No description available.

Nickel and iron-based alloys are important structure and cladding materials for modern nuclear reactors due to their high mechanical properties and high corrosion resistance. To understand the radiative and corrosive environment influence on the mechanical strength, computer simulation works are conducted. In particular, this dissertation is focused on multiscale modeling of the effects of radiation-induced solute segregation and oxidation on grain boundary (GB) strength in nickel-based and iron-based alloys. Besides the atomistic scale density functional theory (DFT) based calculations of GB strength, continuum-scale cohesive zone model (CZM) is also used to simulate intergranular fracture at the microstructure scale. First, the effects of solute or impurity segregation at GBs on the GB strength are studied. Thermal annealing or radiation induced segregation of solute and impurity elements to GBs in metallic alloys changes GB chemistry and thus can alter the GB cohesive strength. To understand the underlying mechanisms, first principles based DFT calculations are conducted to study how the segregation of substitutional solute and impurity elements (Al, C, Cr, Cu, P, Si, Ti, Fe, which are present in Ni-based X-750 alloys) influences the cohesive strength of Σ3(111),Σ3(112),Σ5(210) and Σ5(310) GBs in Ni. It is found that C and P show strong embrittlement potencies while Cr and Ti can strengthen GBs in most cases. Other solute elements, including Si, have mixed but insignificant effects on GB strength. In terms of GB character effect, these solute and impurity elements modify the GB strength of the Σ5(210) GB most and that of the Σ3(111) least. Detailed analyses of solute-GB chemical interactions are conducted using electron localization function, charge density map, partial density of states, and Bader charge analysis. The results suggest that the bond type and charge transfer between solutes and Ni atoms at GBs may play important roles on affecting the GB strength. For non-metallic solute elements (C, P, Si), their interstitial forms are also studied but the effects are weaker than their substitutional counterparts. Nickel-base alloys are also susceptible to stress corrosion cracking (SCC), in which the fracture mainly propagates along oxidized grain boundaries (GBs). To understand how oxidation degrades GB strength, the next step is to use density functional theory (DFT) calculations to study three types of oxidized interfaces: partially oxidized GBs, fully oxidized GBs, and oxide/metal interface, using Ni as a model system. For partially oxidized GBs, both substitutional and interstitial oxygen atoms of different concentrations are inserted at three Ni GBs: Σ3(111) coherent twin, Σ3(112) incoherent twin, and Σ5(210). Simulation results show that the GB strength decreases almost linearly with the increasing oxygen coverage at all GBs. Typically, substitutional oxygen causes a stronger embrittlement effect than interstitial oxygen, except at the Σ3(111). In addition, the oxygen-induced mechanical distortion has a much smaller contribution to the embrittlement than its chemical effect, except for oxygen interstitials at the Σ3(111). For the fully oxidized GBs, three NiO GBs of the same types are studied. Although the strengths of Σ3(112) and Σ5(210) NiO GBs are much weaker than the Ni counterparts, the Σ3(111) NiO GB has a higher strength than that in Ni, indicating that Σ3(111) GB may be difficult to fracture during SCC. Finally, the strength of a Ni/NiO interface is found to be the weakest among all interfaces studied, suggesting the metal/oxide interface could be a favorable crack initiation site when the tensile stress is low. Furthermore, the effects of co-segregation of oxygen and solute/impurity elements on GB strength are studied by DFT, using the 5(210) GB in an face-centered-cubic (FCC) Fe as a model system. Four elements (Cr, Ni, P, Si) that are commonly present in stainless steels are selected. Regarding the effects of single elements on GB strength, Ni and Cr are found to the increase the GB strength, while both P and Si have embrittlement effects. When each of them is combined with oxygen at the GB, the synergetic effect can be different from the linear sum of individual contributions. The synergetic effect also depends on the spatial arrangement of solute elements and oxygen. If they are aligned on the same plane at the GB, the synergetic effect is similar to the linear sum, although P and Si show stronger embrittlement potencies when