Atomistic Simulations of Interstellar Carbon Nanostructures and Multiscale Modeling of Spinodal Decomposition in Metal Alloys

Abstract

Computational materials science (CMS) techniques provide a ready platform for design and deployment of tailored materials as well as for understanding the underlying conditions under which materials were processed. Using electronic structure methods (e.g. density functional theory, DFT), atomistic modeling techniques (e.g. molecular dynamics, MD, microstructural evolution methods, e.g., phase field, PF) and thermodynamic modeling frameworks (e.g. CALculation of PHAse Diagram, CALPHAD), structure-property relations of materials ranging from atomic- to macro-scales can be thoroughly investigated, enabling important insights for characterizing a wide variety of materials, including materials for engineering and technological applications as well as naturally occurring materials (e.g. planetary materials and geomaterials). In this context, three distinct problems are successfully addressed in this dissertation namely: (Project 1) using MD and DFT to unravel the underlying kinetics of carbon nanostructures synthesis (e.g. C_60 fullerenes and carbon nanotubes) from solid state precursors in astrophysical settings; (Project 2) using a multiscale DFT-CALPHAD-PF approach to model spinodal decomposition in binary metal alloy systems; and (Project 3) using a data-driven computational thermodynamics framework to predict miscibility gap boundary in body-centered cubic (BCC) systems. Project 1: the synthesis pathways that underlie the formation of carbon nanostructures, such as fullerenes and carbon nanotubes in the interstellar medium is an open scientific question. Current understanding of how structures like carbon nanotubes (CNT) and C_60 fullerene structures can be synthesized and incorporated into diffuse clouds is very limited. In this regard, using a combination of MD and DFT calculations, a new ‘top-down’ mechanism for in situ synthesis of C_60 and single-wall CNT from few-layer thick graphite grains, is demonstrated; notably, the identified pathways show that gas-phase precursors are not required for carbon nanostructure formation in astrophysical settings. Specifically, in finite-sized graphite grains, the atoms at and near the sheet-edges are sufficiently perturbed due to thermal and mechanical stimuli that are typical of circumstellar shells, leading to interlayer bonding between successive sheets. As a result, the graphite sheet-edges undergo restructuring leading to the formation of different carbon nanostructures. In the case of C_60 formation, the underlying chemical pathways consist of a series of steps that involve bond-breakage and subsequent local rearrangement of atoms, with the activation energy barriers of the rate-limiting step(s) being comparable to the energetics of Stone-Wales rearrangement reactions. We have published our work on the C60 formation. In the case of formation of CNTs, curling of successive graphite sheets occurs in conjunction with inter-sheet bonding, ultimately resulting in tubular structures. The identified mechanism preferentially leads to the formation of armchair tubes (n,n) over zigzag and other chiral CNT. Our work on CNT formation mechanism is yet to be submitted for publication. Project 2: First-principles DFT calculations along with CALPHAD framework and PF modeling were conducted to study spinodal decomposition in a Mo-V binary alloy system. Mo and V are irreplaceable alloying agents in tailoring microstructures and enhancing mechanical and electrochemical properties of dual-phase steels, high-entropy alloys, and ultrahigh-strength steels. Recent experiments have confirmed the miscibility gap in isothermally treated Mo-V binary systems below certain temperatures and identified spinodal decomposition and nucleation growth as two possible phase decomposition mechanisms. The experimental considerations were extended to incorporate the role of cooling rates, compositionally-generated elastic stresses, and applied tractions on the decomposition kinetics in the Mo-V binary alloy system. The PF simulations were performed for temperatures within the miscibility gap, various cooling rates, and several loading conditions. Interestingly, the PF simulations show the ability to tailor the microstructures by varying the cooling rates in conjunction with external applied tractions, providing new pathways for stress driven microstructural engineering of metal alloys. We have not published this work yet. Project 3: In this work, a new approach that incorporates machine learning with the cluster variation method was developed to calculate phase diagrams by explicitly considering short range order (SRO). Specifically, for the chosen clusters, SRO values were expressed through internal variables referred to as correlation functions (CFs). Determination of these CFs at each thermodynamic state of the system requires solving a set of nonlinear equations using numerical methods. Towards this end, a neural network model was developed to predict the values of the CFs. The NN models were trained for the bcc phase under the tetrahedron approximation for both ordering and phase separating systems, enabling accurate prediction of the values of the CFs and thereby Helmholtz energy and the phase diagram with significantly less computational burden than that of conventional methods used. Such methods hold enormous promise for accelerating the prediction of phase diagrams of multicomponent multiphase systems. We have already published this work.