In this thesis, we employ colloids and video microscopy to study phase transitions with single particle dynamics. Specifically, we explore the melting transitions and nucleation of solid-solid transitions in thin-film colloidal crystals.

In the first set of experiments, we studied the melting of multilayer colloidal crystals composed of diameter tunable microgel spheres with short-ranged repulsive interactions confined between two walls. We tune the temperature to change volume fraction of samples to drive the melting transition. The film thickness effects on the melting process and on the phase behaviors of single crystal and polycrystalline films are explored: Thick films ( 4 layers) are observed to melt heterogeneously, while thin films (≤ 4 layers) melt homogeneously, ev...[ Read more ]

In this thesis, we employ colloids and video microscopy to study phase transitions with single particle dynamics. Specifically, we explore the melting transitions and nucleation of solid-solid transitions in thin-film colloidal crystals.

In the first set of experiments, we studied the melting of multilayer colloidal crystals composed of diameter tunable microgel spheres with short-ranged repulsive interactions confined between two walls. We tune the temperature to change volume fraction of samples to drive the melting transition. The film thickness effects on the melting process and on the phase behaviors of single crystal and polycrystalline films are explored: Thick films (> 4 layers) are observed to melt heterogeneously, while thin films (≤ 4 layers) melt homogeneously, even for polycrystalline films. Grain-boundary melting dominates other types of melting processes in polycrystals when films are thicker than 12-layers. A novel heterogeneous melting at dislocation is discovered. Thin film melting is qualitatively different: thin films homogeneously melt by generating many small defects which need not nucleate at grain boundaries or dislocations without hexatic phase. The melting of 3D single crystals and buckled phases is also studied by direct observation.

In another set of experiments, we investigated the nucleation of solid-solid transitions between square (⬜) and triangular (△) lattices in colloidal thin films composed of thermally sensitive micro-spheres. We superheated the interior of crystals, including with vacancies, with dislocations and on a grain boundary, by local heating techniques. Two types of nucleation processes were directly observed by video microscopy and studied at the single-particle level. Without flow, the nucleation is a two-step diffusive process: square-lattice(⬜) crystal → a liquid nucleus → triangle-lattice(△) nucleus and its precursor is local particle-exchange loops. The middle liquid nuclei are well explained by small liquid-⬜ interface tension. A post-critical liquid nucleus is circular but become faceted after transforming to △. The growth rate of different facets depends on their coherence. However, under small flow rates the nucleus of the triangle lattice forms by a martensitic ( diffusionless) mechanism and its precursor is a pair of dislocations. Our observation shows the microscopic process of generating a pair of dislocations and triggering more pairs. The initial nucleus is semi-coherent and loses coherency as nucleus growth. Furthermore, we studied the growth stage of post-critical nuclei and found that incoherent facets with higher interfacial energy grow much faster than coherent facets. Nearby nuclei can merge with each other by devoloping an liquid-like channel, distorting ⬜-lattice between them and attracting smaller nuclei by the strain field in the parent matrix.

In addition, I will report my theoretical work on the self-similarity of phase space networks in four models. These models includes: (1) the antiferromagnetic Ising model on a two-dimensional triangular lattice (1a) at the ground states and (1b) above the ground states and (2) the six-vertex model. The two lattice gas models were (3) the one-dimensional lattice gas model and (4) the two-dimensional lattice gas model. We mapped their phase spaces into networks and used box-covering and cluster-growing methods to illustrate their self-similar properties. According to the box-covering method, fractal phase spaces are found in Models 1a, 2, and 3 with long-range power-law correlations in real space, while models 1b and 4 with short-range exponential correlations in real space exhibit nonfractal phase spaces. This behavior agrees with one of the untested assumptions in Tsallis nonextensive statistics.