they combine with both interstitial and substitutional oxygen. When they are arranged on a trans-plane structure, only nickel combined with oxygen show larger embrittlement potencies than the linear sum in all cases. Crystal Orbital Hamilton Populations analysis of bonding and anti-bonding states is conducted to interpret how the interaction between solutes and oxygen impacts GB strength. Finally, the continuum-scale CZM method, which is based on the bilinear mixed mode traction separation law, is used to model SCC-induced intergranular fracture in polycrystalline Ni and Fe based alloys in the MOOSE framework. The previous DFT results are used to justify the input parameters for the oxidation-induced GB strength degradation. In this study, it is found that the crack path does not always propagate along the weak GBs. As expected, the fracture prefers to occur at the GB orientations perpendicular to the loading direction. In addition, triple junctions can arrest or deflect fracture propagation, which is consistent with experimental observations. Simulation results also indicate that percolated weak GBs will lead to a much lower fracture stress compared to the discontinuous ones. / Doctor of Philosophy / Iron and Nickel based alloys are important structural materials for nuclear reactors due to their good mechanical properties, corrosion resistance, and radiation resistance. Under radiation and corrosive conditions, those alloys are susceptible to radiation induced segregation (RIS) and stress corrosion cracking (SCC). This dissertation is mainly focused on understanding the influence of the two effects on grain boundary (GB) strength. Systematic atomistic scale density functional theory (DFT) simulations are applied for the nickel and iron systems. Based on the DFT results, cohesive zone model is utilized for the continuum scale fracture simulation in nickel and iron polycrystal. First, DFT calculations are conducted for studying the RIS effect on the GB strength in nickel. Al, Cr, Cu, C, Si, P, Fe, and Ti are chosen as segregated element. Σ3(111), Σ3(112), Σ5(210), Σ5(310) four types of GBs are built for GB strength calculations. It is found that substitutional C and P always embrittle the GB, while substitutional Ti and Cr can strengthen the GB in most cases. Partial density of states (PDOS) analysis indicates the formation of C-Ni and P-Ni covalent bonds is the possible reason for their embrittlement effects. Bader charge analysis shows negatively charged elements likely reduce the GB strength. Interstitial element segregation is applied for non-metal elements (C, P, and Si). The results indicate interstitial elements have weaker effects than substitution ones. On the next stage to study the SCC effect, DFT calculations are performed for nickel Σ3(111), Σ3(112), and Σ5(210) GBs with difference oxygen concentration and oxygen incorporation types. Besides partially oxidized GBs, fully oxidized GBs (NiO GBs) and metal-oxide interface are also constructed for comparison. Simulation results show that the GB strength decreases nearly monotonically as oxygen concentration goes up. Typically, substitution oxygen causes a larger embrittlement effect than interstitial oxygen except at Σ3(111). It is found that the large mechanical distortion in this coherent twin GB contributes significantly to the GB strength drop. NiO GBs can be weak (Σ3(112),Σ5(210)) or strong (Σ3(111)). NiO/Ni interface shows lowest strength compared with partially and fully oxidized GBs, indicating the importance of the metal-oxide interface in the SCC process. Furthermore, the combined effects between segregated elements and oxygen are studied in face center cubic (FCC) iron system. In this part only Σ5(210) GB is selected with substitutional Cr, Ni, P, and Si as segregated elements. The results of single element effects show Cr can strength the GB while P has an opposite effect. Other two elements show little effect. For the co-segregation effects, the trans-plane structures have larger embrittlement potencies than in-plane ones for Ni, suggesting the GB strength can also be affected by the spatial arrangement of segregated elements. Finally, cohesive zone model is applied for fracture simulations in polycrystalline nickel and iron under tensile loading condition. It is found that intergranular fracture depends on both GB strength and orientation. GBs perpendicular to the loading direction have higher chances to crack. It is also found the percolated weak GBs induce larger strength drop than the discontinuous ones.

