The performance and reliability of polycrystalline films are strongly affected by the average grain size and the distribution of grain sizes and orientations. These are often controlled through grain growth phenomena which occur during film formation and during subsequent processing. Abnormal rather than normal grain growth is most common in thin films, and leads to an evolution in the distribution of grain orientations as well as grain sizes, often leading to uniform or restricted crystallographic orientations or textures. Surface and interface energy minimization and strain energy minimization can lead to development of different textures, depending on which is dominant. The final texture resulting from grain growth depends on the film thickness, the deposition temperature, the grain growth temperature, the thermal expansion coefficients of the film and substrate, and the mechanical properties of the film, as well as other factors.

Many materials for engineering applications are used in polycrystalline form and contain grain boundaries with a range of structures and properties. However, most research on grain boundaries to date has focussed exclusively on symmetric coincidence site lattice interfaces. To go beyond descriptions for these simple interfaces and thence to an aggregate of grains and grain boundaries in a polycrystal will require a new approach. Here we discuss two models for properties of polycrystalline materials, including their advantages and drawbacks, and indicate the microstructural variables available to optimize properties.

Recent developments, coupling the scanning electron microscope with image processing and crystallographic analysis, now make it possible to automatically index many thousands of lattice orientations exposed on section planes in polycrystalline materials. Backscattered Kikuchi diffraction patterns obtained from high-gain SIT and CCD cameras are analyzed using the Hough transformation (Cartesian coordinates - polar coordinates) in order to identify diffraction band widths and interplanar angles. From this basic information local lattice orientation can be determined. Information from raw data sets (in excess of 100,000 single orientations in some examples) can be used to construct orientation imaging micrographs which emphasize certain aspects of the exposed field of lattice orientations. Thus, features of the spatial placement of lattice orientation (including grain boundary misorientation, microtexture, and connectivity of the microstructure) are readily studied. In this paper these techniques are reviewed, and recent explorations of the connectivity of grain boundary misorientation structure are presented for an interesting nickel-chromium-iron alloy.

Grain structure and grain growth in thin metallic films are important because of their effects on properties such as yield strength, electrical resistance and electromigration resistance. Since almost all thin films are used in contact with a substrate and many also have contacts with overlayers, it is important to consider how interactions with other materials affect the grain growth process. In this paper we consider the effects of diffusive interactions. We will show that interdiffusion often accompanies grain growth and that it can result in a number of novel grain boundary reactions, driven by a variety of effects. Using TEM techniques, we demonstrate cases of grain growth suppression and grain growth enhancement resulting from interdiffusion of solute atoms in gold thin films. The reasons for the observed effects will be considered with a view to providing a fundamental understanding of the types of systems that might be expected to exhibit the various phenomena.

During grain growth, shrinking columnar grains in thin-film polycrystalline microstructures eventually reach sizes comparable to the film thickness. Due to surface drag, the sides of such grains may bow inward rather than remaining fiat through the bulk of the film. The grain boundaries delimiting such small shrinking grains may become unstable long before the surface of the shrinking grain reaches zero area. We report simulation results demonstrating such an instability in the limit of infinite surface drag. This may lead to extremely rapid disappearance of 4- or 5- sided grains, such as have been recently observed in in situ hot-stage TEM experiments on aluminum thin film polycrystals.

We have simulated strain energy effects and surface- and interface-energy effects on grain growth in thin films, using properties of polycrystalline Ag (p-Ag) on single crystal (001) Ni on (001) MgO for comparison with experiments. Surface- and interface-energy and strain energy reduction drive the growth of grains of specific crystallographic orientations. The texture that will result when grain growth has occurred minimizes the sum of these driving forces. In the elastic regime, strain energy density differences result from the orientation dependence of the elastic constants of the biaxially strained films. In the plastic regime, strain energy also depends on grain diameter and film thickness. In p-Ag/(001) Ni, surface- and interface-energy minimization favors Ag grains with (11) texture. In the absence of a grain growth stagnation, the texture at later times is always (111). However, for high enough strains and large enough thicknesses, the strain energy driving force can favor a (001) texture at early times, which reverts to a (111) texture at later times, once the grains have yielded.

