Dominance of Silylene Chemistry in the Decomposition of

May 12, 2014 - Role of free-radical chain reactions and silylene chemistry in using methyl-substituted silane molecules in hot-wire chemical vapor dep...
0 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/JPCA

Dominance of Silylene Chemistry in the Decomposition of Monomethylsilane in the Presence of a Heated Metal Filament R. Toukabri and Y. J. Shi* Department of Chemistry, University of Calgary, 2500 University Drive NW, Calgary, Alberta T2N 1N4, Canada ABSTRACT: The gas-phase reaction chemistry of the decomposition of monomethylsilane (MMS) has been studied in the presence of a heated metal filament in a hot-wire chemical vapor deposition (HWCVD) reactor. A 10.5 eV vacuum ultraviolet laser single-photon ionization time-of-flight mass spectrometry was employed in combination with isotope labeling and chemical trapping to examine the mechanistic details in the reaction chemistry. We have demonstrated the dominant involvement of the methylsilylene (HSiCH3) intermediate in the gas-phase reaction chemistry. Free radical and silene intermediates do not play a role. Major products are found to be H2, 1,2-dimethyldisilane (DMDS), and 1,3disilacyclobutane (DSCB). The formation of DMDS proceeds by the insertion reaction of methylsilylene, whereas DSCB originates from the dimerization reaction of methylsilylene. Similar reaction chemistry has been observed when using the different filament materials of tungsten and tantalum in the HWCVD reactor. This indicates that changing the filament material from Ta to W does not affect the gas-phase reaction chemistry when using MMS in the HWCVD process. Finally, comparison of the reaction chemistry of MMS with those of dimethylsilane, trimethylsilane, and tetramethylsilane sheds light on the influence of increasing Si−H bonds. A switch in the dominated chemistry from free-radical short-chain reactions to silylene insertion/dimerization reactions occurs as the number of Si−H bonds increases in the four methyl-substituted silane molecules.

■

INTRODUCTION Among the four methyl-substituted silane molecules, monomethylsilane (MMS) has received the most attention as a precursor gas for the formation of silicon carbide thin film materials using chemical vapor deposition (CVD). Reports have been found in using MMS in the processes of plasmaenhanced CVD (PECVD),1−3 low-pressure CVD (LPCVD),4−8 and hot-wire CVD (HWCVD, also known as catalytic CVD).9−12 Amorphous, crystalline, as well as microcrystalline SiC thin films have been produced. Main advantages of using MMS lie in the fact that the direct Si−C bond exists in the molecule and that the molecule has the same Si:C composition as in solid SiC materials. Indeed, most of the above studies3,5,7−10,12 reported that stoichiometric SiC thin films were obtained when using MMS. Despite the wide use of MMS in CVD, only a few studies have focused on its decomposition kinetics in a CVD reactor. In their work on using mass spectrometry to study the composition of a MMS plasma, Delplancke et al.1,2 showed that the majority of the Si−C bonds in MMS were preserved in the gas-phase plasma. Johnson et al.4 studied the thermal decomposition kinetics of MMS and also silacyclobutane (SCB) in a LPCVD reactor using mass spectrometric partial pressure analyses in combination with the characterization of silicon carbide thin films. They found that both MMS and SCB molecules decompose to form the isomeric species methylsilylene (HSiCH3) and silene (H2C = SiH2). Ohshita7 also found that HSiCH3 was an important radical species produced © 2014 American Chemical Society

by thermal decomposition of MMS in the LPCVD process. In accordance with his study, this species is very reactive with a sticking coefficient close to 1.0. Although not many studies can be found on the kinetics of MMS decomposition in a LPCVD reactor, decomposition of MMS in pyrolysis has been extensively studied. The first study was reported by Kohanek et al.13 in a flow system at 520 °C, where the main products were H2, 1,2-dimethyldisilane (DMDS), and dimethylsilane (DMS). Methane was found to be a minor product with an amount of about 2.5% that of the dominant product, H2. Later, Neudorfl et al.14,15 performed static pyrolysis of MMS at 40−400 Torr and 340−440 °C. They observed H2 and DMDS in approximately equal yields and DMS as a minor (∼5%) product. H2 and DMDS were shown to come mainly from a molecular process, 1,1-H2 elimination (eq 1) and silylene insertion reaction (eq 2), respectively. ̇ ̇ + H2 CH3SiH3 → CH3SiH

(1)

̇ ̇ + CH3SiH3 → CH3SiH 2SiH 2CH3 CH3SiH

(2)

