| dc.description.abstract | Connecting macroscopic behavior of matter to the microscopic interactions is the central goal of condensed matter physics. Yet, achieving such a challenging goal in atomic and molecular systems would require tracking all particles in the system up to the length scale of atomic resolution (≈ nm) and timescales of picoseconds. A typical workaround through this daunting challenge is to instead model the system with colloidal suspensions, where length scales (≈ μm)
and timescales(≈ ms) are more accessible to experiments. Over the last two decades, colloidal systems have been used to mimic atomic systems as they exhibit various phases such as liquids, crystals, gases, and glasses which have enabled study of various processes such as nucleation and growth of crystals, solid-solid phase transition, liquid crystals, liquid-liquid phase transitions and many more. In this thesis, we utilize colloids to gain insights into the phenomenon of glass formation as well as relaxation of a metastable glass into a crystal. Unlike crystals which have unique structural signatures that defines their macroscopic properties, glasses possess structure that are akin to liquids. Nevertheless, they exhibit rigidity similar to solids. Physicists have tried to link the properties of glasses to the underlying aperiodic structure over the past three decades but the studies have remained inconclusive. This is attributed to the fact that standard correlation functions that could easily identify ordered structures cannot distinguish between distinct amorphous configurations. This thesis is aimed at gaining insights into the mechanisms of glass formation from experiments on colloidal glasses. It is done by designing analytical tools towards recognizing patterns in amorphous configurations which aid in understanding structure and its relation to dynamics in glassy systems. We identify distinct signatures for these disordered structures in two different ways. We designed a protocol to determine amorphous-amorphous interfaces by defining self-induced pins to critically evaluate the assumptions of Random First Order Transition (RFOT) theory, which is a prominent thermodynamic approach to glass transition. Second, we exploited machine learning algorithms to identify distinct structural features that are responsible for relaxation to predict
where crystallization occurs in glass. The method also helped unveil possible structural connections for another prominent theory for glass transition i.e, Dynamic Facilitation which is a kinetic approach. The thesis is organized into two preliminary chapters followed by work chapters which describe in detail the above findings and a concluding chapter at the end.
In Chapter 1 we introduce colloidal suspensions that are typically used as model systems to study phase behavior in condensed matter systems, which in our case was exploited to study glass transition and devitrification. Further, we
present the formulations behind relevant theories of glass formation and briefly explain the theoretical and simulation developments involving devitrification. In addition, we discuss the principle behind machine learning and describe concisely algorithms used to achieve diverse goals. In particular, we discuss supervised learning approach in detail which we utilize to develop connections between structure and dynamics of glasses.
In Chapter 2 we describe the experimental methods utilized to perform the work presented in this thesis. Experimental systems in chapters 4 and 5 utilize polystyrene particles. We describe the synthesis protocols used to develop such systems in chapter 2. Details of experimental system used to realize devitrification in chapter 3 are described in chapter 2. The chapter also describes confocal microscopy, the instrument used for performing devitrification experiments.
Chapter 3 unravels how glasses transform into crystals, a process known as devitrification. Glasses are inherently unstable to crystallization. However, the transformation remains poorly understood as it occurs whilst the dynamics in the glass stay frozen at the particle scale. In Chapter 3, through single-particle-resolved imaging experiments, we show that due to frozen-in density inhomogeneities, a soft colloidal glass crystallizes via two distinct pathways. In the poorly packed regions of the glass, crystallinity grew smoothly due to local particle shuffles, whereas in the well-packed regions, we observed abrupt jumps in crystallinity that were triggered by avalanches - cooperative rearrangements involving many tens of particles. Importantly, we show that softness - a structural order parameter determined through
machine learning methods - not only predicts where crystallization initiates in a glass but is also sensitive to the crystallization pathway. Such a causal connection between the structure and stability of a glass has so far remained elusive. Devising strategies to manipulate softness may thus prove invaluable in realizing long-lived glassy states.
Chapter 4 deals with measurement of surface tension of amorphous-amorphous interfaces which is one of the basic ingredients for Random First Order Transition (RFOT) theory - a prominent thermodynamic approach to glass transition. The relaxation dynamics in liquids on approaching their glass transition was found to become increasingly cooperative and the relaxing regions also become more compact in shape. Of the many theories of the glass transition, only RFOT anticipates the surface tension of relaxing regions to play a role in deciding both their size and morphology. However, owing to the amorphous nature of the relaxing regions, even the identification of their interfaces has not been possible in experiments hitherto.
In Chapter 4, we devise an analytical method to directly quantify the dynamics of amorphous-amorphous interfaces in bulk supercooled colloidal liquids. Our procedure also helped unveil a non-monotonic evolution in dynamical correlations with supercooling in bulk liquids. We measure the surface tension of the interfaces and show that it increases rapidly across the mode-coupling area fraction. Our experiments support a thermodynamic origin of the glass transition.
Chapter 5 explores the possible structural connections to Dynamic Facilitation theory (DF) - which is a purely kinetic approach to glass transition. Despite decades of intense research, whether the transformation of supercooled liquids into glass is a kinetic phenomenon or a thermodynamic phase transition remains unknown.
In Chapter 5, we analyzed optical microscopy experiments on 2D binary colloidal glass-forming liquids and examined a potential structural origin for localized excitations, which are building blocks of DF. To accomplish this, we utilize machine learning methods to identify a structural order parameter termed softness that has been found to be correlated with reorganization events in supercooled liquids. Both excitations and softness qualitatively capture the dynamical slowdown on approaching the glass transition and motivated us to explore spatial and temporal correlations between them. Our results show that excitations predominantly occur in regions with high softness and the appearance of these high softness regions precedes excitations, thus suggesting a causal connection between them. Thus, unifying dynamical and thermodynamical theories into a single structure-based framework may provide a route to understand the glass transition.
In Chapter6 we discuss our findings and suggest possible future directions. | en_US |
