Surface Chemical Species Investigation by FT-IR ... - ACS Publications

M.-I. Baraton*, W. Chang, and B. H. Kear. Faculty of Sciences, LMCTS URA 320 CNRS, 123 Av A. Thomas, F-87060 Limoges, France, and Department of ...
0 downloads 0 Views 510KB Size
J. Phys. Chem. 1996, 100, 16647-16652

16647

Surface Chemical Species Investigation by FT-IR Spectrometry and Surface Modification of a Nanosized SiCN Powder Synthesized via Chemical Vapor Condensation M.-I. Baraton,*,† W. Chang,‡ and B. H. Kear‡ Faculty of Sciences, LMCTS URA 320 CNRS, 123 AV A. Thomas, F-87060 Limoges, France, and Department of Mechanics and Materials Science, Rutgers UniVersity, Piscataway, New Jersey 08855-0909 ReceiVed: April 25, 1996; In Final Form: June 3, 1996X

The surface of a nanosized silicon carbonitride powder synthesized by chemical vapor condensation from hexamethyldisilazane was analyzed by FT-IR spectrometry. Several chemical species were identified, and their behavior toward probe molecules was studied. Deuterium and methanol additions brought evidence of open pores. A slight oxidation led to pore coalescence and strong modifications in the chemical composition of the first atomic layer even though a complete silica coverage was not reached. Moreover, the SiCN surface further nitrided by ammonia addition at high temperature appeared to have, under those conditions, a behavior different from the silicon carbide surface.

Introduction It is now well-known that the surface of nanostructured powders must be carefully characterized independently from any bulk characterization.1 This is specially relevant to nonoxide ceramics. Indeed, the surface contamination by atmospheric water cannot be avoided, and as a consequence, the presence of oxygen in the first atomic layer may drastically change the properties of these ceramics.2,3 Moreover, the chemical composition of the surface critically depends on the synthesis conditions, and, in return, the surface reactivity is strongly related to the nature of the surface species. Besides, the surfaces of nanoparticles often show unique reactivities and adsorbed species compared to microscale particles.4 The aim of this work was to determine by Fourier transform infrared spectrometry the chemical composition of the surface of a nanostructured Si-C-N powder prepared by chemical vapor condensation.5 The modification of this surface was then studied when the powder was heated under oxygen and ammonia. These results can contribute to a better understanding of problems encountered in applications such as sintering and deagglomeration of nanosized powders. Powder Synthesis 1. Process. The experimental arrangement of the chemical vapor condensation (CVC) process used to synthesize nanostructured SiCN powder is schematically shown in Figure 1. High-purity helium is bubbled through the liquid metalorganic precursor, hexamethyldisilazane (HMDS), at room temperature. The gas stream is introduced at 8.55 × 10-3 mol/min into the dynamically pumped vacuum chamber via a needle valve. The pressure in the chamber is maintained at 113 Pa by high-speed pumping. A Mo-wound furnace tube of Al2O3 (99.8%) provides a heat source for controlled decomposition of the precursor. The temperature of the tubular reactor is fixed at 1673 K. At the outlet of the furnace tube, rapid expansion of the two-phase gas stream (gas + nanoparticles) serves to mitigate particle growth and agglomeration. Finally, the nanoparticles condense out on a rotating liquid N2 cooled substrate from which the particles can be scraped off and collected. †

LMCTS URA 320 CNRS. Rutgers University. X Abstract published in AdVance ACS Abstracts, September 1, 1996. ‡

