Characterization of GeO2 Sub-monolayers on SiO2 Prepared by

GeO2 sub-monolayers on SiO2 surfaces were prepared by the chemical vapor deposition (CVD) reaction of Ge(OMe)4 with OH groups on SiO2 at 393 K, ...
1 downloads 0 Views 272KB Size
Download PDF
Langmuir 1998, 14, 3607-3613

3607

Characterization of GeO2 Sub-monolayers on SiO2 Prepared by Chemical Vapor Deposition of Ge(OMe)4 by EXAFS, FT-IR, and XRD K. Okumura,† K. Asakura,‡ and Y. Iwasawa*,† Department of Chemistry, Graduate School of Science and Research Center for Spectrochemistry, Faculty of Science, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113, Japan Received March 7, 1997 GeO2 sub-monolayers on SiO2 surfaces were prepared by the chemical vapor deposition (CVD) reaction of Ge(OMe)4 with OH groups on SiO2 at 393 K, followed by calcination at 693 K. The characterization of the obtained GeO2/SiO2 was conducted by FT-IR, XRD, and EXAFS. Isolated OH groups on SiO2 preferentially reacted with Ge(OMe)4. The Si-OH peak was significantly depleted by the CVD reaction of Ge(OMe)4 and partially reappeared to half of the original Si-OH peak after calcination. The linear increase of the Ge-OH peak and the linear decrease of the Si-OH peak with increased Ge loading suggest a monolayer growth of GeO2. The coverage of GeO2 at saturation for the sample prepared by the one-cycle CVD process was estimated to be 1/5 monolayer of the SiO2 surface, which corresponds to 7.4 wt % Ge loading. There existed only isolated Ge-OH groups on the GeO2 layers, which is contrasted to the case of bulk GeO2, which shows both isolated and hydrogen-bonded OH groups. The surface of the GeO2 layers on SiO2 has no acidic character, whereas GeO2/Al2O3 showed both Bro¨nsted and Lewis acid sites. EXAFS spectra showed a significantly lower Ge-Ge signal compared with that for bulk GeO2. The local structure around Ge in GeO2/SiO2 was determined to be similar to hexagonal type GeO2.

Introduction Inorganic oxide monolayers developed on the other oxide surfaces have attracted attention in terms of preparation of a new type of surface materials with unique catalytic performance.1 The simplest way to produce monolayer catalysts is to use an appropriate combination of overlayer oxide and substrate oxide such as V2O5 on TiO2 and MoO3 on TiO2. In these systems the overlayer oxides wet well the substrate surfaces and spontaneously spread over the surfaces as monolayers by heat treatment. On the other hand, chemical reactions between surface OH groups and precursor complexes are the other way to synthesize monolayer materials, which do not show the spontaneous spread to monolayer. The surface reaction technique using alkoxides as precursors has successfully been applied to the synthesis of monolayer oxides with well-defined structures, for example, ZrO2/ZSM-5,2 Nb2O5/SiO2,3-5 and TiO2/SiO2.6 These materials have been employed as unique catalysts to convert methanol to isopentane selectively2 and to catalyze the esterification of acetic acid with ethanol and the intramolecular dehydration of ethanol, and they have also been employed as a support for Pt.6 Recently, GeO2 monolayers were prepared on Al2O3 and zeolite surfaces by use of Ge organometallic compounds.7-9 It has been reported that Bro¨nsted acidity was generated by the GeO2 layers deposited on Al2O3, and * To whom correspondence should be addressed. Fax: 81-3-58006892 and 81-3-3814-2627. E-mail: [email protected] and http://www.chem.s.u-tokyo.ac.jp/≈yiwswlab/INDEX.htm. † Department of Chemistry. ‡ Research Center for Spectrochemistry. (1) Xie, Y.-C.; Tang, Y.-Q. Adv. Catal. 1988, 37, 1. (2) Asakura, K.; Aoki, M.; Iwasawa, Y. Catal. Lett. 1988, 1, 395. (3) Asakura, K.; Iwasawa, Y. Chem. Lett. 1986, 859. (4) Shirai, M.; Asakura, K.; Iwasawa, Y. J. Phys. Chem. 1991, 95, 9999. (5) Asakura, K.; Iwasawa, Y. Chem. Lett. 1988, 633. (6) Asakura, K.; Inukai, J.; Iwasawa, Y. J. Phys. Chem. 1992, 96, 829.