Hillert’s model of grain growth consists of a drift term in size space that leads asymptotically to a distribution function and a growth exponent not often observed. Later theories introduce a diffusion term that is either assumed to dominate the drift term or a correction to it. This paper shows that the lower order drift term alone determines asymptotic grain growth behavior. A possible conclusion is that experimental results may need to be reinterpreted.

Several alternatives to Hillert’s work on grain growth have been proposed. Unfortunately some of these theories have been shown to violate mass conservation. This paper introduces a way to enforce conservation for any model via a term representing self-interaction of the grain distribution.

Abnormal grain growth is characterized by the lack of a steady state grain size distribution. In extreme cases the size distribution becomes transiently bimodal, with a few grains growing much larger than the average size. This is known as secondary grain growth. In polycrystalline thin films, the surface energy γs and film/substrate interfacial energy γi vary with grain orientation, providing an orientation-selective driving force that can lead to abnormal grain growth. We employ a mean field analysis that incorporates the effect of interface energy anisotropy to predict the evolution of the grain size/orientation distribution. While abnormal grain growth and texture evolution always result when interface energy anisotropy is present, whether secondary grain growth occurs will depend sensitively on the details of the orientation dependence of γi.

Certain experimental results are presented concerning grain growth in microcrystalline Ag films. Dark field TEM technique was used for the measurement of grain size, trijunction velocity and grain boundary mobility. We found that the activation energy for trijunction motion is 25.0 kJ/g.atom, and the activation energy for the grain boundary motion is 50.0 kJ/g.atom.

This paper addresses the validity of the single particle growth law rn-r0 n=Kt in describing the formation of 3D grains during the vapor deposition of thin metal films, especially with respect to the value of the exponent n. Computer simulations based on the Huygen’s construction showed that the grain size distribution at full surface coverage did not depend significantly on the specific model chosen for initiating growth from nuclei. The particle size distributions obtained experimentally by STM measurements of thin Au films deposited on glass substrates agreed very well with the simulation results for n=2.

Grain boundary free volume, simply defined as the difference between the volume of a bicrystal and that of a single crystal containing an equal number of atoms, provides a good measure of average grain boundary coordination. Free volume is useful because (a) computer calculations suggest that the grain boundary free volume scales with the grain boundary energy and (b) experimental measurement of free volume may be relatively easier and more direct than that of grain boundary energy. The objective of this paper is to compare the predictions from computer models of grain boundary free volume with experimental measurements.

The bit-erase process in phase-change optical storage is based on the amorphous to crystalline transformation. While there has been significant progress developing compositions and multilayered media for phase-change applications, quantitative studies of the crystallization kinetics and microstructural development are generally lacking. This paper describes work quantifying crystallization in GeTe thin films. Microstructural changes during isothermal annealing are measured using in-situ hot-stage optical microscopy. This technique measures the fraction crystallized, the number of crystallites, and crystallite radii as a function of time. These data are sufficient to deconvolute the individual contributions of nucleation and growth. We find an Avrami exponent of ∼4, consistent with time-resolved reflection/transmission studies. This exponent is due to 2-D growth at a constant rate plus transient nucleation. The data are used in a kinetic model to simulate non-isothermal crystallization during focused-laser heating characteristic of the bit-erase process.

Through post-deposition annealing in a differential scanning calorimeter (DSC), we have manufactured both thin (200 nm) and bulk (8000 nm) single phase films of crystalline Ge1–xSnx, using rf sputtering. The Sn concentrations produced ranged up to 31 at.%, well beyond the solid solubility limit of this system. There was a marked difference, in the asdeposited structure, between thick and thin films produced under the same deposition conditions. Quantitative models for both systems are given in this paper and were deduced frorn DSC measurements in conjunction with electron microscopy. The metastable crystalline state in the thin films formed by nucleation and growth from an amorphous phase; whereas in the thick films, the desired phase was already present in the as-deposited films and only growth of preexisting grains was observed upon post-deposition annealing. When annealed to high temperature, the Sn phase separates from the alloys and we postulate here that it does so by nucleation and growth of β-Sn. With this hypothesis, the Sn separation in the 8000 nm thick films was accurately modeled by a two-mechanism process, however, in the 200 nm thick films, only one phase separation mechanism was necessary to accurately fit the data. Both models were corroborated by the subsequent melting behavior of the phase separated Sn which, though it varied depending on the sample being measured, always exhibited a melting endotherm starting 25–35°C lower than the bulk melting temperature of Sn. Speculation on the reasons for this are presented.