<p dir="ltr">Theoretical development and implementation of Partition Density Functional Theory, a quantum density embedding framework for electronic structure simulation.</p>

Yes / Materials-binding peptides offer promising routes to the production of tailored Pd nanomaterials in aqueous media, enabling the optimization of catalytic properties. However, the atomic-scale details needed to make these advances are relatively scarce and challenging to obtain. Molecular simulations can provide key insights into the structure of peptides adsorbed at the aqueous Pd interface, provided that the force-field can appropriately capture the relevant bio-interface interactions. Here, we introduce and apply a new polarizable force field, PdP-CHARMM, for the simulation of biomolecule–Pd binding under aqueous conditions. PdP-CHARMM was parametrized with density functional theory (DFT) calculations, using a process compatible with similar polarizable force-fields created for Ag and Au surfaces, ultimately enabling a direct comparison of peptide binding modes across these metal substrates. As part of our process for developing PdP-CHARMM, we provide an extensive study of the performance of ten different dispersion-inclusive DFT functionals in recovering biomolecule–Pd(111) binding. We use the functional with best all-round performance to create PdP-CHARMM.We then employ PdP-CHARMM and metadynamics simulations to estimate the adsorption free energy for a range of amino acids at the aqueous Pd(111) interface. Our findings suggest that only His and Met favor direct contact with the Pd substrate, which we attribute to a remarkably robust interfacial solvation layering. Replica-exchange with solute tempering molecular dynamics simulations of two experimentally-identified Pd-binding peptides also indicate surface contact to be chiefly mediated by His and Met residues at aqueous Pd(111). Adsorption of these two peptides was also predicted for the Au(111) interface, revealing distinct differences in both the solvation structure and modes of peptide adsorption at the Au and Pd interfaces. We propose that this sharp contrast in peptide binding is largely due to the differences in interfacial solvent structuring. / Air Force Office for Scientfi c Research (Grant #FA9550-12-1-0226)

Simulationen elektronischer Strukturen ermöglichen die Bestimmung grundlegender Eigenschaften von Materialien ohne jegliche Experimente. Sie zählen deshalb zu den Standardwerkzeugen, mit denen Fortschritte in materialwissenschaftlichen und chemischen Anwendungen vorangetrieben werden. In den letzten Jahrzehnten hat sich die Dichtefunktionaltheorie (DFT) aufgrund ihrer ausgezeichneten Balance zwischen Genauigkeit und Rechenkosten als die beliebteste Simulationstechnik für elektronische Strukturen etabliert. Jedoch verlangen drängende gesellschaftliche und technologische Herausforderungen nach Lösungen für immer komplexere wissenschaftliche Fragestellungen, sodass selbst die effizientesten DFT-Programme nicht mehr in der Lage sind, Antworten in angemessener Zeit und mit den verfügbaren Rechenressourcen zu liefern. Daher wächst das Interesse an Ansätzen des maschinellen Lernens (ML), die darauf abzielen, Modelle bereitzustellen, die die Vorhersagekraft von DFT-Rechnungen zu vernachlässigbaren Kosten replizieren. In dieser Arbeit wird gezeigt, dass solche ML-DFT Ansätze bisher nicht in der Lage sind, das Vorhersagen der elektronischen Struktur von Materialien auf DFT-Niveau vollständig abzubilden. Davon ausgehend wird in dieser Arbeit ein neuer Ansatz für ML-DFT Modelle vorgestellt. Es wird ein umfassendes Framework für das Training von ML-DFT-Modellen auf Grundlage einer lokalen Darstellung der elektronischen Struktur entwickelt, welcher auch Details wie Strategien zur Datengeneration und Hyperparameteroptimierung beinhaltet. Es werden Ergebnisse vorgestellt, die zeigen, dass mit diesem Framework trainierte Modelle die breite Palette der Vorhersagefähigkeit sowie Genauigkeit von DFT-Simulationen zu drastisch reduzierten