Equilibrium dopant incorporation in CVD silicon epitaxy

In-situ doping of epitaxial silicon layer with arsenic is controlled either by means of a mixed equilibrium – kinetical mechanism characterised by the influence of layer growth rate, or by means of a pure equilibrium-related mechanism showing no layer growth rate effect. Layer growth rate can be selected to mark the transition from one mode to the other by introducing a characteristic rate v. Equilibrium controls dopant incorporation when v_epi 〈 v 〉 0 is ture. That critical rate depends on temperatur as well as on input partial pressure of arsine and can be expressed by the three different empirical quantities A, B, and E. The former two of them are defined by conventional dopant incorporation theory. The additional quantity E is a function of arsine input pressure. That function not only defines an upper limit E= E_∞ but also a lower limit E= 0 at which v= 0 is valid.     

Optimisation of low-temperature silicon epitaxy on seeded glass substrates by ion-assisted deposition

Using single crystalline Si wafer substrates, ion-assisted deposition (IAD) has recently been shown [J. Crystal Growth 268 (2004) 41] to be capable of high-quality high-rate epitaxial Si growth in a non-ultra-high vacuum (non-UHV) environment at low temperatures of about 600 °C. In the present work the non-UHV IAD method is applied to planar borosilicate glass substrates featuring a polycrystalline silicon seed layer and carefully optimised. Using thin-film solar cells as test vehicle, the best trade-off between various contamination-related processes (seed layer surface as well as bulk contamination) is determined. In the optimised IAD process, the temperature of the glass substrate remains below 600 °C. The as-grown Si material is found to respond well to post-growth treatments (rapid thermal annealing, hydrogenation), enabling respectable open-circuit voltages of up to 420 mV under 1-Sun illumination. This proves that the non-UHV IAD method is capable of achieving device-grade polycrystalline silicon material on seeded borosilicate glass substrates.     

Inelastic light scattering studies of silicon chemical vapor deposition (CVD) systems

Raman and resonance fluorescence spectra, determined by inelastic light scattering measurements, are used to identify molecular species and to measure their concentration gradients on a fine spatial scale throughout a CVD reactor. Raman spectra are also analyzed to give gas temperature and tempetature profiles. The temperature profiles near the leading edge of a horizontal rf heated susceptor in a laminar flow system are adequately described by using Lévêque's solution to an energy balance equation, assuming temperature-independent fluid properties. Raman spectra at room temperature of some of the compounds commonly used as source materials for Si epitaxial growth, SiCl₄, SiCl₃H, SiCl₂H₂ and Si₂Cl₆ indicate that these species are all detectable at the 10–100 ppm level and are distinguishable from each other. Measurements at 500–1300°C of these compounds reveal the presence of a common species, SiCl_x (probably SiCl₂), which exhibits a resonance flourescence spectrum at least 1000 times more intense than typical Raman spectra. SiCl_x density profile measurements above the susceptor indicate a concentration boundary layer thickness of 0.7-0.8 cm for one set of experimental conditions. SiCl_x density measurements as a function of suspector temperature are observed to vary over a range of 4 to 5 orders of magnitude, and are much higher for a SiCl₂H₂ input than for a SiCl₄ input.     

Heterogeneous kinetics and mass transport in chemical vapour deposition processes: Part II application to silicon epitaxy

The results obtained for the epitaxial deposition of Si from SiC1₄ and H₂ in a rotating disc reactor are described by a single equation in terms of kinetics and transport. The kinetic parameters required to fit the experimental data suggest that the rate determining step for the CVD process is the homogeneous conversion of SiC1₄ to SiHCl₃. The kinetic analysis also shows that the commonly reported activation energy for the overall growth is too low as a result of measurements being made under conditions of combined transport and kinetics. Results of other workers obtained with reactors of very different geometries, when normalised, are in excellent quantitative agreement with both the experimental and theoretical curves found in this work.     

Control of slip in horizontal silicon epitaxy with profiled susceptors

The important point in controlling slip in silicon epitaxy is to avoid the build-up of mechanical stress that exceeds the critical value at which the stress is relieved by slip. This critical value depends on the slice perfection and, in particular, is lowered by the presence of local flows. On the other hand, the temperature gradient during the deposition process must be controlled in order to keep the mechanical stress below this critical value. The design of profiled susceptors, to achieve a constant temperature profile across the slice during epitaxy, has been studied. The size and the shape of the pockets are rather critical, and depend on the specific resistivity of the susceptor; the frequency of the rf heating system; the susceptor dimensions; and the slice parameters.     

Transport phenomena analysis for boron CVD in a horizontal cold-wall reactor

The transport phenomena in a horizontal cold-walled semicircular reactor are analyzed for the CVD of boron from BCl₃ and H₂. The mixed problem of energy, momentum, and mass conservations is solved by a simple finite difference method. The concentration of the B-reactant on the deposition surface is substituted by the sum of equilibrium mole fractions of the B-containing gas species. The profiles of temperature, velocity, and reactant concentrations in the CVD reactor are illustrated, and the boron deposition rate profile along the substrate is predicted. The effect of the reactant input composition on the deposition rate is calculated, and compared with the experimental data.     

