Most alloy systems can develop more than one solid phase for a given composition during solidification. For metastable undercooled liquids a thermodynamic analysis can provide insight into the range of structural and compositional options that are possible. The numerous examples of metastable phase formation during solidification of undercooled melts demonstrates that the actual phase selection is dominated by kinetics. Indeed, the observation of undercooling is linked to kinetic control. Different forms of kinetic control can be identified and include suppression of heterogeneous nucleation during slow cooling and constrained growth during rapid quenching. In both cases the kinetic competition involved in phase selection is mediated by the thermal history. Nucleation controlled kinetics can be examined in fine powders of undercooled liquids where crystallization rates and metastable phase diagrams provide a basis for analysis. Similarly, constrained growth can be examined during controlled cooling where microstructures and kinetic models can be used for analysis. The existing kinetics models appear to be adequate but the processing conditions are often dynamic and in nonsteady state conditions so that critical tests are difficult unless specific kinetic transitions are available.

Techniques of containerless processing are applied to undercool and solidify metals and alloys. These techniques allow direct measurements of both the undercooling and the crystal growth velocity. Experimental results are presented for studies of nucleation of metastable crystalline phases and quasicrystals. Measurements of the dendrite growth velocity as a function of undercooling are exemplified for dilute Ni-based alloys and intermetallics. The results are analysed within current theories of rapid crystal growth. Their consequences on the formation of grain refined microstructures are highlighted. In addition, recent experiments on the undercooling of magnetic alloys are discussed revealing the existence of long-range magnetic ordering in an undercooled melt.

A hypereutectic Al-40wt%Ni alloy has been manufactured by melt spinning, and the resulting microstructure examined by transmission electron microscopy. As-melt spun hypereutectic Al-40wt%Ni consists of an Ni aluminide matrix and an A1-rich phase distributed in the form of particles with sizes ∼ 50-100 nm, and as an irregular layer at the cell and grain boundaries. Diffraction analysis of the Ni aluminide matrix is consistent with the ASTM x-ray diffraction standard 2θ values for the orthorhombic NiAl3 phase, a=6.6114 Å, b=7.3662 Å andc=4.8112 Å. The solidification nucleation kinetics of Al-rich particles have been examined by heating and cooling experiments in a differential scanning calorimeter over a range of heating and cooling rates. Solidification of the Al-rich phase at the cell and grain boundaries nucleates catalytically on the surrounding Ni aluminide matrix at an undercooling of ∼ 3 K. Analysis of the solidification nucleation kinetics of the Al-rich phase in Al-40wt%Ni supports the hypothesis [1-4] that the classical spherical cap model of heterogeneous nucleation breaks down at low undercoolings and small contact angles.

From the nucleation data on undercooling of liquid Si or Ge, crystal-melt interfacial tensions are calculated. Only a temperature-dependent tension can account simultaneously for the results of experiments on bulk and thin film Si. The observed temperature dependence can be accounted for by reasonable values of the interfacial entropy and enthalpy. The analysis is used to determine the temperature-dependent interfacial tension for Ge. A comparison of results for Ge and Si indicates that homogeneous nucleation has not been achieved in the undercooling of bulk liquid Si.

Thermodynamic data derived by the CALPHAD method has been extended to undercooled melts to study phase competition and the role of time dependent and steady state nucleation in rapidly solidified materials. The theoretical predictions made about phase competition between hcp α, bcc β and Ll0 γ in some Al(Ti,V) alloys are compared with experimental results. It is shown that high melt cooling rates introduce significant transient nucleation effects on phase nucleation.

In the solidification of Fe-Ni droplets (≤ 30 at.% Ni), the selection of different microstructures is dominated by the competition between the bcc and ccp phases. In drop-tube experiments ccp is the primary phase in some dilute (up to 7at% Ni) alloys although the bcc phase is favoured by a lower free energy and by a lower interfacial energy with the liquid. Competitive dendrite growth is a possible explanation for the formation of primary ccp. Comprehensive thermodynamic (CALPHAD) and kinetic modelling is undertaken to understand the growth competition. The origin of the observed primary phases is discussed.