Ge-O-Ge networks were formed on mordenite. GeO2 has also been deposited on SiO2 by the chemical vapor deposition (CVD) of GeCl4 and its subsequent hydrolysis.10 In this study we have synthesized GeO2 layers on SiO2 by the CVD reaction of Ge(OMe)4 with surface OH groups of SiO2, followed by calcination. The aim of the study is to estimate the reactivity of different kinds of OH groups at the SiO2 surface using Ge(OMe)4, to characterize the structure and growth mode of GeO2 overlayers on SiO2 by FT-IR, XRD, and EXAFS, and to examine the difference in the reactivities of Ge-OH groups on the GeO2 layer and Si-OH groups on the SiO2 support with Ge(OMe)4 to provide a possible way to prepare a layer-by-layer structure of GeO2. Experimental Section Preparation of GeO2/SiO2. Preparation of GeO2/SiO2 was carried out in a closed circulating system. Ge(OMe)4 as precursor was purchased from Soekawa Chemicals Co. (99.999%) and purified by distillation in a vacuum line before use. SiO2 (Aerosil 300 (300 m2 g-1) or ox-50 (50 m2 g-1)) was evacuated at 473 K for 1 h to remove physisorbed water and exposed to given amounts of Ge(OMe)4 vapor at 393 K for 1 h, followed by evacuation at 473 K to remove the unreacted Ge(OMe)4 and the organic products. The obtained sample was calcined at 693 K for 1 h under 20.0 kPa of oxygen in the closed circulating system. The evolved CO2 and H2O were removed by a trap with liquid N2. Maximum Ge loading on SiO2 (Aerosil 300) by the CVD reaction using an excess amount of Ge(OMe)4 was 7.4 wt %. The preparation conditions were varied in the CVD temperature range 353-423 K and in the CVD period from 1 to 5 h, but no change was observed in the Ge loading. To obtain the samples with (7) Hibino, T.; Niwa, M.; Murakami, Y.; Sano, M. J. Chem. Soc., Faraday Trans. 1 1989, 85, 2327. (8) Hibino, T.; Niwa, M.; Murakami, Y.; Sano, M.; Komai, S.; Hanaichi, T. J. Phys. Chem. 1989, 93, 7847. (9) Katada, N.; Niwa, M.; Sano, M. Catal. Lett. 1995, 32, 131. (10) Gun’ko, V. M.; Voronin, E. F.; Zarko, V. I.; Pakhlov, E. M. Langmuir 1997, 13, 250.

S0743-7463(97)00262-X CCC: $15.00 © 1998 American Chemical Society Published on Web 05/30/1998

3608 Langmuir, Vol. 14, No. 13, 1998

Okumura et al.

more than 7.4 wt % loading of Ge, the above procedure was repeated two or three times. The Ge loading was determined using an X-ray fluorescence spectrometer (SEIKO SEA-2010) using weighted GeO2 diluted with SiO2 as reference. FT-IR Measurement. FT-IR spectra were measured on a JASCO FT-IR 230 spectrometer with 2 cm-1 resolution. The spectra were recorded in an in-situ IR cell with two NaCl windows attached to a closed circulating system. The sample was pressed to a wafer 2 cm in diameter. Preparation of GeO2/SiO2 and subsequent treatments were carried out in situ in the cell because of the instability of GeO2/SiO2 to moisture. Typically, a SiO2 (50 mg) disk was placed in the IR cell, evacuated at 473 K, and reacted with a given amount of Ge(OMe)4 vapor at 393 K, followed by evacuation at 473 K and calcination under circulating oxygen (20.0 kPa) at 693 K for 1 h. Ge K-Edge XAFS Measurement. X-ray absorption fine structure (XAFS) spectra at the Ge K-edge were measured at beam line 7C in the Photon Factory of the National Laboratory for High-Energy Physics (KEK-PF) (Proposal No. 94G-203). The beam line is equipped with a sagittal focusing Si(111) doublecrystal monochromator. The monochromator was detuned to 70% in order to remove the higher harmonics. The storage ring energy was 2.5 GeV with a ring current of 250-350 mA. All spectra were recorded at room temperature in transmission mode. The sample was transferred to glass cells with two Kapton windows which were connected to a closed circulating system without contacting air. Two ion chambers filled with N2 100% and Ar 15%/N2 85% were used as X-ray detectors for I0 (before sample) and I (after sample), respectively. For extended X-ray absorption fine structure (EXAFS) analysis, the oscillation was extracted from EXAFS data by a spline smoothing method.11 The oscillation was normalized by edge height around 50 eV above the threshold. The energy dependence of the edge height was corrected by the McMaster equation.12 The Fourier transformation of the k3-weighted EXAFS oscillation from k space to r space to obtain a radial distribution function was performed over the range 30-140 nm-1. The inversely Fourier filtered data were analyzed by a curve-fitting method based on eq 1.

