J. Phys. Chem. C 2008, 112, 3585-3590
3585
Effect of Nitrogen on the Stability of Silicon Nanocrystals Produced by Decomposition of Alkyl Silanes Natalia Zaitseva,* Sebastien Hamel, Zu Rong Dai, Cheng Saw, Andrew Williamson, and Giulia Galli Lawrence LiVermore National Laboratory, 7000 East AVenue, LiVermore, California 94551 ReceiVed: September 21, 2007; In Final Form: December 21, 2007
Si nanocrystals 1-10 nm in size highly resistant to oxidation were prepared by thermal (680 °C) or goldinduced (450-600 °C) decomposition of tetramethylsilane and tetraethylsilane using trioctylamine as an initial solvent. Transmission electron microscopy analysis of samples obtained in the presence of gold showed that Si nanocrystals form via solid-phase epitaxial attachment of Si to the gold crystal lattice. The results of computational modeling performed using first principles density functional theory calculations show that the enhanced stability of nanocrystals to oxidation is due to the presence of N or N-containing groups on the surface of nanocrystals.
Introduction The possibility of using Si-based light-emitting devices for optoelectronic applications1 has stimulated intensive studies aimed at the development of new synthesis and characterization methods for the production of nanosized Si particles. Solution techniques, some of which include gas-phase decomposition of silanes,2-4 ultrasonic dispersion of porous silicon,5-7 electrochemical etching of bulk Si,8-10 direct chemical synthesis,11-13 synthesis in inverse micelles,14 and high-temperature-pressure experiments,15-17 are one of the promising synthetic routes to obtain Si nanocrystals (Si NCs) with relatively monodisperse size distributions. Despite the progress made in developing these techniques, the challenge still remains to create well-defined particles needed for studies of photoluminescence (PL). The origin of PL in Si NCs is the subject of continuing debate, and two competing mechanisms are usually invoked to explain PLquantum confinement effects due to the reduced size of crystalline cores and luminescence involving surface states that can be substantially affected by oxidation processes.18,19 A full understanding and separation of these effects requires further development of existing techniques to obtain NCs with a wellcontrolled surface passivation and high oxidation stability. In this paper, we present experimental results showing that the stability of NCs to oxidation can be substantially increased through the use of alkylamines in the initial precursor mixtures. The computational modeling performed using first principles density functional theory (DFT) calculations explains the observed enhanced stability of Si NCs by the presence of N or N-containing groups at the surface. Experimental Procedures Si NCs were obtained in high-temperature reactor syntheses carried out in a 22 mL metal (Hastelloy) vessel. Tetramethylsilane (TMS, Gelest, 99.9%) and tetraethylsilane (TES, Gelest, 99%) were used as precursors prepared as 5 mL of pure TMS, TES, or their mixtures with trioctylamine (TOA, [CH3(CH2)7]3N) in a mol ratio of about 1:1. Additional experiments conducted with hexylamine [CH3(CH2)5NH2, Aldrich) showed * Corresponding author. E-mail:
[email protected]; tel.: (925) 4233537; fax: (925) 422-6594.
