8462
J. Phys. Chem. B 2005, 109, 8462-8468
Complex Thermal Chemistry of Vinyltrimethylsilane on Si(100)-2×1† Laurent Pirolli and Andrew V. Teplyakov* Department of Chemistry and Biochemistry, UniVersity of Delaware, Newark, Delaware 19716 ReceiVed: July 20, 2004; In Final Form: October 6, 2004
The surface chemistry of vinyltrimethylsilane (VTMS) on Si(100)-2×1 has been investigated using multiple internal reflection-Fourier transform infrared spectroscopy, Auger electron spectroscopy, and thermal desorption mass spectrometry. Molecular adsorption of VTMS at submonolayer coverages is dominating at cryogenic temperatures (100 K). Upon adsorption at room temperature, chemical reaction involving rehybridization of the double bond in VTMS occurs. Further annealing induces several reactions: molecular desorption from a monolayer by 400 K, formation and desorption of propylene by 500 K, decomposition leading to the release of silicon-containing products around 800 K, and, finally, surface decomposition leading to the production of silicon carbide and the release of hydrogen as H2 at 800 K. This chemistry is markedly different from the previously reported behavior of VTMS on Si(111)-7×7 surfaces resulting in 100% conversion to silicon carbide. Thus, some information about the surface intermediates of the VTMS reaction with silicon surfaces can be deduced.
1. Introduction The surface chemistry of vinyltrimethylsilane (VTMS) on a Si(100)-2×1 surface has a wide variety of applications. For example, silicon carbide (SiC), which can be formed by silicon modification with VTMS, is a promising material for hightemperature and high-power electronic devices because it is a wide-band-gap semiconductor characterized by a very high chemical and thermal stability. In this type of application, it is important to utilize VTMS as efficiently as possible to create carbide material. In a different type of process, VTMS is used as a ligand in one of the most used copper deposition precursor molecules: hexafluoroacetylacetonate-copper-vinyltrimethylsilane [(hfac)Cu(VTMS)].1,2 In such applications as chemical vapor deposition (CVD) processing, it is imperative to keep carbon contamination down. Thus, ultimate understanding and control of VTMS chemistry on silicon substrates is crucial in both cases. VTMS is a very attractive ligand for the chemical vapor deposition of copper, particularly in combination with hexafluoroacetylacetonate (hfac). The combination of the volatility of the Cu-CVD precursors such as (hfac)CuI(VTMS) with the volatility of the side product, CuI(hfac)2,1 upon Cu0 deposition, creates a useful set of physical properties required for an efficient production of contaminant-free copper films.3,4 Being at the core of the most widely used deposition process, VTMS ligand itself was obviously investigated previously. However, the chemistry of this compound on clean silicon substrates has only been investigated in detail on Si(111)-7×7 surfaces5 and not on the most common ultra-large-scale integrated device support: Si(100). While some of the chemistry may appear to be similar on both surfaces, the presence of different types of surface reactive sites and the different symmetries of the surfaces may lead to preferential surface reactions that could be different on these two substrates. †
Part of the special issue “George W. Flynn Festschrift”. * Author to whom correspondence should be sent. Tel.: (302) 831-1969. Fax: (302) 831-6335. E-mail:
[email protected].
It is because of the role of Si(100)-2×1 in microelectronics applications that the chemistry of various organic compounds on this surface has been scrutinized for the last 2 decades. In general, the unsaturated organic compounds received significant attention (see, for example, refs 6-12 and references therein). Most importantly, a simple [2 + 2] addition was reported for all of the organic compounds with one double bond studied on this surface. The simplest unsaturated hydrocarbon, ethylene,13-16 as well as more complicated unsaturated hydrocarbons such as cyclopentene11,17,18 and vinyl bromide19 all seem to form a di-σ surface complex upon addition to the Si-Si dimers of the Si(100)-2×1 surface. Clemen et al.13 suggested that chemisorption of ethylene occurs via a mobile precursor mechanism with an activation-energy difference between desorption and chemisorption from the precursor of 12.1 kJ/mol. The saturation coverage of ethylene was found to be one molecule of C2H4 per Si2 dimer. It was also observed that the chemisorbed ethylene desorbs unimolecularly from Si(100) around 550 K with a desorption activation energy of 159 kJ/mol and, for the di-σ surface C2H4-Si2 complex, each Si-C bond has a strength of around 305.4 kJ/mol.13 The low desorption activation energy allows ethylene to leave the surface prior to any significant dissociation (approximately 2% of the monolayer undergoes dissociation), preventing the formation of significant coverages of surface carbon and hydrogen. Unlike ethylene chemisorption, the reaction of vinyl bromide on Si(100)-2×1 does lead to further decomposition, rather than a simple desorption, upon thermal annealing.19 Apparently, substitution of one hydrogen atom for bromine increases the desorption energy significantly so that the surface decomposition reaction can compete efficiently with the desorption process. Interestingly enough, Nagao et al.19 found that the chemisorption of vinyl bromide molecules on a Si(100)-2×1 surface occurs via a three-atom intermediate state (observed between 58 and 90 K). In other words, before the chemisorbed di-σ-bonded adduct is formed, the initial step is the formation of a complex between the CdC double bond and only one of the surface silicon atoms. Such a mechanism can proceed similarly on
