Structural relaxation of amorphous Si is studied in the temperature range 500-850 °C using Raman spectroscopy. The minumum value for the Raman peakwidth that can be obtained is inversely proportional to the anneal temperature. The relaxation process is basically the same in a-Si prepared by ion implantation and by vacuum evaporation.

Structural relaxation of amorphous Si (a-Si) is thought to be fully controlled by annihilation of point defects quenched in during preparation. The density of defects in unrelaxed a-Si is estimated from a wide variety of published experimental data to amount to several atomic %. The enthalpic and entropic contributions of these defects to the excess Gibbs free energy difference of a-Si over c-Si is estimated. The variation in free energy due to defects may lead to afurther melting temperature depression of 40 - 200 K for unrelaxed a-Si in addition to the 200 K depression already present between c-Si and relaxed a-Si.

The dynamics of a photogenerated electron-hole plasma in pure amorphous silicon (a-Si) in different stages of structural relaxation have been studied with sub-picosecond resolution using pump-probe reflectivity measurements. For high plasma densities (> 1020/cm3) the plasma evolution is dominated by Auger recombination. At lower plasma densities (≈ 1018/cm3) the plasma decays exponentially with a time constant τ, suggesting that carrier trapping dominates in this regime. The decay time τ increases with the temperature at which the a-Si has been annealed, ranging from τ = 1 ps for as-implanted a-Si to τ=14 ps for a-Si annealed at 500 °C. This observation is consistent with a reduction in the number of defects in a-Si upon thermal annealing.

Substrate curvature measurements were used to monitor viscous flow in ion beam sputtered amorphous Si for temperatures ranging from 150 to 400 °C. The viscosity increases linearly with time, characteristic of a bimolecular defect annihilation process. This is consistent with the defects governing viscous flow being dangling bonds. The isoconfigurational activation enthalpy for the viscosity is 1.8 ±.3 eV.

According to a recent model [1], the enhancement in the rate of the solid phase epitaxial regrowth (SPER) of silicon produced by implanted impurities is determined by the superposition of reconstruction at sites that capture neutral, and ionized, three-fold coordinated dangling bond states. Considerable support for this model is derived from experiments on ionization-enhanced SPER in silicon. In this paper we discuss how this dangling bond model (DBM) could be used to determine the densities of neutral dangling bonds and ionized impurities in amorphous silicon from these experimental results. Both densities, determined by a self-consistent calculation, are in good agreement with those measured by other types of experiments. This result provides further support for the DBM and indicates that simultaneous SPER and ESR measurements could make it possible to depth profile low concentrations of ionized impurities in amorphous silicon.

We have measured the effects of hydrostatic pressure on the solid phase epitaxial growth (SPEG) rates of undoped Ge(100) and Si(100) into their respective self-implanted amorphous phases. We found that pressure enhances the growth process in both Si and Ge, with activation volumes equal to -3.3 ± 0.3 cm3/mole for Si and -6.3 ± 0.60 cm3/mole for Ge. The results of this and other experiments are inconsistent with all bulk point-defect mechanisms, but are consistent with all interface point-defect mechanisms, proposed to date for thermal SPEG. A kinetic analysis of the Spaepen-Turnbull dangling bond mechanism shows it to be a highly plausible model for the growth process.

Direct picosecond laser measurements of the critical fluence for melting have been performed for the first time, giving unambiguously consistent differences in the energy required for surface melting of relaxed and unrelaxed amorphous silicon. The different optical coupling cannot account for this variation which can only be explained in term of different melting temperatures. Heating of unrelaxed amorphous silicon samples at temperatures close to the melting point may result in relaxation of the material even when the treatment occurs in the nanosecond time scale. However nanosecond UV irradiation of relaxed and unrelaxed amorphous silicon samples have provided informations on the specific heat of the two amorphous states. The melting temperature of unrelaxed amorphous silicon has been derived independently via both picosecond data and via free energy calculations.

This paper discusses the intrusion of H into a-Si layers during solid phase epitaxy and the effect of this H on the growth kinetics. We show that during annealing in the presence of water vapor, H is continuously generated at the oxidizing a-Si surface and diffuses into the amorphous layer, where it causes a reduction in the epitaxial growth rate. The measured variation of growth rate with the depth of the amorphous/crystal interface is correlated with the concentration of H at the interface. The diffusion coefficient for H in a-Si is determined by comparing measured depth profiles with calculated values. Hydrogen intrusion is observed even in layers annealed in vacuum and in inert gas ambients. Thin (<;5000 Åthick) a-Si layers are especially susceptible to this effect, but we show that in spite of the presence of H the activation energy for SPE derived earlier from thin-layer data is in good agreement with the intrinsic value obtained from thick, hydrogen-free layers.

