Thesis (MSc)--Stellenbosch University, 2014. / ENGLISH ABSTRACT: A theoretical and experimental study was performed in order to identify factors that influence the propensity of compounds containing anionic functional groups that are commonly found on pharmaceutical drug compounds to form hydrates. A Cambridge Structural Database (CSD) survey was initially undertaken to determine the propensity of different pharmaceutically acceptable anions to form hydrates. The results showed that hydrate formation will take place more regularly when the polarity of the functional group increases. Furthermore, if the charge distribution is very concentrated over the polar groups, hydrate formation will occur more readily. This observation was further investigated by performing a series of potential energy surface (PES) scans for the hydrogen bond (H-bond) in the structure of N-(aminoiminomethyl)-N-methylglycine monohydrate (creatine monohydrate) with various Density Functional Theory (DFT) and Wave Functional Theory (WFT) methods. WFT is often also referred to as ab initio, which refers to the construction of the wave function from first principles when this theory is applied. The scans revealed that several strong and directional H-bonds with different geometrical parameters between the carboxylate group and the water molecule are possible, which suggests that the H-bond plays an important role in driving the formation of pharmaceutical hydrates. A total of 44 hydrate structures were identified that have pharmaceutically acceptable functional groups. Optimisations in the gas phase and in an implicit solvent polarisable continuum solvent model with a variety of solvents showed that there is a significant dependence of the H-bond interaction energy on the anionic group as well as the steric density of surrounding substituents. It was found that the M06-2X method utilising the 6-311++G(d,p) basis set outperformed the other methods that were tested when compared to optimisations performed with the benchmark MP2/aug-cc-pVTZ level of theory. Furthermore, the strength of the H-bond was measured in the 44 experimentally determined structures by using a total of five generalized gradient approximation (GGA) methods, of which two methods contained the DFT-D3 correction. The results of these DFT methods were subsequently compared to results obtained at the benchmark MP2/aug-cc-pVTZ level of theory. The M06-2X method was identified as the most economical method to calculate H-bond energies. It was also found that the H-bond interaction energy shows a substantial dependence on the electrostatic environment. This was observed by a significant decrease in H-bond strength as the relative permittivity of the solvent increases. The effect of steric density on the H-bond interaction energy was investigated by performing hydrogen bond propensity calculations. These values were then compared to the interaction energies of each structure and the results showed that the presence of large bulky substituents can lead to an increase in bond energy by forcing the anionic functional group closer to the water molecule. Contrastingly, the bulky group can also push the anionic group away from the water molecule and result in a decrease in bond energy. Approximate values for the amount of stabilisation offered to the H-bonding system by the surrounding crystalline environment were calculated by optimising the H-bond geometrical parameters of selected compounds with a combination of the M06-2X and MP2 methods utilising the 6-311++G(d,p) basis set. The H-bond interaction energies were then calculated at the M06-2X/6-311++G(d,p) level of theory and compared to the H-bond interaction energies in geometries that have been fully optimised. After these energies were compared and the crystal packing of each structure was investigated, it was found that the packing of some structures within the crystalline environment limits the number of H-bonds that can be formed between the water and the compound of interest. Full optimisation calculations result in structures with cooperative stabilisation, such that more than one H-bond is found between the two fragments. The effect of substituents on H-bond interaction energy was investigated by the addition of six electron-donating and electron-withdrawing groups on four aromatic compounds with different anionic functional groups, namely carboxylate, nitrogen dioxide, sulfonate and phosphonate. It should also be mentioned that the nitrogen dioxide is not an anionic functional group, but it was included as it is a neutral radical that often forms hydrogen bonds. A total of 80 structures were optimised with a combination of the M06-2X and MP2 methods utilising the 6-311++G(d,p) basis set. This was followed by counterpoise corrected single point calculations at the M06-2X/6-311++G(d,p) level of theory. The results showed that the H-bond interaction energy bears no relationship to the inductive strength or the inductive ability of the substituents, but rather the ability of these substituents to rotate the anionic functional group and allow cooperative stabilisation of the H-bond. Furthermore, AIM analysis was performed for the substituted H-bonded aromatic structure. The results showed that electron-donating groups that are placed at the para position yield stronger H-bonds, which is once again accompanied by cooperative stabilisation. Electron-withdrawing groups with sufficient inductive effects can result in a weaker H-bond when placed at the meta position. The effect of water activity (aw) on the hydrate crystal formation was investigated experimentally by performing a series of crystallisations in various solvent mixtures. These mixtures consisted of water mixed with acetone, ethanol and ethyl acetate. A total of three organic acids were used in crystal formation, namely pyridine-4-carboxylic acid (isonicotinic acid), N-amino-iminomethyl-N-methylglycine (creatine) and benzene-1,3,5-tricarboxylic acid. It was found that water activity affects the formation of the hydrate as well as the anhydrous product. Additionally, nucleation and super saturation plays a large role in crystal formation and can serve as an effective technique when the formation of crystals of an appropriate shape and size is required for further analysis. / AFRIKAANSE OPSOMMING: 'n Teoretiese en eksperimentele studie was uitgevoer om faktore te identifiseer wat die geneigdheid van verbindings met anioniese funksionele groepe wat algemeen gevind word op farmaseutiese dwelm verbindings om die hidraat produk te vorm, affekteer. 