S0022-3654(96)01196-3 CCC: $12.00

2. Bulk Characterization. Carbon rich SiCN powder is produced by this CVC method according to the measurement by Rutherford backscattering spectroscopy with ion beam energies of 3.5 MeV, which gives the SiCxNyOz composition (x ) 1.51, y ) 0.49, and z ) 0.28).6 By X-ray diffraction analysis, the as-synthesized nanostructured Si-C-N powder is amorphous. Surface area measurement on the SiCN powder, as determined by single point nitrogen adsorption (BET), gives a value of 272 m2‚g-1. The density of the SiCN powders is determined to be 2.78 g‚cm-3. The average crystallite size of the SiCN powder determined from TEM is about 10 nm. In Figure 2 is reported the Fourier transform infrared (FTIR) spectrum of SiCN. The SiCN powder is dispersed in a KBr matrix and pressed into a pellet. The spectrum, recorded on a Nicolet 5DX spectrometer with a 4 cm-1 resolution, presents a broad band at 867 cm-1, predominantly due to the presence of SiC bonds. This main band is broadened on its high-wavenumber side by shoulders due to the contributions of the stretching vibrations of SiN bonds at 950 cm-1 and SiO bonds at 1085 cm-1.7,8 The other broad band at about 3430 cm-1 will be discussed in the surface analysis below as well as the weaker band at 1636 cm-1. The Raman spectrum of the SiCN powder (Figure 3) was obtained by using a Dilor microprobe. The two broad bands around 1590 and 1350 cm-1 are the strong proof of the presence of amorphous carbon in the powder.9,10 The vibrational modes of the SiCN compound itself are not visible. Indeed, nanosized powders do not usually present clear Raman spectra, whereas the intense bands of amorphous carbon are easily seen. Surface Characterization Experimental Section All of the spectra were recorded by using a Nicolet 5DX Fourier transform infrared spectrometer in the 400-4000 cm-1 range with a 4 cm-1 resolution. For surface analyses, a specially designed cell was used.1 This cell allows us to run in situ experiments under a vacuum or controlled pressures of gases. Inside the cell a small furnace enables heating the sample from room temperature (rt) to 873 K. The powder to be analyzed was pressed into a grid-supported pellet and placed inside the furnace. Because the surface species contribute in a minor way to the transmission spectrum, the amount of powder needed for obtaining a clear surface spectrum must be sufficient, that is, around 20 mg. © 1996 American Chemical Society

16648 J. Phys. Chem., Vol. 100, No. 41, 1996

Baraton et al.

Figure 1. Schematic of the chemical vapor condensation process, which consists of (1) a precursor delivery system, (2) an ultrahigh vacuum chamber, (3) a tubular reactor, and (4) a cold substrate for powder deposition and collection.

Figure 2. Infrared bulk spectrum of the SiCN powder in a KBr matrix.

Figure 3. Raman spectrum of the SiCN nanosized powder.

Before gas addition, the powder surface was activated. An activation consists of heating the sample at 873 K under dynamic vacuum (∼10-6 mbar) for 2 h and cooling it to room temperature while still under dynamic vacuum. This treatment cleans the surface from all physisorbed and weakly chemisorbed species. All gases used in the experiments (deuterium, oxygen, and ammonia) were from Air Liquide (99.9% purity) and were used without further purification. Methanol from Merck Uvasol was dried on molecular sieves. Results and Discussion 1. Activation. The infared spectrum of a raw SiCN pellet is presented in Figure 4a. The absorption due to the vibrational

Figure 4. Infrared spectra of a pure SiCN pellet: (a) sample at room temperature under vacuum; (b) sample activated at 873 K for 1 h; (c) difference spectrum b - a.

modes of the bulk is very intense and obscures the wavenumber region below 1000 cm-1. On the highest wavenumber side the main broad band absorbs around 3400 cm-1. When the SiCN pellet is heated at 873 K under vacuum for 2 h (Figure 4b), this broad band strongly decreases. Concomitantly, we note an increase of a sharp band at 3744 cm-1 and of another band at 2297 cm-1. Moreover, three groups of bands also decrease at 2960, 1685, and 1585 cm-1. As mentioned above, this thermal treatment of the sample corresponds to an activation of the surface and removes physisorbed and weakly chemisorbed species. An additional view of the evolution is given in Figure 4c and corresponds to the difference between