χ(k) )

∑N F (k) exp(-2σ j

j

2 j

kj2) sin(2krj + φj(k))/krj2 (1)

kj ) (k2 - 2m∆E0j/p2)1/2 where Nj, rj, σj, and ∆E0j represent the coordination number, the bond distance, the Debye-Waller factor, and the difference of the threshold energy between reference and sample, respectively. Fj(k) and φj(k) represent amplitude and phase shift functions, respectively. For the curve fitting analysis, empirical phase shift and amplitude functions were extracted from hexagonal type GeO2 as the reference compound. The error of the analysis was estimated by the R factor (Rf) calculated by eq 2.

Rf )

∫|k χ(k) 3

obs

∫

- k3χ(k)calc|2 dk/ |k3χ(k)obs|2 dk

(2)

The error bars in the present curve-fitting analysis for bond distance and coordination number are estimated to be (0.002 nm and (20%, respectively. The above analysis of EXAFS data was performed using the “REX” program (RIGAKU).

Results Characterization of GeO2 Supported on SiO2 by FT-IR and XRD. Figure 1 shows the relationship between the amount of Ge(OMe)4 exposed to the pretreated SiO2 in a closed circulating system and the Ge weight percent in the obtained GeO2/SiO2 sample. The Ge/SiO2 weight ratio increased linearly until it reached saturation at 7.4 wt %. Introduction of Ge(OMe)4 vapor more than (11) Iwasawa, Y., Ed. X-ray Absorption Fine Structure for Catalysts and Surfaces; World Scientific: Singapore, 1996. (12) McMaster, W. H.; Kerr Del Grande, N.; Mallet, J. H.; Hubell, J. H. Comparison of X-ray Cross Section; National Technical Information Service: Springfield, 1969.

Figure 1. Relationship between given amounts of Ge(OMe)4 exposed to SiO2 at 393 K and Ge/SiO2 weight ratios in the obtained GeO2/SiO2.

1.3 kPa to the system did not raise the Ge loading at saturation significantly, as shown in Figure 1. Thus 7.4 wt % was the maximum Ge loading attained by the onecycle CVD process of Ge(OMe)4 on SiO2. FT-IR spectra after the reaction of Ge(OMe)4 with the SiO2 surface and the calcination of the obtained sample were measured to examine the CVD process on SiO2 pretreated for 1 h at 473 K (Figure 2A) and at 723 K (Figure 2B). In Figure 2A, spectrum a for the 473 K-treated SiO2 shows a sharp peak at 3745 cm-1 and a broad one at 3650 cm-1, which have been assigned to isolated and hydrogen-bonded OH groups, respectively. After the CVD reaction with an excess amount of Ge(OMe)4 (spectrum b), most of the isolated OH groups disappeared, and the peaks at 2989, 2946, and 2843 cm-1 due to the C-H stretching modes and those at 1465 and 1452 cm-1 due to the bending modes developed, which may originate from surface Ge(OMe)x (x e 4), because they coincide with the peaks for Ge(OMe)4.13 The intensity of the broad band for the hydrogen-bonded hydroxyl groups also decreased by the CVD of Ge(OMe)4, and a new peak at 3450 cm-1 appeared, which may be due to OH groups interacting with Ge(OMe)x. After calcination of the Ge(OMe)x/SiO2 sample, the methoxy groups completely disappeared and the two absorption bands appeared at 3745 and 3676 cm-1. The former band is identical to that for the isolated OH groups originally bound on SiO2, which was significantly reduced by the CVD reaction with Ge(OMe)4 and increased upon calcination with half of the original intensity. The new peak at 3676 cm-1 is assignable to isolated OH groups on the GeO2 overlayer deposited on SiO2, because the observed frequency agrees well with that for isolated OH groups on bulk GeO2, as reported in the literature.14,15 Both isolated and hydrogen-bonded OH groups are observed on bulk GeO2, whereas the hydrogenbonded OH groups were not present on the deposited GeO2 layers. The reactivity of hydroxyl groups on SiO2 pretreated at 723 K with Ge(OMe)4 was also examined by FT-IR in Figure 2B. The average number of surface OH groups on silica can be controlled by changing the evacuation temperature.16,17 Most of the physisorbed water is re(13) Johnson, O. H.; Fritz, H. E. J. Am. Chem. Soc. 1953, 75, 718. (14) McManus, J. C.; Matsushita, K.; Low, M. J. D. Can. J. Chem. 1969, 47, 1077. (15) Metcalfe, A.; Shankar, S. U. J. Chem. Soc., Faraday Trans. 1 1980, 76, 489. (16) Iwasawa, Y., Ed. Tailored Metal Catalysts; Reidel: Dortrecht, 1986.

