual ethylene glycol alone. 29Si CP/MAS NMR. 1 H - W CP/MAS NMR spectra of NaSiv (Figure 3a), as well as those of L i S i v dried at 180 OC (Figure
4512 The Journal of Physical Chemistry, Vol. 98, No. 17, 1994
Herreros et al.
* -102.7 ppm
( a ) Na-Si”
l-lIO,l
fast MAS
cs-si”
* *
( b ) Na-Si” static
*
. -20
-60
-100
.
. -140
.
-180
ppm from TMS 0
-50
-100
- 150
-200
ppm from TMS Figure 3. lH-29Si CP NMR spectra of NaSiv: (a) fast MAS; (b)
static. Asterisks denote spinning sidebands. sa), consist of a single line a t -102.7 ppm, which indicates that the samples are pure and monophasic because only one kind of crystallographic site is detected. The weak broad signal at ca. 90 ppm in the spectrum of Li-Siv dried at 180 OC comes from an amorphous impurity. Spinning sidebands were identified by spinning at different rates (the position of the isotropic line is independent of the spinning frequency). In the spectrum of sample Li-Siv dried a t 200 OC (Figure 5b) there are five peaks in the range -99 to -103 ppm. The spectrum of Cs-W (see Figure 4) consists of a single peak at -110.1 ppm, which we assign to monomeric silicate species. The peaks a t ca. -102.7 ppm in the spectra of Na-Wand L i S i v dried at 180 and 200 OC are assigned to dimeric silicate species. Those chemical shifts are typical for 5-coordinateorganosiliconcompounds4J (from-80 to-1 30 ppm). By comparison, peaks from 4-coordinate Si range from 0 to -90 ppm for organosilicates, and those from 6-coordinate Si are found between -130 and -200 ~ p m . ~ J ~ Static lH-29Si N M R spectra with cross-polarization for NaSiv and sideband intensity measurements with slow MAS20 for samples L i S i v and Cs-Siv provide information on symmetry around the silicon atoms. Sideband analysis was performed using the method of Herzfeld.2Ob For Na-Siv, the pattern is typical of an axial symmetry. For both Li-Siv and Cs-Siv, parameters are closer to the general case, which could correspond to a large distortion from the ideal trigonal bipyramid. The parameters of the chemical shift anisotropy tensor are given in Table 1. Because several peaks are present, the parameters for Li-SiV dried a t 200 OC represent an average site. Parameters for Li-SiV dried at 180 O C are not listed because of the presence of amorphous impurities in the sample. The anisotropy of the chemical shift (Aavalues are 70-73 ppm for the dimer and 43 ppm for the monomer) strongly suggests that the silicon is 5-coordinate. When, as is the
Figure 4. lH-29Si CP/MAS NMR spectra recorded with slow MAS.
case for a purely siliceous sodalite, the symmetry of the environment of Si is tetrahedral, Au < 10 ppm.18 ESCA. The materials under consideration are organosilicon compounds and may be thought to be liable to surface deterioration on exposure to air as well as to radiation damage. The question of the possible “compromise” of the sample surface must therefore be addressed. If exposure to air were to be important, ESCA would detect surface byproducts rather than the 5-coordinate silicon species, the presence of which has been verified by NMR, a bulk analysis technique. This potential problem cannot be ignored, but the following argument was used to determine whether the surface is of acceptable purity (less than a few 3’% byproducts). Thus, if the bulk species are not the most prevalent components a t the surface, one may guess which kind of species may be produced upon exposure to air. If none are detected in the ESCA spectra, then extensive degradation has not occurred. In view of the absence of a common choice for cationic or anionic species (other than Na+, H+, 02-,and OH-), one may assume that degradation will result in the surface formation of Si02 or related hydrated In order for surface Si02 to be present in significant quantities, its characteristic ESCA signature must be in evidence. The key binding energies for Si02 are listed in Table 2. We note that for spectra a t the resolution of our ESCA system, the presence of Si02 would produce distortions, if not secondary peak structures, in the Si(2p) and O(1s) spectra for the 5-coordinate silicates containing Na+ and Li+, and probably some distortion for the corresponding Cs+ compound. Since no such feature is detected, we conclude that the surfaces are uncompromised. Since all 5-coordinate silicon compounds under consideration contain glycolate moieties, the C( 1s) and O(1s) spectra should directly reflect its presence and the Si(2p) peak should show certainsecondaryeffects.l5J7 In thecaseoftheO( Is) thepresence of the organic should produce a peak structure near 531 eV.15J7 However, this is also found to be the position of the O(1s) peak
The Journal of Physical Chemistry, Vol. 98, No. 17, 1994 4513
Spectroscopic Studies of 5-Coordinate Si Compounds
TABLE 1: Chemical Shift Anisotropy Parameters for 5-Coordinate Si Compounds sample NaSiV LiSiv dried at 200 OC (averge site) CsSiv
6iKl
611
612
633
-102.7 -102.8 -110.1
-128.7 f 3.3 -141 f 5.7 -87.3 f 3.2
-125 f 3.6 -111.6f4.2 -104.1k2.4
-54.3 f 3.3 -55.7 f 5.7 -138.9f3.2
TABLE 2 Key ESCA Binding Energies and, in Parentheses, Corresponding Line Widths (f0.05 eV) for Compounds with 5-Coordinate Si* material Si(2~) O(l4 cation CsSiV 102.6 531.9 (2.6) 724.6 (Cs(3d~p)) NaSiV 101.4 (2.4) 531.1 (2.1) 1070.4 (Na(1s)) LiSiv (200 "C) 101.4 531.1 54.1 (Li(1s)) LiSiv 180 OC 101.65 531.4 (2.6) 54.5 (Li(1s)) for 12 h aSiO2 103.5 532.9 a Binding energies are referenced to C(1s) = 284.6 eV.
