<div>Data scarcity is intrinsic to many problems in chemical engineering due to physical constraints or cost. This challenge is acute in chemical and materials design applications, where a lack of data is the norm when trying to develop something new for an emerging application. Addressing novel chemical design under these scarcity constraints takes one of two routes: the traditional forward approach, where properties are predicted based on chemical structure, and the recent inverse approach, where structures are predicted based on required properties. Statistical methods such as machine learning (ML) could greatly accelerate chemical design under both frameworks; however, in contrast to the modeling of continuous data types, molecular prediction has many unique obstacles (e.g., spatial and causal relationships, featurization difficulties) that require further ML methods development. Despite these challenges, this work demonstrates how transfer learning and active learning strategies can be used to create successful chemical ML models in data scarce situations.<br></div><div>Transfer learning is a domain of machine learning under which information learned in solving one task is transferred to help in another, more difficult task. Consider the case of a forward design problem involving the search for a molecule with a particular property target with limited existing data, a situation not typically amenable to ML. In these situations, there are often correlated properties that are computationally accessible. As all chemical properties are fundamentally tied to the underlying chemical topology, and because related properties arise due to related moieties, the information contained in the correlated property can be leveraged during model training to help improve the prediction of the data scarce property. Transfer learning is thus a favorable strategy for facilitating high throughput characterization of low-data design spaces.</div><div>Generative chemical models invert the structure-function paradigm, and instead directly suggest new chemical structures that should display the desired application properties. This inversion process is fraught with difficulties but can be improved by training these models with strategically selected chemical information. Structural information contained within this chemical property data is thus transferred to support the generation of new, feasible compounds. Moreover, transfer learning approach helps ensure that the proposed structures exhibit the specified property targets. Recent extensions also utilize thermodynamic reaction data to help promote the synthesizability of suggested compounds. These transfer learning strategies are well-suited for explorative scenarios where the property values being sought are well outside the range of available training data.</div><div>There are situations where property data is so limited that obtaining additional training data is unavoidable. By improving both the predictive and generative qualities of chemical ML models, a fully closed-loop computational search can be conducted using active learning. New molecules in underrepresented property spaces may be iteratively generated by the network, characterized by the network, and used for retraining the network. This allows the model to gradually learn the unknown chemistries required to explore the target regions of chemical space by actively suggesting the new training data it needs. By utilizing active learning, the create-test-refine pathway can be addressed purely in silico. This approach is particularly suitable for multi-target chemical design, where the high dimensionality of the desired property targets exacerbates data scarcity concerns.</div><div>The techniques presented herein can be used to improve both predictive and generative performance of chemical ML models. Transfer learning is demonstrated as a powerful technique for improving the predictive performance of chemical models in situations where a correlated property can be leveraged alongside scarce experimental or computational properties. Inverse design may also be facilitated through the use of transfer learning, where property values can be connected with stable structural features to generate new compounds with targeted properties beyond those observed in the training data. Thus, when the necessary chemical structures are not known, generative networks can directly propose them based on function-structure relationships learned from domain data, and this domain data can even be generated and characterized by the model itself for closed-loop chemical searches in an active learning framework. With recent extensions, these models are compelling techniques for looking at chemical reactions and other data types beyond the individual molecule. Furthermore, the approaches are not limited by choice of model architecture or chemical representation and are expected to be helpful in a variety of data scarce chemical applications.</div>

<div>Organic mixed ion-electron conducting (OMIEC) polymers are capable of transporting both electrons and ions. This unique functionality underpins many emerging applications, including biosensors, electrochemical transistors, and batteries. The fundamental operating principles and structure-function relationships of OMIECs are still being investigated. Computational tools such as coarse-grained molecular dynamics (CGMD), which use simpler representations than in atomistic modeling, are ideal to study OMIECs, as they can explore the slow dynamics and large length scale features of polymers. Nevertheless, methods development is still required for CGMD simulations to accurately describe OMIECs.</div><div><br></div><div>In this thesis, two CGMD simulation approaches have been adopted. One is a so-called "top-down" approach to develop a generic model of OMIECs. Top-down models are phenomenological but capable of exploring a broad space of materials variables, including backbone anisotropy, persistence length, side-chain density, and hydrophilicity. This newly developed model was used to interrogate the effect of side-chain polarity and patterning on OMIEC physics. These studies reproduce experimentally observed polymer swelling while for the first time clarifying several molecular factors affecting charge transport, including the role of trap sites, polaron delocalization, electrolyte percolation, and suggesting side-chain patterning as a potential tool to improve OMIEC performance.</div><div><br></div><div>The second strategy pursued in this thesis is bottom-up CGMD modeling of specific atomistic systems. The bottom-up approach enables CGMD simulations to be quantitatively related to specific materials; yet, the sources of error and methods for addressing them have yet to be systematically established. To address this gap, we have studied the effect of the CG mapping operator, an important CG variable, on the fidelity of atomistic and CGMD simulations. A major observation from this study is that prevailing CGMD methods are underdetermined with respect to atomistic training data. In a separate study, we have proposed a hybrid machine-learning and physics-based CGMD framework that utilizes information from multiple sources and improves on the accuracy of ML-only bottom-up CGMD approaches. </div>