GeO2 Sub-monolayers on SiO2

Langmuir, Vol. 14, No. 13, 1998 3609

Figure 2. (A) FT-IR spectra of SiO2 (Aerosil 300) (a) evacuated at 473 K for 1 h, (b) after CVD reaction with excess Ge(OMe)4 at 393 K followed by evacuation at 473 K, and (c) after calcination of (b) at 693 K for 1 h. (B) FT-IR spectra of SiO2 (Aerosil 300) (a) evacuated at 723 K for 1 h, (b) after CVD reaction with excess Ge(OMe)4 at 393 K followed by evacuation at 473 K, and (c) after calcination of (b) at 693 K for 1 h. (C) FT-IR spectra of SiO2 (ox-50) (a) evacuated at 473 K for 1 h and (b) after calcination of the Ge(OMe)4-deposited SiO2 at 693 K for 1 h.

moved at 473 K, leaving a saturated amount of OH groups (ca. 5 OH nm-2), while strongly hydrogen-bonded OH groups and some weakly hydrogen-bonded OH groups are removed by evacuation at 723 K. As shown in Figure 2B most of the hydrogen-bonded OH groups were removed, and the isolated OH groups were retained on the SiO2 surface treated at 723 K. The preferential reaction of the isolated OH groups on the 723 K-treated SiO2 with Ge(OMe)4 was observed in Figure 2B and was similar to the case for the 423 K-treated SiO2 (Figure 2A). A new peak at 3420 cm-1 appeared, which may be due to OH groups interacting with surface Ge(OMe)x species. By calcination of the obtained Ge(OMe)x/SiO2 sample at 693 K the isolated OH groups increased again, and the new peak at 3678 cm-1 also appeared, which is assigned to isolated OH groups on the deposited GeO2 layers. These features are similar to those observed with the 423 K-treated SiO2. The CVD of Ge(OMe)4 was also performed on lowsurface-area SiO2, ox-50, whose surface area is 1/6 of that of Aerosil 300, to check the influence of surface area on the CVD process. Similar FT-IR spectra were obtained, as shown in Figure 2C, which indicates that the CVD process to form GeO2 overlayers is independent of the kind of SiO2. After the CVD reaction with Ge(OMe)4 at 393 K, the sample was evacuated at 473 K for 2 h to remove the excess unreacted Ge(OMe)4, but the Si-OH peak at 3745 cm-1 remained unchanged and the methoxy peaks retained their intensity. The peak of the isolated Si-OH groups was recovered only after calcination of the sample at 693 (17) Zhuravlev, L. T. Langmuir 1987, 3, 316.

Figure 3. FT-IR spectra of GeO2/SiO2 with different Ge loadings prepared by a one-cycle CVD process with various amounts of Ge(OMe)4 (7.4 wt % and less) and by repeated CVD processes (8.7 wt % and above).

K. The X-ray fluorescence analysis showed no change of the Ge loading in GeO2/SiO2 before and after the calcination, which implies that the growth of the Si-OH peak is not due to desorption of the deposited GeO2. Figure 3 shows typical FT-IR spectra of GeO2/SiO2 with different Ge loadings prepared by varying the amount of Ge(OMe)4 vapor (up to 7.4 wt %) exposed to SiO2. The

3610 Langmuir, Vol. 14, No. 13, 1998

Okumura et al.

Figure 4. Intensity of Ge-OH (3676 cm-1) and Si-OH (3745 cm-1) peaks for GeO2/SiO2 as a function of Ge wt %: (O) GeOH; (×) Si-OH.