Nanosized SiCN Powder Synthesized via CVC the spectra recorded before and after the thermal treatment. The negative bands correspond to disappearing species, whereas the positive bands correspond to appearing or increasing species. The disappearance of the broad band centered at 3355 cm-1 and of the band at 1685 cm-1 shows that physisorbed water molecules are released by heating.10 Indeed, these water molecules are hydrogen bonded on hydroxyl surface groups as on silica, silicon carbide, and silicon nitride.11-14 The corresponding freed hydroxyl groups absorb at 3744 cm-1. This frequency is very close to that of hydroxyl groups on silica (3747 cm-1).15 Therefore, we assigned this sharp band to the ν(OH) stretching mode of SiOH silanol groups. The broader band around 3647 cm-1 which appears on the low-wavenumber side of the 3744 cm-1 band (Figure 4b) is also assigned to SiOH groups. However, its lower absorption frequency and its broadness make it comparable to the band usually assigned to hydrogen bonded hydroxyl groups.11,16 As a consequence, two types of SiOH groups are identified and will be discussed later on. The broad band at 3370 cm-1 revealed by the activation (Figure 4b) is assigned to the ν(NH) stretching vibration in imido groups by comparison with Si3N4.12,14 The corresponding deformation mode overlapped by the bulk vibrations should absorb around 1200 cm-1.17 Besides, the band at 1585 cm-1 clearly seen in the spectrum of the raw sample (Figure 4a) disappears by heating (Figure 4b). On the difference spectrum (Figure 4c), this disappearing band can be pointed as a shoulder on the low-wavenumber side of the 1685 cm-1 negative band. This 1585 cm-1 band is attributed to the bending mode of NH3 molecules coordinated on the surface. The stretching vibrations of these NH3 molecules absorb in the same region as the ν(OH) stretching vibration of H-bonded water and cannot be discriminated from them. As expected, gaseous NH3 is released from the surface by heating, like in other nitride compounds.18 In the 2900 cm-1 region several bands are characteristic of the CH stretching vibrations (Figure 4a) and decrease by heating (Figure 4b). The weak band at 3064 cm-1, which may already exist in the raw sample, has a too high absorption frequency to be attributed to a CHx aliphatic group. It may be assigned rather to ν(CH) with the carbon atom in the sp2 state, since exhaust gases during the synthesis could be C2H6, C2H4, or/and C2H2.19 The corresponding δ(CH) bending vibrations of all of these CHx groups should fall in the 1450 cm-1 region where the bulk vibrations are very intense. Moreover, a band appears at 2297 cm-1 during the activation (Figure 4b). In the 2000-2300 cm-1 region absorb the ν(SiH) stretching modes,20-22 whose absorption frequencies are very dependent upon the electronegativity of the other three atoms bonded to silicon.23-28 But this region also corresponds to the stretching modes of triple bonds such as -CtN, -NtC, and CtC. It precisely turns out that, as mentioned above, the ν(CH) absorption frequency above 3000 cm-1 could be related to -CtC- groups. However, the presence of -CtN and -NtC groups may not be discarded a priori. As a conclusion, this activation process allowed us to identify several chemical groups present in the SiCN powder, namely, SiOH, SiNH, Csp2H, Csp3H which are revealed by the elimination of water and ammonia. A band at 2297 cm-1 also brought the information that other species exist on the surface, possibly SiH and/or triple bonds. 2. Deuteration. In order to check the location of the previously identified chemical groups, an H/D isotopic exchange was performed on an activated powder. Indeed the hydrogen atoms will exchange only if they are accessible to deuterium, that is, if they are located either on the surface or in open pores.

J. Phys. Chem., Vol. 100, No. 41, 1996 16649

Figure 5. Infrared spectra of a pure SiCN pellet: (a) sample activated at 873 K for 1 h; (b) sample after deuterium addition at 873 K (three subsequent doses: 100 mbar for 1 h, 150 mbar for 1 h, 100 mbar for 2 h); (c) difference spectrum b - a.