Minor contributions to the two products were from a short-chain free radical reaction, which was surface catalyzed. The minor product DMS originated from the free radical process. CH4 was Received: March 20, 2014 Revised: May 8, 2014 Published: May 12, 2014 3866

dx.doi.org/10.1021/jp502795u | J. Phys. Chem. A 2014, 118, 3866−3874

The Journal of Physical Chemistry A

Article

tetramethylsilane (TMS).26 We have shown that short freeradical chain reactions dominate the HWCVD chemistry when using TMS and TriMS, whereas the contributions from silene/ silylene species are minor. With an increasing number of Si−H bonds in DMS, the silylene and silene chemistry becomes dominant at low filament temperatures (≤1200 °C) and short reaction time. The prevalence of free-radical reactions only revealed itself when the filament temperature is increased to ≥1500 °C. As part of the ongoing research efforts on the gasphase reaction chemistry of organosilicon molecules in HWCVD, we have examined the decomposition chemistry of MMS. It aims to answer the important question of the effect of a further increase in Si−H bonds in the precursor gas molecules on the gas-phase reactions, hence on the identity of the film growth precursors. In this work, the gas-phase reaction chemistry of MMS in the presence of a heated metal filament (W or Ta) was studied under practical pressures in a HWCVD reactor. Nonresonant singlephoton ionization (SPI) using a 10.5 eV vacuum ultraviolet (VUV) photon coupled with time-of-flight-mass spectrometry (TOF-MS)27,28 was employed to detect the chemical species produced from the gas-phase reactions. Experiments on the decomposition of methylsilane-d3 (CH3SiD3, MMS-d3) were performed to help with the identification of reaction products. 1,3-Butadiene was used to trap silylene and silene species. This serves to differentiate the mechanism involving silylene/silene intermediates from the free-radical processes, and therefore, provide insights into the mechanistic details in the reaction chemistry. The effect of filament materials on the gas-phase reaction chemistry was also investigated. Finally, the reaction chemistry using MMS was compared with those of DMS, TriMS, and TMS to shed light on the effect of the increasing number of Si−H bonds in the source molecules.

not observed as a product in their study. In search for CH4, Davidson et al.16 investigated a low-pressure pyrolysis at 0.01−0.1 Torr and 569−727 °C. H2 (∼96%) and CH4 (∼4%) were the only two products detected by mass spectrometry. Methane was formed by the molecular elimination process (eq 3), whereas the contributions from the free-radical mechanism were discounted on thermochemistry grounds. CH3SiH3 → : SiH 2 + CH4

(3)

17−19

Sawrey et al. revisited the decomposition of MMS in order to determine the nature and yields of the primary decomposition processes. They have carried out the MMS pyrolysis both in a shock tube at a total pressure of 4700 Torr in the temperature range of 852−977 °C and in a static system at 1500 Torr and 427 °C. H2 and CH4 were reported as the major products, together with small amounts of SiH4 and DMS. DMDS was not observed. Aside from the occurrence of 1,1-H2 elimination and methane elimination as suggested by previous studies, they also showed the presence of 1,2-elimination to produce silene species. CH3SiH3 → CH 2SiH 2 + H 2

(4)

The relative efficiency of 1,1-H2 elimination, 1,2-H2 elimination, and CH4 elimination for the static pyrolysis at 427 °C was determined to be 0.78, 0.16, and 0.06, respectively. Therefore, the 1,1-elimination reaction leading to the formation of molecular hydrogen and methylsilylene has been accepted by most research groups to be the dominant step in MMS pyrolysis. The occurrence of other steps differs from one study to another. Under the pyrolysis conditions, the cleavage of Si−H [BDE (CH3SiH2−H) = 89.6 kcal mol−1]20 and Si−C [BDE (SiH3−CH3) = 88.2 kcal mol−1]20 bond to form free radicals would require high energy. Hence, they only contribute to the overall decomposition to a small extent. HWCVD differs from the conventional LPCVD in that it uses a resistively heated metal filament, typically made of tungsten (W) or tantalum (Ta), to decompose the precursor gas. The species thus produced are then transported to a substrate placed near the heated filament but kept at a relatively low temperature. The actual film growth precursors come from the primary decomposition species on the filament and/or from the secondary reaction products in the gas phase. However, very few studies can be found on characterizing the gas-phase species from the decomposition of MMS in the presence of a metal filament. In their work on the detection of free radicals during the hot-wire decomposition of MMS, along with other four organosilicon molecules, Zaharias et al.21 found that Si radicals were the major product on both W and rhenium (Re) filaments. The CH3 radicals were also observed. The formation of methyl radicals was also demonstrated in our previous work on the decomposition of methyl-substituted silanes, including MMS, on W and Ta filaments.22 H radicals and H2 molecules were also detected in our work. It should be noted that all these studies were performed at very low pressures from 5 × 10−6 to 1 × 10−5 Torr that ensure a collision-free condition. In practice, the pressures used for the deposition of SiC films by CVD using MMS as a precursor gas range from several tens of milliTorrs to several Torrs.1−7,10−12 At these pressures, the collision-free condition is voided. The secondary gas-phase reactions between the primary radicals themselves and between the radicals and parent source molecules become important. However, there is a lack of knowledge for the nature of secondary gas-phase reactions when using MMS in a HWCVD reactor. Recently, our lab has studied the gas-phase reaction chemistry of DMS,23 trimethylsilane (TriMS),24,25 and

