Phase transformation generally begins with nucleation, in which a small aggregate of atoms organizes into a different structural symmetry. The thermodynamic driving forces and kinetic rates have been predicted by classical nucleation theory, but observation of nanometer-scale nuclei has not been possible, except on exposed surfaces. We used a statistical technique called fluctuation transmission electron microscopy to detect nuclei embedded in a glassy solid, and we used a laser pump-probe technique to determine the role of these nuclei in crystallization. This study provides a convincing proof of the time- and temperature-dependent development of nuclei, information that will play a critical role in the development of advanced materials for phase-change memories.

We measured the transverse and longitudinal coherence properties of the Linac Coherent Light Source (LCLS) at SLAC in the hard x-ray regime at 9 keV photon energy on a single shot basis. Speckle patterns recorded in the forward direction from colloidal nanoparticles yielded the transverse coherence properties of the focused LCLS beam. Speckle patterns from a gold nanopowder recorded with atomic resolution allowed us to measure the shot-to-shot variations of the spectral properties of the x-ray beam. The focused beam is in the transverse direction fully coherent with a mode number close to 1. The average number of longitudinal modes behind the Si(111) monochromator is about 14.5 and the average coherence time τ(c)=(2.0±1.0) fc. The data suggest a mean x-ray pulse duration of (29±14)  fs behind the monochromator for (100±14) fc electron pulses.

Fluctuation electron microscopy (FEM) is explicitly sensitive to 3-and 4body atomic correlation functions in amorphous materials; this is sufficient to establish the existence of structural order on the nanoscale, even when the radial distribution function extracted from diffraction data appears entirely amorphous. However, it remains a formidable challenge to invert the FEM data into a quantitative model of the structure. Here, we quantify the FEM method for a-Si by forward simulating the FEM data from a family of high quality atomistic models. Using a modified WWW method, we construct computational models that contain 10-40 vol% of topologically crystalline grains, 1-3 nm in diameter, in an amorphous matrix and calculate the FEM signal, which consists of the statistical variance V (k) of the dark-field image as a function of scattering vector k. We show that V (k) is a complex function of the size and volume fraction of the ordered regions present in the amorphous matrix. However, the ratio of the variance peaks as a function of k affords the size of the ordered regions; and the magnitude of the variance affords a semi-quantitative measure of the volume fraction. We have also compared models that contain various amounts of strain in the ordered regions. This analysis shows that the amount of strain in realistic models is sufficient to mute variance peaks at high k. We conclude with a comparison between the model results and experimental data.

Phase change memory devices are based on the rapid and reversible amorphous-to-crystalline transformations of phase change materials, such as Ge2Sb2Te5 and AgInSbTe. Since the maximum switching speed of these devices is typically limited by crystallization speed, understanding the crystallization process is of crucial importance. While Ge2Sb2Te5 and AgInSbTe show very different crystallization mechanisms from their melt-quenched states, the nanostructural origin of this difference has not been clearly demonstrated. Here, we show that an amorphous state includes different sizes and number of nanoscale nuclei, after thermal treatment such as melt-quenching or furnace annealing is performed. We employ fluctuation transmission electron microscopy to detect nanoscale nuclei embedded in amorphous materials, and use a pump-probe laser technique and atomic force microscopy to study the kinetics of nucleation and growth. We confirm that melt-quenched amorphous Ge2Sb2Te5 includes considerably larger and more quenched-in nuclei than its as-deposited state, while melt-quenched AgInSbTe does not, and explain this contrast by the different ratio between quenching time and nucleation time in these materials. In addition to providing insights to the crystallization process in these technologically important devices, this study presents experimental illustrations of temperature-dependence of nucleation rate and growth speed, which was predicted by theory of phase transformation but rarely demonstrated.

No abstract