Kosten replizieren. Weiterhin wird die allgemeine Nützlichkeit dieses Ansatzes demonstriert, indem Modelle über Längenskalen, Phasengrenzen und Temperaturbereiche hinweg angewendet werden.:List of Tables 10 List of Figures 12 Mathematical notation and abbreviations 14 1 Introduction 19 2 Background 23 2.1 Density Functional Theory 23 2.2 Sampling of Observables 35 2.3 Machine Learning and Neural Networks 37 2.4 Hyperparameter Optimization 46 2.5 Density Functional Theory Machine Learning Models 50 3 Scalable Density Functional Theory Machine Learning Models 59 3.1 General Framework 59 3.2 Descriptors 67 3.3 Data Generation 69 3.4 Verification of accuracy 78 3.5 Determination of Hyperparameters 87 4 Transferability and Scalability of Models 99 4.1 Large Length Scales 100 4.2 Phase Boundaries 108 4.3 Temperature Ranges 117 5 Summary and Outlook 131 Appendices 136 A Computational Details of the Materials Learning Algorithms framework 137 B Data Sets, Models, and Hyperparameter Tuning 145 Bibliography 161 / Electronic structure simulations allow researchers to compute fundamental properties of materials without the need for experimentation. As such, they routinely aid in propelling scientific advancements across materials science and chemical applications. Over the past decades, density functional theory (DFT) has emerged as the most popular technique for electronic structure simulations, due to its excellent balance between accuracy and computational cost. Yet, pressing societal and technological questions demand solutions for problems of ever-increasing complexity. Even the most efficient DFT implementations are no longer capable of providing answers in an adequate amount of time and with available computational resources. Thus, there is a growing interest in machine learning (ML) based approaches within the electronic structure community, aimed at providing models that replicate the predictive power of DFT at negligible cost. Within this work it will be shown that such ML-DFT approaches, up until now, do not succeed in fully encapsulating the level of electronic structure predictions DFT provides. Based on this assessment, a novel approach to ML-DFT models is presented within this thesis. An exhaustive framework for training ML-DFT models based on a local representation of the electronic structure is developed, including minute treatment of technical issues such as data generation techniques and hyperparameter optimization strategies. Models found via this framework recover the wide array of predictive capabilities of DFT simulations at drastically reduced cost, while retaining DFT levels of accuracy. It is further demonstrated how such models can be used across differently sized atomic systems, phase boundaries and temperature ranges, underlining the general usefulness of this approach.:List of Tables 10 List of Figures 12 Mathematical notation and abbreviations 14 1 Introduction 19 2 Background 23 2.1 Density Functional Theory 23 2.2 Sampling of Observables 35 2.3 Machine Learning and Neural Networks 37 2.4 Hyperparameter Optimization 46 2.5 Density Functional Theory Machine Learning Models 50 3 Scalable Density Functional Theory Machine Learning Models 59 3.1 General Framework 59 3.2 Descriptors 67 3.3 Data Generation 69 3.4 Verification of accuracy 78 3.5 Determination of Hyperparameters 87 4 Transferability and Scalability of Models 99 4.1 Large Length Scales 100 4.2 Phase Boundaries 108 4.3 Temperature Ranges 117 5 Summary and Outlook 131 Appendices 136 A Computational Details of the Materials Learning Algorithms framework 137 B Data Sets, Models, and Hyperparameter Tuning 145 Bibliography 161

No / The terahertz frequency spectrum of pentaerythritol tetranitrate (PETN) is calculated using Discover[1] with the COMPASS[2] force field, CASTEP[3] and PWscf.[4] The calculations are compared to each other and to terahertz spectra (0.3-3 THz) of crystalline PETN recorded at 4 K. A number of analysis methods are used to characterise the calculated normal modes.