As a consequence of the exchange, the vibrational frequencies involving hydrogen will shift toward lower wavenumbers. Several doses of deuterium were consecutively added at 873 K (100 mbar for 1 h, 150 mbar for 1 h, 100 mbar for 2 h), and then the spectrum was recorded after pumping and cooling at room temperature (Figure 5a,b). The result is quite clear (Figure 5a-c): all of the OH and NH groups are very easily exchanged. The difference spectrum shows the frequency shifts for ν(OH) and ν(NH) stretching bands. The ν(OH)/ν(OD) and ν(NH)/ ν(ND) frequency ratios are close to 1.356, as expected in the harmonic oscillator approximation.29 The CHx groups are not modified. Thus, they may be located partly or totally inside the bulk. However, it must be noted that usually CHx groups are not easily exchanged even though they are right on the surface.12,13 The 2297 cm-1 band decreases under deuteration. This fact may prove that the corresponding vibrational mode involves the hydrogen atoms. But, on the other hand, considering that this band can also be assigned to triple bonded species, deuterium may also act as a reducing gas. Therefore, hydrogen was added to the sample under the same conditions as deuterium to check this possibility. The hydrogen addition also results in a decrease of this 2297 cm-1 band (spectrum not shown) and in a slight transformation of the very weak absorption bands in the 2000-2300 cm-1 region. As a consequence, at this stage of the experiment, the assignment of the 2297 cm-1 band to either SiH or triple bonded species cannot be considered as completely sure. But, it is quite probable that triple bonded species exist and, at least, give rise to weak absorption bands in the 2000-2300 cm-1 region. The lack of appearing ν(SiD) stretching frequency around 1690 cm-1 also argues in favor of triple bonded species. 3. Methanol Addition. When methanol is added to an activated silica surface, the OH groups can link to surface silanols by hydrogen bonds. On another hand, it is well-known that a reaction11,30 between hydroxyl and silanol groups occurs on silica, leading to the formation of new methoxy groups and elimination of water. Since on the SiCN activated surface two types of silanol groups have been identified, their behavior toward methanol was checked.

16650 J. Phys. Chem., Vol. 100, No. 41, 1996

Baraton et al.

Figure 7. Infrared spectra of a pure SiCN pellet: (a) sample activated at 873 K for 1 h; (b) sample after heating at 873 K under 50 mbar of oxygen; (note: In this case, these two spectra haVe not been translated for comparison purposes); (c) difference spectrum b - a. Figure 6. Infrared spectra of a pure SiCN pellet: (a) sample activated at 873 K for 1 h; (b) sample after CH3OH addition (9 mbar, rt, 1 h); (c) sample after evacuation at 573 K for 1 h; (d) difference spectrum b - a; (e) difference spectrum c - a.

At room temperature, 9 mbar of gaseous CH3OH was added to the activated pellet (Figure 6). In the difference spectrum (Figure 6d), the broad positive band appearing at 3325 cm-1 along with the negative sharp one at 3744 cm-1 are characteristic of a hydrogen bond formation. The free silanols (3744 cm-1) transform into silanols hydrogen-bonded to the methanol molecules (3325 cm-1). Concomitantly several bands appear around 2900 cm-1 and below 1500 cm-1. They are assigned to the methanol vibrations.29 In the difference spectrum, it is obvious that the 3647 cm-1 band we have above assigned to another type of silanol groups (cf. activation section) is not affected by the hydrogen bond. Consequently these silanol groups are not accessible to methanol. Since they are exchanged by deuterium, we can conclude that the size of methanol molecules prevents them from approaching close enough to these silanols to form hydrogen bonds. In other words, these silanol groups absorbing at 3647 cm-1 are located inside open pores. The broadness of the 3647 cm-1 band compared to the 3744 cm-1 band results from the perturbation of the OH bonds caused by the electrostatic force field inside the pores. After pumping at 573 K, the hydrogen bonds no longer exist (Figure 6c,e). The bands at 2884 and 2956 cm-1 (shoulder at 2977 cm-1) are assigned to ν(CH) stretching vibrations in SiOCH3 surface groups30 with the corresponding deformation frequencies below 1500 cm-1. The methoxylation of the SiOH “outer” groups is exactly similar to that observed on pure silica.30 The surface is irreversibly modified since the methoxyl vibrational modes are still visible after evacuation at 873 K. 4. Oxidation. In order to investigate the modifications undergone by the SiCN surface during oxidation, an activated pellet was subjected to 50 mbar of dry oxygen at 873 K for 2 h. The spectra recorded before (Figure 7a) and after (Figure 7b) oxidation show striking differences (Figure 7c). The first noticeable evolution is the decrease of the average absorbance of the sample which falls from 0.79 to 0.25 at 4000 cm-1 and from 0.45 to 0.24 at 2400 cm-1. This can be explained by the elimination of free amorphous carbon whose presence has been proven by micro-Raman analysis (cf. Bulk

