Silicon atom number density profiles have been measured using laser-induced fluorescence during the chemical vapor deposition of silicon from silane. Measurements were obtained in a rotating-disk reactor as a function of silane partial pressure and the amount of hydrogen added to the carrier gas. Absolute number densities were obtained using an atomic absorption technique. Results were compared with calculated density profiles from a model of the coupled fluid flow, gas-phase and surface chemistry for an infinite-radius rotating disk. An analysis of the reaction mechanism showed that the unimolecular decomposition of SiH2 is not the dominant source of Si atoms. Profile shapes and positions, and all experimental trends are well matched by the calculations. However, the calculated number density is up to 100 times smaller than measured.

This paper presents a summary of the current state of our understanding of the absolute reactivity of transient silicon hydride species, such as SiH, SiH2 and SiH3 in the gas phase and at the surface of thin films.

This paper describes the application of ab-initio molecular orbital (MO) theories in conjunction with reaction rate theory to obtain thermochemistry and energetics of nucleation processes. The specific example used for the illustration of this approach is the nucleation of SiO. MO computations on the equilibrium structures have shown the polymers up to the tetramer to be planar rings and exothermic to addition of the monomer. Transition state analysis has shown that subsequent addition of the monomer (SiO) most likely proceeds without an energy barrier. Reaction rate theory analysis of the polymerization shows the rate coefficients to be very pressure dependent, with the dimer formation process rate limiting. Oxidation of clusters however showed a substantial barrier which should result in oxygen deficient clusters.

This study explored the surface chemistry and the promise of the binary reaction scheme:

(A) Si-OH+SiCl4 → Si-Cl + HCl

(B) Si-Cl + H2O → Si-OH + HCl

for controlled SiO2 film deposition. In this binary ABAB… sequence, each surface reaction may be self-terminating and ABAB… repetitive cycles may produce layer-by-layer controlled deposition. Using this approach, the growth of SiO2 thin films on Si(100) with atomic layer control was achieved at 600 K with pressures in the 1 to 50 Torr range. The experiments were performed in a small high pressure cell situated in a UHV chamber. This design couples CVD conditions for film growth with a UHV environment for surface analysis using laser-induced thermal desorption (LITD), temperature-programmed desorption (TPD) and Auger electron spectroscopy (AES). The controlled layer-by-layer deposition of SiO2 on Si(100) was demonstrated and optimized using these techniques. A stoichiometric and chlorine-free SiO2 film was also produced as revealed by TPD and AES analysis. SiO2 growth rates of approximately 1 ML of oxygen per AB cycle were obtained at 600 K. These studies demonstrate the methodology of using the combined UHV/high pressure experimental apparatus for optimizing a binary reaction CVD process.

Gas-phase Fourier transform infrared spectroscopy has been used to monitor reactants and products in the study of O3 decomposition and the reaction of O3 with tetramethylsilane and tetraethoxysilane. The results confirm the interpretation that O3 decomposition is heterogeneous, and a dominant factor in SiO2 deposition using these chemistries. Product analysis shows that the higher reactivity of TEOS with O3 is probably due to the ability of TEOS to undergo α-hydride elimination of one ethoxy group, per molecule, and one oxidative acetic acid elimination, per molecule, during the deposition process. Results also suggest elimination via the Sicenter involving a single ethoxy group rather than via a 6-center elimination involving two ethoxy groups.

Diethyldiethoxysilane (DEDEOS) is a potential single molecular precursor for SiO2 growth. Additionally, DEDEOS may be applicable in a deposition scheme that will allow precise atomic layer control of SiO2 film thicknesses. Deposition of SiO2 on silicon using DEDEOS was studied in three regimes. In the first regime, the adsorption and decomposition of DEDEOS were studied using temperature-programmed desorption (TPD) and Fourier-transform infrared spectroscopy (FTIR) under UHV conditions. TPD and FTIR results indicated that DEDEOS deposits ethyl and ethoxy species on the surface which decompose by a β-hydride elimination mechanism. In the second regime, step-wise SiO2 growth using DEDEOS was studied by laser-induced thermal desorption (LITD) and TPD as a function of repetitive growth cycles composed of DEDEOS deposition at 300 K followed by thermal annealing to 820 K. The amount of SiO2 growth per cycle initially decreased as a function of cycle until reaching a constant value after approximately 10 cycles. Finally, DEDEOS deposition of SiO2 was studied in a high pressure chamber at surface temperatures between 873 K and 1003 K and at a pressure of 0.5 Torr. The deposition of SiO2 as a function of exposure was found to display a fast initial growth step followed by a slower step which continued indefinitely. The activation energy for SiO2 growth by DEDEOS in the slower growth step was found to be 49 ± 6 kcal/mol. This activation energy is very similar to the activation energy of 45 kcal/mol observed for SiO2 growth by tetraethoxysilane (TEOS).