10.1021/jp0467853 CCC: $30.25 © 2005 American Chemical Society Published on Web 01/29/2005
Thermal Chemistry of Vinyltrimethylsilane Si(100)-2×1 and Si(111)-7×7 surfaces. In fact, the three-atom intermediate has been considered by Andersohn et al. for VTMS on Si(111)-7×7.5 It was demonstrated that VTMS adsorption onto the Si(111)-7×7 surface is site-specific.5 VTMS interacts primarily with the center adatoms (most-depleted of negative charge), and the resulting adsorbed molecules are mostly immobile. At room temperature, the adsorption is nondissociative with an overall initial sticking coefficient in excess of 0.03. Upon annealing, the molecule dissociates completely, yielding on the surface Si and C, which form SiC clusters. As will be shown below, the chemistry of the trimethylsilyl fragment of VTMS is also important in understanding the surface decomposition of VTMS on silicon. Craig et al. have extensively studied trimethylsilane on Si(100).20-22 Physisorbed trimethylsilane desorbs molecularly at 155 K at a monolayer coverage. At higher temperatures, the trimethylsilane-covered Si(100) shows hydrogen desorption from silicon monohydride states at 780 K, whereas the formation of the silicon dihydride species was suppressed by the presence of surface carbon supplied by trimethylsilane. Thus, the decomposition of this fragment indeed leads to the formation of surface silicon carbide. This paper presents and analyzes the differences in the surface chemistry of VTMS on Si(100)-2×1 as compared to Si(111)7×7. As in previous studies of VTMS on Si(111)-7×7, this molecule is expected to utilize its CdC double bond to interact with Si-Si dimers on the Si(100)-2×1 surface. The mechanism of this addition reaction is expected to be similar to the ethylene chemisorption on the same surface. If a [2 + 2] addition is considered to be a major mechanism on Si(111)-7×7, although the exact nature of this addition process is suggested to be sitespecific, it is also expected to play a major role on a Si(100)2×1 surface. The studies below confirm this assessment. The thermal chemistry of VTMS on Si(100)-2×1 is much more complex than that on Si(111)-7×7. It was proposed that the only surface reaction pathway for VTMS on Si(111)-7×7 is surface decomposition leading to the nanostructured silicon carbide formation and the release of hydrogen. Si(100)-2×1 exhibits far more complicated behavior. While the topographic analysis of the resulting surface is a subject of a separate study, this paper presents the analysis of a complex surface thermal chemistry of VTMS on Si(100)-2×1. 2. Methods 2.1. Experimental Methods. The two ultrahigh-vacuum (UHV) chambers used in the studies described here are located at the University of Delaware. Both chambers have a base pressure of around 5 × 10-10 Torr. They are both equipped for Auger electron spectroscopy (AES), low-energy electron diffraction (LEED), and surface cleaning using an ion gun. One chamber is coupled to an infrared spectrometer (Nicolet, Magna 560) set up in a multiple internal reflection (MIR) mode utilizing a liquid-nitrogen-cooled external mercury-cadmium-telluride (MCT) detector. A total of 2048 scans were collected in the infrared measurements with a resolution of 4 cm-1. First, the background spectrum was collected; then the appropriate dose of the compound of interest was introduced into the ultrahigh vacuum (UHV) chamber using a leak valve, and a brief annealing to the desired temperature was performed if needed. The spectrum was then collected at the same temperature as that of the background. The unshielded mass spectrometer (SRS 200) was used to determine the cleanliness of the dosing compounds in situ. The other UHV chamber was largely used for temperature-programmed desorption (TPD) studies. It is equipped with a shielded, differentially pumped mass spec-
J. Phys. Chem. B, Vol. 109, No. 17, 2005 8463 trometer (Hiden Analytical) pumped by a dedicated ion pump. During the TPD studies, the crystal was positioned ∼2 mm from a 2 mm aperture in the shield. A 25 × 20 × 1 trapezoidal sample of Si(100)-2×1 with 45° beveled edges (Harrick Scientific) was used for the MIRinfrared spectroscopy studies. This sample, polished on both sides, was mounted on a manipulator capable of cooling the sample to 90 K with liquid nitrogen and heating it to above 1150 K using an e-beam heater (McAllister Technical Services). Another 1 × 1 cm2 silicon sample was cut from a Si(100) wafer (Semiconductor International) and mounted in a different chamber for temperature-programmed reaction/desorption (TPR/ D) studies. This sample was polished on one Si(100) surface. The heating rate in the TPR/D studies was 2 