Mechanistic Exploration of Two-Dimensional Materials from First-Principles to Data-Driven Simulations

Abstract

It has long been assumed that “It is wrong to think the task of physics is to find out how nature is. Physics concerns what we can say about nature.” as proclaimed by Dr. Niels Bohr, and that “There is nothing that living things do that cannot be understood from the point of view that they are made of atoms acting according to the laws of physics” as emphasised by Dr. Richard Feynman. These views converge upon a certitude that underpins this dissertation since the practical innovation in materials science is rooted from a fundamental, atomistic understanding of mechanisms, the energetics, forces, and structural responses that govern materials at their most elementary scale. Two-dimensional (2D) materials, celebrated for their exceptional surface-area-to-volume ratio, tuneable band structures, and quantum-level responsiveness, offer fertile ground for both technological exploration and foundational inquiry. However, it is no longer sufficient to describe what these materials can achieve and rather, it is essential to discern why they behave as they do, what governs their chemical reactivity, structural stability, phase transitions, or emergent interfacial phenomena. This dissertation pursues a computationally meticulous approach, integrating first-principles theory, ab initio molecular dynamics, and machine learning force fields, to explore 2D material behaviour through the lens of atomistic mechanisms. In doing so, it aims to bridge the descriptive and the explanatory, the phenomenological and the predictive. The first segment of this body of work addresses a long-standing simplification in edge energy calculations of graphene. Whilst conventional models treat edge contributions as mere superpositions of armchair (AC) and zigzag (ZZ) segments, this study demonstrates, via density functional theory (DFT), that kink energies take a nontrivial, crucial modulator in stabilising chiral edges. These kink- mediated junctions, often overlooked, prove indispensable for accurately reproducing edge formation energies and equilibrium morphologies, especially under functionalisation. This mechanistic insight is pivotal for guiding edge-specific engineering in graphene nanostructures and supplement the incompetence of the approximations when local site interactions are involved. The focus then shifts to the synthesis and phase control of tin selenides (SnSex) through a low- temperature metal-organic chemical vapour deposition (LT-MOCVD) strategy. This work links the phase-selective growth of SnSe2 and SnSe to thermodynamic stability maps and kinetic barriers, providing mechanistic clarity regarding how precursor ratios and growth temperatures dictate diffusion pathways, energy landscapes, and phase purity. Notably, SnSe2 facilitates two-dimensional in-plane growth owing to its lower diffusion barrier, whilst SnSe, with higher barriers, requires a post-synthetic transformation for the planar phase. These findings reveal how synthetic control at the atomic scale depends upon understanding diffusional anisotropy and interfacial energetics. Following this, the dissertation investigates the oxidation dynamics of transition metal dichalcogenides (TMDs), specifically WS2 and MoS2. Through a series of DFT-calculated substitutional energies, vacancy formation barriers, and phase diagrams consisting of W, S, and O, it is shown that oxidation is not uniform but defect-mediated and rather stable in terms of energetics. Even a single adsorbed oxygen atom significantly lowers the barrier for sulphur removal, resulting in vacancy formation and following substitutional doping. With increasing O-coverage, a cascade of transformations emerges, WSO, WO2, and finally, WO3, driven by local energetics and facilitated by interlayer oxygen intercalation. This mechanistic model resolves discrepancies in oxidation behaviours observed experimentally and clarifies how phase evolution arises not from abrupt global changes but from local energetic thresholds being crossed. The final study examines the intercalation-driven lifting and phase separation of graphene on Pt (111). Contrary to prior assumptions that oxygen intercalation is responsible for lifting, this work presents compelling theoretical and thermodynamic evidence that carbon monoxide (CO) is the true driver of such morphological transitions. CO intercalation induces a particular binding energy tendency at intermediate coverage (~ 40-60%), fostering patchwise lifting and domain formation. By contrast, oxygen, lacking comparable energetic favourability, induces only mild vertical expansion without phase separation. To capture these mesoscale effects, machine learning force fields (MLFF) trained on ab initio data were deployed in large-scale molecular dynamics simulations. These simulations, spanning over 12,000 atoms and nanosecond timescales, revealed how CO-based intercalation spontaneously partitions the graphene overlayer into selectively lifted regions, an emergent mosaic driven by local energy minimisation. This study not only reconciles longstanding experimental ambiguities but demonstrates the potency of intercalation as a tool for programmable nano-structuring at the interface of 2D materials and metals. In summary, this dissertation affirms that to meaningfully engineer, grow, or functionalise 2D materials, one needs to comprehend first the microscopic mechanisms, bonding motifs, defect energetics, diffusion dynamics, and interfacial forces, that govern their behaviour. Computational approaches, when meticulously constructed and prudently applied, provide not merely an alternative to experimentation, but a complementary and explanatory framework, capable of describing phenomena, which seem unable to understand to the outcomes of the measurement directly. By grounding the study of 2D materials in energetic principles and atomic interactions, this work offers a roadmap for mechanism-driven materials design, where theory informs practice, and the microscopic informs the macroscopic. In doing so, it echoes the assertion from Dr. Niels Bohr and Dr. Richard Feynman that the laws of physics, when applied with fidelity to atomic realities, yield not only elegant models but also practical, predictive tools for advancing science and technology.