Au 72.6 f 4.9 70.6 f 8.6 43.2k4.9
1)
0.08 f 0.1 1 0.63 f 0.1 1 0.58f0.11
Survey scan Na-Si"
SI (2s)
-102.8 ppm
I
180
160
140
120
100
80
60
20
40
0
Binding energy (eV) Figure 6. ESCA survey scan (C-200 eV) of NaSiv.
L i - S i v 180'C
Na-Si"
-40
-60
-80
-100
-120
-140
-160
295
290
285 280 Binding energy (eV)
275
Figure 7. High-resolution C( 1s) ESCA spectrum of NaSiv.
- 96
i
1
-98
-1 00
-1 02
-104
-106
ppm from TMS
Figure 5. 1H-29SiCP/MAS NMR spectra of LiSiv: (a) dried at 180 OC for 12 h; (b) dried at 200 OC for 1 h. Spectrum (b) is shown on an expanded scale in order to show its structure. from the Si-0 units. Thus, the presence of C-0 apparently leads to little more than a broadening of the O( 1s) peak. This feature may be revealed in the Cs+ and Li+ silicon compounds in that their O(1s) peaks are noticeably broader than the other peaks, but this is not the case for Na-Siv. The signature of the C(1s) spectra is different from that for O( 1s). First, the spectrum reflects an increase in carbonaceous content above that typical of adventitious carbon alone.15J7 Second, glycolatosilicates contain a mixture of hydrocarbon C,H, units and C - O S i bonds. The ESCA spectrum of the latter is not readily interpretable, but it is well-known that C(1s) binding energies of C-0-C linkages in ethers and epoxides are shifted from the hydrocarbon C( 1s) peaks by 1&1.5 eV (with an average
a t ca. 1.2 eV). In the present case we find the C(1s) peak for the C - O s i units to be shifted from 0.8 eV in the hydrocarbon to 1.O eV. This small but significant decrease in C ( 1s) binding energy is consistent with our arguments that the substitution of the more ionic Si-0 bond for the C-0 bond in an ether should enhance the covalency of the remaining C-O bond, thus decreasing the C(1s) binding energy for the carbon i n v o l ~ e d . I ~The 1 ~ fact that some of the C-0 linkages in Cs-Siv are alcoholic should not shift the resulting peaks markedly15 and is apparently undetected at our level of resolution. Although few differences were found for the carbonaceous parts of these materials, several noticeable differences were detected for the silicate parts of the 5-coordinate silicon compounds. The results reported in Table 2 and Figures 6-8 were obtained during the binding energy analysis. We see that creation of these 5-coordinate silicon compounds has resulted in a noticeable reduction in binding energies for the Si(2p) and O(1s) lines compared to the results for SiO2.'6J7 In fact, the results for the systems formed with Li+ dried a t 200 "C and Na+ cations both result in nearly identical sets of framework S i 4 binding energies, while the Si(2p) and O(1s) peaks in compounds formed with the much larger Cs+ cation are noticeably shifted to higher binding energies. A similar positive shift occurs when the Li-Siv compound is heated at 180 OC for 12 h in an oven.
Herreros et al.
4514 The Journal of Physical Chemistry, Vol. 98, No. 17, I994
found that the introduction of large cations into the interstitial space of a variety of zeolites produces an increase in the binding energies for the frameworkatoms.2122 It is therefore not surprising that the binding energies for 5-coordinate silicates containing Cs+ are larger than the corresponding energies for the Li+ and Na+ compounds. The fact that the binding energies for 5-coordinate silicon compounds are reduced compared to those for Si02 suggests that factors 2 and 3 are more important than (1).
Acknowledgment. We are grateful to the Fulbright Commission for a Professorial Fellowship for T.L.B. and to Unilever Research, Port Sunlight, for a Research Studentship for B.H. The 29Si NMR spectra were obtained at the ULIRS solid-state NMR Facility at University College London.
References and Notes
106
102
98
94
Binding energy (eV) Figure 8. Narrow scans of Si(2p) ESCA peaks. The binding energy scales have been adjusted to C(ls) = 284.6 eV.