A structure composed of both partitionless face-centered cubic (α-fcc) and body-centered cubic (β-bcc) phases has been observed following high undercooling solidification of nickel-vanadium (Ni-V) alloys in the composition range 47-51.7 at.% V. Containerless processing and rapid solidification suppresses the formation of the equilibrium σ phase and results in the simultaneous formation of partitionless fcc and bec phases. Transmission electron microscopy analysis has identified regions in which 0.1-1 μηι a and β grains co-exist in a duplex structure. A nucleation and growth kinetics analysis has been developed to describe the conditions under which this structure may form. Thermodynamic and kinetic arguments have been used to provide bounds on various kinetic parameters. These bounds indicate that the nucleation rates for a and β must be larger than about 1027 m−3 s−3 to account for the number of grains observed in the time available for nucleation, and the growth rates are constrained to be larger than about 1 m/s to account for complete solute trapping.

For Cu - 30 at.% Ni droplets processed by containerless solidification under lévitation, the crystallographic texture correlates strongly with the microstructure seen in optical metallography and with the undercooling at which solidification started. The variation in texture is useful in understanding the development of the microstructure, in particular at the extremes of the undercooling range where grain-refined microstructures are observed.

Ternary titanium-aluminides with compositions of Ti51Al47Fe2, Ti51A147Cr2 and Ti51Al47Mn2 were investigated with respect to the correlation of their solidification front velocity v and bulk undercooling ΔT. The observation of the solidification front during the recalescence event has been realized using a high speed video system capable of recording up to 12,000 pictures per second. The temperature measurement was carried out by pyrometry, avoiding contact with the sample. The comparison of the experimental data with the LKT-theory (Lipton, Kurz, Trivedi; [1]) refers to a primary (hcp) β-Ti solidification for undercoolings below ΔT≈ 130 K and primary (bcc) α-Ti solidification for ΔT≥ 130 K. For undercoolings ≥ 150 K the theory differs greatly from the experimental results.

The maximum undercoolings achieved were 268 K (Ti51Al47Fe2), 285 K (Ti51Al47Cr2) and 280 K (Ti51Al47Mn2), corresponding to a solidification front velocity v ≈ 9-10 m/s for all alloys.

The solid-state transformations following rapid solidification of intermetallics can include ordering of primary disordered phase(s), resulting in anti-phase domain boundaries (APDBs). The present work focuses on the Ni-Al system. A qualitative model for solid-state ordering, originally developed by Chen and Khachaturyan [1] has been modified to simulate the APD formation during solidification of a disordered phase. An interpretation is suggested for the formation of various APD patterns observed in sequentially ordered Ni3Al.

Several samples of metals and alloys have been undercooled during scanning in a high temperature DSC. Liquid Ni was undercooled of about 230 K when previously fluxed in molten B2O3. From enthalpy data of melting and solidification as a function of temperature, the excess heat capacity of liquid metals and alloys was evaluated. The specific heat of the liquid is definitely higher than that of the corresponding crystalline phases for glass-forming alloys, whereas it is close to that of the solid for pure metals. The Ni-B system has been studied in detail around the Ni-Ni3B eutectic. On undercooling, a new metastable phase (Ni23B6) was produced. A metastable Ni-B phase diagram has been drawn using data of thermal analysis of several alloys containing the metastable phase.

In this investigation, a mechanism for supercooling in organic liquids is formulated. By comparing the melting temperature and spontaneous freezing temperature of the chemicals evaluated with their molecular characteristics, the factors promoting supercooling are developed. The results obtained indicate that the following molecular characteristics promote supercooling in organic liquids:

sharing of electrons between the atoms of a molecule;

A three-dimensional chemical structure;

A permanent, three-dimensional, and partially charged pocket within the chemical structure;

A partially charged projection having a charge opposite that of the pocket and located on the side of the molecule opposite that of the pocket.