■

EXPERIMENTAL METHODS Details of the HWCVD sources and the VUV laser SPI-TOF mass spectrometer used to study the gas-phase reaction chemistry in the HWCVD processes have been provided elsewhere.29,30 Only the essential features will be described here. Two types of HWCVD sources were used in this work. A collision-free HWCVD source was used to monitor the primary decomposition products of MMS directly released from the hot metal filament. Here, a resistively heated W or Ta filament (10 cm length, 0.5 mm diameter) was placed in the main chamber where the operating pressure was maintained at 1 × 10−5 Torr to ensure collision-free conditions. The second type is a cylindrical HWCVD reactor (volume 7.1 L) housing the heated metal filament. The total pressure of a mixture of MMS and helium in the reactor, monitored by a capacitance manometer (MKS Baratron, type 626A), was maintained at 12 Torr using a mass flow controller (MFC) (MKS, type 1179A). The typical flow rate was 1.40 sccm. This gives a partial pressure of MMS (or MMS-d3) in the reactor of 0.48 Torr when using a 4% sample mixture in He. In the reactor setup, the collision-free conditions were voided. This allows for the detection of the products from secondary gas-phase reactions between the primary decomposition species themselves and between the decomposition species and the precursor molecules. The reactor was connected to the main chamber through a 0.15 mm diameter pinhole. The filament temperature was measured by a two-color pyrometer. Gas-phase reaction products were examined at filament temperatures varying from 900 to 1800 °C at increments of 100 °C. For each temperature, a mass spectrum was recorded every 1 min for 30−60 min. 3867

dx.doi.org/10.1021/jp502795u | J. Phys. Chem. A 2014, 118, 3866−3874

The Journal of Physical Chemistry A

Article

MMS on Ta and W filaments is initiated by Si−H bond cleavage. This is followed by Si-CH3 bond breaking to form methyl radical or by recombination of H adsorbates to form the H2 molecule. The formation of the ·H radical was also demonstrated when MMS and other methyl-substituted silane molecules were used as precursor gases. In order to understand the concomitant products with the formation of H2, the intensity ratio of possible complementary fragment peaks to that of the parent ion peak (m/z 46) were examined. Figure 1 shows the intensity ratios of CH3SiH+

The 118 nm VUV laser radiation (10.5 eV) used to ionize the chemical species produced from hot-wire decomposition and from secondary gas-phase reactions was generated by frequency tripling the 355 nm UV output from an Nd:YAG laser in a gas cell containing 210 Torr of 10:1 Ar:Xe gases. A lithium fluoride (LiF) lens was placed at the exit end of the gas cell to focus the 118 nm VUV light at the center of the ionization source in the TOF mass spectrometer, whereas the 355 nm UV beam was focused beyond. The ions produced were detected by a microchannel plate (MCP) detector located at the end of a field-free flight tube. Signals from the MCP detector were preamplified, averaged over 512 laser pulses, and saved into a computer for analysis. W (99.9+%, Aldrich) and Ta (99.9+%, Aldrich) were used as filament materials as received. MMS (Gelest, 99.9%) was used without further purification. Methylsilane-d3 (MMS-d 3 , CH3SiD3) was synthesized by the reduction of methyltrichlorosilane (CH3SiCl3) with an excess of lithium aluminum deuteride (LiAlD4, CDN Isotopes, 98 atom % D).31 Due to the high volatility of MMS-d3, its isotope purity was examined mainly with our TOF mass spectrometer. The ionization energy (IE) of MMS is higher than the single photon energy of 10.5 eV used in the SPI mode,32 therefore, the peaks from the parent MMS ion and its fragment ions are very weak. Due to the low intensity of the parent ion peaks, the isotope purity of MMS-d3 cannot be determined. However, the observed peak patterns in the mass spectrum of MMS-d3 are consistent with the molecular structure of CH3SiD3. Gaseous mixtures of MMS (or MMS-d3) in helium (99.999%, Praxair) were prepared by diluting the room-temperature MMS (or MMS-d3) sample in He in a 2.25 L sample cylinder.