Discussion Some of the observed binding energy shifts cannot be attributed to a specific effect because as many as three different effects may be simultaneously present. The following factors have to be considered: (1) The presence of C-O bonds which are more covalent than the Si-O bonds17.23 should enhance the ionicity of the Si-0 bond, and all other factors being constant, the Si(2p) binding energy should increase compared to that for Si02.23324 (2) The substantial cation effects caused by the presence of a cation should decrease the Si(2p) binding energy compared to that for Si02.17,21 (3) Thereshould bea significant binding energy shift due to silicon being 5-coordinate. The direction of this shift is not obvious, except that it is the conversion from 4- to 5-coordination which brings about the negative charge. Thus factors 2 and 3 are intertwined. An example of factor 3 is the transition from 6-coordinate A1 in A1203 to 4-coordinate A1 in NazA1204 which produces a negative shift in the binding energy of Al(2p).17.24Such shifts are more pronounced for silicon than for aluminum.17~2l~22 It thus seems natural to find an apparent negative shift in the Si(2p) and O(1s) binding energies induced in the 5-coordinate Na+ and Li+ compounds compared to the case of Si02. In addition, we have
(1) Liebau, F. Structural Chemistry ofSilicates; Springer, Berlin, 1985. (2) Thomas, J. M.; Gonzales-Calbert, J. M.; Fyfe, C. A,; Gobbi, G. C.; hichol, M. Geophys. Res. Lett. 1983,10,91. (3) Grimmer, A,-R.; Wieker, W.; Lampe, F. V.; Fechner, E.; Peter, R.; Molgedey, G. Z . Chem. 1980,20,403. (4) Holmes, R. R. Chem. Rev. 1990, 90, 17 and references therein. Kumara Swamy, K. C.; Chandrasekhar, V.;Harland, J. J.; Holmes, J. M.; Day, R. 0.;Holmes, R. R. J. Am. Chem. SOC.1990, 112,2341. (5) Corriu, R. J. P.; Young, J. C. In The Chemistry of Organic Silicon Compounds; Patai, S.,Rappaport, Z., Eds.; Wiley: Chichester, U.K., 1989; Chapter 20 (see also references therein). (6) Tandura, S.N.; Voronkov, M. G.; Alekseev, N. V. Top. Curr. Chem. 1986,131,99. (7) Corriu, R. J. P.; Guerrin, C.; Henner, B. J. L.; Wong Chi Man, W. W. C. Oreanometallics 1988. 7. 237. (8) Eeiters, D. A.; Holmes; R. R.; Holmes, J. M. J . Am. Chem. Soc. 1988, 110,7672. (9) Johnson, S. E.; Deiters, D. A.; Day, R. 0.; Holmes, R. R. J. Am. Chem. Soc. 1989, 1 I I, 3250. (10) Corriu, R. J. P.; Young, J. C. In The Chemistry of Organic Silicon Compounds; Patai, S.,Rappaport, Z., Eds.; Wiley: Chichester, U.K., 1989; p 305. (11) Corriu, R. J. P.; Guerrin, C.; Moreau, J. J. E. Adu. Organomet. Chem. 1982,20,265. (12) Laine, R. M.;Blohoviac, K. Y.; Robinson, T.R.; Hoppe, M. L.; Nardi, P.; Kampf, J.; Uhm, J. Nature 1991, 353,642. (13) Blohoviac, K. Y.; Laine, R. M.; Robinson, T. R.; Hoppe, M. L.; Kampf, J. Inorganic and Organometallic Polymers with Special Properties; Laine, R. M., Ed.; Kluwer: Dordrecht, 1992;pp 99-1 11. (14) Barr, T.L. In PracticalSurface Analysis, 2nd ed.; Briggs, D., Seah, M. P., Eds.; Wiley: Chichester, U.K., 1990; Chapter 8. (15) Briggs, D., Seah, M. P., Eds. Practical Surface Analysis, 2nd ed.; Wiley: Chichester, U.K., 1990. (16) Barr, T. L. Crit. Reu. AMI. Chem. 1991,22,113. (17) Barr, T. L.Modern ESCA: The Principles and Practise of X-ray Photoelectron Spectroscopy; CRC Press: Boca Raton, FL (in press). (18) Herreros, B.;He, H.; Barr,T. L.;Klinowski, J. J . Phys. Chem. 1994, 98,1302. (19) Hoppe, M. L.;Laine, R. M.; Kampf, J.;Gordon, M. S a n d Burggraf, L. W. Angew. Chem., Int. Ed. Engl. 1993,32,287.Evans, D. F.; Wong, C. Y. Polyhedron 1991, 10, 1131. (20) (a) Dec, S.F.; Fitzgerald, J. J.; Frye, J. S.;Shatlock, M. P.; Maciel, G. E. J. Magn. Reson. 1991,93,403.(b) Herzfeld, J.; Berger, A. E.J . Chem. Phys. 1980,73,6021. (21) Barr, T. L. Appl. Surf. Sci. 1983,15, 1. (22) Barr, T. L.; Lishka, M. A. J . Am. Chem. Soc. 1986,108,3178. Barr, T.L.; Chen, L. M.; Mohsenian, M.; Lishka, M. A. J . Am. Chem. SOC.1988, 11 0,7962. (23) Barr, T. L. J . Vac. Sci. Technol. 1991, A9, 1793. (24) Barr, T.L.Zeolites 1990,10,760.