The discovery of new inorganic molecules is an interesting problem since it implies an extended understanding of two contingent events: the first one is the formation of a new molecule, and the second is its crystallisation. The reason for that is that in the case of the product formation (and in order to make a discovery) the conditions under which the building blocks assemble have to be found, while in the case of crystallisation the conditions under which the product aggregates into crystals (which can be a subregion of the potential synthesis conditions) need to be identified in order to be isolated and characterised. There are a vast number of combinations of the experimental conditions and the coordination modes of the transition metals taking part in the building blocks, which means that a full exploration of the chemical space of any given compound would be impossible. As a result, the intuition of highly trained and experienced chemists is required in order to design the appropriate experiments that will determine the right conditions for the isolation of any new products. Unfortunately, intuitions of the experimenters can be biased by both the current knowledge of the field and their frame of mind, which makes important discoveries difficult to achieve. The work presented in this thesis is focused on the field of polyoxometalate chemistry and is exploring a multidisciplinary approach to probe the interaction of artificial intelligence methods with the human intuition during the process of exploring the crystallisation space. Our fundamental difference with relative work in the field is the application of active learning methods (which consist of methodologies capable of deciding what experiments to perform next in order to collect data that will improve the understanding of our system) in contrast to the data mining methods and simulations that have been employed so far. This algorithm method is compared to how human experimenters approach the exploration of the crystallisation space, and their performances are evaluated in terms of prediction accuracies and volume coverage. In this case, the human experimenters are allowed to follow whichever exploration strategy they see fit in order to address the task at hand. Finally, the same algorithm method is extended into collaborating with the human experimenters, and we will study the way the inherent biases can affect the search of the experimental space. This interaction is accomplished with the algorithm suggesting a set number of experiments and the human experimenter selecting the ones they seem appropriate to complete their task.

The work presented in this thesis focused on the automation of multistep organic syntheses on a laboratory scale in batch. Automation has received much attention in the chemical sciences and many automated synthesis solutions are commercially available or in development, yet all those solutions are focused on narrow aspects of the wider problem of automating chemical synthesis. In particular, a gap in the currently employed technologies was identified where synthesis in batch on a gram scale was concerned. Furthermore, many available solutions are monolithic and cannot easily be adapted for new applications, partially due to shortcomings of the hardware design, and partially due to the bespoke software controlling them. To address this need for a new approach to automated synthesis, a novel strategy comprising hardware modules dedicated to individual unit operations (as opposed to other solutions built around specific chemical transformations) and a modular flexible control software was developed. To enable the hardware development, liquid handling hardware was built and optimised, and four major modules for the unit operations of mixing under heating or cooling, liquid/liquid separation, filtration, and evaporation as well as several auxiliary modules were developed and tested. The control software orchestrating the operation of the synthesis platform was modelled after a compiler in modern computer science, separating the synthetic operations from the physical hardware of the platform. This way, synthetic procedures can be transferred between different platforms, and new hardware modules can be added to the system at will. To enable the average synthetic chemist to use the system, a rudimentary scripting language for chemical operations was developed. To prove the capabilities of the platform, three Active Pharmaceutical Ingredients (APIs) were synthesised in a fully automated fashion in yields and purities comparable to those obtained by hand. The automated reactions included a Grignard reaction and a chlorosulfonation, to name but a few. Additionally, the synthesis of one of the APIs was repeated on two physically different platforms simply by executing the same code on both systems.