<p> Metal-stabilized carbenes, most commonly formed through the decomposition of  diazoacetates, are extensively employed in organic synthesis. However, several classes of carbenes,  such as vinylidenes, are challenging to utilize in transition metal catalysis due to the instability of  the required diazo precursors. To overcome this challenge, most transition metal-catalyzed  vinylidene transfer and insertion methods rely on alkynes as vinylidene precursors. Only catalysts  that form stable M=C multiple bonds and weak M(π-C≡C) interactions can promote this alkyne  isomerization, and the resultant metal(vinylidene) species is often less reactive compared to free  vinylidenes. The discovery of 1,1-dihaloalkenes as precursors to transition metal vinylidene  complexes has significantly expanded the scope of vinylidene transfer and insertion reactions.  Dinuclear catalysts were found to promote the reductive cyclization of 1,1-dichloroalkenes  containing pendant alkenes to form methylenecycloalkenes, and mechanistic studies are consistent  with the formation of a Ni2(vinylidene) species. Furthermore, these catalysts promote reductive  three-component cycloaddition reactions with 1,1-dichloroalkenes and aldehydes to generate  methylenedioxolanes, which upon treatment with aqueous acid provides access in one step to new,  unsymmetrical aliphathic α-hydroxy ketones that would be difficult to access with existing  methods. Under dilute conditions, an enone byproduct is formed and a DFT model is presented  that accounts for concentration-based reaction selectivity.</p>

Use of Melatonin Derivatives as Spin Traps for Hydroxyl Radicals: A Computational Studies. Aaron Teye Caesar and Dr. Scott Jeffery Kirkby, Department of Chemistry, College of Arts and Sciences, East Tennessee State University, Johnson City, TN. Free radicals, especially reactive oxygen species, have been implicated in several deleterious processes which result in degenerative and cardiovascular diseases. Melatonin (N-acetyl-5-methoxytryptamin, MLT) is a naturally occurring antioxidant which has shown some potential for use as a spin trap. Spin traps react with short lived radicals such as hydroxy (.OH) or superoxide (O2-) to produce more stable products called spin adducts which may be characterized by electron paramagnetic resonance spectroscopy. This work examines whether MLT derivatives show improved spin adduct stability which may enhance their spin trapping characteristics. Electronic structure calculations of MLT, selected derivatives and 2-OH radical products were performed at the HF/6-31G(d), cc-pVDZ and DFT/B3LYP/6-31G(d) and cc-pVDZ levels of theory using NWChem. The stabilization energy was calculated using; ∆Estabilization = Espin adduct – (Espin trap + Ehydroxy radical). In units of hartrees, the results of 2-OHMLT, 2-OHMLT-Me and 2-OHMLT-CN are -0.43738, -1.60054, -1.60380 for HF/6-31G(d); -1.46071, -1.44788 and -1.46173 for DFT/6-31G(d) respectively. Also, HF/cc-pVDZ and DFTB3LYP/cc-pVDZ respectively gave -1.61268, -1.60233, -1.61409 and -1.44929, -0.26318, -1.45521.

Plants and bacteria both have specialized light-harvesting pigment-protein complexes, composed of a network of chromophores encompassed by a protein scaffold, that are involved in photosynthesis. While chromophore, as well as protein, composition and arrangement vary in these light-harvesting complexes, chromophores transfer energy as molecular excitation energy through their complex multi-chromophoric network with near perfect efficiency. Understanding the efficiency of this excitation energy transfer process has been the focus of many interdisciplinary studies. By elucidating the mechanisms involved in efficient excitation energy transfer in biological systems, we are able to guide the design of novel organic materials for their application in photovoltaic systems. Interdisciplinary studies of light-harvesting biological systems leverage advanced spectroscopic techniques and theoretical models to help explain the interaction be- tween excited electronic states. Difficulties in assigning the origin of spectral features in spectroscopy experiments arise from both homogeneous and inhomogeneous effects. Various computational studies have been able to provide theoretical models that help disentangle these effects and provide insight into the origin of some these spectral features. In this work, we present a computational approach that is used to calculate an ensemble of model Hamiltonians for a light-harvesting pigment-protein complex found in algae. To verify the reliability of our model, we compare various computed spec- tra with experimental measurements. Next, we extend our computational approach for parameterizing an ensemble of Hamiltonians for two configurationally unique or- ganic dimers. Finally, we examine the error of some of the approximations made while partitioning “system” and “bath” degrees of freedom when computing molecu- lar properties. Using these methods we are able to provide mechanistic interpretations and explanations of spectral signatures observed in various linear and nonlinear ex- perimental spectra.

