Decades of work by a wide variety of techniques were required to establish unambiguously the essential role of simple – and sometimes not so simple – “point” defects in mediating bulk diffusion in crystalline solids. Amorphous solids present new problems for establishing basic diffusion mechanisms. Most experimental techniques which work well for study of diffusion in crystalline solids are useless for study of amorphous materials because of their inherent nonequilibrium structures. A survey of some current results also gives a strong impression that more complex basic mechanisms than simple point defects may be required to account for volume diffusion in these materials.

Atomic migration in crystals is obscured by the crystal itself, and it has therefore taken ingenious indirect methods to study the diffusion process. In contrast, diffusion at surfaces is amenable to fairly direct examination, on the atomic level, through the use of the field-ion microscope. By combining measurements of the mean-square displacement with determinations of the distance distribution function, considerable detail about atomic jumps in diffusion can be obtained. Here we will briefly review results obtained by field-ion microscopic examination of single metal atoms and metal clusters diffusing on metal crystals.

What is noteworthy in the diffusion of single metal atoms is the variety of behavior which is dictated by the nature of the substrate, and the occasional complexity of the diffusion process. Examples of substrate-specific effects in atomic migration will be given for both fcc and bcc metals. Another surprising aspect of diffusion on metal surfaces is the role of clusters. These are stable at relatively high temperatures, and often diffuse at rates comparable to those of single atoms. The status of such cluster studies will be briefly examined, not only for their significance in surface transport, but also for their interest in understanding cohesion at metal surfaces.

We present measurements of interdiffusion in amorphous silicon/amorphous germanium compositional multilayer structures using Raman scattering and optical absorption spectroscopies. We find that interdiffusion coefficient D* can be described by D* = Do exp (−ED/kT) with ED = 3.3 eV and Do = 104 cm2s−1. The high value of ED suggests that the diffusion process is substitutional.

Turnbull and co-workers have shown that the noble metals Cu, Ag, and Au diffuse into single crystals of Pb and Sn by an interstitialcy mechanism with measured diffusion coefficients of ∼10−6cm2s−1. These experiments have been extended to study the diffusion rates of the noble metals along the interphase boundaries of single crystal of Pb-Sn eutectic alloys and to measure the diffusion coefficient of silver along the interphase boundaries of Ag-Cu eutectic single crystals.

The measured interphase boundary diffusion coefficients of the noble metals in the Pb-Sn single crystals are extremely high, being 2.6 × 10−3cm2s−1for Cu and ∼10−4cm2s−1for Ag and Au at room temperature. These results again imply an interstitialcy mechanism of diffusion at the interphase boundaries. In contrast, the diffusion coefficient of Ag110in a Ag-Cu eutectic single crystal worked out to be ∼8×10−9cm2s−1at 500°C, two orders of magnitude lower than the value obtained by Turnbull and Hoffman for self-diffusion along low-angle (8∼10°) boundaries in silver.

The atomic structure of representative tilt boundaries in gold has been determined by high resolution transmission electron microscopy. Characteristic and varying regions of decreased density and coordination have been identified and related to mechanisms of grain boundary diffusion and migration

The precipitation of Pb3Au from dilute Pb(Au) alloys and the subsequent Ostwald ripening of the precipitate phase have been studied. The Au diffusivity which limits the precipitation from Pb(500 at. ppm Au) specimens is in good agreement with Dp = 45 exp[(−20.8 kcal/mol)/RT], found previously to control the precipitation rate in Pb(900–1000 at. ppm Au) alloys (Rossolimo and Turnbull, 1973). In contrast, precipitation from Pb(100 at. ppm Au) alloys and Ostwald ripening in the Pb(500 at. ppm Au) specimens appear to be governed by the high temperature Au tracer diffusivity, DAu = 0.0041 exp[(−9.35 kcal/mol)/RT], which is about 103 times greater than Dp at the temperatures of interest, 75 to 112°C, and is attributed to the motion of a Au interstitial defect. In the 100 ppm alloys the Au is incorporated most easily at the ends of the Pb3Au crystals and a ribbon- or whisker-like morphology results. The particles which form at the grain boundaries in the 500 ppm Au alloy specimens are at least six times larger in volume than the matrix particles and ripen at their expense. The Ostwald ripening kinetics of the grain boundary particles scale with t1/2 rather than with t1/3 as is usual. The Pb3Au formed whiskers, blades, ribbons, or equiaxed particles depending upon the Au concentration and precipitation temperature. The precipitation and ripening kinetics are critically dependent on the morphology of the second phase.