Figure 1. Intensity ratio of CH3SiH+ (m/z 44) to that of the parent of CH3SiH3+ (m/z 46) for pure MMS versus the Ta filament temperature at a chamber pressure of 1 × 10−5 Torr.

■

RESULTS AND DISCUSSIONS The room-temperature 10.5 eV SPI mass spectrum of MMS has been described in our previous work.22 The fragment ion peak of HSiCH3+ (m/z 44) is the base peak. Other peaks observed are the parent ion peak (m/z 46, 35.9 ± 3.9%) and its fragments SiHx+ (x = 0−3), SiCH3+, H2SiCH3+. As mentioned earlier, the IE of MMS is greater than the photon energy of 10.5 eV in the SPI mode used in this work. The observation of the parent MMS ion peak and its photofragment peaks is attributed to the minor electron ionization (EI) due to a small amount of photoelectrons caused by scattered UV radiation in the ionization region. In order to quantify the EI contribution relative to SPI, we have previously used the intensity ratio of the He+ peak (from EI) to that of the base peak at m/z 73 using TriMS (from SPI). From it, the ionization cross section ratio of He to TriMS has been determined to be 3.8 × 10−4 using the 10.5 eV SPI.25 Therefore, the contribution from EI is present, but minor, in the SPI mode used in this work. This explains the weakness of the observed parent MMS ion peak and its fragment ion peaks. 1. Primary Decomposition on W and Ta Filaments at Low Pressures. Under the collision-free conditions, H2 molecule and ·CH3 radical were clearly observed as the products. The formation of H2 agrees well with the results from previous studies of MMS pyrolysis13−19 that H2 is the dominant product, as described in Introduction. However, the methyl radical formation that was not observed and discounted in the MMS pyrolysis did occur in the presence of a heated W or Ta filament. The mechanism for the formation of H2 molecule and methyl radical from MMS along with other three methylsubstituted silane molecules has been discussed in detail in our previous work.22 It has been shown that the dissociation of

(m/z 44) to that of the parent MMS ion at different Ta filament temperatures at a chamber pressure of 1 × 10−5 Torr. The ratio is greater than the room temperature ratio of 2.8. It increases with Ta filament temperature (same for W), indicating the presence of a species with an elemental composition of CH4Si in the gas phase. This species, being the coproduct of H2, can be dimethylsilylene or methylsilene (both having a mass of 44 amu), depending on whether the second H is cleaved from Si−H or C−H bonds (eqs 1 or 4). In the session below, identification of the products from secondary gas-phase reactions helps in elaborating the nature of the coproduct of H2 released from the hot filament. 2. Secondary Gas-Phase Reactions in a HWCVD Reactor with Ta at Relatively High Pressures. To study the secondary gas-phase reactions, 12 Torr of 4% MMS diluted in He was introduced in the HWCVD reactor. The mass spectra recorded for this sample at 2 min after the filament was turned on at different temperatures ranging from 900 to 1600 °C are shown in Figure 2. As can be seen from the figure, the intensity of the predominant fragment peak of the parent MMS at m/z 44 decreases with increasing filament temperatures. It was found that the intensity of the peak at m/z 44 started to decrease at a filament temperature of 1000 °C, and simultaneously, two new peaks appeared at m/z 88 and 90, respectively, after the filament was turned on for 3 min and longer. This temperature is similar to that for DMS. However, it is lower compared to the lowest temperature of 1400 °C for the decomposition of TMS26 and 1200 °C for TriMS.24 This is probably due to the increasing number of Si−H bonds present in the molecule, which facilitates its dissociative adsorption on the filament to produce the reactive intermediates for the 3868

dx.doi.org/10.1021/jp502795u | J. Phys. Chem. A 2014, 118, 3866−3874

The Journal of Physical Chemistry A

Article

Figure 2. 10.5 eV SPI TOF mass spectra of 0.48 Torr of MMS in the reactor at Ta filament temperatures between 900 and 1600 °C after the filament was turned on for 2 min.