that other amines with shorter alkyl chains and lower boiling points also can be used with similar results. When gold was used as a catalyst, a few drops (about 50 µL) of Au nanoparticle solutions (2 and 5 nm, Ted Pella) were vacuum-dried in the reactor prior to the introduction of the precursor. The precursor was introduced into the reactor in an Ar glovebox, sealed hermetically, and heated in a furnace to the reaction temperature. The duration of the reactions varied from 15 min to 1 h in different experiments. After cooling to room temperature, the reactor was opened in air, and the final product was washed several times by ethanol combined with centrifuging and drying under nitrogen. The dry powder did not dissolve in any of the regular solvents (e.g., toluene, hexane, or chloroform) but was easily dispersed in N,N-dimethylformamide [DMF, HCON(CH3)2, Aldrich], which was used as a solvent for sample preparation. Transmission electron microscopy (TEM) images were obtained at UC Berkeley using a Tecnai-12 TEM with accelerating voltages of 100 kV. High-resolution TEM (HRTEM) and EDS (energy dispersive spectroscopy) analysis were performed with a 300 kV Philips CM300FEG. A high-angle annular detector dark field scanning TEM microscope (HAADF STEM) with an accelerating voltage of 200 kV was used for the elemental analysis of individual nanoparticles and clusters. X-ray diffraction (XRD) was acquired using a CPS 120 INEL curved position sensitive detector system utilizing Cu KR radiation. Fourier transform infrared (FTIR) spectroscopy was performed using a Nicolet 850 FTIR spectrometer with pellet samples of Si nanoparticles mixed with KBr. Simulation Methods We performed a series of first principles DFT calculations on 1 nm silicon clusters, terminated with hydrogen and various carbon- or nitrogen-containing groups. The calculations were made within the Perdew-Burke-Ernzerhof20 (PBE) generalized gradient approximation to the exchange-correlation functional. For each cluster, the atomic coordinates were relaxed to the closest local minimum of the potential energy surface, and then the electronic properties were calculated for the relaxed structure. The calculations use norm conserving, Troullier-Martins pseudo-potentials and a plane-wave basis with a 70 Ry cutoff for the clusters involving carbon and nitrogen. The simulation
10.1021/jp0776255 CCC: $40.75 © 2008 American Chemical Society Published on Web 02/19/2008
3586 J. Phys. Chem. C, Vol. 112, No. 10, 2008
Zaitseva et al.
Figure 1. TEM images of Si NCs synthesized in a metal reactor: (a) - 680 °C, TMS with TOA, no catalyst and (b) -600 °C, TES with TOA, 2 nm gold particle catalyst; an electron diffraction pattern (upper inset) corresponds to d-spacings of 3.13 Å (111), 1.93 Å (220), and 1.63 Å (311) of the cubic diamond structure of Si; lower insets show histograms of NC sizes based on measurements of 400 particles from different regions of TEM grids. (c) HAADF STEM image of Si NCs shown in panel a with an EDS pattern of Si with only traces of oxygen in individual NCs (point 1) and groups of nanoparticles (area 1); copper signal is scattered from the grid. (d) XRD pattern of a final solid.
supercell sizes were chosen to place more than 10 Å of vacuum between the periodic replica of the Si NCs, so as to remove any interactions. The atomic relaxations were performed until the residual force on all atoms was lower than 3 × 10-3 eV/Å. The QBOX21 code was used for the structural relaxations. Results and Discussion Formation of NCs. Si NCs obtained by thermal decomposition of TMS in TOA are shown in Figure 1a. According to the previous reports,22-24 the decomposition of both TMS and TES yielded the formation of elemental Si and gas products that essentially formed upon pyrolysis of ethane (C2H4, CH4, H2, and C). In our experiments made in the absence of added catalysts, Si NCs did not form when TES was used as a precursor, even if the syntheses were carried out at 700 °C. Formation of crystalline Si was observed only in the syntheses with the decomposition of TMS at temperatures above 680 °C (Figure 1a). Multiple analyses made using a combination of EDS (Figure 1c), electron diffraction (inset in Figure 1a), and XRD (Figure 1d) confirmed the formation of Si NCs with a cubic diamond structure. The most typical size of the NCs was about 5 nm with the size distribution within a factor of 2. The formation of large, macroscopic-scale Si crystals was not observed in any of the syntheses, even if the reaction time exceeded 1 h. HRTEM analyses (Figure 2) revealed highly crystalline structures with clearly seen planes and individual columns of atoms even in the smallest NCs with sizes close to 1 nm [i.e., nanocrystals with only about 35 Si atoms (estimated from our DFT calculations)]. However, despite the remarkable quality of the individual NCs presented in Figure 2, it should be noted that the high temperature required for decomposition of TMS posed difficulties concerning the control of the reaction conditions. These difficulties were introduced by the large impact of