'n Opname van strukture in die Cambridge Strukturele Databasis (CSD) is onderneem om die geneigdheid van verskillende farmaseutiese aanvaarbare anione om hidrate te vorm te bepaal. Die resultate het getoon dat hidraatvorming meer gereeld plaasvind indien die polariteit van die funksionele groepe toeneem. Verder is daar ook opgemerk dat 'n gekonsentreerde ladingsverspreiding op die polêre groepe ook tot 'n toename in hidraat vorming sal lei. Hierdie waarneming is verder ondersoek deur 'n reeks potensiële energie oppervlak (PES) skanderings van die waterstof binding (H-binding) vir die struktuur van N-amino-iminometiel-N-metielglisien monohidraat (kreatien monohidraat) met verskeie Digtheids-Funksionele Teorie (DFT) en Golffunksie Teorie (WFT) metodes uit te voer. Die skanderings het getoon dat verskeie sterk, gerigte H-bindings met verskillende geometriese parameters tussen die karboksilaatgroep en die watermolekule kan vorm. Hierdie bevindinge lê klem op die belangrike rol wat H-bindings in die vorming van farmaseutiese koolhidrate speel. 'n Totaal van 44 hidraat strukture met farmaseutiese aanvaarbare funksionele groepe was geïdentifiseer. Optimaliserings is in die gas fase asook in 'n implisiete kontinuum polariseerbare oplosmiddel model met 'n verskeidenheid oplosmiddels uitgevoer. Die resultate het 'n beduidende afhanklikheid van die H-binding interaksie-energie op die anioniese groep asook die steriese afkskerming van omringende groepe getoon. Daar is bepaal dat die M06-2X metode wat saam met die 6-311++G(d,p) basisstel die mees akkuraatste resultate gelewer het in vergelyking met die ander DFT metodes asook die MP2/aug-cc-pVTZ maatstaf. Die H-binding se sterkte is vir hierdie strukture bereken deur vyf GGA metodes te gebruik, waarvan twee metodes van die DFT-D3 korreksie gebruik maak. Die resultate van die berekeninge met hierdie DFT metodes is daarna vergelyk met resultate verkry met die MP2/aug-cc-pVTZ maatstaf. Daar is gevolglik bepaal dat die M06-2X metode die mees ekonomiese metode is om H-binding energië te bereken. Die H-binding interaksie energie toon 'n aansienlike afhanklikheid op die diëlektriese konstante van die oplosmiddel aan. Hierdie waarneming is op grond van 'n beduidende afname in die H-binding interaksie-energie indien die relatiewe permittiwiteit van die oplosmiddel verhoog word gemaak. Die effek van steriese digtheid is ondersoek deur waterstofbindinggeneigdheid waardes te bereken. Hierdie waardes is met die interaksie-energië van elke struktuur vergelyk. Die resultate dui daarop dat steries digte groepe tot 'n toename in interaksie energie kan lei wanneer die anioniese funksionele groep nader aan die water molekule gestoot word. Verder is dit ook moontlik vir hierdie steries digte groepe om die anioniese groep weg van die water molekule te stoot en gevolglik 'n afname in interaksie energie te veroorsaak. Benaderde waardes vir die hoeveelheid stabilisering wat die omringende kristallyne omgewing aan die H-binding bied is bereken deur die H-binding geometriese parameters van geselekteerde verbindings met die M06-2X en MP2 metodes en die 6-311++G (d,p) basisstel te optimaliseer. Die H-binding interaksie-energië is gevolglik by die M06-2X/6-311++G(d,p) vlak van teorie bereken en met die H-binding energië in strukture wat volledige optimaliseer is vergelyk. Nadat hierdie waardes vergelyk is, is daar gevind dat die pakking van strukture in the kristallyne omgewing verhoed dat sekere H-bindings tussen die water molekule en die verbinding van belang kan vorm. Strukture wat volledig optimaliseer is, lei tot strukture wat in staat is om koöperatiewe stabilisering te ondergaan. Koöperatiewe stabilisering word gekenmerk deur die vorming van meer as een H-binding tussen twee fragmente. Die effek van substituente op die H-binding interaksie energie is ondersoek deur die bevoeging van ses elektrondonor- en elektronontrekkendegroepe op vier aromatiese verbindings, naamlike die karboksilaatgroep , stikstofdioksied , sulfonaat en fosfonaat. Dit moet ook genoem word dat stikstofdioksied nie 'n anioniese funksionele groep is nie, maar dit was wel ingesluit omdat dit ‘n neutrale radikaal groep is wat dikwels waterstofbindings vorm. 'n Totaal van 80 strukture optimiserings was uitgevoer met 'n kombinasie van die M06-2X en MP2 metodes wat gebruik maak van die 6-311++G(d,p) basisstel. Dit is gevolg deur interaksie-energie berekeninge op die M06-2X/6-311++G(d,p) vlak van teorie. Die resultate het getoon dat daar geen verband tussen die induktiewe vermoë van die substituente en die sterkte van die H-binding is nie, dit is eerder die vermoë van hierdie substituente om die anioniese funksionele groep te laat roteer wat toelaat dat koöperatiewe stabilisering van die H-binding kan geskied. Die AIM analise is op 'n gesubstitueerde H-binding struktuur toegepas. Die resultate het getoon dat elektrondonorgroepe wat by die para posisie geplaas word tot sterker H-bindings sal lei, wat weereens met koöperatiewe stabilisering vergesel word. Elektrononttrekkendegroepe met sterk induktiewe effekte kan tot 'n swakker H-binding lei indien hulle by die meta posisie geplaas word. Die effek van water aktiwiteit (𝑎w) op hidraatkristalvorming is deur die uitvoering van 'n reeks kristallisasies in verskeie oplosmiddelmengsels ondersoek. Hierdie oplosmiddel mengsels bestaan uit water met asetoon, etanol of etielasetaat gemeng. Kristallisasies is vir drie organiese sure, naamlik piridien-4-karboksielsuur, N-amino-iminometiel-N-metielglisien monohidraat en 1,3,5-benseen tri-karboksielsuur uitgevoer. Daar is gevind dat water aktiwiteit 'n invloed op die vorming van die hidraat en watervrye produkte kan hê. Daarbenewens, speel water aktiwiteit 'n belangrike rol in die nukleasie fase van kristalvorming en kan as 'n effektiewe tegniek dien om kristalle van 'n toepaslike vorm en grootte vir verdere analise te verkry.