We have studied the effects of low energy ion bombardment on thin copper films. Evaporated, sputtered and CVD copper films (∼50 nm) were exposed to Magnetically Enhanced (ME) Ar plasmas. The microstructural changes (grain size) in the films were studied using Transmission Electron Microscopy (TEM).

Grain growth is observed in thin Cu films when the films are exposed to low energy (87 eV) Ar plasmas. The microstructural changes in sputtered and evaporated films are quite significant whereas the plasma bombardment has less effect on CVD films. These changes occur very rapidly and cannot be attributed solely to the thermal effects, especially at low RF power levels (500 W). The initial microstructure of the film has a significant effect on grain growth during plasma exposure.

Stimulable phosphor thin films are being investigated for use as optical data storage media. We have successfully applied atomic force microscopy (AFM) to the measurement of the surface texture of these films. Determination of the surface texture of the films is important for evaluating the effect of surface quality on optical scatter. In other thin film material systems it has been found that the surface “bumps” revealed by AFM correspond to grains in the film. This is not the case for the stimulable phosphor films used in our study. We have determined the grain size of our phosphor films by transmission electron microscopy (TEM) and x-ray diffraction (XRD). The grain size from TEM and XRD does not correlate with the size of the AFM surface “bumps.” For example, in two of the five films studied, the XRD derived grain size varies by a factor of two but the size of the surface “bumps” remains the same. We conclude that the texture of the film surface is not directly determined by the grain size of the phosphor material.

Nanostructured materials are being extensively studied because their ∼1–100nm grain size can dramatically affect properties. Most nanocrystalline synthesis methods produce particulate or flake. The process of consolidation also allows coarsening, contamination, and the introduction of porosity. The effect of nanocrystallinity on mechanical properties must be deconvoluted from these extrinsic artifacts. Most synthesis routes also produce small quantities of material. Reproducibly making enough specimens to explore more than a few properties is thus difficult. This paper describes thin-film processes to produce nanostructured materials. Thin-film deposition can easily produce many specimens, free from extrinsic artifacts, with identical composition and processing history. Many methods are now well established to study a variety of thin-film mechanical properties. We show examples of nanostructured films generated by controlling deposition and/or post-deposition processing.

X-ray diffraction was used to study the influence of ultra-fine dispersions of CeO2 on the texture development and epitaxial relationships in oxide films, formed at high temperatures. The substrate orientation exerted the essential effect on microstructure and growth rate of oxide films on both pure as well as coated nickel. NiO grown on (100)Ni was polycrystalline with grains randomly oriented. Applying of CeO2 resulted in a marked decrease of the oxide growth rate without significant changes in the texture. The NiO formed on pure (111)Ni exhibited strong (111) texture. During initial stages of oxidation the presence of CeO2 caused the nucleation of randomly oriented oxide. However, the oxide developed during further exposure had the same character as that grown on pure (111)Ni face. X-ray measurements are compared with analysis conducted by TEM and electron microdiffraction. The role of chemically active element in inhibition of diffusion processes during the growth of oxides on both crystal faces is discussed.

The texture formation during the electrodeposition process was simulated using a Monte Carlo technique. The simulation uses a two dimensional hexagonal lattice to map the microstructure of the deposit. The criteria for the texture formation was based on the minimization of the system’s free energy. The anisotropy of surface-energy was taken into account. Since a metal’s surface energy is influenced by hydrogen adsorption, the texture of metal deposits may vary with hydrogen co-deposition.

We fabricated CulnSe2 and Cu(In,Ga)Se2 thin films by two different pathways using physical vapor deposition. In the first we formed a Cu-Se precursor and then reacted it with a flux of (In,Ga) + Se. These films had large grains but were too rough for optimal device performance. In the other pathway, we first formed a smooth precursor of (In,Ga)2Se3 and then exposed it to a flux of Cu+Se. We overshot the optimal film composition to allow recrystallization of the film by a secondary CuxSe phase. We then consumed the excess CuxSe in a third stage deposition of (In,Ga) + Se. The recrystallization step increased the grain sizes, and the resulting films remained smooth. Photovoltaic solar cells made from these films have produced the highest total-area efficiencies of any non-single-crystal, thin-film solar cell.