On the adsorption- and equilibrium-referred interpretation of the doping element incorporation during the growth of CVD epitaxial silicon

For investigating the partial steps of CVD-silicon doping: gas transport, adsorption-desorption, incorporation, a solution reflecting the kinetic influence of the layer growth in a satisfactory way, is obtained only if specific forward and backward reactions of the incorporation step are taken into account. Regardless of assuming a thermodynamical decomposition equilibrium for the doping source material and the incorporation equilibria for each of the individual components of the decomposition equilibrium, these components are not interchangeable in representing the total incorporation flow of dopants. The incorporation process is dominated by that component which exhibits the largest relative equilibrium partial pressure.     

Modeling particle growth and deposition in a tubular CVD reactor

A computational model is developed to simulate the particle formation, growth, and deposition in a tubular CVD reactor. The model takes into account the momentum, heat, and mass transfers, chemical reaction, Brownian coagulation, Brownian diffusion, thermal diffusion, and thermophoresis that occur within the reactor. The chemical system is illustrated by the formation of TiO₂ particles through oxidation of TiCl₄. A set of coupled partial differential equations is solved with the finite volume method to give spatial distributions of velocities, pressure, temperature, TiCl₄ concentration, and sizes of TiO₂ particles, from which the deposition mass fluxes and size distributions of exit TiO₂ particles can be readily evaluated. In the model, the growth of particles is simulated by the discrete-sectional model, which can cover particle growth from monomers up to the submicron size. Among other things, the simulation predicts a maximum deposition flux near the reactor entrance and a rebound in deposition flux near the reactor exit, which have been observed in several experimental studies. The model also reveals a delay in evolution of size distribution near the reactor wall as compared to that of the bulk. This delay signifies that the size distribution of depositing particles at the wall is in general different from that of the bulk, which normally is the measured one. The effects of important operating parameters such as temperature, pressure, inlet TiCl₄ concentration, and inlet stream velocity on various system performance indicators are also investigated.     

Influence of boron doping on roughness microcrystalline silicon

We have studied roughness of boron-doped microcrystalline Si (μc-Si) surfaces with an emphasis on the influence of heavy doping. μc-Si films were prepared using plasma-enhanced chemical vapor deposition (PECVD) with different boron concentrations in gas phase from 0% to 2%. Growth-induced roughening of μc-Si surfaces was monitored ex situ using an atomic force microscope (AFM). With an increase in the deposition time, the surface width (rms roughness), w, of undoped μc-Si surface exhibited usual behaviors; first, (a) w increased, (b) slightly dropped, (c) rose again, and then (d) gradually increased. In the case of B-doped μc-Si, w differently behaved; (a) w increased very soon, (b) slightly dropped, (a′) rose again, (b′) slightly dropped again, (c) rose, and finally (d) gradually increased. The quick increase in w indicates that boron doping promotes the nucleation, and the repeated nucleation is responsible for the behavior (a′)–(b′). Additionally, the nucleation density, that was derived using the lateral correlation length of surface heights, monotonically increased with an increase in the boron concentration. The effects of boron doping are discussed with the catalytic effects and the formation of the surface-covering layer.     

Doping process control in silicon epitaxy (I). System identification

The transition behaviour of the phosphorus incorporation in silicon epitaxial layers grown in a CVD reactor has been investigated, considering the reactor as a linear control system with u = lg p⁰_PH3 (t) as the input and y = lg N(t) as the output. The response of system to both upward and downward step inputs has been studied experimentally, using SiH₄ and PH₃ sources. The dopant system of a horizontal silicon epitaxial reactor has been identified and a mathematical model relating to the transient behaviour of the system has been found. The parameters of the model have been estimated from layer growth experiments. The step response functions found can be approximated by an exponential function relating to n time constants T, all equal to each other. It was found for the system investigated that the second order model is of sufficient accuracy for the optimal control calculations described in the next part of this series.     

Equilibrium and kinetics in the chemical vapour deposition of silicon

In the growth of silicon layers on various substrates it appears that the growth rate is not uniquely determined by substrate temperature and input parameters of the gases used. An analysis of this situation is given and essential points in the chain reaction of steps in the growth process will be indicated. Differences in growth rate reported for different experimental situations can be explained on the basis of this analysis where the temperature gradient normal to the growing interface will appear to be of special importance. Some examples are given of resulting surface morphology as a function of growth and etch conditions. Another point of interest in the growth process is the incorporation of dopant which determines the electrical properties of the layers. Here also equilibrium and kinetics show an interplay. For low growth rates there appears to be a good correlation between the concentration of impurities in the solid and the partial pressure of dopant in the gas phase. For growth rates exceeding a critical-value kinetic effects can be expected as those found in liquid phase epitaxy. It appears that an n-type dopant as phosphorus shows this effect. In this case the surface concentration of ionized donors exceeds the bulk concentration of these centres and trapping occurs at higher growth rates.     