This thesis presents electronic structure calculations on problems related to the bonding in inorganic coordination compounds and clusters. A wide range of molecules with the ability to exist in different structural forms or electronic states has been selected and density functional theory is systematically applied in order to gain detailed insight into their characteristics and reactivity at the electronic level. First, we address the question of redox non-innocent behaviour of bipyridine in a series of 1st row transition metal complexes. Complexes of the type [M(2,2'-bipyridine)(mes)₂]⁰ (M = Cr, Mn, Fe, Co, Ni; mes = 2,4,6-Me₃C6H₂) and their one-electron reduced forms have been explored. The results clearly show that the anions are best described as complexes of the monoanionic bipyridine radical (S_bpy = 1/2), giving a rationale for the observed structural changes within the ligand. Likewise, we have identified dianionic bipyridine in both the complexes [Zn2(4,4'-bpy)(mes)₄]²⁻ and [Fe(2,2'-bpy)₂]²⁻. In no case have we found evidence for significant metal-to-ligand backbonding. The subject of redox-noninnocence is further revisited in a comparative study of the two complexes [M(o-Clpap)₃] (M = Cr, Mo; o-Clpap = 2-[(2-chloro-phenyl)azo]-pyridine), and their associated electron transfer series. The results indicate that all electron transfer processes are primarily ligand-based, although in the case of the Mo analogue these are coupled to substantial electron density changes at the metal. The ability of pap to form radical anions finds a contrasting case in the di- nuclear Rh complex [Rh₂(μ-p-Clpap)₂ (cod)Cl₂], where the two ligand bridges act as acceptors of strong dπ∗ backbonding from a formally Rh^–I centre. We then direct our attention to the endohedral Zintl clusters [Fe@Ge₁₀]³⁻ and [Mn@Pb₁₂]³⁻, which reveal peculiar topologies. We have probed the electronic factors that influence their geometric preferences, and propose a model based on the shift of electron density from the endo- hedral metal to the cage to account for the observed geometries. Subsequently, we reassess the electronic structure of the xenophilic clusters Mn₂(thf)₄(Fe(CO)₄)₂ and [Mn(Mn(thf)₂)₃(Mn(CO)₄)₃]^–. We conclude that these are best viewed as exchange coupled Mn^II centres bridged by closed- shell carbonylate fragments. In the closing chapter the reduction of NO₂^– to NO by the complex [Cu(tct)(NO₂)]⁺ (tct = cis,cis-1,3,5-tris(cinnamylideneamino)cyclohexane) is studied, a process that mimics the enzyme-catalysed reaction. Two viable pathways for the reaction have been traced and key inter-mediates identified. Both direct release of NO or via decomposition of a Cu-NO complex are kinetically and thermodynamically feasible.

Radiation chemistry is concerned with understanding the chemical kinetics following the application of ionising radiation. There are two main methods for modelling recom- bination and spin dynamics in radiation chemical systems: The Monte Carlo random flights algorithm, in which the trajectories of the diffusing species are followed ex- plicitly and the Independent Reaction Times (IRT) algorithm, where reaction times are sampled from appropriate marginal distribution functions. This thesis reports develop- ments to both methods, and applies them to better understand experimental findings, particularly spin relaxation effects. Chapter 4 introduces current simulation techniques and presents newly developed algorithms and simulation programs (namely Hybrid and Slice) for modelling spatially dependent spin effects. A new analytical approximation for accurately treating ion-pair recombination in low-permittivity solvents in also presented in this chapter. Chapter 5 explores the photodissociation of H₂O₂, where there is some controversy in the literature on the spin state of the precursor. This chapter explores the possibility of reproducing the observed spin polarisation phase using the Radical Pair Mechanism. Chapter 6 presents two new algorithms for treating reactive products in the IRT framework. These have been tested for two chemical systems: (i) photodissociation of H₂O₂ where the ·OH are scavengeable; (ii) water photolysis which produces H⁺, ·OH and e⁻__aq. In the latter case a careful handling of three body correlations is required. Chapter 7 presents simulation results which suggest a strong correlation between scavenging and ion recombination in low permittivity solvents. A path decomposition method has been devised that allows IRT simulations to be corrected for this effect. Chapter 8 presents evidence for spin-entanglement and cross-recombination to act as an extra source of relaxation for ion-recombination in low permittivity solvents. It is hypothesised this effect contributes to the anomalous relaxation times observed for certain cyclic hydrocarbons. Chapter 9 presents an extension of the IRT simulation method to micelles. The kinetics are shown to be accurately described using the mean reaction time and the exponential approximation.