Thermal decomposition of tetraethoxysilane(TEOS) has been investigated in a vertical type LP-CVD reactor. The mass spectrum of reacting gas was found to have a feature corresponding to hexaethoxysiloxane ((H5C2O)3SiOSi(OC2H5)3). Trapped byproducts were found to include ethanol and ethoxy-based silica. Kinetic analysis of deposition rate profile indicated the existence of two intermediates. Feed gas residence time, as well as substrate temperature. directly affected the film quality.

A vapor phase reaction model was proposed to explain these results. According to this model. TEOS decomposes into ethylene and triethoxysilanol. The latter reacts with TEOS to form hexaethoxysiloxane as observed by mass spectrometry. The triethoxysilanol, hexaethoxysiloxane and TEOS should be actual deposition precursors. Similar reaction continues to form ethoxybased siloxane polymers which are trapped from the exhaust gas.

In order to understand and successfully model the plasma processing used in device fabrication, it is important to determine the role played by plasma-generated radicals. We have used the IRIS technique (Imaging of Radicals Interacting with Surfaces) to obtain the reactivity of NH(X3∑-) and OH(X2II) at a silicon nitride film surface while the film is exposed to a plasma-type environment. The reactivity of NH was found to be zero both during exposure of the surface to an NH3 plasma and during active deposition of silicon nitride from a SiH4/NH3 plasma. No NH surface reaction was detectable for any rotational states of NH and over a surface temperature range of 300-700 K. OH radicals generated in an H2O plasma were found to have a reactivity of 0.57 on a room temperature oxidized silicon nitride surface. The OH reactivity falls to zero as the temperature of the substrate is raised.

Insight into how dopants can enhance deposition rates has been obtained by comparing the reactivities of tetraethyl orthosilicate (TEOS, Si(OCH2CH3)4) with silanol and boranol groups on SiO2. This comparison has direct relevance for boron-doped SiO2 film growth from TEOS and trimethyl borate (TMB, B(OCH3)3) sources since boranols and silanols are expected to be present on the surface during the thermal chemical vapor deposition (CVD) process. A silica substrate having coadsorbed deuterated silanols (SiOD) and boranols (BOD) was reacted with TEOS in a cold-wall reactor in the mTorr pressure regime at 1000K. The reactions were followed with Fourier transform infrared spectroscopy. The use of deuterated hydroxyls allowed the consumption of hydroxyls by TEOS chemisorption to be distinguished from the concurrent formation of SiOH and BOH that results from TEOS decomposition at this temperature. It was found that TEOS reacts with BOD at twice the rate observed for SiOD, given equivalent concentrations of BOD and SiOD. This demonstrates that hydroxyl groups bonded to boron increase the rate of TEOS chemisorption. In contrast, additional results show that surface ethoxy groups produced by the chemisorption of TEOS decompose at a slower rate in the presence of TMB decomposition products. Possible dependencies on reactor geometries and other deposition conditions may determine which of these two competing effects will control deposition rates. This has significant implications for microelectronics fabrication since the specific dependencies would be expected to affect process reliability. In addition, this may explain (in part) why the rate enhancement effect is not always observed in boron-doped SiO2 CVD processes.

Recently, TiSi2 has been the object of considerable study because of its low resistivity among the transition metal silicides and its compatibility with existing ULSI technology [1,2]. Film growth by CVD offers the potential for selective area deposition and high production throughput. However, selective CVD of TiSi2 from gas phase SiH4 and TiCl4 is usually accompanied by a competing reaction which consumes intolerable amounts of the Si substrate [3,4]. Controlling this consumption is crucial in TiSi2 growth; however, no quantitative correlation exists between silicon consumption and growth conditions or film thickness. Additionally, the reaction mechanism for TiSi2 growth is poorly understood, and some disagreement even exists about the reaction stoichiometry [5,6]. The combined CVD/UHV approach we have developed fills many gaps in the current understanding of TiSi2 CVD.

The thermal desorption of hydrogen from a Si(100)-(2×l) monohydride surface has been reported previously to follow first-order kinetics. Pairing of hydrogen atoms on a Si-Si dimer prior to desorption is most likely the cause of this unusual behavior. To examine the degree of hydrogen pairing at various surface temperatures, this study examines the blocking of adsorption sites by hydrogen for subsequent acetylene adsorption. Temperature programmed desorption and Auger electron spectroscopy were used to measure the absolute saturation coverage of acetylene for various coverages of preadsorbed atomic deuterium. The observed linear decrease in saturation coverage of acetylene with deuterium coverage, for atomic deuterium adsorption between 550 and 150 K, indicates that deuterium is paired on Si-Si dimers. The observation that pairing occurs even at 150 K suggests that the diffusion of chemisorbed hydrogen is not responsible for pairing.

Classical Molecular dynamics simulation of silicon cluster growth (up to 1000 atoms) have been conducted using the Stillinger-Weber 3-body interaction potential. The cluster binding energy has been fit to an expression that separates the surface and bulk contribution to the energy over a wide temperature and size range. Cluster growth simulations show that large heat release results from new bond formation at gas kinetic rates (i.e. sticking coefficient = unity). Temperature was found to be the primary controlling process parameter in the evolution of cluster morphology from an aggregate to a coalesced cluster below 1000 K, with the impact parameter playing a secondary role.