samples with Ge loadings of more than 7.4 wt % were prepared by further CVD of the 7.4 wt % GeO2/SiO2 sample with Ge(OMe)4 vapor at 393 K, followed by calcination at 693 K. The intensity of the Ge-OH peak at 3676 cm-1 increased with an increase of Ge loading, while the intensity of the Si-OH peak at 3745 cm-1 decreased as Ge loading increased. Figure 4 shows the plots of their intensities against the Ge wt %. The linear increase of Ge-OH intensity and reversely the linear decrease of SiOH intensity were observed in the range of Ge loading 0-7.4 wt %. The isolated Si-OH intensity for the sample with the Ge loading 7.4 wt % decreased to half of the original Si-OH intensity before Ge deposition, at which the Ge-OH intensity was nearly the same as the Si-OH intensity. The loading of 7.4 wt % Ge corresponds to occupation of 1/5 monolayer of the silica surface, assuming that the two-dimensional [GeO2] unit area is 0.1 nm2. We have tried to obtain a full monolayer of GeO2 on the SiO2 surface by repeated Ge(OMe)4 CVD-calcination cycles. To know the reactivity of Ge(OMe)4 with the SiO2 surface and the GeO2 layer (GeO2/SiO2; 7.4 wt %) obtained by one CVD-calcination cycle using excess Ge(OMe)4, a small amount of Ge(OMe)4 vapor (67 Pa) was introduced to the 7.4 wt % GeO2/SiO2 at 393 K. Figure 5 shows the drastic decrease of the Ge-OH peak by the reaction of Ge(OMe)4 with the GeO2/SiO2. On the contrary, the SiOH peak reduced little. It suggests that Ge(OMe)4 vapor reacts preferentially with the Ge-OH groups on the GeO2 sub-monolayers rather than the Si-OH groups on the exposed SiO2 surface. After calcination of the GeO2/SiO2 samples produced by two or three cycles of CVD processing, we found little change in the peak heights of Si-OH and Ge-OH. The break in the lines in Figure 4 suggests a change in the growth mode of the GeO2 layer. Figure 6 shows the XRD pattern of GeO2/SiO2. No diffraction line was detected with the one to three CVD cycles samples, which demonstrates that the deposition of GeO2 on SiO2 did not grow to three-dimensional crystals. Figure 7 shows FT-IR spectra of the GeO2/SiO2 sample preserved under moisture, together with the fresh GeO2/ SiO2. Moisture significantly decreased the Ge-OH peak intensity (3676 cm-1). This change is possibly caused by crystallization of the GeO2 overlayers and decrease of the GeO2 surface area. The crystallization of the GeO2 overlayers was evidenced by the growth of a diffraction line at 2θ ) 25.5°, as shown in Figure 6d. The ease of destruction of the GeO2 structure with moisture may be related to the crystal structure of GeO2, which was similar to the water-soluble hexagonal type GeO2 as characterized

Figure 5. FT-IR spectra (a) of GeO2/SiO2 (Ge: 7.4 wt %) and (b) after exposure of (a) to a small amount (67 Pa) of Ge(OMe)4 at 393 K.

Figure 6. X-ray diffraction patterns of GeO2/SiO2 prepared by repeated CVD processing: (a) one-cycle CVD; (b) two-cycle CVD; (c) three-cycle CVD; (d) GeO2/SiO2 exposed to moisturesaturated air.

by EXAFS (discussed hereinafter). It was expected that the crystallization of the GeO2 sub-monolayer to the threedimensional GeO2 would cause the reappearance of the SiO2 surface which was covered by GeO2 layers and hence the growth of Si-OH groups, but the Si-OH peak intensity did not increase on exposure to moisture. However, the peak became somewhat broader due to the overlap of a new peak at 3741 cm-1 with the original Si-OH peak at 3745 cm-1. Morrow et al. observed the appearance of a band at 3742 cm-1 during rehydration when a small quantity of water was added to fully dehydrated silica.18-20 They attributed the new band to a weakly interacting pair of vicinal single silanols. Mol et al., on the other hand, attributed the peak at 3742 cm-1 to geminal (18) Morrow, B. A.; Cody, I. A. J. Phys. Chem. 1975, 79, 761. (19) Morrow, B. A.; Cody, I. A.; Lee, L. S. M. J. Phys. Chem. 1976, 80, 2761. (20) Morrow, B. A.; McFarlan, A. J. J. Phys. Chem. 1992, 96, 1395.

GeO2 Sub-monolayers on SiO2

Langmuir, Vol. 14, No. 13, 1998 3611

Figure 7. FT-IR spectra of (a) GeO2/SiO2 (Ge: 7.4 wt %) and (b) GeO2/SiO2 exposed to moisture-saturated air.

Figure 9. Ge K-edge EXAFS oscillations (A) and Fourier transforms (B): (a) as-deposited Ge(OMe)4 (no calcination); (b) GeO2/SiO2 (3.2 wt %); (c) 7.4 wt %; (d) 8.3 wt %; (e) GeO2 (hexagonal type).