Reductive functionalization of methyl ligands by 3d metal catalysts and two possible side reactions has been studied. Selective oxidation of methane, which is the primary component of natural gas, to methanol (a more easily transportable liquid) using organometallic catalysis, has become more important due to the abundance of domestic natural gas. In this regard, reductive functionalization (RF) of methyl ligands in [M(diimine)2(CH3)(Cl)] (M: VII (d3) through CuII (d9)) complexes, has been studied computationally using density functional techniques. A SN2 mechanism for the nucleophilic attack of hydroxide on the metal-methyl bond, resulting in the formation of methanol, was studied. Similar highly exergonic pathways with very low energy SN2 barriers were observed for the proposed RF mechanism for all complexes studied. To modulate RF pathways closer to thermoneutral for catalytic purposes, a future challenge, paradoxically, requires finding a way to strengthen the metal-methyl bond. Furthermore, DFT calculations suggest that for 3d metals, ligand properties will be of greater importance than metal identity in isolating suitable catalysts for alkane hydroxylation in which reductive functionalization is used to form the C—O bond. Two possible competitive reactions for RF of metal-methyl complexes were studied to understand the factors that lower the selectivity of C—O bond forming reactions. One of them was deprotonation of the methyl group, which leads to formation of a methylene complex and water. The other side reaction was metal-methyl bond dissociation, which was assessed by calculating the bond dissociation free energies of M3d—CH3 bonds. Deprotonation was found to be competitive kinetically for most of the 1st row transition metal-methyl complexes (except for CrII, MnII and CuII), but less favorable thermodynamically as compared to reductive functionalization for all of the studied 1st row transition metal complexes. Metal-carbon bond dissociation was found to be less favorable than the RF reactions for most 3d transition metal complexes studied. The first dinitrogen dicopper (I) complex has been characterized using computational and experimental methods. Low temperature reaction of the tris(pyrazolyl)borate copper(II) hydroxide {iPr2TpCu}2(µ-OH)2 with triphenylsilane under a dinitrogen atmosphere gives the µ -N2 complex, {iPr2TpCu}2(µ -N2). X-ray crystallography reveals an only slightly activated N2 ligand (N-N: 1.111(6) Å) that bridges between two iPr2TpCuI fragments. While DFT studies of mono- and dinuclear copper dinitrogen complexes suggest a weak µ-backbonding between the d10 CuI centers and the N2 ligand, they reveal a degree of cooperativity in the dinuclear Cu-N2-Cu interaction.

Submitted by Angela Maria de Oliveira (amolivei@uepg.br) on 2017-10-25T13:39:14Z No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) Januario Kordiak.pdf: 1907122 bytes, checksum: a65c6272637170bbfa2b17ce2175c777 (MD5) / Made available in DSpace on 2017-10-25T13:39:14Z (GMT). No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) Januario Kordiak.pdf: 1907122 bytes, checksum: a65c6272637170bbfa2b17ce2175c777 (MD5) Previous issue date: 2017-08-25 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / O ensino de ligações químicas é uma das fases mais importantes no processo de ensino-aprendizagem da Química. Todavia, existe uma grande necessidade de introduzir aos educandos do Ensino Médio a conexão entre a ligação química e a sua natureza quântica. Para alcançar esse objetivo utilizou-se o método de Hückel, um método simples desenvolvido em pesquisas em Química Teórica, antes do início da química computacional. Como metodologia avaliativa o método foi contraposto com as concepções prévias e posteriores dos educandos e, separadamente, com as concepções de educandos ensinados pelo método tradicional, ou seja, quadro, giz e livro. Foi observado que o uso de conceitos atuais, principalmente com o uso do computador, melhorou as argumentações por parte dos educandos ao definirem as ligações químicas, indicando que as aplicações da química quântica aliada a programas computacionais contribuem para resultados expressivos no processo de ensino-aprendizagem da Química. / The nature of chemical bond is essential to understand chemistry concepts, mainly, at initial process in high school. From quantum mechanics concepts was possible describe the chemical bond basing in electronic density, which around the nucleus between two atoms. Hückel method was the first method presented to solve the Schrodinger Equation in approximate level and it reached good qualitative results for aromatic compounds indicating the feasibility of the computational chemistry. However, the Hückel method based on matrix is in accordance with molecular geometry proposed; then, the resolution of such method is possible for students in high level. From this idea, we presented the Hückel method as fundamental to understand the chemical bond for students in the first and second years of high school and we applied questions to evaluate the chemical bond knowledge before/after the Hückel method. We observed that there was increase and reinforce of the chemical bond concept for students.