Yes / Cocrystallization is an attractive approach to improving the physicochemical properties of active pharmaceutical ingredients (APIs), which have great potential in drug development. Accordingly, there is a growing need to understand the physicochemical changes that occur upon co-crystallisation. This work focuses on the combined use of spectroscopy and density functional theory (DFT) calculations to understand the molecular structure, hydrogen bond interactions and physicochemical properties of a pharmaceutical cocrystal. Solid-state NMR (ssNMR) spectroscopy can provide detailed molecular structure information on pharmaceutical cocrystals and complexes. It is non-destructive and usually provides deep structural insights that complement well with vibrational spectroscopy. In this work, a cocrystal of an antibiotic drug, nitrofurantoin (NF), with 4-hydroxybenzoic acid (4HBA) is examined to understand the capability of multiple spectroscopic techniques such as infrared (IR), Raman and solid-state NMR spectroscopies, and to confirm the molecular structure and hydrogen bonding of cocrystal systems. The results of IR and Raman spectroscopy showed that for the cocrystal formation, NF and 4HBA molecules interact through N–H⋯O–H interactions between the imide N–H of nitrofurantoin and the phenolic –OH of 4-hydroxybenzoic acid, and these interactions are also confirmed by natural bond orbital (NBO) and quantum theory of atoms in molecules (QTAIM) analyses. It is critical to understand whether a given cocrystal, upon conceiving a modified crystalline structure compared to that of its API, shows enhanced physical and chemical properties or not. Computationally, it is found that the NF–4HBA cocrystal shows softer (more reactive) behaviour in comparison to NF as its cocrystal, NF–4HBA, has a low band gap in comparison to the API, NF. These results demonstrate that the quantum chemical approach predicts accurately how to relate cocrystal with its physical and chemical properties. / BSR meritorious fellowship scheme. The Newton-Bhabha PhD placement award (2017). The Royal Society Seed Corn Research Grant (2018-19)

Quantum chemistry methodologies can be used to address a wide variety of chemical problems. Key to the success of quantum chemistry methodologies, however, is the selection of suitable methodologies for specific problems of interest, which often requires significant assessment. To gauge a number of methodologies, the utility of density functionals (BLYP, B97D, TPSS, M06L, PBE0, B3LYP, M06, and TPSSh) in predicting reaction energetics was examined for model studies of C-O bond activation of methoxyethane and methanol. These species provide excellent representative examples of lignin degradation via C-O bond cleavage. PBE0, which performed better than other considered DFT functionals, was used to investigate late 3d (Fe, Co, and Ni), 4d (Ru, Rh, and Pd), and 5d (Re, Os, and Ir) transition metal atom mediated Cβ -O bond activation of the β–O–4 linkage of lignin. Additionally, the impact of the choice of DFT functionals, basis sets, implicit solvation models, and layered quantum chemical methods (i.e., ONIOM, Our Own N-layered Integrated molecular Orbital and molecular Mechanics) was investigated for the prediction of pKa for a set of Ni-group metal hydrides (M = Ni, Pd, and Pt) in acetonitrile. These investigations have provided insight about the utility of a number of theoretical methods in the computation of thermodynamic properties of transition metal hydrides in solution. As single reference wavefunction methods commonly perform poorly in describing molecular systems that involve bond-breaking and forming or electronic near-degeneracies and are typically best described with computationally costly multireference wavefunction-based methods, it is imperative to a priori analyze the multireference character for molecular systems so that the proper methodology choice is applied. In this work, diagnostic criteria for assessing the multireference character of 4d transition metal-containing molecules was investigated. Four diagnostics were considered in this work, including the weight of the leading configuration of the CASSCF wavefunction, C02; T1, the Frobenius norm of the coupled cluster amplitude vector related to single excitations and D1, the matrix norm of the coupled cluster amplitude vector arising from coupled cluster calculations; and the percent total atomization energy, %TAE. This work demonstrated the need to have different diagnostic criteria for 4d molecules than for main group molecules.