secondary gas-phase reactions. With increasing filament temperatures, the two dominant product peaks at m/z 88 and 90 increased in their intensities, reaching the highest intensity at 1300 °C for the filament-on time of 2 min. Along with these two products, other new peaks were observed at m/z 2 (H2), m/z 132, m/z 134, m/z 176, and m/z 178 at increasing temperatures. The product distributions depend also on the filament-on time. Figure 3 shows the intensity distributions of the two dominant product peaks at m/z 90 and 88 as a function of time for the temperatures of 1000 and 1200 °C. At 1000 °C when the formation of these two products started, the peak intensities first increase with time and reach a plateau at t ≥ 20 min, indicating a net formation at the beginning before reaching a steady state. As the filament temperature is increased to 1100−1400 °C, the intensity decreased after about 1−3 min of turning on the filament, as represented for 1200 °C in Figure 3b. At temperatures higher than 1500 °C, all product peaks disappeared rapidly within the first 3 min after the filament was turned on. 2.1. Dominant Involvements of Silylene Intermediates. The peak at m/z 90 is the dominant product in the high-mass region throughout the entire temperature range and reaction time when using MMS in the reactor. This peak is assigned to 1,2-dimethyldisilane (DMDS). Two reactions can be responsible for the formation of this species: (1) insertion of methylsilylene into the Si−H bond in the parent MMS molecule as represented by eq 2 and (2) a recombination reaction of methylsilyl radicals (from Si−H bond cleavage), as represented by eq 5. ̇ 2 + CH3SiH ̇ 2 → CH3SiH 2SiH 2CH3 CH3SiH (5)

Figure 3. Intensity distributions of the peaks at m/z 88 and 90 as a function of reaction time when using 0.48 Torr of MMS and at a Ta filament temperature of (a) 1000 and (b) 1200 °C.

Another source for DSCB can be from the dimerization of 1-methylsilene (eq 7), where 1-methylsilene is a possible coproduct of H2 from the 1,2-H2 elimination (eq 4)

To confirm the assignments of the two peaks at m/z 90 and 88, experiments with the MMS-d3 sample were carried out under the same experimental conditions. Figure 4 shows the mass spectrum of 12 Torr of 4% MMS-d3/He mixture recorded at a filament temperature of 1300 °C after 2 min of reaction time. When MMS was replaced with MMS-d3, the peak at m/z 90 was clearly shifted to m/z 94, and the dominance of this peak is preserved in the spectrum. Both methylsilylene insertion (eq 2) and free radical recombination (eq 5) reactions produce the same deuterated 1,2-dimethyldisilane, CH3SiD2SiD2CH3 (DMDS-d4), with a mass of 94 amu. Therefore, the isotope labeling experiment is, in general, useful in confirming the identity of the peak at m/z 90 to be DMDS. However, it is not helpful in differentiating the exact mechanism for its formation. Fortunately, the two pathways leading to the formation of the

Aside from the insertion reaction, methylsilylene can also dimerize to form 1,2-dimethyldisilene, which then isomerizes to 1,3-disilacyclobutane (DSCB, m/z 88).33

3869

dx.doi.org/10.1021/jp502795u | J. Phys. Chem. A 2014, 118, 3866−3874

The Journal of Physical Chemistry A

Article

Figure 4. 10.5 eV SPI TOF mass spectrum of 0.48 Torr of MMS-d3 in the mass region of 70−98 amu recorded 2 min after the Ta filament was turned on at 1300 °C.

peak at m/z 88 show different mass shifts when using MMS-d3. Specifically, if the peak was formed by methylsilylene dimerization according to eq 6, switching to MMS-d3 should result in a mass shift of 2 amu and a mass peak at m/z 90 should be detected. On the other hand, silene dimerization (eq 7) will lead to a mass shift of 4 amu when using MMS-d3. As illustrated in Figure 4, the peak at m/z 90 was obviously detected, whereas the one at m/z 92 was barely discernible from the baseline. This strongly suggests that dimerization of methylsilylene is the major channel for the formation of the DSCB molecule (m/z 88), and that dimerization of silene does not play a significant role in the formation of DSCB. As mentioned above, the formation of DMDS could originate from either the silylene insertion reaction (eq 2) or the free radical recombination reaction (eq 5). In order to isolate the mechanism responsible for the formation of DMDS, 1,3-butadiene was used to trap the silylene species.34,35 A five-membered-ring adduct, 1-methyl-silacyclopent-3-ene with a mass of 98 amu, forms as a result of the reaction between 1,3-butadiene and methylsilylene.36

Figure 5. 10.5 eV SPI TOF mass spectra of (a) 12 Torr of 4% DMS + 32% 1,3-butadiene in He in the reactor at a Ta filament temperature 1300 °C and 4 min after turning the filament on (note: the peak at m/z 54 is overloaded). Inset: Intensity of peak at m/z 98 from this sample versus the filament-on time at a filament temperature of 1300 °C and (b) 12 Torr of 32% 1,3-butadiene in He in the reactor at a Ta filament temperature of 1300 °C and 5 min after turning the filament on. Inset: Zoomed-in picture in the mass region of 20−110 amu. The peak at m/z 54 is overloaded.