In this paper, we first describe and then demonstrate the soundness of a new experimental method that is designed to deliver a wide range of accurately measured nucleation rates of solids in supercooled liquid Si. The method utilizes thin-film processes, as well as laser-quenching and detection techniques, in order to first induce isothermal transformation and then extract the corresponding solid nucleation rates over a frequency range potentially greater than ten orders of magnitude. The technological and scientific gains that can result from acquiring such information are discussed.

Corrections to the standard expressions for the volume fraction of the crystalline phase which arise due to time-dependent nucleation effects are considered both for the isothermal (transient) and nonisothermal (quench) situations. Analytical results are tested against numerically exact solutions of the Turnbull-Fisher nucleation equations. Physical consequences of the obtained expressions are discussed.

We have theoretically investigated the nature and kinetics of solid nucleation in supercooled liquid Si within the framework of the classical nucleation theory, and corresponding to the cases in which the liquids are quenched at extremely high quenching rates (from 109 to 1011 K/s). In doing so, we identify and draw a general conclusion that in addition to the well-treated phenomenon of transient nucleation, one must also consider the mechanism of athermal nucleation in order to properly elucidate the situations that are encountered at such high quenching rates. Moreover, contrary to the common notion that the transient effect is relevant at low temperatures where sluggish kinetics prevail, it is noted that the effect can also become prominent at near-equilibrium conditions due to the increase in the time needed by the embryos to reach the exceedingly large critical size.

Monte Carlo simulations based on a Spin-1 Ising Model for binary alloys have been used to investigate the non-equilibrium partition coefficient (kneq ) as a function of solid-liquid interface velocity and orientation. In simulations of Si with a second component kneq is greater in the [111] direction than the [100] direction in agreement with experimental results reported by Aziz et al. The simulated partition coefficient scales with the square of the step velocity divided by the diffusion coefficient of the secondary component in the liquid.

During rapid solidification, deviations from local interfacial equilibrium are manifested by solute trapping and interfacial undercooling. Both the solute trapping function and the interface velocity-temperature function have been measured in the Si:As alloy system following pulsed laser melting, permitting us to test models for nonequilibrium interface kinetics. The results are consistent with the Continuous Growth Model “without solute drag” of Aziz and Kaplan and are inconsistent with models that incorporate solute drag effects during solidification.

Ion implantation of carbon into single-crystal silicon followed by excimer laser irradiation was used to create supersaturated, epitaxial SixC1-x. films. Crystallization proceeded from the underlying single-crystal silicon through the carbon containing layers at velocities of approximately 5 m/s. Characterization by high-resolution x-ray diffraction and Fourier-transform infrared absorption indicate that the carbon is found predominantly on substi-tutional lattice sites for concentrations up to 1.4 at.% C. Secondary-ion mass spectrometry profiles and numerical mass transfer calculations were used to estimate the diffusion coefficient of carbon in the liquid as 2-3 × 10−4cm2/s with a segregation coefficient greater than 0.4. Unusual diffusion behavior was observed for the carbon at 1.4 at.% C. At higher concentrations, evidence of SiC precipitates was observed in transmission electron microscope images and FTIR absorption spectra.

The assumption that dendrite tips are parabolic bodies of revolution pervades many of the theories and experiments addressing dendritic growth. This assumption, while reasonable, is known to become less valid as regions of interest further from the tip of the dendrite are considered. Experimental measurements were made on pure succinonitrile dendrites at several super coolings. The equation that describes the dendrite tip profile is extended from a second order polynomial (paraboloidal) form to one that includes higher-order terms. The deviation of a dendrite tip from a parabolic body of revolution can be characterized by a parameter obtained from the coefficient of the fourth-order term describing the profile. This dimensionless parameter, Q, is found to be a function of the profile orientation only, independent of supercooling.

A numerical model of heat transfer by combined conduction, radiation and convection was developed using the FIDAP finite element code for NASA's Advanced Automated Directional Solidification Furnace (AADSF). The prediction of the temperature gradient in an ampoule with HgCdTe is a necessity for the evaluation of whether or not the temperature set points for furnace heaters and the details of cartridge design ensure optimal crystal growth conditions for this material and size of crystal. A prediction of crystal/melt interface shape and the flow patterns in HgCdTe are available using a separate complementary model.