K/s. The temperature range in this chamber was 130-1150 K. VTMS (Aldrich, 99%) was prepurified by at least 10 freezepump-thaw cycles before introduction into the chamber. The purity of the compound was verified in situ by mass spectrometry, and the mass spectra obtained were compared with the available database mass spectrum of VTMS.23 Argon (Matheson, 99.999%) for surface cleaning was used without additional purification. The silicon crystals were prepared by sputtering with Ar+ for 40 min at room temperature, followed by annealing for 20 min at 1150 K. This procedure leads to a clean and well-ordered Si(100)-2×1 surface, as confirmed by AES and LEED. 2.2. Computational Methods. Electronic-structure calculations were done using the B3LYP hybrid density functional,24,25 as implemented in the Gaussian 98 suite of programs.26 The Stevens-Basch-Krauss27,28 effective core potentials were employed with a polarized triple-ζ valence basis set (CEP121G*). Geometries were optimized without any symmetry constraints, with the exception of structure b in Figure 6, vide infra. A two-dimer cluster model (Si15H16) of the Si(100)-2×1 surface was used. Zero-point-energy corrections were accounted for. This approach has been successfully used before to investigate the chemical properties of Si(100)-2×1 and the processes of ordering of hydrogen and iodine atoms on this surface.29 3. Results and Discussion The study of the CVD of VTMS has been conducted in two parts: cryogenic temperature adsorption and chemistry and room-temperature adsorption and chemistry. 3.1. Adsorption of VTMS on Si(100)-2×1 at Cryogenic Temperatures. First and foremost, the bonding and reactivity of VTMS with Si(100) have been explored using TPD and MIR-FTIR (FTIR ) Fourier transform infrared) techniques starting at cryogenic temperatures of 100-130 K. A series of TPD spectra for varying exposures of VTMS onto a clean and annealed Si(100)-2×1 sample at cryogenic temperature are shown in Figure 1. A mass-to-charge ratio (m/z) ) 85 was followed in this set of studies because this is the most intense peak in the mass spectrum of VTMS; however, other m/z’s were also followed to confirm that the desorbing molecule is indeed VTMS (m/z ) 59, 100, etc. not shown here). The cooling system of the manipulator in the TPD chamber allows the sample to be cooled to 130 K for adsorption. For a submonolayer coverage of VTMS, the major desorption feature was observed at 164 K, with a much smaller feature observed at 209 K. (The 209 K feature increases linearly with exposure above the 3 L dose of VTMS and is likely the background signal from the desorption of VTMS from the manipulator and sample support.) The activation energy calculated for the 164 K feature
8464 J. Phys. Chem. B, Vol. 109, No. 17, 2005
Figure 1. TPR/D studies of VTMS adsorbed on a Si(100)-2×1 surface at cryogenic temperatures (below 140 K). m/z ) 85 was followed as a function of VTMS exposure. The inset shows the integrated peak areas as a function of the initial exposure.
yields a value of 42.1 kJ/mol when the Redhead method with a pre-exponential factor of 1013 s-1 is used.30 This is characteristic of a physisorbed molecule and is remarkably similar to the ∆H of sublimation for VTMS, ∆Hs ) 40.7 kJ/mol, which can be calculated on the basis of the ∆H of fusion, ∆Hfusion ) 7.6 kJ/mol,31 and the ∆H of evaporation, ∆Hvap ) 33.1 kJ/mol.32 Upon increasing the exposure, we recorded a combination of the multilayer feature and the monolayer desorption at 164 K as one desorption peak at 155 K for a 3 L dose. Higher-exposure studies were not reproducible in our experimental setup because adsorption temperatures lower than 130 K would be needed for an accurate measurement. It is clear, however, that 1 L exposure corresponds to the submonolayer coverage, where only desorption from the monolayer is observed at 164 K with a smaller feature at 209 K, while the 3 L exposure desorption suggests the presence of the multilayer in the adsorbed material. The inset for Figure 1 shows the integrated area of the desorption peak as a function of coverage. At least up to the exposure of 3 L, this yields a straight line within our error of measurement, suggesting that the majority of VTMS adsorbed on Si(100)2×1 desorbs molecularly upon annealing. The AES study supported this statement because no measurable amount of carbon was detected on the surface following TPD. MIR-FTIR studies of VTMS adsorption on Si(100)-2×1 are presented in Figure 2. The instrument for these studies allows for cooling the sample to 100 K and, thus, a multilayer of VTMS can be easily condensed on the silicon sample. The coverage dependence of the absorption spectra and the molecular desorption of VTMS following adsorption at 100 K were confirmed by MIR-FTIR (not shown), supporting the conclusions reached after the TPD studies summarized in Figure 1. When the infrared absorption features in Figure 2b-e are compared to the literature data for gaseous and solid VTMS in Table 1,4,33 it is clear that the adsorption is molecular, with no
Pirolli and Teplyakov
Figure 2. MIR-FTIR studies of VTMS on the Si(100)-2×1 surface at (a) 300 K and (b-e) 100 K as a function of VTMS exposure. 15003500 cm-1 spectral region.