The need for renewable and cleaner sources of energy has made biofuels an interesting alternative to fossil fuels, especially in the case of butanol isomers, with their favorable blend properties and low hygroscopicity. Although C4 alcohols are prospective fuels, some key reactions governing their pyrolysis and combustion have not been adequately studied, leading to incomplete kinetic models. Butanol reactions kinetics is poorly understood. Specifically, the unimolecular and H-assisted tautomerism of propen-2-ol to acetone, which are included in butanol combustion kinetic models, are assigned rate parameters based on the analogous unimolecular tautomerism vinyl alcohol ↔ acetaldehyde and H addition to the double bound of iso-butene, respectively. In an attempt to update current kinetic models for tert- and 2-butanol, a theoretical kinetic study of the unimolecular and H-assisted tautomerism, i-C3H5OH⟺CH3COCH3 and i-C3H5OH+Ḣ⟺CH3COCH3+Ḣ, was carried out by means of CCSD(T,FULL)/aug-cc-pVTZ//CCSD(T)/6-31+G(d,p) and CCSD(T)/aug-cc-pVTZ//M062X/cc-pVTZ ab initio calculations, respectively. For H-assisted tautomerism, the reaction takes place in two consecutive steps: i-C3H5OH+Ḣ⟺CH3ĊOHCH3 and CH3ĊOHCH3⟺CH3COCH3+Ḣ. Multistructural torsional anharmonicity and variational transition state theory were considered in a wide temperature and pressure range (200 K – 3000 K, 0.1 kPa – 108 kPa). It was observed that decreasing pressure leads to a decrease in rate constants, describing the expected falloff behavior for both isomerizations. Results for unimolecular tautomerism differ from vinyl alcohol ↔ acetaldehyde analogue reactions, which shows lower rate constant values. Tunneling turned out to be important, especially at low temperatures. Accordingly, pyrolysis simulations in a batch reactor for tert- and 2-butanol with computed unimolecular rate constants showed important differences in comparison with previous results, such as larger acetone yield and quicker propen-2-ol consumption. In the combustion and pyrolysis batch reactor simulations, using all the rate constants computed in this work, H-assisted reactions are limited because H radicals become abundant once the propen-2-ol has been consumed by other reactions, such as the non-catalyzed tautomerism i-C3H5OH⟺CH3COCH3, which becomes one of the main source of acetone. The intermediate radical (CH3ĊOHCH3) is formed exclusively from tert-butanol, with its concentration in 2-butanol oxidation being smaller because the secondary alcohol is unable to produce the radical directly. In all cases, the intermediate is converted effectively to acetone.

Since our society became industrialised, the levels of carbon dioxide in our atmosphere have been steadily rising, to the point where it in early 2020 at is 413 ppm. The high concentration is causing several troubling effects worldwide because of the increase in mean temperature that it creates, which causes longer draughts, more severe floods, and rising seawater levels to name a few. There are a few measures that can be taken to reduce carbon dioxide in the atmosphere, among which there are a number of methods that currently are being researched and/or used. The prospect of capturing carbon dioxide and using it as a carbon building block to make methanol is one solution that is particularly interesting, since it in theory could provide a fuel for combustion engines that is net neutral regarding carbon emission. Methanol can be synthesised from carbon dioxide using a heterogeneous catalyst consisting of copper, Cu, and zinc oxide, ZnO. This research is focused on one of the components of the catalyst, the metal oxide ZnO in the form of crystallites or nanoparticles (ZnO)n. Quantum chemistry is a branch of computational chemistry which is centered on solving the Schrödinger equation for molecular systems. Density functional theory, DFT, is an approach to quantum theory which in this study was used to calculate the geometry and energy of the particles. The supercomputer Tetralith in the National Supercomputer Centre, NSC, was used to carry out the calculations. The DFT calculations utilized the functional B3LYP and the basis set 6-31G (d,p). One of the largest particle sizes studied, (ZnO)20, with a structure that has a large, flat surface, was found to be the most energetically favourable. According to studies, the presence of an oxygen vacancy on the surface of ZnO reduces the amount of activation energy required for CO2 to bond to the particle, which increases the chance of forming CO and thus continuing the process of forming methanol. Two structures of (ZnO)20 were investigated in this regard, where oxygen atoms were removed at different locations, creating four versions of Zn20O19 in total. This proved yet again that the version with a large, flat surface yields the lesser amount of energy when an O atom is removed from the centre of its surface. The adsorption of CO2 to the ZnO clusters was studied by calculating the energy of adsorption, and this showed that it was the second version of (ZnO)20, without an O vacancy, that yielded the least amount of energy, thus being the most favourable species to engage in physisorption with CO2. Lastly, the activation energy was investigated, and a diagram of the reaction process of CO2 adsorbing to Zn20O19 forming (ZnO)20 and CO is presented in this paper, which shows that the required activation energy is 127 kJ/mol.