Recent theoretical calculations and experiments have considered that metal-organic frameworks are promising for storing molecular hydrogen (H2). Optimizing the geometry and the interaction energy of storing for enormous H2 storage is of great current interest. In this work, we used specific category of MOFs, Zeolitic Imidazole Frameworks (ZIFs). We carried out calculations through high-accuracy electronic structure calculations (MP2, CCSD and CCSD(T)) levels of theory, with controlled errors. Also we established and calibrated a computational protocol for accurately predicting the binding energy and structure of weakly bound complexes. Then, we applied the protocol to a number of models for metal-organic frameworks. For example, we have built many systems of noncovalently bound complexes [H2...benzene, H2....imidazole, CO.... imidazole, N2... imidazole, NH3...imidazole and H2O ...imidazole] and we have optimized geometries of these systems through calculating numerical gradients at MP2/CP level and LMP2 level of theory and extrapolated from aug-cc-PVTZ and aug-cc-PVQZ basis set to evaluate the binding energy by using Hobza's scheme to obtain correct interaction energies. We found that NH3 and H2O with imidazole prefer to form hydrogen bonds rather than physical adsorption (London dispersion force). Also, the perpendicular position of hydrogen has the lowest potential energy surface, while the parallel hydrogen position has the highest potential energy surface. We have confirmed that by using a high level of basis set at MP2 such as ccpVXZ (x= Q, 5, 6) and aug-cc-pVXZ (x=D, T, Q, 5, 6), and by using the same basis sets at CCSD and CCSD(T) as the high level of theory. Also, it is clear from these results that the binding energies are sensitive to improvement of the size of basis sets. In terms of applying Hobza's scheme to obtain correct interaction energies, we found that this scheme CCSD(T)/ [34] = MP2/ [34] + (CCSD(T)/ [23] – MP2 [23]) achieved the highest accurate of interaction energy for CO...imidazole. On the other hand, this scheme CCSD(T)/ [34] = MP2/ [34] + [CCSD(T)/AVDZ– MP2/AVDZ] produced the highest accurate of interaction energy for H2...imi, N2...imi and H2...Benzene. Regarding to Basis Set Superposition Error (BSSE) and counterpoise examination (CP), Ab initio and Force field investigations of physical hydrogen adsorption in Zeolitic Imidazole Frameworks we found that the MP2/CP and LMP2 methods yield very similar results at the basis set limit and the convergence of MP2 and LMP2 with increasing size of basis sets is different since the BSSE in LMP2 is reduced. Furthermore, we found that the extrapolation to the CBS limit cannot offer an alternative to the counterpoise correction where the differences in the values of bending energies are large so we need to use both techniques together to overcome the BSSE problem. Then to confirm our result regard to the potential energy surface, we calculated corresponding potential energy surfaces using several popular force fields potential, and compare critically with best ab initio results, where we focused on the adsorption of H2 on imidazole as the organic linker in ZIFs. We carried out ab initio calculations at the MP2/CCSD(T) levels with different basis sets, basis set extrapolation and Lennard-Jones potential for the three directions X, Y and Z for 294 positions of H2. Also, we have fitted ab initio binding energy at the MP2/CCSD(T) levels with different basis set and basis set extrapolation to Lennard-Jones (12-6 LJ) binding energy by applying the nonlinear least squares method. Then we estimated the fitted binding energy using Hobza’s schemes to reduce the errors. We found that the 12-6 LJ formula produced unreasonable fit for ab initio calculated potential energy surface PES, for both the equilibrium and attractive regions, to improve this fitting, we found the good fit is only achieved by the exponential formula of repulsion region. It is hoped that this study could facilitate the search for a “good” application to store the H2 molecule conveniently and safely.

Nowadays polycyclic aromatic hydrocarbons (PAHs) have been the object of study in the search for novel semiconductor materials. Synthetic research in this field is ongoing since the beginning of the century, but it was only in recent years that technological development caused an increased interest for the optoelectronic properties of such systems. As a result, spectroscopic studies revealed interesting properties of peri-xanthenoxanthene (PXX) and its derivatives that will be extensively presented and discussed in this thesis. Concerning the tuning and tailoring of the optoelectronic properties of PAHs, many strategies can be applied in the quest for novel and better performing materials. One of the most common and efficient techniques is the atom doping that consists of a replacement of a carbon atom with a heavier one, such as oxygen or sulphur. In Chapter 1, before addressing the detailed investigation of this thesis work, a brief introduction on the nature and applications of organic semiconductor materials is given. Optoelectronic properties of well-known perylene diimides (PDIs) are compared and the discussion eventually moves to PXX derivatives that are the core of this thesis. In Chapter 2, a variety of synthetic pathways is explored in order to prepare PXX imide derivatives bearing electron-withdrawing groups in the peri position. The bottom-up approach is used to afford novel PXX systems. The synthesis of the desired systems is shown and discussed exploiting the key hydroxynaphthalene anhydride substrate. As a conclusion of this thesis, Chapter 3 deals with the characterisation of the optoelectronic properties of PXX imide derivatives. Furthermore, PXX substrates have been screened as photoredox systems to perform dehalogenation reactions and the mechanism of the photo-triggered chemical transformation has been investigated.