chemical transformation taking place. In particular, both the 3047 cm-1 absorption feature corresponding to the νdCH2 stretch and the peak at 1595 cm-1 corresponding to νCdC are clearly observed, even at the lowest exposure studied, 0.1 L. Thus, the double bond of the VTMS molecule has not reacted with the Si(100)-2×1 surface at 100 K. It should be noted that, in our studies, the 25-mm-long trapezoidal silicon crystal was transparent enough in the region of 1000-1500 cm-1 at cryogenic temperatures. Silicon is a strong absorber of infrared light below 1500 cm-1 at room temperature; however, at the temperature of 100 K, the spectral window of our sample has opened significantly, allowing for the assessment of the absorption bands in the wide region of 1000-4000 cm-1. The data in Table 1 summarize all of the absorption signatures, including the fingerprint region of the surface species. The comparison of 100, 3, and 0.1 L doses condensed on a Si(100)-2×1 surface at 100 K obtained in our studies correlates very well with the previous investigation of VTMS. The set also exhibits remarkable similarities among all three spectra. In fact, the only absorption features that cannot be deduced for low exposures (because of the lower signal-to-noise ratio) are the 2 × CdC stretch (which has very low intensity and can barely be observed even for the dose of 100 L), some CH3 deformations, and one of the two CH2 scissors modes. The integrated areas of all of the absorption features increase linearly with exposure. On the basis of the first part of our study, the mass spectrometry and MIR-FTIR investigations starting at cryogenic temperatures suggest that VTMS is adsorbed molecularly on Si(100)-2×1 at temperatures below 130 K. This adsorption leaves the double bond of the vinyl group unreacted. Upon heating, all of the VTMS molecules desorb intact without any noticeable chemical interaction with the Si(100)-2×1 surface.
Thermal Chemistry of Vinyltrimethylsilane
J. Phys. Chem. B, Vol. 109, No. 17, 2005 8465
TABLE 1. Summary of Vibrational Frequencies with Approximate Assignments for Gas-Phase VTMS,4 Solid VTMS,33 VTMS Adsorbed on SiO2,4 VTMS Condensed on a Si(100)-2×1 Surface at 100 K (100, 3, and 0.1 L Exposure; This Study), and VTMS Adsorbed on a Si(100)-2×1 Surface at Room Temperature (Saturation Exposure; This Study) gas-phase VTMS4
solid VTMS33
VTMS on SiO24
100 L VTMS/ Si(100) (100 K)
3 L VTMS/ Si(100) (100 K)
0.1 L VTMS/ Si(100) (100 K)
3193 3061 3020 2968 2908 2860
3178 3048 3009 2956 2894 2854
3184 3053 3012 2960 2900 2852
3185 3047 3010 2955 2895 2839
3182 3047 3010 2956 2895 2839
3044 3002 2959 2892 2838
2813 1600
2796 1594
2802 1596
2796 1595
2797 1590
1445
1439 1427 1420 1413 1401 1395 1364 1341 1277 1268 1257 1249 1240
1443
2796 1595 1458 1441 1421 1416 1404 1396 1365 1342 1279 1269
1402
1402
1409
1405
1250 1240
VTMS/Si(100) (300 K)
2957 2901
approximate assignment 2 × CdC stretch CH2 asym stretch CH stretch CH3 asym stretches CH3 sym stretches overtone/combination of CH3 deformations
2070 (Si-H) CdC stretch CH3 asym deformation CH3 asym deformation CH3 asym deformation CH3 asym deformation CH3 asym deformation CH2 scissors CH2 scissors
1363 1275 1277 1261
1262
1246
1248
3.2. Thermal Chemistry of VTMS Adsorbed on a Si(100)2×1 Surface at Room Temperature. The MIR-FTIR spectrum in Figure 2a shows a room-temperature Si(100)-2×1 surface exposed to a saturation exposure of VTMS. It clearly exhibits the symmetric and asymmetric stretches of the methyl group at 2901 and 2957 cm-1, while the absorption peak corresponding to the νCdC double bond of the VTMS as well as that of the νC-H bond at 3047 cm-1 indicative of the terminal dCH2 group are absent. A noticeable νSi-H stretch mode at 2070 cm-1 suggests that partial decomposition of VTMS takes place even at room temperature. An estimation based on a comparison with the saturated monohydride surface (infrared spectrum not shown) suggests that less than 10% of the surface is covered with hydrogen as a result of the partial VTMS decomposition at room temperature. Coverage dependence of the roomtemperature adsorption is summarized in Figure 3. On the basis of the comparison of the integrated areas of the C-H stretch region of the infrared spectra versus the exposure of VTMS shown in the inset for Figure 3, it is clear that, even though the initial sticking coefficient is quite high, it is smaller than unity and it decreases with coverage. VTMS saturates the surface at approximately 10 000 L, with