Surface self-assembly, the spontaneous aggregation of molecules into ordered, sta- ble, noncovalently joined structures in the presence of a surface, is of great importance to the bottom-up manufacturing of materials with desired functionality. As a bulk phenomenon informed by molecular-level interactions, surface self-assembly involves coupled processes spanning multiple length scales. Consequently, a computational ap- proach towards investigating surface self-assembled systems requires a combination of quantum-level electronic structure calculations and large-scale multi-body classical simulations. In this work we use a range of simulation approaches from quantum-based methods, to classical atomistic calculations, to mean-field approximations of bulk mixed phases, and explore the self-assembly strategies of simple dipoles and polyaromatic hydrocarbons on symmetric surfaces. / Graduate

Everything in the universe is made up of elements from the periodic table. Each element has its own role that it plays in the formation of things it makes up. For instance, pencil lead is graphite. A series of honeycomb-like structures made up of carbon stacked on top of one another. Carbon’s neighbor to the left, boron doesn’t like to form such stacked honeycomb-like structures. But, what if there was a way to make boron act like carbon so it did like to form such structures? That question is the basis of the electronic transmutation concept presented in this dissertation. Electronic transmutation states that an element, such as boron, can behave structurally like carbon (form stacked honeycomb structures) if you make them valence (outer most) isoelectronic (“iso”- same; “electronic”- electrons), so both would have the same number of outer most electrons. As a result, chemists would have a new tool to aid in the rational design of new materials.

The main aim of work presented here is to understand photophysical processes of organic dyes and to design better organic molecules/systems which can be applied in many applications such as solar cells, photodynamic therapy, and fluorescence sensors. Developments of novel multichromophore organic materials for the above mentioned applications were made using computational tools. A brief description of the history of computational chemistry was given based on the photochemistry of organic dyes in the introductory chapters and also the importance of basis sets and functionals was discussed in order to produce accurate computational results. Density functional theory (DFT) and time-dependent DFT (TDDFT) calculations were performed to understand the photophysical processes in the porphyrin-perylene bisimide (HTPP-PDI) dyad that exhibited long-lived triplet states. The DFT results show that breaking the rigidity of PDI in HTPP-PDI was responsible for the generation of long-lived triplet states. Furthermore, six porphyrin derivatives were designed by introducing a 4,4’-dicarboxybutadienyl functional group to the porphyrin moiety and studied to investigate the substituent effects on the non-coplanarity, molecular orbitals, and excitation wavelength of the porphyrin donor. Five of the six proposed porphyrin derivatives are promising donors in the HTPP-PDI dyad to replace HTPP for its potential use in photodynamic therapy. Six donor- accepter(s) systems were designed for their potential application in solar cells. Four D-A1-A2 architectural triads, MTPA-TRC-AEAQ, MTPA-TRC-HTPP, MTPA-TRC-PDI, and MTPA-TRC-PBI were designed. The cascade electronic energy levels were obtained and experimentally observed, which lead to sequential electron transfers from 1MTPA* to TRC and then to AEAQ (HTPP/PDI/PBI) module as well as a hole transfer from 1AEAQ*(1HTPP*/1PDI*/1PBI*) to MTPA module. Therefore, all the D-A1-A2 systems we have designed are ambipolar. Interestingly, the lifetime of charge separated states of the newly designed MTPA*+-TRC-AEAQ*- was elongated to 650 ns, an eightfold of that of the donor-acceptor MTPA-TRC parent molecule (80 ns). However, different charge separated state lifetimes were obtained for MTPA*+-TRC-PDI*-(22ns) and MTPA*+-TRC-PBI*-(75ns). The photophysical results suggested that the charge separated state may decay to the triplet state when the charge separated state exhibits a higher energy level than the triplet state. Further, the photovoltaic tests indicated potential applications of MTPA-TRC-AEAQ in solar cells. DFT and TDDFT calculations were performed together with experimental studies to explore the nature of fluorescence enhancement in the anthracene-based sensor after the addition of Zn2+. A 23-fold fluorescence emission was quenched via non-radiative decay pathway in the absence of Zn2+. However, when the Zn2+ chelated to the sensor fluorescence intensity was increased remarkably. A 32-fold fluorescence increase was overserved and calculation results suggested this could be due to the inhibition of the electron-transfer pathway and enhanced rigidity of sensor-Zn2+ complex. The response selectivity of Zn2+ over Ca2+, Mg2+, Cu2+, and Hg2+ ions was also studied using DFT calculations and it was found that Zn2+ has a strong binding affinity to the sensor, which could be a potential application in the detection of Zn2+.