Figure 8. FT-IR spectra (a) of GeO2/SiO2 (Ge 7.4 wt %), (b) after exposure of (a) to 6.6 kPa of pyridine, followed by evacuation at room temperature, and (c) after evacuation of (b) at 373 K.

silanediol.21,22 In the present case, we do not know the correct origin for the new peak, but the peak at 3741 cm-1 may originate from the Si-OH groups created by the hydration and destruction of a Si-O-Ge bond. It seems that the generated SiO2 surface was reconstructed or locally rearranged upon the hydration. Figure 8 shows the FT-IR spectra of pyridine adsorbed on GeO2/SiO2 to examine the acidic character of GeO2/ SiO2. There are neither strong Bro¨nsted nor Lewis sites on SiO2 surfaces.23 The spectrum after the adsorption of pyridine and subsequent evacuation at room temperature shows two bands at 1445 and 1596 cm-1 in the region of the ring vibration of pyridine. It is known that adsorption of pyridine on Bro¨nsted acid sites yields pyridinium ion with an absorption band at 1540 cm-1.24 Pyridine (21) Van Roosmalen, A. J.; Mol, J. C. J. Phys. Chem. 1978, 82, 2748. (22) Van Roosmalen, A. J.; Mol, J. C. J. Phys. Chem. 1979, 83, 2485. (23) Haukka, S.; Root, A. J. Phys. Chem. 1994, 98, 1695.

coordinated on Lewis acid sites has a band at 1440-1465 cm-1, and the frequency depends on the strength of bonding. By evacuation at 373 K the adsorbed pyridine disappeared. Thus there is no pyridine strongly adsorbed on GeO2/SiO2, indicating no Bro¨nsted and Lewis acid sites on the surface. This is contrasted to the report on GeO2/ Al2O3, where both types of acidic sites were detected from the IR measurement of adsorbed pyridine.9 The reason of neutrality of the GeO2 layers on SiO2 may be ascribed to the neutral character of the Ge-O-Si bond at the interface and also to the equivalent valency of Ge and Si atoms. Characterization of GeO2 Supported on SiO2 by EXAFS. EXAFS spectra for GeO2/SiO2 were measured to characterize the local structure around Ge atoms. Figure 9 shows the EXAFS oscillations (k3χ(k)) at the Ge K-edge and their Fourier transformations for the GeO2/SiO2 samples with different Ge loadings and for bulk GeO2 (hexagonal type). The spectra of Figure 9a-d are similar to each other, but the observed spectrum is considerably different from that of the bulk GeO2 (Figure 9e). The peak for GeO2/SiO2 at 0.28 nm (phase shift uncorrected) was attributed to the Ge-Ge bond by the curve-fitting analysis. The peak intensity was remarkably smaller than that for bulk GeO2. The peak height and position of the Ge-O bond observed at 0.14 nm (phase shift uncorrected) did not change in these GeO2/SiO2 samples. Similar (24) Parry, E. P. J. Catal. 1963, 2, 371.

3612 Langmuir, Vol. 14, No. 13, 1998

Okumura et al.

Table 1. Curve-Fitting Results of Ge K-Edge EXAFS Spectra for GeO2 and GeO2/SiO2 sample SiO2 (Aerosil 300) reacted with Ge(OMe)4 1.7 wt % GeO2/SiO2 (ox-50) 3.1 wt % GeO2/SiO2 (Aerosil 300) 7.4 wt % GeO2/SiO2 (Aerosil 300) 8.7 wt % GeO2/SiO2 (Aerosil 300) GeO2 (hexagonal)f GeO2 (tetragonal)f

scatterer atom CNa rb/nm

∆E0c/ eV σd/nm Rfe/%

O Ge

4.1 0.172 -2.4 0.0050 2.3 0.311 -10.0 0.0089

O Ge O Ge O Ge O Ge O Ge O O Ge

4.6 2.8 4.1 2.0 3.9 2.0 4.5 2.4 (4) (4) (4) (2) (8)

0.172 -1.2 0.0061 0.307 -16.6 0.0098 0.173 -0.2 0.0052 0.306 -19.9 0.0105 0.173 -1.5 0.0052 0.311 -10.0 0.0086 0.173 -1.5 0.0059 0.313 -6.2 0.0083 (0.174) (0.315) (0.187) (0.190) (0.342)

5.9 3.8 3.5 4.4 2.6

a Coordination number. b Bond distance. c Energy difference in the origin of a photoelectron between the reference and the sample. d Debye-Waller factor. e Residual factor. f Data from X-ray crystallography.