no noticeable changes observed at higher exposures. Upon thermal annealing, we observed multiple reaction pathways. According to AES studies comparing the amount of carbon on a surface right after roomtemperature exposure of the Si(100)-2×1 surface to a saturation exposure of 10 000 L and after annealing of this surface to 1000 K, 35% of the carbon is lost in thermal chemistry, with the remaining 65% likely forming silicon carbide, as in the previous studies of VTMS on the Si(111)-7×7 surface.5 A combination of AES, MIR-FTIR, and thermal desorption studies was used to address the thermal chemistry of VTMS adsorbed on a Si(100)-2×1 surface at room temperature. Figure 4 shows the changes in the integrated C-H stretch-region absorption of a Si(100)-2×1 surface saturated with VTMS at room temperature (Figure 2a) and briefly annealed to the indicated temperatures. Each point in Figure 4 represents a separate experiment so that no interference from multiple annealings is observed. Because the shape of the absorption features in the infrared spectra did not change significantly
CH bend (in plane) CH3 sym deformation CH3 sym deformation CH3 sym deformation CH3 sym deformation
during the course of annealing, the integrated area of the absorption in the region between 2600 and 3100 cm-1 (C-H stretch region) should serve as a quantitative indicator of the carbon loss. As will be shown below, in thermal desorption studies, all of the products desorbing from the surface contain methyl groups and the absorption bands corresponding to symmetric and asymmetric stretches of CH3 are the most prominent bands in the spectra of the VTMS chemisorbed on a Si(100)-2×1 surface at room temperature. Thus, if the amount of surface carbon is proportional to the integrated infrared absorption intensity in the C-H stretch region, the following correlation can be obtained: out of the 35% of carbon lost during thermal annealing, the first 5% of carbon is lost around 360 K, then 10% more around 500 K, and finally 20% between 650 and 800 K. These data are in agreement with the AES studies within the error of measurement. Thermal desorption was used to identify some of the species desorbing from the surface exposed to 10 000 L of VTMS upon annealing. Figure 5 summarizes these studies. A molecular desorption of VTMS occurs around 360 K, as shown in Figure 5a, tracing m/z ) 85 (m/z ) 43, 45, 59, 73, and 100 were used to confirm this assignment). Upon further annealing between 450 and 550 K, a complex surface reaction takes place. We estimate that the majority of the desorbing species are propylene based on m/z ) 39, 40, 41, and 42, with the rest likely consisting of VTMS and other silicon-containing products. Previous studies of propylene adsorption and reaction with the Si(100)-2×1 surface by Yates’ group34 suggest that this hydrocarbon partially decomposes on this silicon surface while a portion of it desorbs around 550 K, which is consistent with our findings. A mechanistic explanation of the propylene formation is presented below in section 3.3. Finally, at even higher annealing temperatures, complex silicon-containing products are evolved, as confirmed by following traces of m/z ) 59, 73, 85, and 100. While the assignment of the two lower temperature desorption features is relatively easy based on the mass spectra of propylene and VTMS,23 we could not specifically assign the last desorption feature around 800 K to a single molecule. It is likely that the compounds desorbing from the surface within this temperature range form a mixture. The only signature of this mixture is that
8466 J. Phys. Chem. B, Vol. 109, No. 17, 2005
Pirolli and Teplyakov
Figure 5. TPR/D studies of VTMS adsorbed on a Si(100)-2×1 surface at room temperature. The evolution of (a) hydrogen (m/z ) 2), (b) vinyldimethylsilane (m/z ) 85), (c) propylene (m/z ) 42), and (d) silicon-containing compounds (m/z ) 45) are shown. m/z ) 85 was followed for VTMS evolution (most prominent fragmentation peak).
Figure 3. MIR-FTIR studies of VTMS adsorbed on the Si(100)-2×1 surface at 300 K as a function of VTMS exposure. 2600-3100 cm-1 spectral region. The inset shows the integrated peak areas as a function of the initial exposure.
Figure 4. Integrated area of the 2600-3200 cm-1 absorption band of VTMS as a function of the surface annealing temperature. The initial exposure was 10 000 L of VTMS at room temperature to produce a saturated monolayer. Each point corresponds to a separate experimental measurement.