Microcrystalline silicon films were formed at room temperature without hydrogen dilution by ECR PECVD. Microwave power more than 400 W was necessary to get crystalline films and the crystallinity increased with the power thereafter. Addition of hydrogen and argon enhanced the crystalline phase formation and the deposition rate, the reason of which was found that hydrogen etched silicon films and argon addition drastically increased the etch rate. Annealing of the films showed that microcrystalline silicon films formed by ECR PECVD have a small fraction of amorphous phase. TFT's using silicon nitride and doped/undoped microcrystalline silicon films were fabricatedd with whole processes at room temperature.

Silicon films were deposited on a fluororesin surface. The process was divided into two steps: surface modification process and silicon CVD onto the modified parts. In the modification process, SiH4 and B(CH3)3 mixed gases were used with ArF excimer laser. Fluorine atoms of the surface were pulled out by boron atoms which were photo—dissociated from B(CH3)3 and were replaced with silicon atoms released from SiH4. In the CVD process, SiH4 gas was used with high—density excited ArF excimer laser. Silicon films were deposited onto the nuclei by photodecomposition of SiH4.

Chemical compositions of the modified layers and the deposited parts were inspected by XPS analysis. 1000 Å thickness of the deposited silicon films was confirmed by the surface roughness interference–meter.

Maskless deposition of silicon from silane on Si monocrystalline wafer using copper bromide vapor laser (CBVL) is investigated. Morphology and geometric parameters of the stripes obtained are studied and some conclusions for the process mechanism are made. Applying Kirchoff's transform and Green's function analysis nonlinear heat diffusion problem for different pulse shapes was solved. The influence of pulse shape on the temperature distribution and its time evolution was studied. Nonisothermal and non-stationary deposition kinetic models using the obtained results were developed.

The effect of post adsorbed atomic hydrogen on the adsorption, desorption, and decomposition of ethylene on Si(100)-(2×l) has been studied using high-resolution electron energy loss spectroscopy (HREELS), temperature programmed desorption (TPD), and low-energy electron diffraction (LEED). Exposures to atomic hydrogen of more than 1015 atoms/cm2 convert the initial (2×l) reconstruction of sp3-hybridized, di-σ bonded ethylene to a (l×l) structure. Furthermore, after post exposure to atomic hydrogen, the thermal desorption peak of molecular ethylene is shifted up by approximately 100 K and reduced in intensity. HREEL spectra for deuterated ethylene show the formation of a C-H bond after exposure to atomic hydrogen, whereas the C-C bond remains intact. We explain our data by an atomic hydrogen-driven conversion of the di-σ bonded ethylene to a mono-σ bonded surface ethyl. Thermal activation after post exposure to atomic hydrogen leads to decomposition of about 60% of the initial ethylene in contrast to the observed molecular desorption in the absence of hydrogen.

Experimental measurements of the decomposition of methyltrichlorosilane (MTS), a common silicon carbide precursor, in a high-temperature flow reactor are presented. The results indicate that methane and hydrogen chloride are major products of the decomposition. No chlorinated silane products were observed. Hydrogen carrier gas was found to increase the rate of MTS decomposition. The observations suggest a radical-chain mechanism for the decomposition. The implications for silicon carbide chemical vapor deposition are discussed.

We examine the dynamic behavior of a heterogeneous reaction model for the chemical vapor deposition of carbon, silicon, and silicon carbide from the precursors generated by the thermal decomposition of methyltrichlorosilane (MTS, CH3SiCl3 ). All reactions are treated as reversible in order to account for the strong inhibitory effect of the reaction by–products on the deposition process that was observed in our experiments and in other studies. The equilibrium constants of the adsorption steps of the reactions are treated as model parameters, and those of the other reactions are calculated from the thermodynamic constants of a set of overall deposition reactions. Results are presented on the influence of the various model parameters on the reaction rate, the stoichiometry of the deposit, and the variation of these quantities with the distance in a plug flow deposition reactor.

We describe the first measurements of reaction kinetics between diamond and a gas phase species—H atoms—involved in its formation. We develop a remarkably simple method to measure H atom concentrations and use the method to measure γd, the destruction probability of H atoms on diamond at 20 torr and 1200 K. We find γd = 0.12. This value is close to that estimated from gas phase alkane rate constants.

A novel reactor has been constructed to investigate the fundamental surface kinetics of diamond chemical vapor deposition(CVD). The molecular beam reactor permits independent control of the atomic hydrogen flux, the methyl radical flux and the substrate temperature. The low pressure in the growth chamber (≈ 1 mTorr) minimizes the impact of gas-phase chemistry. The reactive mixture impinging the substrate is sampled through an orifice and analyzed by mass spectroscopy. Differential pumping in the two adjacent chambers quenches the beam, allowing quantitative analysis of radical species such as H and CH3. In preliminary experiments deposition was achieved onto seeded molybdenum substrates.