No description available.

In the automotive industry, a kind of fuel additives, known as surfactant, is used to protect metallic surfaces. Its efficiency strongly depends on factors such as temperature, solvent properties and the presence of other surfactants in the system. In this thesis, both classical and ReaxFF molecular dynamics (MD) simulations are used in studying the impacts of these factors on the adsorption of organic surfactants at the fluid-solid interface. Firstly, a classical MD simulation study of competitive adsorption is carried out on a multi-functional phenol and amine surfactant model with ethanol at the oil/iron oxide interface. As the concentration of ethanol increases, the ethanol molecules effectively compete for the adsorption sites on the iron oxide surface. This observation concurs with the experimental findings of similar oil/iron oxide systems. Unlike most MD interfacial studies, ReaxFF MD uses a fully flexible and polarizable solid surface. The second part of the thesis includes a study on the effect of polarity of organic molecules on the structure of iron oxide using ReaxFF-based MD simulations. The simulation results suggest that care must be taken when parameterising empirical and transferable force fields because the fixed charges on a solid slab may not be a perfect representation of the real system, especially when the solid is in contact with polar compounds. Lastly, but not the least, missing ReaxFF interaction parameters for Fe/N have been developed to simulate the adsorption of amine based surfactant on iron oxide. The parameterisation of the force field is done by fitting these interaction parameters to a set of quantum mechanical data involving iron-based clusters. These newly developed parameters are able to capture chemisorption and proton transfer between hexadecylamine and iron oxide.

Density functional theory (DFT) has become the most widely used first-principles computational method to simulate different atomic, molecular, and solid phase systems based on electron density assumptions. The complexity of describing a many-body system has been significantly reduced in DFT. However, it also brings in potential error when dealing with a system that involves the interactions between metallic and non-metallic species. DFT tends to overly-delocalize the electrons in metallic species and sometimes results in the overestimation of reaction energy, metallic properties in insulators, and predicts relative surface stabilities incorrectly in some instances. There are two approaches to overcoming the failure of DFT using standard exchange-correlation functionals: One can either use a higher level of theory (and thus incur a greater computational cost) or one can apply an efficient correction scheme. However, inaccurate corrections and improper calculation models can also lead to more errors. In the beginning of this dissertation, we introduce the correction methods we developed to accurately model the structure and electron density in material surfaces; then we apply the new methods in surface reactivity studies under experimental conditions to rationalize and solve real life problems. We first investigate the post-DFT correction method in predicting the chemisorption energy (Echem) of a NO molecule on transition metal surfaces. We show that DFT systematically enhances back-donation in NO/metal chemistorption from the metal d-band to NO 2π* orbital, and relate the back-donation charge transfer to the promotion of an electron from the 5σ orbital to the 2π* orbital in the gas-phase NO G2Σ-←X2Π excitation. We establish linear relationships between Echem and ΔEG←X and formulate an Echem correction scheme to the (111) surfaces of Pt, Pd, Rh and Ir. As a precursor to further optimization of DFT corrections on transition metal oxide surfaces, we systematically compare the alumina (α-Al2O3) and hematite (α-Fe2O3) (0001) surfaces to study how the atomic positions treatment during geometry optimizations would affect the electronic structure and modeled reactivity, since they are often reported to have a minimal effect. Our results suggest that both can vary significantly in quantitative and qualitative ways between partially constrained or fully relaxed slab models. We continue to use the α-Fe2O3 (0001) surfaces to optimize the Hubbard U method implemented in DFT that determines the Coulomb repulsion correction (Ud) to localize Fe d-electrons. It successfully restores the insulating properties of bulk hematite, but underestimates the stability of the oxygen-terminated surface. It is mainly due to the fact that all the chemically distinct surface Fe atoms were treated the same way. Here we develop a linear response technique to derive specific Ud values for all Fe atoms in several slab geometries. We also find that in a strongly correlated system, the O p-orbitals also need the Hubbard correction (Up) to accurately predict the structural and electronic properties of bulk hematite. Our results show that the site-specific Ud, combined with Up as Ud+p, is crucial in obtaining theoretical results for surface stability that are congruent with the experimental literature results of α-Fe2O3 (0001) surface structure. Besides methodology development, we continue to apply our specific Ud+p method in the engineered application of the Chemical Looping Combustion (CLC) process in which transition metal oxides play the role of oxidizing fuel molecules for full CO2 capture. Current molecular dynamic studies use partially constrained surface models to simulate the CH4 reaction on hematite surfaces without the detailed comparison of the early stage adsorption products. Here we use hematite (α-Fe2O3) and magnetite (Fe3O4) surfaces as analogous to systematically study the early adsorption products of CH4. Our results show that the reaction favors the homolytic pathway on O-terminated surface, and that as a reduced form of hematite, the magnetite surface also shows excellent reactivity on CH4 dissociation. Knowing how to simulate DFT surface model properly we continue to enrich our theoretical methods for more complicated systems under aqueous conditions. We focus on various structures of the lithium-ion battery material, LiCoO2 (LCO) (001) surface, involving hydroxyl groups. We assess the relative stabilities of different surface configurations using a thermodynamic framework, and a second approach using a surface-solvent ion exchange model. We find that for both models the –CoO–H1/2 surface is the most stable structure near the O-rich limit, which corresponds to ambient conditions. We also found that this surface has nonequivalent surface geometry with the stoichiometric –CoO–Li1/2 surface, leading to distinct band structures and surface charge distributions. We go on to probe how those differences affect the surface reactivity in phosphate anion adsorption. All of the work presented in this dissertation reveals the importance of accurately modeled material structures in theoretical studies to achieve correct physical properties and surface reactivity predictions. We hope our DFT correction schemes can continue to contribute to future surface studies and experimental measurements, and to enlighten new ideas in future DFT methodology improvements.