organic pyrolysis25 and possible reactions with the material of the reactor for temperatures above 600 °C (temperature of the commercial specification of the Hastelloy reactor). To decrease the crystallization temperature, 2 nm Au nanoparticles were used, with the initial expectation that catalytic decomposition of the precursors would result in the formation of Si nanowires according to the vapor-liquid-solid (VLS) mechanism.26,27 With the use of gold nanoparticles, the decomposition of both TMS and TES occurred in the lower temperature region (SI-1). Si NCs could be occasionally detected in the final products of precursor decomposition at temperatures as low as 450 °C, with a noticeably increasing yield at higher temperatures of 550600 °C. However, almost no nanowires were observed, and the NCs had rather isometric shapes (Figure 1b) similar to those shown in noncatalytic syntheses. Additionally, in many cases, Si NCs were found to be far apart from Au nanoparticles not detected by EDS analysis of the corresponding areas. Experiments made with larger Au nanoparticles (5 nm), to increase the resolution of TEM analysis, showed that the goldcatalyzed formation of crystalline Si (c-Si) observed in our experiments did not occur by the VLS mechanism but rather relates to the process known as metal-induced crystallization of amorphous Si (a-Si).28 HRTEM analysis of individual particles showed that, in addition to the free-standing Si and Au NCs (Figure 3a,b), the final products of the syntheses contained Au particles with a-Si shells (Figure 3c,d) or Si NCs attached to the gold surface (Figure 3e,f). A finding that makes it difficult to explain formation of c-Si by the VLS mechanism is that most of the adjacent Au and Si NCs have the same crystallographic orientation resulting from the epitaxial connection between their crystal lattices. In the VLS mechanism,26 the deposition of Si atoms occurs from the liquid interface between Au and Si. Atomic Si produced in the catalytic decomposition of a precursor first dissolves in a liquid gold
Effect of N on Stability of Si Nanocrystals
J. Phys. Chem. C, Vol. 112, No. 10, 2008 3587
Figure 2. HRTEM images of NCs synthesized at 680 °C in a metal reactor in the presence of TOA. Measured distances between the planes correspond to (111) and (220) d-spacings of cubic diamond Si.
particle and then deposits on a Si substrate that determines the crystallographic structure of the resulting crystal or nanowire. Although the classical VLS mechanism requires the presence of a liquid gold particle, it has been shown that it may also proceed at temperatures much lower than the Au melting point, due to the formation of thin liquid eutectic layers on the surface of solid gold nanoparticles.29,30 However, since the growing Si crystal is separated from the gold particle by a liquid eutectic layer, the crystal lattice of Si should form independently of the crystallographic structure of the catalyst. The situation is different in the NCs obtained in our experiments. HRTEM images in Figure 3 present clear evidence of the epitaxial connection between Au and Si crystal lattices, in which the (111) planes of c-Si directly continue the same simple-index planes of Au, without the formation of intermediate layers of Au-Si alloys. The structure of a typical boundary between two single crystals periodically contains three (111) Si planes formed on four corresponding Au planes. The extra planes in the gold crystal lattice (shown by arrows in Figure 3g) represent classical examples of misfit dislocations formed to compensate the difference in the inter-planar distances (25%) between two crystal lattices formed as a result of the epitaxial growth of silicon on gold. The epitaxial connection between two crystallographic lattices is possible only if during the crystallization process initial Si atoms attach directly to the Au crystal lattice and if further growth of Si NCs proceeds not by deposition from the liquid interface between Au and Si, as it occurs in the VLS process,
Figure 3. HRTEM images of Si formed by decomposition of TES catalyzed by 5 nm Au nanoparticles in a TOA solution at 450 °C: (a and b) free-standing Si NC; (c) Au particle with an amorphous Si shell; (d) Au particle with an amorphous Si shell partially crystallized on the Au crystal lattice; large size of Au particle might result from the hightemperature coagulation of smaller nanopaticles; (e and f) joined Au and Si NCs with the same orientation of (111) planes; and (g) an epitaxial boundary between Au and Si crystallographic lattices showing the formation of {111} Si planes on the {111} Au planes; the difference between d-spacings is compensated for by the formation of misfit dislocations (shown by arrows as extra planes in the gold lattice); corresponding area is squared in the lower magnification image of a joint Au/Si NC (inset). Scale bars are 5 nm in panels a-f.