at least some of these products contain silicon. This statement is supported by the absence of methane, ethane, and ethylene
at any temperature during desorption and by the fact that some of the products desorbing from the surface above 600 K show m/z ratios as high as 100 amu, while the formation of carbonbased oligomeric molecules is unwarranted. Thus, on the basis of a combination of thermal desorption and vibrational spectroscopy, we can estimate that approximately 5% of the VTMS adsorbed at room temperature on a Si(100)2×1 surface desorbs into the gas phase around 360 K. On the basis of the Redhead method with a pre-exponential factor of 1013 s-1, this desorption temperature yields 93 kJ/mol as the activation energy. A total of 10% of the surface carbon is lost as propylene; however, it should also be noted that, in the previous studies by Yates et al.,34 it was shown that only 35% of the adsorbed propylene is released back into the gas phase upon annealing, while 65% forms surface carbon. This means that, although there may be more than 10% conversion of the chemisorbed VTMS to propylene, we can only register the propylene that desorbs from the surface. 3.3. Mechanistic Explanation of the VTMS Reactions on Si(100)-2×1. The key observations of the low-temperature interaction of VTMS with a Si(100)2×1 surface, based on thermal desorption and infrared measurements, suggest that the VTMS molecule is physisorbed on this surface at cryogenic temperatures. The main difference in room-temperature adsorption, as compared to the 100-130 K interval, is that this adsorption definitely results in the formation of covalent bonds and the rehybridization of the double bond of VTMS. By itself, this conclusion is not unexpected and is consistent with multiple previous studies of the unsaturated hydrocarbons on Si(100)2×1. It is also similar to the proposed reaction of VTMS on Si(111)-7×7.5 However, it is the further thermal transformations of the adsorbed VTMS that seem to differ so much on these two surfaces. While on Si(111)-7×7 the only chemical reaction was found to be the formation of surface silicon carbide and no species other than hydrogen were detected to desorb into the gas phase upon annealing, the chemistry of VTMS on a
Thermal Chemistry of Vinyltrimethylsilane
Figure 6. Models for adsorption and reactions of VTMS on Si(100)2×1. VTMS adsorbed by a simple [2 + 2] addition to a silicon dimer of the Si(100)-2×1 surface (a) is compared to a similar addition product involving two neighboring silicon dimers (b). The stability of these products is compared to those of the models that are obtained by chemically plausible transformations of configuration a when (c) a trimethylsilyl group transfers a methyl group to the adsorbed vinyl entity to form an adsorbed propylene molecule and the resulting Si(CH3)2 group bridges-over between two silicon atoms of a neighboring silicon dimer and when (d) a trimethylsilyl group of the VTMS molecule chemisorbed on a silicon dimer releases a methyl group to one of the silicon atoms of a neighboring dimer while bridging the vinyl entity with the other silicon atom of the same neighboring dimer. Structure e shows the [2 + 2] product depicted in (a) with all of the silicon atoms, including the atoms representing surface silicon dimers, saturated with hydrogens, and structure f shows the [2 + 2] product depicted in (b) with all of the silicon atoms, including the atoms representing surface silicon dimers, saturated with hydrogens.
Si(100)-2×1 surface is more elaborate, but it also offers several insights into the nature of surface silicon carbide formation. Because previous studies of VTMS on Si(111)-7×75 did not extend down to the cryogenic temperatures, the comparison of VTMS behaviors on different silicon surfaces could not be complete. However, as shown in these studies, VTMS adsorbs molecularly on a Si(100)-2×1 surface at cryogenic temperatures of 100-130 K. Upon heating, molecular desorption is the only observed reaction pathway, suggesting that the barrier for chemisorption is significantly higher than that for desorption. At room temperature, the first step of the VTMS interaction with the Si(100)-2×1 surface is most likely the [2 + 2] attachment of a double bond of the vinyl group to a surface silicon dimer to form a Si2C2 ring with two Si-C covalent bonds, shown in Figure 6a. While we cannot comment on the mechanism of this transformation based on the results presented here, the VTMS behavior is consistent with the formation of a physisorbed intermediate followed by chemical interaction either by a diradical intermediate model or via a three-atom intermediate described previously.5,19 Because of a difference in the