Characterization, Figure 3). Concomitantly CO2 is detected in the gas phase. On the other hand, all ν(CH) bands in the 3000 cm-1 region are eliminated, as indicated by the negative bands in the difference spectrum (Figure 7c) which could be related to the negative band at 1357 cm-1 assigned to the δ(CH) vibration. It is indeed known that CH groups are vulnerable to oxidative attack.31 Moreover, the increasing band at 1745 cm-1 denotes new CdO groups. Another important modification is the strong increase of the ν(OH) sharp band at 3747 cm-1 corresponding to the formation of new SiOH free silanols. This time, their absorption frequency is exactly the same as that of silanols on the silica surface. Thus, we can conclude that the silicon atom in these silanol groups is bonded to three oxygen atoms as in pure silica, proving then the SiCN surface oxidation. Besides, only one type of silanol remains on the surface since the broad band at 3647 cm-1 completely disappears. An explanation can be the particle size increase under oxidation, leading to pore coalescence as also observed during silicon carbide oxidation.32 As a consequence, the silanol groups inside these pores are eliminated. The strong band at 1240 cm-1 revealed by the difference spectrum (Figure 7c) might be assigned to the stretching mode of either new SiO surface bonds33 or new SiOC or COC groups resulting from the oxidation.13 The weak band at 2297 cm-1 is strongly modified after oxidation and transforms into several bands at 2306, 2216, 2135 (sh), and 2068 cm-1. It must be noted that in silicon carbide where no -CtN or -NtC species is possible, the SiH surface groups are still present in the 2000-2300 cm-1 region after an oxidation performed in the same conditions as in the present case.25 New SiH bands may result from oxidation if, on one hand, one considers the formation of SiHx groups (x ) 2 or 3)13 in which hydrogen comes from either the disappearing CHx groups or a possible migration from the bulk, or, on the other hand, if one considers the change in the silicon electronic environment caused by oxidation which leads to shifts of the ν(SiH)25 according to the number of oxygen and nitrogen atoms bonded to silicon. For a better understanding we performed the same oxidation experiment on a deuterated sample. Even though the 2297 cm-1 band partly decreases by heating under deuterium, the oxidation also results in similar strong modifications of the 2000-2300 cm-1 range (spectra not shown).

Nanosized SiCN Powder Synthesized via CVC

J. Phys. Chem., Vol. 100, No. 41, 1996 16651 surface.13 Indeed in this latter compound these species increase or are modified, whereas they decrease on the SiCN surface. This result shows that SiC and SiCN surface reactivities are not identical, even though the surface species are not so different regardless of nitrogen presence. Conclusion

Figure 8. Infrared spectra of a pure SiCN pellet: (a) sample activated at 873 K for 1 h; (b) sample after heating at 873 K under 13 mbar of ammonia; (c) difference spectrum b - a.