but by the attachment of atoms to the Si layers on the side opposite of the gold nanoparticle. Si atoms needed for such an attachment may come from the amorphous layer around gold nanoparticles (Figure 3c,d), similar to the process of metalinduced crystallization of thin amorphous Si layers.31 It is known that when small metal particles (Au, Ni, and Fe) are present, crystallization of Si from the amorphous phase, which in the
3588 J. Phys. Chem. C, Vol. 112, No. 10, 2008 absence of metal catalysts takes place in the temperature range of 600-700 °C, occurs at lower temperatures of 130600 °C.28,32 One explanation for this phenomenon has been attributed to an interaction of the free electrons of the metal with the covalent Si bonds near the growing interface.32,33 Since it is known that Au has a high diffusion rate in Si,34 such an interaction, enhanced by the small size of Au particles and increasing temperature, may explain the formation of isolated Si NCs in our syntheses at elevated temperatures above 550 °C. Epitaxial crystallization of a-Si on crystal lattices of the metal or silicon-metal alloys was observed previously for the Ni-Si system.33 The images of Figure 3 show that such growth can also be initiated by the presence of solid gold particles. A new observation made in our experiments is that the epitaxial growth of Si can proceed directly on the gold crystal lattice, in the absence of silicon-wafer substrates used in the previous studies to initiate the Si crystallization process.33,35 The limited availability of a-Si located only within close vicinity of gold particles can explain as to why macroscopic-sized crystals did not form, and the absence of the Si substrate, which is a necessary part of the VLS growth, can be a reason as to why crystalline Si nanowires did not form in our experiments. PL. Final products obtained with all syntheses described previously produced visible PL upon excitation by UV light (SI-2). Since blank experiments made with TOA and some linear chain alkanes (pentane and hexane) under the same conditions did not result in luminescent products, this PL should be related to the presence of Si or silicon-containing compounds. There are, however, certain issues that do not allow us to attribute the PL only to quantum effects in Si NCs. The main problem relates to the difficulties in the separation of NCs from the amorphous fraction simultaneously formed during syntheses. Purification procedures typically produced final samples in amounts of 500600 mg. However, XRD patterns of these samples (Figure 1d) always showed the presence of a large amorphous fraction (up to 90%, estimated from TEM analysis), which according to EDS analysis, contained Si as its main constituent. The fact that PL could be observed in macroscopic measurements indicated that the amorphous material introduced a major contribution to the total, observed luminescence. Since small amorphous Si clusters (1-5 nm) are known to show luminescence,36-39 the broad spectra that were obtained could result from combined PL produced by Si NCs and amorphous clusters present in large numbers in the final products of the reactions. Stability of NCs. The formation of Si NCs in our experiments was detected in syntheses conducted both with and without amines. However, when the decomposition of pure TMS or TES occurred in the absence of initial solvents, the final products exposed to air after opening the reactor had much larger fractions of SiOx-containing amorphous particles. Multiple TEM analyses of the same grids exposed to air showed that Si NCs found in the mixtures with the oxidized products could remain stable for a few weeks, after which they also were transformed into amorphous structures with an increasing concentration of oxygen. The situation was noticeably different when TOA or hexylamine was used as the solvent in the initial precursor mixtures. HRTEM images in Figure 2, for example, were taken from the TEM grids exposed to air for 5 months after they were first detected by low-resolution TEM. These images show that even very small NCs with sizes close to 1 nm preserved their perfect structure without oxidation. The stability of the NCs increased with the increasing temperature of the synthesis. Numerous analyses made by HAADF STEM imaging showed that Si NCs synthesized in the presence of amines in the
Zaitseva et al.
Figure 4. FTIR spectra of Si NCs synthesized by decomposition of pure TMS (1) and TMS in TOA solution (2).