This dissertation applies covariant density functional theory (CDFT), one of the modern theoretical approaches for describing finite nuclei and neutron stars, to investigate the density distribution of nuclei, which is a manifestation of the nodal structure of the single-particle states in physical phenomena, including charge radii and bubbles. A systematic global investigation of differential charge radii has been performed within the CDFT framework for the first time. Available experimental data is compared with theoretical charge radii across the neutron shell closures at N = 28, 50, 82, and 126. Odd-even staggering (OES) in charge radii are believed to be primarily caused by the pairing. Our research proposes a new approach where a considerable contribution to OES in charge radii is provided by the fragmentation of the single-particle content of the ground state in odd-mass nuclei due to particle-vibration coupling. The proton-neutron interaction explained with the nodal structure of the products of the proton and neutron wave functions. However, proton core is responsible for a major contribution to the buildup of differential charge radii. This interaction between protons and neutrons causes a rearrangement of the single-particle density of occupied proton states, which affects the charge radii. According to our microscopic analysis, the shape of the proton potential, the overall proton density, and the energies of the single-particle proton states are all influenced by self-consistency effects, but they have a minimal impact on the differential charge radii. A detailed and microscopic analysis of bubble physics strongly suggests that single-particle processes are primarily responsible for the creation of bubble shapes in superheavy nuclei. The creation of bubble structure is also influenced by nuclear saturation processes and self-consistency effects, and it is dependent on the availability of low-�� single-particle states for occupation since single-particle densities. For the first time, we investigated how nuclear bubbles are formed in the central classically prohibited area at the bottom of the wine bottle potentials, resulting in decreased s state densities at r = 0.

We integrate theoretical, computational and data-driven approaches for the sake of understanding, design and discovery of metal based catalysts. Firstly, we develop theoretical frameworks for predicting electronic descriptors of transition and noble metal alloys, including a physics model of d-band center, and a tight-binding theory of d-band moments to systematically elucidate the distinct electronic structures of novel catalysts. Within this framework, the hybridization of semi-empirical theories with graph neural network and attribution analysis enables accurate prediction equipped with mechanistic insights. In addition, novel physics effect controlling surface reactivity beyond conventional understanding is uncovered. Secondly, we develop a computational and data-driven framework to model high entropy alloy (HEA) catalysis, incorporating thermodynamic descriptor-based phase stability evaluation, surface segregation modeling by deep learning potential-driven molecular simulation and activity prediction through machine learning-embedded electrokinetic model. With this framework, we successfully elucidate the experimentally observed improved activity of PtPdCuNiCo HEA in oxygen reduction reaction. Thirdly, a Bayesian optimization framework is employed to optimize racemic lactide polymerization by searching for stereoselective aluminum (Al) -complex catalysts. We identified multiple new Al-complex molecules that catalyzed either isoselective or heteroselective polymerization. In addition, feature attribution analysis uncovered mechanistically meaningful ligand descriptors that can access quantitative and predictive models for catalyst development. / Doctor of Philosophy / In addressing the critical issues of climate change, energy scarcity, and pollution, the drive towards a sustainable economy has made catalysis a key area of focus. Computational chemistry has revolutionized our understanding of catalysts, especially in identifying and analyzing their active sites. Furthermore, the integration of open-access data and advanced computing has elevated data science as a crucial component in catalysis research. This synergy of computational and data-driven approaches is advancing the development of innovative catalytic materials, marking a significant stride in tackling environmental challenges. In my PhD research, I mainly work on the development of computational and data-driven methods for better understanding, design and discovery of catalytic materials. Firstly, I develop physics models for people to intuitively understand the reactivity of transition and noble metal catalysts. Then I embed the physics models into deep learning models for accurate and insightful predictions. Secondly, for a class of complex metal catalysts called high-entropy alloy (HEA) which is hard to model, I develop a modeling framework by hybridizing computational and data-driven approaches. With this framework, we successfully elucidate the experimentally observed improved activity of PtPdCuNiCo HEA in oxygen reduction reaction which is a key reaction in fuel cell technology. Thirdly, I develop a framework to virtually screen catalyst molecules to optimize polymerization reaction and provide potential candidates to our experimental collaborator to synthesize. Our collaboration leads to the discovery of novel high-performance molecular catalysts.