A comparison is made of, ordering kinetics for Cu3Au, (Tc = 388 ° C), Ni3Mn (Tc = 500° C) Ni3Fe (Tc = 517°C) and Ni3Al (Tc = 1450 ° C), which all form the L12 superlattice, and these kinetics are correlated with diffusion data for the same alloys. Comparing different alloys, it is shown that for a fixed relaxation time for the establishment of order, the lower is the self-diffusivity, Ds, of the slowest-diffusing species, the higher the critical temperature, Tc, for the first appearance of order. Ni3A1, in particular, has an estimated Ds of about 5 orders of magnitude lower, on average, than the three other alloys. The implications of this inverse relation between Tc and Ds are discussed in relation to published models for diffusion in ordered alloys and ordering kinetics.

Metallic glass formation by solid state reactions has been observed in multilayer Zr-Co diffusion couples. The kinetics of the reaction are limited by the diffusion of the Co-atoms in the growing amorphous layer, at least for longer times, as shown by cross-sectional transmission electron microscopy and resistance measurements. The latter one provides the interdiffusion constant and the activation energy of about 1.1 eV. Deposition of the crystalline layers at 77 K results in an enhanced amorphization process in the first stage of the reaction and gives preliminary answers about the nucleation of the amorphous phase.

The formation of an amorphous Ni-Zr alloy, by heating to cause interdiffusion in the temperature range 250°C to 325°C, was demonstrated in a self-supporting lateral-diffusion couple. Micro-area diffraction in a scanning transmission electron microscope (STEM) indicates that microcrystalline ordering in the amorphous phase must be on a scale less than 5 nm. The samples were heated in a vacuum furnace to form the amorphous phase and later heated in the STEM for in situ observation of crystalline compoundformation. The amorphous to crystalline transformation occurs first in theZr-rich regions at 370°C. A crystalline phase begins to consume the Ni rich amorphous material, at 450° C; but at 325° C, the amorphous phase is stable up to 20.5 hours, even in the presence of crystalline compounds.

The scientific contributions by David Turnbull and his co-workers toward understanding crystal growth span over 35 years from the late 1940's to the present. Turnbull's early attempt (1950) to correlate interfacial energies derived from droplet supercooling measurements with other thermochemical data still provides a significant data base for estimating such energies in a variety of materials. His work with Hillig in 1956 on quantification of the screw dislocation mechanism for interfacial molecular attachment remains as a predictive theory for defect-assisted growth of faceted, and therefore kinetically hindered, interfaces. The Hillig-Turnbull screw dislocation mechanism is now ranked among such notable kinetic models for crystal growth as the Wilson-Frenkel random attachment theory and the nucleationlimited layer spreading models of Burton, Cabrera, and Frank. Turnbull's contributions and interest in elucidating crystal growth kinetics have continued throughout his long and productive career, as evidenced by his recent work with Coriell (1982) on estimates of collision limited rapid crystal growth in highly supercooled transition metals. Progress in unraveling the kinetic contributions of interfacial attachment from those of heat and solute transport will also be reviewed to provide a current context of Professor Turnbuli's contributions to the field of rapid crystallization.