Therefore, we must consider that the 2297 cm-1 band is due to triple bonded species transforming into silyl isocyanate (SiNCO), which absorbs at 2310 cm-1, cyanamide, isocyanamide, or possibly silyl cyanate (SiOCN) species.34-36 Because of the large amount of carbon-nitrogen and carbon-nitrogen-oxygen species absorbing in this 2000-2300 cm-1 region,34 the assignments can only be tentative. We note also the complete disappearing of the CHx groups which should imply the concomitant elimination of -CtCbonds. It is worth mentioning that no drastic changes in the ν(NH) absorption band is noted. The slight intensity increase can be due to either the formation of some new NH groups resulting from a rearrangement of the nitrogen distribution on the surface or an increase of the ν(NH) extinction coefficient by modification of the electronic environment. The presence of SiNH and carbon-nitrogen groups on the surface after heating under oxygen is a strong proof that the surface is not completely oxidized. In other words, after this treatment the SiCN grains must not be considered as entirely covered by a silica layer. 5. Nitridation. The treatment of SiCN by heating in NH3 environment at 873 K causes less drastic changes than the oxidation does (Figure 8a,b). The difference spectrum between the spectra recorded before and after ammonia addition is reported on Figure 8c. We note a slight decrease of the 3744 cm-1 band along with an increase of the bands in the 3500 cm-1 region. These two features can be related to the appearing band at 1545 cm-1. Indeed a part of the “outer” silanol groups are nitrided leading to the formation of SiNH2 whose stretching bands are located around 3500 cm-1 while the bending band is at 1545 cm-1.13 Moreover CdO groups are either desorbed or reacted as indicated by the negative band at 1720 cm-1.13 The band appearing at 3045 cm-1 may correspond to new Csp2H groups. This formation may be explained by a concomitant reduction of carbonyl-like species and their reaction with ammonia acting as a hydrogenating agent.37 The band at 1276 cm-1 would be tentatively assigned to a rearrangement of the oxidized species after nitridation.13 Besides it must be noted that the 2297 cm-1 band completely disappears after reaction with ammonia. The behavior of SiOH surface groups under ammonia is different from that of the same species on the SiC

The FT-IR surface analysis of the SiCN powder brought evidence of several surface species, namely, SiOH, SiNH, CHx, CtC, CtN, and CdO. Two types of SiOH silanol groups were identified. One type is located on the outer surface and is like the silanol groups on the silica surface. The other silanol type is on the inner surface, that is, on the surface of open pores, and cannot form a hydrogen bond, even though the pore size does not prevent deuterium exchange. The SiNH groups are quite inert and are perturbed neither by methanol nor under oxidation. However, oxidation causes strong modifications, yielding to new SiOH and isocyanate groups and pore coalescence as well. Moreover, the surface is not entirely covered by a silica layer as proven by the SiNH persistency on the surface. The evolution of the surface under ammonia during heat treatment shows a decrease of the oxygen-containing species (CdO and SiOH) and corresponds to a nitridation of the surface. References and Notes (1) Baraton, M.-I. J. High Temp. Chem. Process. 1994, 3, 545-554. (2) Rahaman, M. N.; Boiteux, Y.; de Jonghe, L. C. Am. Ceram. Soc. Bull. 1986, 65 (8), 1171-1176. (3) Boutonnet-Kizling, H.; Gallas, J. P.; Binet, C.; Lavalley, J. C. Mater. Chem. Phys. 1992, 30, 273-277. (4) Itoh, H.; Utamapanya, S.; Stark, J. V.; Klabunde, K. J.; Schlup, J. R. Chem. Mater. 1993, 5, 71-77. (5) Chang, W.; Skandan, G.; Danforth, S. C.; Kear, B. H.; Hahn, H. NanoStruct. Mater. 1994, 4 (5), 507-520. (6) Chang, W.; Skandan, G.; Hahn, H.; Danforth, S. C.; Kear, B. H. NanoStruct. Mater. 1994, 4 (3), 345-. (7) Chang, W.; Skandan, G.; Danforth, S. C.; Rose, M.; Balogh, A. G.; Hahn, H.; Kear, B. H. NanoStruct. Mater. 1995, 6 (1-4), 321-324. (8) Cauchetier, M.; Croix, O.; Luce, M.; Baraton, M.-I.; Merle, T.; Quintard, P. J. Eur. Ceram. Soc. 1991, 8, 215-219. (9) Nakamizo, M.; Kammereck, R.; Walker, P. L. Carbon 1974, 12, 259-267. (10) Bonnot, A. M. Phys. ReV. 1990, 41 (9), 6040-49. (11) Hair, M. L. Infrared Spectroscopy in Surface Chemistry; Dekker: New York, 1967. (12) Busca, G.; Lorenzelli, V.; Porcile, G.; Baraton, M.-I.; Quintard, P.; Marchand, R. Mater. Chem. Phys. 1986, 14, 123-140. (13) Ramis, G.; Quintard, P.; Cauchetier, M.; Busca, G.; Lorenzelli, V. J. Am. Ceram. Soc. 1989, 72, 1692-1697. (14) Ramis, G.; Busca, G.; Lorenzelli, V.; Baraton, M.-I.; Merle, T.; Quintard, P. In Surfaces and Interfaces Ceramic Materials; Dufour, L. C., et al., Ed.; Kluwer: Dordrecht, The Netherlands, 1989; pp 173-184. (15) Morrow, B. A.; Cody, I. A. J. Phys. Chem. 1973, 77, 1465-1469. (16) Morrow, B. A. In Spectroscopic Characterization of Heterogeneous Catalysts, Part A; Fierro, J. L. G., Ed.; Elsevier: Amsterdam, 1990; pp A161-A224. (17) Boher, P.; Renaud, H.; Van Ijzendoorn, L. J.; Hily, Y. Appl. Phys. Lett. 1989, 54 (6), 511-513. (18) Merle, T.; Baraton, M.-I.; Quintard, P.; Lorenzelli, V. J. Chem. Soc., Faraday Trans. 1993, 89 (16), 3111-3115. (19) Borsella, E.; Botti, S.; Fantoni, R.; Alexandrescu, R.; Morjan, I.; Popescu, C.; Dikonimos-Makris, T.; Giorgi, R.; Enzo, S. J. Mater. Res. 1992, 7 (8), 2257-2268. (20) Ru¨bel, H.; Fo¨lsch, J.; Schade, H. Solid State Commun. 1993, 85 (7), 593-596. (21) Brodsky, M. H.; Cardona, M.; Cuomo, J. J. Phys. ReV. B 1977, 16 (8), 3556-3571. (22) Parsons, G. N.; Lucovsky, G. Phys. ReV. B 1990, 41 (3), 16641667. (23) Lucovsky, G. Solid State Commun. 1979, 29, 571-576. (24) Low, M. J. D.; Severdia, A. G. J. Mol. Struct. 1982, 80, 209-212. (25) Baraton, M.-I. NanoStruct. Mater. 1995, 5 (2),179-192. (26) Hayashi, S.; Kawata, S.; Kim, H. H.; Yamamoto, K. Jpn J. Appl. Phys. 1993, 32, 4870-4877.