The experiments were performed using the mixture of MMS with 8-fold of 1,3-butadiene. Prior to each run with the mixture, a control experiment with 4% MMS and 32% 1,3-butadiene trapping agent, respectively, was done separately as a reference under the same experimental conditions. Figure 5a shows the TOF mass spectrum of 12 Torr of 4% MMS: 32% 1,3-butadiene in He recorded at a Ta filament temperature of 1300 °C after 4 min of reaction time. The control experiment using 32% 1,3-butadiene performed under the same experimental conditions shows that this molecule does not decompose to form new species at 1300 °C when the filament-on time is shorter than 10 min, as illustrated in Figure 5b) for the 5 min time. At longer filamenton time, new peaks appear at m/z 66, 68, 70, 78, 80, and 92 from the decomposition of 1,3-butadiene at 1300 °C. From the examination of the mass spectra collected during the trapping experiment with 1,3-butadiene, it was noticed that the DMDS peak at m/z 90, along with all other high-mass peaks disappeared completely in the presence of 1,3-butadiene. At the same time, a new peak at m/z 98, representing the adduct

of 1,3-butadiene and methylsilylene (eq 8), was observed. The intensity of this adduct peak increases with increasing reaction time, as shown in the inset of Figure 5a. Therefore, the formation of DMDS is unambiguously proven to be due to the insertion of methylsilylene into the Si−H bond of the MMS molecule. The recombination reaction of methylsilyl radicals can be ruled out. Our results agree very well with the previous studies13−15 on the pyrolysis of MMS where DMDS was also observed as a product originating from the methylsilylene insertion reaction. Aside from silylene, 1,3-butadiene also reacts with the silene species.35,37 The reaction between 1,3-butadiene and silene leads to the formation of silacyclohex-3-ene with a mass of 98 amu, as represented in eq 9.

3870

dx.doi.org/10.1021/jp502795u | J. Phys. Chem. A 2014, 118, 3866−3874

The Journal of Physical Chemistry A

Article

species also react with the filament to form metal silicides40,41 and carbides.42,43 The nature of the metal alloys formed was not characterized in this work with MMS; however, visible changes were observed in the appearance of the filaments after their exposure to MMS. In addition, there was an increase in the electric power supplied to maintain a constant temperature during the 0.5−1 h period in our experiments, indicating a change in the filament composition. The alloying causes the filament to age and affects the reaction rates in the reactor. At short reaction time and low filament temperature, the gas-phase reactions are fast and filament alloying is relatively slow, therefore there is a net formation of the two products of DMDS and DSCB. This causes an increase in the peak intensity at m/z 90 and 88 with time and temperature. With increasing time, the filament alloying process consumes the reactive intermediates and also slows down the gas-phase reactions since the metal alloy is not as good a catalyst as the metal itself. This leads to a decrease in the product peak intensities at a later time and higher temperatures. Similar observations have been obtained in our previous study on the reaction chemistry with 1,1,3,3tetramethyl-1,3-disilacyclobutane.39 2.2. Formation of H2 Molecules. All previous studies on the pyrolysis of MMS13−19 have agreed on H2 as the major product, although other products vary. In the HWCVD reactor with MMS as a precursor gas, the formation of H2 has been observed. As shown in Figure 2, a peak at m/z 2, representing H2 appeared at relatively high temperatures. At a filament-on time of 2 min, the H2 peak started to appear at 1200 °C and reached a maximum at 1400 °C. The H2 peak was weak since its high IP of 15.4 eV44 is greater than the ionization photon energy of 10.5 eV. The observation of this peak is due to the minor UV photon-induced EI present in the SPI mode. The intensity ratio of the H2 peak (m/z 2) to that of DMDS (m/z 90) ranges from 0.1 to 4.9 at different filament temperatures and reaction time. The IE of DMDS was calculated to be 8.75 eV (