J. Phys. Chem. B, Vol. 109, No. 17, 2005 8467 geometrical arrangement of potential reactive sites on Si(100)2×1 and Si(111)-7×7 surfaces, another possibility should also be considered for Si(100)-2×1. Instead of interacting with two silicon atoms of the same dimer on a Si(100)-2×1 surface, another bonding configuration, with the double bond of a VTMS molecule interacting with two neighboring dimers, can be formed. However, this configuration, depicted in Figure 6b, is much less favorable energetically compared tostructure a. It should be noted that, to prevent the model cluster from collapsing when calculating structure b, the positions of five silicon atoms representing the third and fourth layers, as well as two central atoms in the second layer, were fixed. This was the only structure whose geometry was restricted during the optimization procedure. The real difference between the two bonding configurations corresponding to the [2 + 2] addition can be significantly smaller; however, for the purposes of this discussion, it is more important that the experimental observations can be explained on the basis of reasonable surface intermediates of the VTMS reaction with the Si(100)-2×1 surface. To explore these differences further, the hydrogensaturated models e and f in Figure 6 were also examined. These structures represent the adducts of types (a) and (b) on a silicon surface when extra hydrogen atoms from partial decomposition processes participate in surface chemistry. Structures e and f in Figure 6 are identical to structures a and b except that the surface silicon atoms are also terminated with hydrogen to prevent the collapse of the two-dimer cluster model. Structure f proved to be 43.58 kJ/mol more stable than structure e, so as surface decomposition pathways produce significant amounts of hydrogen on a Si(100)-2×1 surface, the stability of the initial [2 + 2] adducts may change. This result indicates that thermal chemistry of the silicon carbide formation from VTMS on the Si(100)-2×1 surface is sensitive to the amount of hydrogen present and, thus, the mechanism of the decomposition can change as the decomposition proceeds. One of the most important findings presented here is the fact that the carboncarbon double bond rehybridizes upon VTMS adsorption on Si(100)-2×1 at room temperature. Thus, the formation of a stable dative-bond intermediate, where the carbon-carbon double bond remains intact, suggested as a possibility by Andersohn et al.5 for VTMS on Si(111)-7×7 can be ruled out. At the same time, a similar type of surface adsorbate may be responsible for the infrared spectrum of VTMS physisorbed on a Si(100)-2×1 surface at cryogenic temperatures. Potentially, both structures a and b could be formed on a Si(100)-2×1 surface, and without a detailed analysis of the transition states leading to these configurations, neither of these models can be excluded. However, we will focus on the lowerenergy structure, a, which is similar to the structures produced by other mono-unsaturated compounds on a Si(100)-2×1 surface, to examine the feasibility of the produced surface intermediates in terms of the thermal chemistry results obtained in thermal desorption studies. The thermal desorption and infrared spectroscopy studies prove the formation of propylene and silicon-containing gas-phase products together with the fact that methyl groups are stable during the whole course of the thermal reaction. Keeping this in mind, the two most likely surface transformations of configuration a should lead to the structures depicted in parts c and d of Figure 6. In principle, similar structures resulting from transformation starting with configuration b can also be examined; however, here, we only analyze a set of plausible structures explaining the experimental results, starting with the chemisorption configuration, a. A computational comparison of models c and d with the initial
8468 J. Phys. Chem. B, Vol. 109, No. 17, 2005 attachment product, a, suggests that both of them are much more stable than this initial adduct. Structure c is consistent with the production of propylene at elevated temperatures. This structure is 31.13 kJ/mol more stable than the [2 + 2] adduct, a. Even more stable is structure d, which is understandable because structure c has a strained C2Si cycle. Structure d is 155.15 kJ/ mol more stable than the initial [2 + 2] adduct. Structure d can be produced by a shift of a methyl group from the trimethylsilyl entity of the adsorbed VTMS to the neighboring silicon dimer, while Si(CH3)2 is bridged from the R carbon of the vinyl group to the other silicon atom of the same dimer. Although other surface intermediates involving more silicon dimers are certainly not impossible to form, only configurations c and d are stable entities that explain the formation of both propylene and siliconand methyl-group-containing products at elevated temperatures. Further thermal transformations of both structures are probably much more complicated. However, this study presents an energetically plausible explanation for the mechanism of the initial stages of silicon carbide formation. The main difference between Si(100)-2×1 and Si(111)-7×7 surfaces with respect to their reaction with VTMS appears to be in the geometrical arrangement of the reactive sites. Despite a complicated 7 × 7 reconstruction pattern on a Si(111) surface, there is essentially only one type of adsorbed species that VTMS can form: a Si3C2 cycle involving one rest atom and one adatom.5 This adsorption is essentially irreversible, and it seems that the available neighboring sites can only facilitate a complete decomposition upon thermal annealing. On a Si(100)-2×1 surface, the presence of a neighboring surface dimer after the initial chemisorption of VTMS offers a variety of reaction pathways, including the formation of propylene. The propylene formation is explained by the computational optimization of the suggested models in Figure 6. It should also be noted that the adsorption of VTMS on Si(100)-2×1 is not completely irreversible. According to the TPD studies, the desorption process for a chemisorbed VTMS on Si(100)-2×1 competes with the formation of more thermodynamically stable surface intermediates of the kind depicted in Figure 6c,d. Further computational studies are needed to explain the unique decomposition pathway on Si(111)-7×7. 