Date of Award
2025
Degree Type
Dissertation
Degree Name
Doctor of Philosophy in Mechanical Engineering and Applied Mechanics
Department
Mechanical, Industrial and Systems Engineering
First Advisor
Ashutosh Giri
Abstract
As device dimensions shrink in the current era of miniaturization, effective thermal management at the material level has become critical. In ultrasmall device lengths, traditional electronic heat transport becomes severely limited as boundary scattering effects curtail the electronic transport because the electronic mean free paths substantially exceed those of phonons. The resulting high-power- density devices generate thermal hot spots that compromise both performance and long-term reliability. This challenge has intensified the search for materials with superior phonon-mediated heat transport, a key requirement for effective thermal management in next-generation nanoelectronics. Accurately predicting thermal transport properties with near-experimental accuracy is therefore essential. Ab initio methods such as density functional theory (DFT) provide the necessary quantum-mechanical accuracy to examine phonon-driven thermal transport. However, these ab initio calculations often demands simulations of hundreds to thousands of atoms, making ab initio calculations prohibitively expensive. Classical molecular dynamics (MD) provides computational efficiency, but the traditional empirical potentials used in classical MD lack the accuracy and transferability needed to reliably model complex thermal transport behavior across diverse conditions. To overcome this limitation, this thesis leverages first-principles data and machine-learned interatomic potentials (MLPs) to enable accurate MD simulations at scales and timescales far beyond the reach of DFT. This thesis aims to provide a comprehensive understanding of thermal transport under extreme temperatures and pressures conditions that have remained largely inaccessible to conventional experimental and computational methods. Gaining insight into these regimes is crucial for tuning thermal conductivity and understanding the underlying energy-carrier dynamics, with direct relevance to wide array of applications, including next-generation interconnects, biomedical devices, sensors, and thermal circuits.
This thesis presents a combination of first-principles calculations and machine- learning-interatomic-potential-based molecular dynamics simulations for accurate investigations of phonon-mediated heat transport across metals and polar insulators while examining adsorptive as well as thermal properties for polymeric frameworks using machine learning based modeling. MLPs are trained on comprehensive ab initio datasets that capture the full range of atomic configurations and interactions relevant to thermal transport under extreme conditions of temperatures as well as pressures. Rigorous benchmarking against DFT ensures that ab initio accuracy is preserved while enabling efficient large-scale MD simulations. Using the Green-Kubo formalism, we compute phonon thermal conductivity and uncover new pressure-induced transport regimes in metals and polar insulators.
Our results show that in tungsten, heat conduction transitions from electron-dominated (~70% electronic at ambient conditions) to phonon-dominated (~70% phononic at 100 GPa). Iridium displays a similar effect, with lattice thermal conductivity rising from ~30 W m-1 K-1 to ~120 W m-1 K-1 under compression. In lithium halides (LiBr, LiI), we observe dramatic stiffening where bulk modulus increases by ~15x and thermal conductivity rises by ~100x up to 90 GPa, revealing an ultra-stiff, ultra-conductive phonon transport regime. Complementing these studies, we develop machine-learning models to predict adsorptive and thermal properties of covalent organic frameworks (COFs), enabling the screening from large virgin COF databases containing thousands of diverse COFs. Our models identify ~214 3D-COFs with thermal conductivity exceeding 1 W m-1 K-1 and volumetric deliverable capacities above 175 vSTP/v, including 18 high-performance COFs with both K > 1 W m-1 K-1 and VDC > 200 vSTP/v.
Overall, this thesis establishes a comprehensive and accurate phonon-engineering framework that reveals new regimes of thermal transport in metals and polar insulators under extreme conditions, while enabling large-scale screening of polymeric frameworks for adsorption and thermal applications. This work provides predictive pathways for designing next-generation materials optimized for heat dissipation at the nanoscale, where phonons are harnessed to play the dominant role in thermal management while accurately modeling the adsorptive and thermal properties in COFs using ML models.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Bhatt, Niraj, "MACHINE LEARNING GUIDED INSIGHTS INTO PHONON SCATTERING MECHANISMS FOR TUNABLE THERMAL TRANSPORT IN MATERIALS FROM FIRST-PRINCIPLES" (2025). Open Access Dissertations. Paper 4525.
https://digitalcommons.uri.edu/oa_diss/4525