results were obtained on GeO2 supported on a low-surfacearea SiO2 (ox-50). The low intensity of the Ge-Ge peak for GeO2/SiO2 was independent of Ge loading for loadings below 7.4 wt % Ge. The Ge-Ge contribution in the EXAFS spectra increased a little above this loading (two or three CVD cycles samples), which is apparent by the rapid wave observed around 98 and 118 nm-1 in the k3χ(k) spectra, as indicated by arrows (Figure 9d). Table 1 summarizes the curve-fitting results of the EXAFS data for GeO2/SiO2 as well as the crystallographic data for hexagonal and tetragonal GeO2.25,26 Bond distances determined by the curve-fitting analysis were 0.172-0.173 nm for Ge-O and 0.306-0.313 nm for Ge-Ge. The curve-fitting analysis showed again the increase of Ge-Ge coordination number of two CVD cycles sample (8.7 wt %). Such an increase is caused by the growth of the second layer of GeO2 on the first layer of GeO2. The GeO2/SiO2 (ox50) also had a larger Ge-Ge coordination number, which will be discussed later. The EXAFS data for the as-deposited CVD sample before calcination in Figure 9a are essentially the same as those for the calcined samples. This suggests that the fundamental structure of the GeO2 layer on SiO2 was similar before and after calcination and that the Ge-Ge bonding already existed before calcination. The Si-O-Ge(OMe)3 species may be assembled through the methoxy linkage. The formation of a similar network structure has been proposed in the process of deposition of Si(OMe)4 on mordenite by IR and gas-phase analysis.27 Discussion Structure of GeO2 Overlayer. FT-IR spectra indicated that the CVD process of Ge(OMe)4 consumed the isolated OH groups on the SiO2 surface and that the produced GeO2 overlayers possessed isolated Ge-OH groups. The number of isolated Ge-OH groups per Ge atom in GeO2/SiO2 can be estimated from comparison of the Ge loading with the intensity of the Ge-OH groups observed in the FT-IR spectra, assuming that the absorption coefficients of O-H stretching peaks are similar in Ge-OH groups on GeO2 and Si-OH groups on SiO2. The density of isolated OH groups on SiO2 evacuated at 473 (25) Bauer, W. H.; Khan, A. A. Acta Crystallogr. 1971, B272, 133. (26) Smith, G. M.; Isaacs, P. B. Acta Crystallogr. 1964, 17, 842. (27) Niwa, M.; Kawashima, Y.; Hibino, T.; Murakami, Y. J. Chem. Soc., Faraday Trans. 1 1988, 84, 4327.

K was estimated to be 2 OH nm-2 by 1H MAS NMR.23 According to their estimation, on GeO2/SiO2 (7.4 wt %) the amount of Si-OH groups decreased to 1 OH nm-2 and simultaneously the same number of Ge-OH groups grew in. From the number of Ge-OH groups and the Ge loading it can also be estimated that one Ge-OH group exists per two Ge atoms in the GeO2/SiO2 sample because 7.4 wt % Ge loading on 300 m2/g SiO2 corresponds to 2 [GeO2] units per 1 nm2, assuming the [GeO2] unit area is 0.1 nm2. EXAFS spectra revealed the presence of Ge-O-Ge in the GeO2/SiO2 overlayer. Comparing the Ge-O and GeGe distance of GeO2/SiO2 with those of two GeO2 reference samples (hexagonal and tetragonal) shown in Table 1, we concluded that the GeO2 has a similar structure to the hexagonal one. Hibino et al. also found a similar structure in the GeO2 thin layer on mordenite prepared by CVD of the Ge(OMe)4.7 The height of Ge-Ge peak in the Fourier transform of EXAFS for GeO2/SiO2 samples seems to be independent of Ge loading for loadings below 7.4 wt %. This observation suggests the existence of Ge-O-Ge networks even when the GeO2 loading is as small as 3.2 wt %, which corresponds to only 0.1 monolayer. If Ge(OMe)4 monomer randomly reacted with the OH group, we should observe Ge monomers at the small loading and the increase of the Ge-Ge interaction with the increase of the Ge loading. Thus the constant height of the Ge-Ge peak for the samples with Ge loadings below 7.4 wt % implies that the GeO2 sub-monolayers grow as islands with a similar dimension and that the number of the islands increases with the increase of Ge loading. The 7.4 wt % loading is the maximum amount of Ge deposited by the one-cycle CVD reaction of Ge(OMe)4 with surface silanols. Further introduction of Ge(OMe)4 by two and three CVD cycles is necessary for more loading, which involves the reaction of Ge(OMe)4 with the OH groups on GeO2, as shown in Figures 3 and 4. This is supported by the EXAFS observation. The Ge-Ge coordination increased a little above the loading of 7.4 wt % shown in Figure 9. It may be difficult to obtain a full monolayer of GeO2 on SiO2 surfaces under the present CVD conditions. However, these results present a possible way for layer-by-layer growth of GeO2 by the repeated deposition procedure. The Ge-Ge coordination number of 1.7 wt % GeO2/SiO2 (ox-50), which has a smaller surface area, is larger than that of 7.4 wt % GeO2/SiO2 (aerosil 300). This may be due to the difference in surface area between the SiO2 groups. The surface area of ox-50 SiO2 is only 17% of that of aerosil 300 SiO2, while the loading of Ge on ox-50 SiO2 corresponds to 24% of that on the aerosil SiO2. Excess GeO2 more than 1/5 monolayer on ox-50 SiO2 may be present as the second layer of GeO2. CVD Reactions between Ge(OMe)4 and Silanols. The results that the saturated amount of Ge deposited in the one-cycle CVD process by using excess Ge(OMe)4 was about 1/5 monolayer of the SiO2 surface and that the SiOH peak was significantly depleted by the CVD reaction with Ge(OMe)4 at 393 K and regained by calcination of the Ge(OMe)x/SiO2 at 693 K may be explained as follows. As the amount of the Si-OH groups depleted after calcination was equal to the amount of Ge loading, as discussed above, it seems that the reaction of Si-OH with Ge(OMe)4 at 393 K took place stoichiometrically as follows:

Si-OH + Ge(OMe)4 f Si-O-Ge(OMe)3 + MeOH (3) MeOH can react with Si-OH to form Si-OMe as follows:

GeO2 Sub-monolayers on SiO2

Langmuir, Vol. 14, No. 13, 1998 3613

Si-OH + MeOH f Si-OMe + H2O

(4)

Calcination of the Si-OMe group results in the formation of the Si-OH group again. It is possible that the SiO-Ge(OMe)3 species further react with Si-O-Si to produce (Si-O)2Ge(OMe)2 + Si-OMe without involving hydroxyl groups as follows: Si–O Si–O–Ge(OMe)3 + Si–O–Si

OMe

+ Si–OMe

Ge Si–O

OMe

(5)

Alternatively, the CVD reaction may occur as follows: Si–O 2Si–OH + Ge(OMe)4

OMe

+ 2MeOH

Ge Si–O

OMe

(6)

The obtained (Si-O)2Ge(OMe)2 species should be transformed to Si-OH (recovered) and Si-O-GeOx by calcination. The concentration of the isolated Si-OH groups on the SiO2 surface is suggested to be as small as 2 OH/nm2, so the probability that a Ge(OMe)4 molecule reacts exclusively with two Si-OH groups might be small. The overall steps are unlikely to fit the stoichiometry observed experimentally. The third plausible CVD processing involves the reaction between Si-OH and the [Ge(OMe)4]n oligomer/assembly. The alkoxide compounds are likely to form an oligomer/assembly. When n ) 2, the CVD reaction may simply be shown as follows:

Si-OH + [Ge(OMe)4]2 f Si-O-[Ge(OMe)3][Ge(OMe)4] + MeOH (7) Si-O-[Ge(OMe)3][Ge(OMe)4] can further react with SiOH and Si-O-Si. The [Ge(OMe)4]n CVD process produces the as-deposited surface species with Ge-Ge bonding,

which was found in the EXAFS data as shown in Figure 9. The [Ge(OMe)4]n CVD process (eq 7) is plausible under the present conditions, but other reactions (eqs 1-6) may also contribute to the CVD process. The stoichiometric features observed in the CVD processing are assumed to be due to the surface-limited chemical processes. If Ge species are incorporated into the SiO2 bulk, the intensities of Si-OH and Ge-OH peaks would not change linearly as a function of Ge loading, unlike the result of Figure 4. The 1/5 monolayer coverage of GeO2 at saturation in the one-cycle CVD process indicates that the coverage of GeO2 layers is limited by the Si-OH quantity and the large molecular size of [Ge(OMe)4]n under the present CVD conditions. Conclusions Formation of GeO2 overlayers on SiO2 was achieved by the chemical reaction of Ge(OMe)4 vapor with OH groups on SiO2 surfaces. The reaction occurred preferentially with the isolated OH groups of SiO2. Monolayer growth of GeO2 (Ge loading e 7.4 wt %) on SiO2 was characterized by FT-IR, XRD, and Ge K-edge EXAFS. Neither Bro¨nsted nor Lewis acidic sites were detected on GeO2/SiO2. The GeO2/SiO2 showed Ge-O-Ge networks and the existence of one isolated Ge-OH group per two Ge atoms. The local structure of Ge in GeO2/SiO2 was similar to that for hexagonal GeO2 as characterized by EXAFS. The maximum Ge loading obtained by the one-cycle CVD process on SiO2 (Aerosil 300) was 7.4 wt %, which corresponds to 1/ of the GeO full monolayer over the SiO surface. 5 2 2 Repeated CVD processing cycles resulted in the multilayer growth of GeO2 on the first GeO2 overlayer, as suggested by FT-IR. Acknowledgment. This work has been supported by CREST (Core Research for Evolutional Science and Technology) of Japan Science and Technology Corporation (JST). LA970262N