temperature range of 550-680 °C could freely resist oxidation during time periods of over a year (SI-3). The enhanced stability against oxidation of the Si NCs points at the presence of a passivating layer on the surface of the clusters that hampers the reaction of Si with oxygen. Unfortunately, since the K-edge of the nitrogen X-ray absorption is in the absorption range of the EDS detector, it is not easy to perform direct analysis of nitrogen-containing surface layers using TEM. FTIR analysis, which can be more informative in this case, showed that the composition of these layers can be different depending on the synthesis conditions. Typical peaks in the range of 2890-2960 cm-1 characteristic for C-H stretching modes of CH2 and CH3 groups correspond to the presence of organic species in all NCs (Figure 4). Measurements made with the samples synthesized in the absence of amines (spectrum 1 in Figure 4) reveal a strong band at 1070-1080 cm-1 typical of a Si-O stretching-bending mode,40 thus indicating that these organic species may be not efficient in preventing NC oxidation. This band disappeared in the samples produced in the presence of amines (spectrum 2 in Figure 4), being replaced by a number of different peaks, among which the strongest (830 cm-1) could be attributed to the Si-N stretching mode typically observed41,42 in the range of 830880 cm-1. According to the blank experiments made with TOA (SI-1), above 550 °C, amines go through the transformations, resulting in the breaking of C-C and C-N bonds with the formation of lower amines, nitrogen, or ammonia, which also was detected by its characteristic smell upon opening the reactors. All these nitrogen-containing compounds can readily attach to the Si surface.43 Our DFT electronic structure calculations (see next section) show that when present, N is more likely to be on the surface than in the core of a Si NC, and if synthesis conditions allow for N to be on the surface, multiple N coverage is likely to occur, resulting in a key role in passivating the Si NC surface and consequently in preventing the NC oxidation. Computational Modeling of Silicon Nanocrystals To identify relevant Si NC structures, we first addressed the solubility of nitrogen in crystalline silicon at the nanoscale. Nitrogen is known to have a very low solubility in bulk Si.44 Previous calculations45,46 have shown that a single N interstitial shows a fast diffusion through crystalline Si, until it is trapped by other defects such as a second N interstitial. This di-interstitial nitrogen pair was identified as the most common defect in bulk
Effect of N on Stability of Si Nanocrystals
J. Phys. Chem. C, Vol. 112, No. 10, 2008 3589
Figure 5. Comparison of structures of Si NCs containing a diinterstitial nitrogen defect (a) in the core and (b) on the surface.
silicon.47 In the case of NCs, the high mobility of a single N interstitial would quickly move it to (or keep it on) the surface of a growing Si NC. One could envision a pair of N interstitials becoming trapped in the core of the cluster (Figure 5a); however, our calculations show that at the surface (Figure 5b), diinterstitials are energetically favored by 0.7 eV. This is due to minimization of the stress on the crystalline core when the diinterstitial pair of N atoms is at the surface. Considering the low solubility of defects in Si NCs, we concentrated our study on surface N groups. A representative set of the N-doped Si NCs calculated in this study is shown in Figures 5-7. Our reference H-terminated 1 nm Si NC consists of a diamond crystalline core of 35 Si atoms, where dangling bonds are capped with 36 H atoms. For all the nitrogen-doped clusters, nitrogen is three-fold coordinated.48 We focused on two types of surface N groups (a NH2 substituting a H atom and a bridged NH substituting two H atoms), and we considered the effect of multiple N groups on the surface of Si NCs. In the case of oxygen dopants, Puzder et al. have shown18 that the energy differences associated with the addition of a second oxygen group were within 0.1 eV of that for the addition of the first oxygen and that this remained true until the cluster was completely covered with either double bonded or bridged oxygen atoms. Therefore, from a thermodynamic standpoint, if the chemical potentials of hydrogen and oxygen are such that it is favorable to replace surface hydrogen atoms with oxygen atoms, the oxygenation reaction is likely to continue until the NC is completely covered with oxygen. We follow the same approach as that of ref 18 in the case of N groups. We first calculated the difference in the total energy of the reference clusters Si35H36 and Si35H35NH2, where the amine group replaced a hydrogen of the original cluster (Figure 6a,b). We then calculated the difference in total energy between a cluster with one and two amine groups, with the amine groups either alongside or on opposite sides of the cluster (Figure 6cf). This provides an estimate of the steric repulsion between the amine groups, which turns out to be lower than 0.1 eV. This is comparable to the energy difference associated with the addition of a second amine group, and this remained so up to six amine groups (SI-4 and SI-5). This means that it is just as favorable to add the sixth amine group as it is to add the first. Calculations with the bridged N configuration showed similar results. Therefore, from purely thermodynamic considerations, if the chemical potentials of nitrogen and hydrogen are such that it is favorable to replace surface hydrogen atoms with a nitrogen-containing group, we predict that multiple coverage with nitrogen groups is likely to occur. This conclusion holds for small nitrogen groups (NH and NH2). For larger surface groups, steric interactions between the groups may play a more important role and affect the results found here for small groups. The precursor molecules (TMS and TES) used during the colloidal growth of Si NCs contain numerous carbon atoms.