16652 J. Phys. Chem., Vol. 100, No. 41, 1996 (27) He, L.; Kurata, Y.; Inokuma, T.; Hasegawa, S. Appl. Phys. Lett. 1993, 63 (2), 162-164. (28) Danisheskii, A. M.; Trapeznikova, I. N.; Terukov, E. I.; Tsolov, M. B. Semiconductors 1994, 28 (10), 1001-1005. (29) Herzberg, G. Molecular Spectra and Molecular Structure, Vol. II: Spectra of Polyatomic Molecules; Van Nostrand: Princeton, 1963. (30) Morrow, B. A. J. Chem. Soc., Faraday Trans. 1974, 70, 15271545. (31) Eldridge, J. M.; Moore, J. O.; Olive, G.; Dunton, V. J. Electrochem. Soc. 1990, 137 (7), 2266-2271. (32) Vix-Guterl, C.; Lahaye, J.; Ehrburger, P. Carbon 1993, 31 (4), 629635.

Baraton et al. (33) Borsella, E.; Botti, S.; Fantoni, R.; Alexandrescu, R.; Morjan, I.; Popescu, C.; Dikonimos-Makris, T.; Giorgi, R.; Enzo, S. J. Mater. Res. 1991, 6 (11), 2442-2451. (34) Morrow, B. A.; Cody, I. A. J. Chem. Soc., Faraday Trans. 1 1975, 71, 1021-1032. (35) Eley, D. D.; Kiwanuka, G. M.; Rochester, C. H. J. Chem. Soc., Faraday Trans. 1 1973, 69, 2062-2073. (36) Ruttenberg, F. E.; Low, M. J. D. J. Am. Ceram. Soc. 1973, 56, 241-244. (37) Bartram, M. E.; Michalske, T. A.; Rogers, J. W.; Mayer, T. M. Chem. Mater. 1991, 5, 953-960.

JP961196L