A necessity of managing scientific data is the ability to maintain experimental legacy information without continually modifying the applications that create and use that information. By facilitating the management of scientific data we hope to give scientists the ability to effectively use additional modeling applications and experimental data. We have demonstrated that an extensible interpreter, using a series of stored directives, allows the loading of data from computational chemistry applications into a generic database. Extending the interpreter to support a new application involves supplying a list of directives for each piece of information to be loaded. This research confirms that an extensible interpreter can be used to load computational chemistry experimental data into a generic database. This procedure may be applicable to the loading and retrieving of other types of experimental data without requiring modifications of the loading and retrieving applications.

Techniken des computergestützten Wirkstoffdesigns spielen eine wichtige Rolle bei der Entwicklung neuer Wirkstoffe. Die vorliegende Arbeit befasst sich sowohl mit der Entwicklung als auch mit der praktischen Anwendung von Methoden des strukturbasierten Wirkstoffdesigns. Die Arbeit glieder sich daher in zwei Teile. Der erste Teil beschäftigt sich mit der Entwicklung von empirischen Scoring-Funktionen, die eine Schlüsselrolle im strukturbasierten computergestützen Wirkstoffdesign einnehmen. Grundlage dieser Arbeiten sind die empirischen Deskriptoren und Scoring-Funktionen aus dem SFCscore-Programmpaket. Dabei wurde zunächst untersucht, wie sich die Zusammensetzung der Trainingsdaten auf die Vorhersagen von empirischen Scoring-Funktionen auswirkt. Durch die gezielte Zusammenstellung eines neuen Trainingsdatensatzes wurde versucht, die Spannweite der Vorhersagen zu vergrößern, um so vor allem eine bessere Erkennung von hoch- und niedrig-affinen Komplexen zu erreichen. Die resultierende Funktion erzielte vor allem im niedrig-affinen Bereich verbesserte Vorhersagen. Der zweite Themenkomplex beschäftigt sich ebenfalls mit der verbesserten Separierung von aktiven und inaktiven Verbindungen. Durch den Einsatz der Machine Learning-Methode RandomForest wurden dazu Klassifizierungsmodelle abgeleitet, die im Unterschied zu den klassischen Scoring-Funktionen keinen genauen Score liefern, sondern die Verbindungen nach ihrer potentiellen Aktivität klassifizieren. Am Beispiel des mykobakteriellen Enzyms InhA konnte gezeigt werden, dass derartige Modelle den klassischen Scoring-Funktionen im Bezug auf die Erkennung von aktiven Verbindungen deutlich überlegen sind. Der RandomForest-Algorithmus wurde im nächsten Schritt auch verwendet, um eine neue Scoring-Funktion zur Vorhersage von Bindungsaffinitäten abzuleiten. Diese Funktion wurde unter dem Namen SFCscoreRF in das SFCscore-Programmpaket implementiert. Die Funktion unterschiedet sich in einigen wesentlichen Punkten von den ursprünglichen SFCscore-Funktionen. Zum einen handelt es sich beim RF-Algorithmus um eine nicht-lineare Methode, die im Unterschied zu den klassischen Methoden, die zur Ableitung von Scoring-Funktionen eingesetzt werden, nicht von der Additivität der einzelnen Deskriptoren ausgeht. Der Algorithmus erlaubt außerdem die Verwendung aller verfügbaren SFCscore-Deskriptoren, was eine deutlich umfassendere Repräsentation von Protein-Ligand-Komplexen als Grundlage des Scorings ermöglicht. Für die Ableitung von SFCscoreRF wurden insgesamt 1005 Komplexe im Trainingsdatensatz verwendet. Dieser Datensatz ist somit einer der größten, die bisher für die Ableitung einer empirischen Scoring-Funktion verwendet wurden. Die Evaluierung gegen zwei Benchmark-Datensätze ergab deutlich bessere Vorhersagen von SFCscoreRF im Vergleich zu den ursprünglichen SFCscore-Funktionen. Auch im internationalen Vergleich mit anderen Scoring-Funktion konnten für beide Datensätze Spitzenwerte erreicht werden. Weitere ausgiebige Testungen im Rahmen einer Leave-Cluster-Out-Validierung und die Teilnahme am CSAR 2012 Benchmark Exercise ergaben, dass auch SFCscoreRF Performanceschwankungen bei der Anwendung an proteinspezifischen Datensätzen zeigt - ein Phänomen, dass bei Scoring-Funktionen immer beobachtet wird. Die Analyse der CSAR 2012-Datensätze ergab darüber hinaus wichtige Erkenntnisse im Bezug auf Vorhersage von gedockten Posen sowie über die statistische Signifikanz bei der Evaluierung von Scoring-Funktionen. Die Tatsache, dass empirische Scoring-Funktionen innerhalb eines bestimmten chemischen Raums trainiert wurden, ist ein wichtiger Faktor für die protein-abhängigen Leistungsschwankungen, die in dieser Arbeit beobachtet wurden. Verlässliche Vorhersagen sind nur innerhalb des kalibrierten chemischen Raums möglich. In dieser Arbeit wurden verschiedene Ansätze untersucht, mit denen sich diese ``Applicability Domain'' für die SFCscore-Funktionen definieren lässt. Mit Hilfe von PCA-Analysen ist es gelungen die ``Applicability Domain'' einzelner Funktionen zu visualisieren. Zusätzlich wurden eine Reihe numerischer Deskriptoren getestet, mit den die Vorhersageverlässlichkeit basierend auf der ``Applicability Domain'' abgeschätzt