Figure 6. Silicon atoms are red, nitrogen atoms are blue, and hydrogen atoms are light gray. (a, c, and e) Amine groups are replacing H atoms from SiH sites (face of the Si NCs). (b, d, and f) Amine groups are replacing H atoms from SiH2 sites (summit of the Si NCs). In panels g and h, a NH group is replacing two H atoms from SiH sites (faceface) and one H atom from a SiH2 site and one from a SiH site (summit-face).
Figure 7. Silicon atoms are red, nitrogen atoms are blue, and hydrogen atoms are light gray. Panel a has six amine groups on SiH2 sites. Panel b has six NH groups bonded to a SiH2 site and a SiH site.
Therefore, in order for the NCs to grow, some silicon-carbon bonds must be broken. While the amount of carbon greatly outnumbers the amount of nitrogen (coming from the TOA), if kinetic effects allow for Si-C bonds to be substituted by Si-N bonds during growth, the latter will be robust to further
3590 J. Phys. Chem. C, Vol. 112, No. 10, 2008 substitutions, as Si-N bonds are expected to be stronger than Si-C bonds at the surface of a small Si NC.51 We estimate from DFT calculations on small molecules (SiH3NH2 and SiH3CH3) that the binding energy is ∼0.6 eV larger for Si-N than for Si-C, reinforcing the hypothesis that our colloidal Si NCs have some nitrogen groups on their surfaces. This surface passivation by nitrogen may help to explain the stability of the Si NCs to oxidation when they are exposed to air. Conclusion Si NCs with sizes of 1-10 nm were synthesized in metal reactors by thermal (680 °C) and Au-induced (450-600 °C) decomposition of tetramethylsilane and tetraethylsilane dissolved in alkylamines (trioctylamine and hexylamine). HRTEM analysis of the final samples showed that the formation of Si NCs proceeds by metal-induced crystallization of amorphous Si and by solid-phase epitaxial attachment of Si atoms to the crystal lattice of gold. Computational first-principles modeling addressing the structural stability of Si NCs with different surface groups indicated that both C-- and N--containing functional groups are strongly bound to the surface of colloidal Si NCs. The calculated stronger binding energy of N-containing groups correlates with the experimental observation that introducing N-containing solvents into the synthesis process increases the stability of Si NCs to oxidation. Acknowledgment. This work was performed under the auspices of the U.S. Department of Energy at the University of California/Lawrence Livermore National Laboratory under Contract W-7405-Eng-48. Supporting Information Available: (SI-1) P-T diagram of precursor decomposition; (SI-2) typical photoluminescence spectra of final powders; (SI-3) TEM and HAADF STEM images of Si nanocrystals exposed to air; and (SI-4 and SI-5) multiple coverage by amine and bridged groups. This information is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Ossicini, S.; Pavesi, L.; Priolo, F. Light Emitting Silicon for Microphotonics; Springer: Berlin, 2003. (2) Littau, K. A.; Szajowshki, P. J.; Muller, A. J.; Kortan, A. R.; Brus, L. E.; J. Phys. Chem. 1993, 97, 1224-1230. (3) Li, X.; He, Y.; Talukdar, S. S.; Swihart, M. T. Langmuir 2003, 19, 8490-8496. (4) Fojtik, A.; Henglein, A. Chem. Phys. Lett. 1994, 221, 363-367. (5) Heinrich, J. L.; Curtis, C. L.; Credo, G. M.; Kavanagh, K. L.; Sailor, M. J. Science (Washington, DC, U.S.) 1992, 255, 66-68. (6) Swerida-Krawiec, B; Cassagneau, T; Fendler, J. H. J. Phys. Chem. B 1999, 103, 9524-9529. (7) Bley, R. A.; Kauzlarich, S. M.; Davis, J. E. Lee, H. W. H. Chem. Mater. 