Improvements in silicon doping of InP and GaInAs in metalorganic molecular beam epitaxy

We have investigated the Si doping of InP and GaInAs in metalorganic molecular beam epitaxy (MOMBE) by using a conventional Si effusion cell. In order to reduce the formation of SiC promoted by the background gases in MOMBE, we introduced a liquid nitrogen cooled baffle between the cell and the mechanical shutter. The results show that the passivating reaction can be substantially suppressed by a proper treatment of the source cell. The doping efficiency remains constant over a long period of operation corresponding to a large total layer thickness (>100 μm). The comparison of SIMS analysis with Hall data reveals an electrical activation of Si in InP up to 100% and about 65% for Si in GaInAs. These results and the investigations on doping profiles show that Si is a suitable donor in InP and GaInAs in the MOMBE process.     

Mechanisms of chemical vapor deposition of silicon

The distribution in the silicon epitaxial growth from SiCl₄ and hydrogen are observed in situ by IR absorption spectroscopy. Two methods are used complementarily, one is IR spectroscopy of reactants extracted from the reactor by a fine quartz tube which is not disturbing the reactions, and gives knowledge about the local distribution, the other is direct IR spectroscopy of hot reactants in the reactor which is useful to ascertain the results at the real high temperature situation. The intermediate species are SiHCl₃, SiH₂Cl₂ which is estimated from the induced emission bands at 500 and 570 cm^-1. HCl is a dominant waste product and contributes to reverse reactions. To investigate the reaction, HCl is intentionally injected into the reacting gas. This kind of injection method may also be very effective to analyze the reactions using other reactants such as SiCl₄, SiHCl₃ and SiH₂Cl₂.     

Cracking assisted nucleation in chemical vapor deposition of silicon nanoparticles on silicon dioxide

Deposition of sub-monolayer silicon on SiO₂/Si(1 0 0) greatly facilitates nucleation in subsequent thermal chemical vapor deposition (CVD) of silicon nanoparticles. Sub-monolayer seeding is accomplished using silicon atoms generated via disilane decomposition over a hot tungsten filament. The hot-wire process is nonselective towards deposition on silicon and SiO₂, is insensitive to surface temperature below 825 K, and gives controlled coverages well below 1 ML. Thermal CVD of nanoparticles at 1×10⁻⁴ Torr disilane and temperatures ranging from 825 to 925 K was studied over SiO₂/Si(1 0 0) surfaces that had been subjected to predeposition of Si or were bare. Seeding of the SiO₂ surface with as little as 0.01 ML is shown to double the nanoparticle density at 825 K, and densities are increased twenty fold at 875 K after seeding the surface with 30% of a monolayer.     

Synthesis and characterization of boron silicon oxycarbide glass fibers

Boron silicon oxycarbide fiber mats have been obtained through pyrolysis in inert atmosphere of sol–gel derived hybrid materials. The characterization of the preceramic hybrid fibers and the pyrolysis process reveals different boron incorporation depending on the composition. The ceramic conversion involves a series of reactions governed by the presence of the mixed bonds that implies larger phase segregation and higher graphite nanodomain size with the decrease of the number of borosiloxane bridges in the preceramic material. The formation of a mixed structure, where both Si and B are taking part on the ceramic network occurs with the excess of B that precipitates as boric acid in the surface of the fibers. These precipitates can cause pores in the surface of the pyrolyzed fiber.     

Doping process control in silicon epitaxy (II). Calculation of optimum control

A procedure is developed to obtain desired dopant profile in epitaxial layer growth. Based an the results of system identification, linear-quadratic optimal control theory is used to determine the optimal input PH₃ concentration as a function of time. In the performance index an auxiliary weighting coefficient must be incorporated. It is discussed how to select this weighting coefficient.     

New insights in microcrystalline silicon deposition with expanding thermal plasma chemical vapor deposition

We report on further insights in the microcrystalline silicon (μc-Si:H) deposition using expanding thermal plasma chemical vapor deposition. We have shown before that the refractive index at 2 eV of μc-Si:H layers increased if the silane (SiH₄) was injected close to the substrate, while the deposition rate remained the same. We argued that at high injection-ring position, the SiH₄ travels a long way to the substrate and therefore has a long interaction time with the plasma, in particular atomic hydrogen. In this way, the SiH₄ injection position influences the number of hydrogen atoms stripped from the SiH₄ as well as the consumption of atomic hydrogen. In this paper, we present an analysis of the growth flux of depositing particles as function of the radical production rate. The data suggest that there is no dependence on the SiH₄ injection position, implying that the mix of depositing radicals is not changed. However, the data also show the microcrystalline-to-amorphous transition shifts to higher SiH₄ flows for lower injection positions. We therefore now think that it is not the interaction time between the SiH₄ and the arc plasma determining the material properties, but the interaction of excess atomic hydrogen with the μc-Si:H growth surface.     

Suppression of epitaxial silicon layer doping with boron in the presence of hydrogen chloride

The present paper informs about thermodynamic calculations of the B Cl H system, carried out irrespective of data, already published by other authors. It further deals with recent experimental results on silicon doping with boron in the presence of hydrogen chloride, the layer growth rate being varied. Finally the possibility of quantitatively interpreting the suppressing influence of hydrogen chloride on the doping of silicon with boron is discussed.