werden könnte. Die RF-Proximity hat sich hier als vielversprechender Ausgangspunkt für weitere Entwicklungen erwiesen. Der zweite Teil der Arbeit beschäftigt sich mit der Entwicklung neuer Inhibitoren für das Chaperon Hsp70, welches eine vielversprechende Zielstruktur für die Therapie des multiplen Myeloms darstellt. Grundlage dieser Arbeiten war eine Leitstruktur, die in einer vorhergehenden Arbeit entdeckt wurde und die vermutlich an einer neuartigen Bindestelle in der Interface-Region zwischen den beiden großen Domänen von Hsp70 angreift. Die Weiterentwicklung und Optimierung dieser Leitstruktur, eines Tetrahydroisochinolinon-Derivats, stand zunächst im Vordergrund. Anhand detaillierter Docking-Analysen wurde der potentielle Bindemodus der Leitstruktur in der Interfaceregion von Hsp70 untersucht. Basierend auf diesen Ergebnissen wurde eine Substanzbibliothek erstellt, die von Kooperationspartnern innerhalb der KFO 216 synthetisiert und biologisch getestet wurde. Die Struktur-Wirkungsbeziehungen, die sich aus diesen experimentellen Daten ableiten lassen, konnten teilweise gut mit den erstellten Docking-Modellen korreliert werden. Andere Effekte konnten anhand der Docking-Posen jedoch nicht erklärt werden. Für die Entwicklung neuer Derivate ist deswegen eine umfassendere experimentelle Charakterisierung und darauf aufbauend eine Verfeinerung der Bindungsmodelle notwendig. Strukturell handelt es sich bei Hsp70 um ein Zwei-Domänen-System, dass verschiedene allostere Zustände einnehmen kann. Um die Auswirkungen der daraus folgenden Flexibilität auf die Stabilität der Struktur und die Bindung von Inhibitoren zu untersuchen, wurden molekulardynamische Simulationen für das Protein durchgeführt. Diese zeigen, dass das Protein tatsächlich eine überdurchschnittlich hohe Flexibilität aufweist, die vor allem durch die relative Bewegung der beiden großen Domänen zueinander dominiert wird. Die Proteinkonformation die in der Kristallstruktur hscaz beobachtet wird, bleibt jedoch in ihrer Grundstruktur in allen vier durchgeführten Simulationen erhalten. Es konnten hingegen keine Hinweise dafür gefunden werden, dass die Mutationen, welche die für die strukturbasierten Arbeiten verwendete Kristallstruktur im Vergleich zum Wildtyp aufweist, einen kritischen Einfluss auf die Gesamtstabilität des Systems haben. Obwohl die Interface-Region zwischen NBD und SBD also in allen Simulationen erhalten bleibt, wird die Konformation in diesem Bereich doch wesentlich durch die Domänenbewegung beeinflusst und variiert. Da dieser Proteinbereich den wahrscheinlichsten Angriffspunkt der Tetrahydroisochinolinone darstellt, wurde der Konformationsraum detailliert untersucht. Wie erwartet weist die Region eine nicht unerhebliche Flexibilität auf, welche zudem, im Sinne eines ``Induced-Fit''-Mechanismus, durch die Gegenwart eines Liganden (Apoptozol) stark beeinflusst wird. Es ist daher als sehr wahrscheinlich anzusehen, dass die Dynamik der Interface-Region auch einen wesentlichen Einfluss auf die Bindung der Tetrahydroisochinolinone hat. Molekuardynamische Berechnungen werden deswegen auch in zukünftige Arbeiten auf diesem Gebiet eine wichtige Rolle spielen. Die Analysen zeigen zudem, dass die Konformation der Interface-Region eng mit der Konformation des gesamten Proteins - vor allem im Bezug auf die relative Stellung von SBD und NBD zueinander - verknüpft ist. Das untermauert die Hypothese, dass die Interface-Bindetasche einen Angriffspunkt für die Inhibtion des Proteins darstellt. / Methods of computational drug design play a crucial role in the development of new pharmaceutical drugs. The work presented here comprises the methodological development and the practical application of structure-based techniques in computational drug design. The first part of this dissertation focuses on the development of empirical scoring functions, which play an essential part in structure-based computer-aided drug design. The basis for this work are the empirical descriptors and scoring functions of the SFCscore software package. First, the influence of the training data composition on the prediction of empirical scoring functions was analyzed. A new training data set was created to spread the prediction range of the function and thus achieve a better separation of high and low affinity binders. The resulting function indeed yielded better predictions in the low affinity area compared to the original functions. In another approach, which also addresses the issue of discriminating active and inactive compounds, the Machine Learning method RandomForest (RF) was used to derive a classification model. Different to classical empirical scoring functions, this model no longer predicts a precise value but classifies the compounds according to their potential affinity as 'active' or 'inactive'. The example of the mycobacterial enzyme InhA showed that such models are clearly superior to different classical scoring function in terms of separating active and inactive compounds. The RandomForest algorithm was also used to derive a new scoring function for the prediction of binding affinities. This new function was implemented into the SFCscore software package under the