1996, 8, 1881-1888. (8) Nayfeh, M.; Barry, N.; Therrien, J.; Akcakir, O.; Gratton, E.; Belomoin, G. Appl. Phys. Lett. 2001, 78, 1131-1133. (9) Belomoin, G.; Therrien, J.; Nayfeh, M. Appl. Phys. Lett. 2000, 77, 779-781. (10) Lie, L. H.; Duerdin, M.; Tuite, E. M.; Houlton, A.; Horrocks, B. R. J. Electroanal. Chem. 2002, 538-539, 183-190. (11) Baldwin, R. K.; Pettigrew, K. A.; Garno, J. C.; Power, P. P.; Liu, G. Y.; Kauzlarich, S. M. J. Am. Chem. Soc. 2002, 124, 1150-1151. (12) Pettigrew, K.; Liu, Q.; Power, P. P.; Kauzlarich, S. M. Chem. Mater. 2003, 15, 4005-4011. (13) Zhang, X.; Neiner, D.; Wang, S.; Louie, A. Y.; Kauzlarich, S. M. Nanotechnology 2007, 18, 95601. (14) Wilcoxon, J.; Samara, G.; Provencio, P. Phys. ReV. B: Condens. Matter Mater. Phys. 1999, 60, 2704-2714. (15) Heath, J. R. Science (Washington, DC, U.S.) 1992, 258, 11311133. (16) Holmes, J. D.; Ziegler, K. J.; Doty, R. C.; Pell, L. E.; Johnston, K. P.; Korgel, B. J. Am. Chem. Soc. 2001, 123, 3743-3748.
Zaitseva et al. (17) English, D. E.; Pell, L. E.; Yu, Z.; Barbara, P. F.; Korgel, B. A. Nano Lett. 2002, 2, 681-685. (18) Puzder, A.; Williamson, A.; Grossman, J.; Galli, G. J. Chem. Phys. 2002, 117, 6721. (19) Zhou, Z.; Brus, L.; Friesner, R. A. Nano Lett. 2003, 3, 163-167. (20) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. ReV. Lett. 1996, 77, 3865-3868. (21) Gygi, F. Lawrence Livermore National Laboratory: Livermore, CA. (22) Helm, D. F.; Mack, E. J. Am. Chem. Soc. 1937, 59, 60-62. (23) Waring, C. E. Trans. Faraday Soc. 1940, 36, 1142-1153. (24) Voronkov, M. G.; Dolgov, B. N.; Karpenko, G. B. SoV. J. Gen. Chem. 1954, 24, 273-275. (25) Pell, L.; Schricker, A.; Miculec, F.; Korgel, B. Langmuir 2004, 20, 6546-6548. (26) Wagner, R. S.; Ellis, W. C. Appl. Phys. Lett. 1964, 4, 89-90. (27) Holmes, J. D; Johnston, K. P.; Doty, R. C.; Korgel B. A. Science (Washington, DC, U.S.) 2000, 287, 1471-1473. (28) Hultman, L.; Robertsson, A.; Hentzell, H. T. G.; Engstrom, I.; Psaras, P. A. J. Appl. Phys. 1987, 62, 3647-3655. (29) Bokhonov, B; Korchagin, M. J. Alloys Compd. 2000, 312, 238250. (30) Wu, Y.; Yang, P. J. Am. Chem. Soc. 2001, 123, 3165-3166. (31) Hultman, L.; Robertsson, A.; Hentzell, H. T. G.; Engstro¨m, I.; Psaras, P. A. J. Appl. Phys. 1987, 62, 3647-3655. (32) Spaepen, F.; Nygren, E.; Wagner, A. V. In Crucial Issues in Semiconductor Materials and Processing Technologies; NATO AS1 Series E: Applied Sciences, Vol. 222; Coffa, S., Priolo, F., Rimini, E., Poate, J. M., Eds.; Kluwer: Dordrecht, The Netherlands, 1992; p 483. (33) Hayzelden, C.; Batstone, J. L. J. Appl. Phys. 1993, 73, 8279-8289. (34) Wakayama, Y.; Tanaka, S. J. Cryst. Growth 1997, 181, 