We made high precision cw laser interferometric measurements of the variations of the rate of solid phase epitaxial regrowth (SPER) of amorphous layers on (100) silicon implanted with both boron and phosphorus. Depth profiles of SPER were correlated with the implanted boron and phosphorus distributions measured by secondary ion mass spectroscopy (SIMS). The results showed that: (1) the minimum (SPER) rate did not occur at the depth where the implanted impurity concentrations were equal; (2) the maximum activation energy for SPER (∼2.9 eV; ≈0.18 eV greater than for SPER in intrinsic Si) occurred at the depth where the regrowth rate was a minimum; (3) the regrowth rates in the dual implanted sample were different from those of the samples doped only with phosphorus at the same net phosphorusc concentration; and (4) the rate at the depth where the impurity concentrations were equal was different from the intrinsic rate. Further interpretation of the results suggests that the SPER rate in the dual implanted samples is equal to the value for intrinsic silicon at a depth where the net ionized impurity concentration is compensated. The SPER rate was a minimum at a depth where the net ionized impurity concentration was slightly p type.

In this work we searched for evidence of low level photoionization effects in the solid phase epitaxial regrowth (SPER) of intrinsic amorphous silicon on (100) silicon during isothermal furnace annealing. We used in situ cw laser interferometry to measure the changes in the rate at 500 °C as the laser power was varied from 20 mW-80 mW. Calculation showed that laser heating increased the sample temperature by a maximum of 6 °C at 80 mW. The measured change of the SPER rate with laser power in this range was always smaller than the change computed from an Arrhenius calculation using the measured activation energy, and the calculated value of the laser-produced increment in the sample temperature. The result indicates that there are negligible low level photoionization effects in silicon SPER.

In this work we measured the functional dependence of the solid phase epitaxial regrowth (SPER) of amorphous silicon on NAI, the concentration of implanted aluminum (p-type). The SPER rates of self-ion amorphized layers in silicon wafers with (100) substrate orientation were measured by in situ high precision, isothermal, cw laser interferometry for temperatures from 470 °C to 550 °C, and concentrations in the range 3×1018 cm−3 ≤NAI≤ 4.7×1020 cm−3 obtained from samples implanted with three different doses.

In the concentration range 3×1018 cm−3 ≤NAI≤ 2.3×1019 cm−3, we observed a “compensation effect” in which, with increasing NAI, the SPER rate decreased below the regrowth rate in intrinsic silicon and the activation energy of SPER increased to 2.85 eV, compared to 2.72 eV for intrinsic silicon. In the range 3.3×1019 cm−3 ≤NAI≤ 5.6×1019 cm−3, the regrowth rate increased linearly with NAI as previously observed for SPER in boron, phosphorus, and arsenic implanted samples. However, due to the compensation effect, the aluminum data could not be fit to the normalized equation; V/Vi = 1 + N/Ni, as was done previously for data obtained for boron, phosphorus, and arsenic. The regrowth rate increased nonlinearly to the maximum implanted concentration of 4.7× 1020 cm−3 at which the regrowth rate was more than double the previously observed maximum rate in boron doped silicon. In the high concentration range, the SPER rate enhancement could be fit by a quadratic equation whose curvature was positive as was the case for boron. This contrasts with the negative curvature required to fit the nonlinear dependence of the SPER rate on the concentration of donor impurities such as phosphorus and arsenic.

The competition between solid phase epitaxy and random nucleation during thermal annealing of amorphous Si implanted with the fast diffusers Cu and Ag has been studied. For low concentrations of these impurities, solid phase epitaxy proceeds with small deviations from the intrinsic rate and with the impurity remaining in the shrinking amorphous layer. At a critical metal concentration in the amorphous layer of ∼ 0.12 at.% rapid random nucleation occurs, halting epitaxy and transforming the remaining amorphous material to polycrystalline Si via grain growth. The nucleation rate is at least 8 orders of magnitude greater than the intrinsic homogeneous rate. At higher Cu concentrations nucleation is observed below the temperature needed for epitaxy (400°C). This nucleation, clearly caused by the presence of Cu or Ag in the layer, may be induced by the impurities exceeding the absolute stability concentration and starting to phase separate, leading to enhanced crystal Si nucleation in the metal rich regions.

The structural stability of silicon in the amorphous state has been studied by TE1 on in-situ heat treated non-supported layers of a-Si:F, a-Si:D:H and double-layers of a-Si:F/Pd and a-Si:D:H/Pd structures. The a-Si:F and a-Si:D:H films, 500-800A thick, were evaporated by plasma decomposition of a(SiH4+D2)-mixture and SiF2, at 200-260°C onto cleaved rock salt. Pd film was deposited on a-Si by electron beam evaporation at room temperature. The amorphous-to-crystalline transition for the a-Si was quantitatively described by the delay time, t0, before the onset of crystallization. This parameter was found to decrease exponentially with the temperature. Influence of the Pd on t0 (metal induced crystallization) is discussed. This is the first report on the influence of F on the metal/a-Si system. It is also the first time a direct method is used to determine the temperature dependence of the fundamental parameter-delay time preceding a-Si crystallization.