4. Conclusions Complex thermal transformations of VTMS on a Si(100)2×1 surface have been investigated by thermal desorption, infrared, and Auger electron spectroscopies and by computational analysis. At the cryogenic temperatures of 100-130 K, VTMS is physisorbed on this surface, while room temperature adsorption leads to the chemisorbed species. A small percentage of VTMS itself is released back into the gas phase by 400 K, propylene desorbs around 500 K, a combination of several silicon-containing products desorbs around 800 K, and finally residual hydrogen is released at 800 K. Approximately 65% of the carbon initially adsorbed on the Si(100)-2×1 surface is retained as silicon carbide. An explanation of this chemistry was given using a computational analysis of the most probable adsorption models. Important differences in the chemical behavior of VTMS on Si(100)-2×1 and Si(111)-7×7 can be understood on the basis of the arrangement of the reactive sites on these silicon surfaces. While essentially only one configuration that further leads to a complete decomposition and silicon carbide formation can be formed on a Si(111)-7×7 surface,
Pirolli and Teplyakov neighboring dimer sites on a Si(100)-2×1 surface offer a possibility for a more complicated set of surface reactions. Acknowledgment is made to the National Science Foundation (Grant CHE-0313803) for the support of this research. A.V.T. also thanks Dr. Olga Dmytrenko (Department of Chemistry and Biochemistry, University of Delaware) for her help with computational structure optimization and Professor Douglas J. Doren and Jeffrey Frey (Department of Chemistry and Biochemistry, University of Delaware) for useful discussions. References and Notes (1) Hampden-Smith, M. J.; Kodas, T. T. Polyhedron 1995, 14 (6), 699732. (2) Murarka, S. P. Mater. Sci. Eng., R 1997, R19, 87-151. (3) Norman, J. A. T.; Muratore, B. A.; Dyer, P. N.; Roberts, D. A.; Hochberg, A. K. J. Phys. IV 1991, 1, C2-C271. (4) Dubois, L. H.; Zegarski, B. R. J. Electrochem. Soc. 1992, 139, 3295. (5) Andersohn, L.; Kochanski, G. P.; Norman, J. A. T.; Hinch, B. J. J. Vac. Sci. Technol., B 1996, 14 (2), 1032-1037. (6) Buriak, J. M. Chem. ReV. 2002, 102, 1272-1308. (7) Hovis, J. S.; Liu, H.; Hamers, R. J. Surf. Sci. 1998, 402-404, 1-7. (8) Bent, S. F. Surf. Sci. 2002, 500, 879-903. (9) Bent, S. F. J. Phys. Chem. B 2002, 106, 2830-2842. (10) Hamers, R. J.; Coulter, S. K.; Ellison, M. D.; Hovis, J. S.; Padowitz, D. F.; Schwartz, M. P.; Greenlief, C. M.; Russell, J. N., Jr. Acc. Chem. Res. 2000, 33, 617-624. (11) Hamers, R. J.; Wang, Y. Chem. ReV. 1996, 96, 1261-1290. (12) Wolkow, R. A. Annu. ReV. Phys. Chem. 1999, 50, 413-441. (13) Clemen, L.; Wallace, R. M.; Taylor, P. A.; Dresser, M. J.; Choyke, W. J.; Weinberg, W. H.; Yates, J. T., Jr. Surf. Sci. 1992, 268, 205-216. (14) Casaletto, M. P.; Zanoni, R.; Carbone, M.; Piancastelli, M. N.; Aballe, L.; Weiss, K.; Horn, K. Phys. ReV. B: Condens Matter Mater. Phys. 2000, 62 (24), 17128-17133. (15) Lu, X. J. Am. Chem. Soc. 2003, 125, 6384-6385. (16) Pan, W.; Zhu, T.; Yang, W. J. Chem. Phys. 2004, 107 (10), 39813985. (17) Konecny, R.; Doren, D. J. Surf. Sci. 1998, 417, 169-188. (18) Machida, S.; Hamaguchi, K.; Nagao, M.; Yasui, F.; Mukai, K.; Kato, H. S.; Okuyama, H.; Kawai, M. J. Phys. Chem. B 2002, 106, 16911696. (19) Nagao, M.; Mukai, K.; Yamashita, Y.; Yoshinobu, J. J. Phys. Chem. B 2004, 108, 5703-5708. (20) Ascherl, M. V.; Campbell, J. H.; Lozano, J.; Craig, J. H., Jr. J. Vac. Sci. Technol., A 1995, 13 (6), 2721-2725. (21) Wang, P. W.; Sulak, J.; Bater, S.; Zhang, L. P.; Craig, J. H., Jr. Thin Solid Films 1997, 304, 160-165. (22) Lozano, J.; Campbell, J. H.; Craig, J. H., Jr. Appl. Surf. Sci. 1998, 136, 159-165. (23) National Institute of Standards and Technology (NIST) Chemistry WebBook. http://webbook.nist.gov. (24) Becke, A. D. J. Chem. Phys. 1993, 98, 1372-1377. (25) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B: Condens. Matter Mater. Phys. 1988, 37, 785. (26) Frisch, M. J. Gaussian 98, revision A.9; Gaussian, Inc.: Pittsburgh, PA, 1998. (27) Stevens, W. J.; Basch, H.; Krauss, M. J. Chem. Phys. 1984, 81, 6026-6033. (28) Stevens, W. J.; Krauss, M.; Basch, H.; Jasien, P. G. Can. J. Chem. 1992, 70 (2), 612-630. (29) Bulanin, K. M.; Shah, A. G.; Fitzgerald, D. R.; Doren, D. J.; Teplyakov, A. V. J. Phys. Chem. B 2002, 106, 7286-7289. (30) Redhead, P. A. Vacuum 1962, 203-211. (31) Lebedev, B. V.; Lebedev, N. K.; Khotimskii, V. S.; Durgar’yan, S. G.; Nametkin, N. S. Dokl. Akad. Nauk SSSR 1981, 259, 629. (32) Voronkov, M. G.; Baryshok, V. P.; Klyuchnikov, V. A.; Danilova, T. F.; Pepekin, V. I.; Korchagina, A. N.; Khudobin, Y. I. J. Organomet. Chem. 1988, 345, 27-38. (33) Durig, J. R.; Natter, W. J.; Johnson-Streusand, M. Appl. Spectrosc. 1980, 34 (1), 60-65. (34) Bozack, M. J.; Choyke, W. J.; Muehlhoff, L.; Yates, J. T., Jr. Surf. Sci. 1986, 176, 547-566.