304-307. (35) Wakayama, Y.; Fujinuma, H.; Tanaka, S. J. Mater. Res. 1998, 13, 1492-1496. (36) Dal Negro, L.; Yi, J. H.; Kimerling, L. C.; Hamel, S.; Williamson, A.; Galli, G. Appl. Phys. Lett. 2006, 88, 183103. (37) Allan, G.; Delerue, C.; Lannoo, M. Phys. ReV. Lett. 1997, 78, 31613164. (38) Rinnert, H.; Vergnat, M.; Burneau, A. J. Appl. Phys. 2001, 89, 237-243. (39) Franzo, G.; Boninelli, S.; Pacifici, D.; Priolo, F.; Iacona, F.; Bongiorno, C. Appl. Phys. Lett. 2003, 82, 3871-3873. (40) Yang, B; Xiong, Q; Lei, J. R.; Liu, Y. G.; Wang, D. Z.; Huang, N. K. J. Mater. Sci. Lett. 2002, 21, 1073-1075. (41) Yin, Z.; Smith, F. W. Phys. ReV. B: Condens. Matter Mater. Phys. 1990, 42, 3666-3675. (42) Cheng, S.; Ren, Z. X.; Liang, R.; Lu, Q.; Liu, W.; Ning, Z. Plasma Sci. Technol. 2000, 2, 213-218. (43) Chen, M.; Zheng, A.; Hao, L.; Zhou, M. J. Phys. Chem. A 2002, 106, 3077-3083. (44) Jennings, H. M. J. Mater. Sci. 1983, 18, 951-967. (45) Schultz, P. A.; Nelson, J. S. Appl. Phys. Lett. 2001, 78, 736-738. (46) Goss, J. P.; Hahn, I.; Jones, R.; Briddon, P. R.; O ¨ berg, S. Phys. ReV. B: Condens. Matter Mater. Phys. 2003, 67, 45206. (47) Jones, R.; O ¨ berg, S.; Rasmussen, R. B.; Nielsen, B. B. Phys. ReV. Lett. 1994, 72, 1882-1885. (48) This is consistent with known molecules containing N, Si, and H atoms such as disilazane, trisilylamine, and studies of the dissociative adsorption of ammonia on silicon surfaces.49,50 (49) Lee, S.-H.; Kang, M.-H. Phys. ReV. B: Condens. Matter Mater. Phys. 1998, 58, 4903-4908. (50) Bengio, S.; Ascolani, H.; Franco, N.; Avila, J.; Asensio, M. C.; Bradshaw, A. M.; Woodruff, D. P. Phys. ReV. B: Condens. Matter Mater. Phys. 2004, 69, 125340. (51) Consider the following exothermic reactions: Si + C (graphite) f SiC, 3SiC + 2N2 f Si3N4 + C (graphite), and Si3N4 + 3O2 f 3SiO2 + 2N2. Taken together, these imply that as compared to the various elements in their standard states, silicon oxide is more stable than silicon nitride, which is more stable than silicon carbide, which is more stable than crystalline silicon. Consider also the enthalpies of formation of H3SiNH2, H3SiCH3, and other small molecules (found in the HiTempThermo online database of results of high-level ab initio-based computed properties: public.ca.sandia.gov/HiTempThermo/). At 300 K, the ∆H values (in kcal/ mol) are as follows: CH4 (-17.9), H3SiCH3 (-7.3), H2 (0.0), H2O (-57.8), NH3 (-11.0), H3SiOH (-65.8), SiH4 (8.2), and SiH3NH2 (-11.4). Hence, we can compare CH3SiH3 + NH3 f SiH3NH2 + CH4, (-18.3 f -29.3) and NH2SiH3 + H2O f SiH3OH + NH3, (-69.2 f -76.8), and in molecular species as well as in crystalline species, silicon-oxygen bonds are stronger (more stable) than silicon-nitrogen bonds, which are stronger than silicon-carbon bonds, which are the same as or slightly stronger than silicon-hydrogen bonds. Under the conditions at which the clusters were made, N replaced some or all of the carbons. At room temperature and pressure, the barrier to replace Si-N bonds by the more stable Si-O bonds is too large (-0.4 kcal/mol or 201 K), and so oxidation is effectively prevented.
