Colloidal particles with spatially anisotropic (direction-dependent) interactions are known to self-aggregate into a rich variety of ordered and disordered structures, driven by induced dipoles or patterned surface interactions. In this thesis, we investigate the self-aggregation dynamics of spherical colloidal particle systems in bulk that interact via anisotropic interactions, using Brownian dynamics simulations. The central focus lies on understanding how local structural motifs emerge during aggregation and how transitions between these motifs govern the pathway toward larger-scale assemblies. A primary objective of this work is to elucidate the role of non-equilibrium effects in shaping colloidal self-aggregation and the associated structural transitions. To this end, we study two distinct classes of non-equilibrium driving: activity, modeled via self-propelled (active) Brownian particles, and non-reciprocal inter-particle interactions. In the first part of the thesis, we investigate the impact of non-reciprocal interactions in anisotropically interacting colloidal systems. While most previous studies of non-reciprocity have focused on isotropic interactions, its role in anisotropic self-assembly remains largely unexplored. We extend a model of field-responsive colloidal particles subjected to orthogonal electric and magnetic fields by incorporating tunable non-reciprocal interactions. We show that even weak non-reciprocity leads to effective self-propulsion of particle pairs, forming active colloidal dimers that exhibit both translational and rotational motion. These active units profoundly alter aggregation kinetics and long-time behavior, leading to faster assembly, dynamic restructuring, and, under certain conditions, analogues of motility-induced phase separation. Importantly, the independent tunability of anisotropy and non-reciprocity in our model allows us to demonstrate that strong anisotropy suppresses such phase separation and can even destroy ordered aggregates under sufficiently strong non-reciprocal driving. Overall, non-reciprocal interactions are shown to destabilize kinetically trapped structures and open new non-equilibrium pathways for self-assembly. In the second part of this thesis, we introduce self-propulsion to triblock Janus colloidal particles to investigate its role in facilitating the successful self-aggregation of Kagome structures. Building on earlier work mallory2019activity, we perform extensive simulations over a broader range of temperatures and densities, constructing complete state diagrams for both passive and active systems. Motivated by the observation that activity strongly enhances Kagome yield at selected state points, we propose an activity-time protocol in which self-propulsion is applied only for a finite duration and subsequently switched off, allowing the system to relax intrinsically toward equilibrium. This approach enables direct comparison with equilibrium Monte Carlo studies romano2011two. We find that activity-assisted assembly yields well-developed Kagome structures across a significantly broader parameter range than purely passive systems, and that the resulting crystallization boundary closely matches that obtained from equilibrium simulations, despite differences in the underlying models. Finally, we develop a coarse-grained, discrete-state model to describe the aggregation dynamics of anisotropic colloidal systems, using the field-responsive dipolar system. Inspired by Markov state models (MSM) developed for protein folding and nanocluster assembly, we identify a small set of discrete local structural states and extract transition probabilities between them directly from Brownian dynamics data. Unlike existing approaches, our framework is applicable to systems containing hundreds to thousands of particles that self-aggregate into multiple clusters simultaneously. We demonstrate a clear separation of time scales between cluster growth and local structural transitions, and incorporate this effect by modeling transition rates as dependent on the size of the largest cluster. By training Gaussian process regressions on size-dependent transition matrices, the model accurately predicts the time evolution of structure populations across different regimes, including scenarios with time-dependent interaction parameters. The resulting reduced description reproduces the full simulation results and provides mechanistic insight into non-equilibrium effects, including transient violations of detailed balance during early aggregation stages. In summary, this thesis makes three primary contributions: (i) it demonstrates how non-reciprocal interactions induce active-like dynamics and accelerate self-assembly in anisotropic colloidal mixtures; (ii) it introduces and validates an activity-time control protocol that significantly enhances the assembly of complex target structures in patchy colloids; and (iii) it presents a novel MSM-based coarse-graining framework capable of predicting long-time self-assembly kinetics and revealing the underlying structural transition mechanisms.
Salman Fariz Navas (Thu,) studied this question.