Ion beam induced epitaxial crystallization of Au and Ag doped amorphous Si results in segregation and trapping of the impurity. Combining the measured interface velocity and impurity profiles in segregation simulations provides a measure of the segregation coefficient k during growth. To adequately match the experimental profiles, k must increase during the early stage of growth until saturating at a temperature dependent value. This segregation process cannot be explained within standard models where k depends on the inteface velocity (kinetic trapping) or the interface impurity concentration (thermodynamic solubility). Instead the data suggests that the number of trapping sites at the interface increases during the initial stages of ion beam induced growth. We present several possible mechanisms for this trapping increase and discuss their significance in ion beam and thermal epitaxy models.

In situ electron microscopy has been used to observe crystal nucleation and growth in amorphous Si films. Results demonstrate that a repeated intermediate temperature ion irradiation/thermal annealing cycle can lead to suppression of nucleation in amorphous regions without inhibition of crystal growth of existing large crystals. Fundamentally, the experimental results indicate that the population of small crystal clusters near the critical cluster size is affected by intermediate temperature ion irradiation. Potential applications of the intermediate temperature irradiation/thermal anneal cycle to lateral solid epitaxy of Si and thin film device technology are discussed.

The interface growth dynamics of Si leading to laser-induced amorphization have been studied as a function of incident energy density and crystallographic orientation using tran-smission electron microscopy (TEM). Using lithographic and dry etching techniques, cross sectional TEM specimens across the entire melt region were produced. These samples allowed observation of the evolution of the amorphous thickness and crystal defect structure from shallow anmorphous Si (a-Si) formation near the melt threshold energy density to crystallized material (c-Si) at higher energy densities. For surfaces near [111] orientations, marked differences in the solidification behavior were observed as compared to [001] or [110] orientations. The transition to a-Si growth at the moving liquid/crystal interface is discussed in terms of interface response function kinetics and crystallization models. A growth instability model, arising from the shape of the interface response function, is proposed to explain the observed nmicrostructure.

Transient time-resolved picosecond reflectivity measurements were performed on silicon melted by a 40 ps ultraviolet (308 nm) beam. These were compared to simulations of reflectivity versus amorphous and liquid layer depth and of amorphous interface velocity for various models including transition state theory and density limited growth, both with and without thermal activation.

A study of chromium-enriched domain growth occurring in binary Fe-Cr alloys quenched from above to various temperatures within the miscibility gap has been made. We present kinetic sequences of in-situ small angle neutron scattering (SANS) data for ageing times up to 75 hours on alloys containing 20, 30 and 40 atomic percent chromium. The SANS measurements are compared with partial structure functions obtained from computer simulations performed on a distributed array processor (DAP). We use a pair-potential lattice model, but simulate large systems containing up to 16 million lattice sites. We find good agreement between the scaled structure factors for our SANS data and computer simulated system.

The small angle neutron scattering has been investigated in situ at 450° and 500°C for a polycrystalline, duplex Fe-27.5 at.% Cr - 5.6 at.% Ni steel. A broad diffuse maximum in the scattering function is the signature of the α' phase formation, and this maximum is superimposed on a strong, temperature-dependent component due mainly to critical magnetic scattering. The time dependence of the shift in the peak intensity position to lower scattering vectors and the increase in peak intensity obey power law scaling behavior. Furthermore, the structure function exhibits dynamical scaling, after about three hours annealing. It is suggested that this behavior could be utilized to predict the microstructure, and hence some of the properties, after significantly longer annealing times.

We have investigated the kinetics of the B2/DO3 transition in Fe3AI (28 at. %) using in situ time-resolved x-ray scattering. In these experiments, the evolution of the diffuse and/or Bragg intensity near the (½ ½ ½) DO3 superlattice peak is observed after the temperature of the sample is abruptly changed. H-ere we present results for the kinetics of re-equilibration of short-range order within the disordered (B2) phase, and of short- and long-range order within the ordered (DO3) phase. The short-range order is characterized by the diffuse peak intensity IDIFFUSE and correlation length ξ;the long-range order is characterized by the Bragg intensity IBRAGG. For quenches within the disordered phase, IDIFFUSE and ξ both relax exponentially at the same rate. The temperature dependence of the relaxation time shows evidence of a divergence at the critical temperature. For shallow quenches within the ordered phase, IBRAGG, IDIFFUSE and ξ all relax exponentially, but with different rates. However, for deep quenches within the ordered phase, IDIFFUSE and ξ do not show simple exponential relaxation. Instead, coarsening of short-range order into long-range order is seen, as in quenches from the disordered phase into the ordered phase. Investigation of up-quenches and down-quenches to the same temperature within the ordered phase indicates that disordering is faster than ordering.