The condensed state phase transition energetics and kinetics are reported for specific compositions within the Al-Si-Mn system which demonstrate crystal → crystal, quasicrystal→ crystal, and glass.→crystal structural transformations. Our observations, and their interpretation, for the solid state crystalline AI(Mn)→Al+Al6 Mn reaction and the crystallization of glassy Al54Si 30Mn16 are consistent with the anticipated mode for these transformations. The nature of the quasicrystal→ crystal transformation, consistent with our observations, is an enigma.

The prediction by Turnbull and his colleagues that amorphous Si and Ge undergo first order melting transitions at temperatures Taℓ substantially beneath the crystalline melting temperature Tcℓ has stimulated much work. Structural, calorimet:ic and transient conductance measurements show that, for Si, Tcℓ – Taℓ lies in the range 225–250°K. Studies of the pulsed laser melting of the Si amorphous-liquid transition have resulted in the following findings, an estimate of the undercooling rate of 15°K/m/sec, an understanding of the mechanism mediating explosive crystallization, the formation of internal melts and segregation of dopants at the liquid-amorphous interface.

The fraction of impurity atoms in the liquid at the solid-liquid interface that join the crystal, known as the segregation coefficient k, during rapid crystal growth is known to deviate away from the equilibrium value towards unity as the interface speed v increases. Several plausible models have been proposed that account qualitatively for this behavior with different functional forms of k(v). We report measurements of the segregation behavior during rapid solidification following pulsed laser melting of Bi-implanted Si. The velocity dependence and the orientation dependence of the segregation coefficient of Bi in Si has been determined to sufficient accuracy to allow us to distinguish between models. Implications for the mechanism of solute trapping are discussed.

Normal grain growth in thin films leads to columnar grains with sizes roughly equal to the film thickness. In subsequent grain growth, a minor fraction of the grains continue to grow at appreciable rates, leading initially to a bimodal grain size distribution and ultimately to a monomodal distribution of grains with sizes much larger than the film thickness. Those grains which continue to grow, secondary grains, often have uniform or restricted texture, suggesting that in such cases surface energy minimization plays an important role in driving their preferential growth. Surface-energy-driven secondary grain growth can, in principle, lead to single crystal films. Turnbull and co-workers were among the first to apply rate theory to the analysis of grain boundary motion, grain growth, and recrystallization. These models have been adapted to describe surface-energy-driven secondary grain growth in thin films. Recent experiments on secondary grain growth in thin metallic and semiconductor films are reviewed. It has been shown that film thickness, film composition and surface topography have pronounced effects on growth rates, final grain sizes and orientations.

In this paper we discuss some features of the microstructure of large-grainsized very low voltage varistors and the effects of heat treatments on the pre-breakover electrical characteristics. It is shown that a second phase precipitates in the bismuth-rich segregation phase and is thought responsible for the reduced device leakage properties.

Cu-2.9 at.% Ti single crystals were homogenized at various temperatures (780 °C ≤ TH ≤ 960°C) and quenched. Subsequent isothermal aging at 350 °C led to phase separation, the kinetics of which have been followed by employing small-angle neutron scattering (SANS). According to comple-mentary transmission electron and analytical field ion microscopy studies, the resulting transformation products of this first order phase transition are stoichiometrically ordered ellipsoidal Cu4Ti particles, the aspect ratio of which changes with aging time (t) as revealed by two-dimensional SANS-detection. In the early stages of phase separation, the decomposition kinet-ics are strongly influenced by the quenching rate via quenched-in excess vacancies. During aging the structure factor S(K,t) develops a maximum, the height (Sm) of which increases and the position (Km) of which decreases with t. Neither Sm(t) nor Km(t) follow a power law as predicted by several recent theories on spinodal decomposition. On the other hand, the time evolution of the mean Ti-rich cluster size (R), their number density (Nv), and the supersaturation (Δc) as inferred from the SANS-data and the diffuse Laue-scattering, are well predicted by a precipitation model which describes nucleation, growth and coarsening as competing processes.