name SFCscoreRF. This new function differs from the original SFCscore functions in several essentials points. On the one hand, the RF-algorithm is a non-linear method, which - in contrast to classical methods used for the derivation of empirical scoring functions - does not assume the additivity of the single descriptors. On the other hand, the algorithm allowes for using the whole set of available SFCscore descriptors and is therefore able to utilize a more comprehensive representation of a protein ligand complex as the basis for the prediction. Additionally, the training data set used to derive SFCscoreRF comprised 1005 complexes. This training set is one of the largest data sets used to train an empirical scoring function. The evaluation against two widely-used benchmark sets confirmed that SFCscoreRF yields superior predicitons as compared to the original functions. The comparison with other functions tested for these benchmarks shows that SFCscoreRF also achieves top results on an international level. Further analyses using a leave-cluster-out validation scheme and the participation in the CSAR 2012 Benchmark Exercise revealed that - similar to other scoring functions - SFCscoreRF shows varying performances when applied to protein-specific data sets. Additionally, by analysing the results of the CSAR 2012 data sets, valuable insight were gained regarding the prediction of docking poses and the statistical significance for the evaluation and comparison of scoring functions. The fact that empirical scoring functions are trained within a certain chemical space, is an important reason for the target-dependent performance observed in this work. Reliable predictions can only be expected within the calibrated area. Different approaches for the definition of this ``applicability domain'' are presented in this work. PCA analyses have been used to create a two dimensional representation of the ``applicability domain''. Additionally, different numerical descriptors have been tested to estimate the reliability of SFCscore predicitons. The RF-proximity has been found to be a promising starting point for future research. The development of new inhibitors for the molecular chaperone Hsp70 - a promising target in the therapy of multiple myeloma - comprises the second part of this dissertation. The basis for this work was a lead structure that was found in a previous work and attacks a novel binding pocket in the interface between the two domains of the Hsp70 protein. The optimization and development of that lead structure - a tetrahydroisochinolinone - was the primary focus of the present work. Potential binding poses in the interface were elucidated by detailed docking analyses. Based on that information, a compound library was compiled, which was synthezised and biologically analyzed by cooperation partners within the CRU 216. The resulting structure activity relationships can partially be explained on the basis of the corresponding docking poses. However, some of the effects remain unexplained. For the further development of new derivatives a comprehensive experimental characterization of the current compounds is needed. This information can be used as a basis for the refinement of the existing binding models. Hsp70 is a two-domain system, which can visit different allosteric states. To further investigate the effects of the resulting flexibility on the stability of the structure and on inhibitor binding, molecular dynamics simulations were conducted. These simulations show an above-average felxibility of the protein, which is primarily dominated by the movement of the two domains NBD and SBD relatively to each other. However, the basic conformation that is observed in the crystal structure hscaz, which was used in this work, remains stable in all simulations. Furthermore, the trajectories showed no evidence that the mutations, in which hscaz differs from the wild type protein, have a significant effect on the overall protein conformation. Although, the overall conformation of the interface between NDB and SBD remains stable, the exact conformation in this area is signficantly influenced by the domain movement. As this region includes the binding pocket of the tetrahydroisochinolinones, the conformational space of this area was analyzed in detail. The analyses expectedly reveal a high flexibility in the interface area that is dominated by the SBD-NBD movement. Furthermore, it could be shown that the conformation and dynamics can be influenced by a bound ligand (apoptozole), in terms of an induced fit mechanism. It is highly probable that the binding of the tetrahydroisochinolinones trigger similar effects, influencing the binding mechanism of this compound class. Thus, molecular dynamics simulations should play a crucial role in the future development of new compounds. The analyses also show that the dynamics of the interface region have large effects on the overall structure of the protein and vice versa. Especially, the relative orientation of NBD and SBD has a large impact on the binding pocket. This underlines the hypothesis that the interface region constitutes a promising target area for the inhibition of Hsp70.