This thesis describes an investigation into performing organic synthesis under mechanochemical conditions. Procedures were developed for the selective mono- and difluorination of 1,3-dicarbonyls and the one-pot, two-step synthesis of fluorinated pyrazolones under ball milling. Attempts to perform a two-step mechanochemical synthesis of difluoromethylthioethers led to exciting results demonstrating that ball milling can lead to alternative reactions occurring. Finally, some initial results into the generation and reaction of organomanganese reagents under mechanochemical conditions are reported. Initial investigations into the use of mechanochemistry for organic synthesis focused on the mechanochemical formation of the C-F bond, with a particular focus on differences in selectivities observed under different milling conditions. It was found that electrophilic fluorination of 1,3-dicarbonyls could be achieved under ball milling conditions using Selectfluor. The selectivity of this process could be significantly enhanced using Liquid Assisted Grinding with acetonitrile as an additive. The possible causes of this observed change in selectivity were investigated. Further work developing a one-pot, two-step mechanochemical process was performed. A procedure for the synthesis of fluorinated pyrazolones was developed and some of the key considerations when attempting one-pot mechanochemical procedures were established by a careful optimisation. Conditions compatible with both the heterocycle formation step and the fluorination step were found and a range of fluorinated pyrazolones successfully synthesised by this method. It was observed that mechanochemistry could be used to alter the chemoselectivity of a reaction while attempting the synthesis of difluoromethylthioethers. After detailed study, a hypothesis to the origin of this alteration in selectivity was proposed. Finally, some initial results into the use of mechanochemical methods to activate manganese metal for applications in synthesis are presented.

This thesis describes the development of new routes towards the synthesis of cyanamides. Cyanamides are present in a range of biologically active compounds and are useful functional groups for the synthesis of many interesting compounds such as guanidines, ureas, isoureas and many varieties of heterocycles. A range of methods for the synthesis of cyanamides exist, however the most common technique is utilising cyanogen bromide and amines. The technique is effective and vast arrays of cyanamides can be accessed in one step. However, cyanogen bromide is highly toxic and poses a significant safety risk. In recent times new methods have been developed to avoid cyanogen bromide, however many of these techniques are operationally complex or use other highly toxic compounds. In this work three new methods for the synthesis of cyanamides have been developed. A new method for cyanamide synthesis using trichloroacetonitrile as a less toxic and safer to handle cyano source has been developed. A range of cyanamides can be formed in an operationally simple one-pot two-step procedure. This technique provides complementary selectivity to cyanogen bromide. It has also been applied to the synthesis of a biologically active PDE4 inhibitor. The one-pot deoxycyanamidation of alcohols has been developed using N-cyano-N-phenyl-p-methylbenzenesulfonamide (NCTS) as a sulfonyl transfer reagent and cyano source accessing a range of tertiary cyanamides. An array of tertiary cyanamides were accessed, including aniline type, which could not be accessed with TCAN. This approach exploits the under-developed desulfonylative (N-S bond cleavage) reactivity pathway of NCTS. A novel cyanamide and allenamide moiety, N-allenyl cyanamides have been synthesised. Utilising the deoxycyanamidation process, propargyl alcohol and a range of sulfonamides could be reacted to access an array of aryl substituted N-allenyl cyanamides. In addition, this moiety was investigated as a novel chemical building block accessing a range of otherwise challenging to access bespoke cyanamides by hydroarylation, hydroamination, [4+2] and [2+2] cycloadditions.

The research presented within this thesis highlights aspects of crystal structure determination from the combined use of powder X-ray, synchrotron and neutron diffraction and also computational crystal structure prediction from molecular structure only. The use of DE enabled the crystal structure of 2,4-dichloro-5-sulfamoylbenzoic acid and oxamic acid to be examined from conventional laboratory X-ray diffraction. In the case of 2,4-dichloro-5-sulfamoylbenzoic acid two comparable structures were identified each of which refined to similar extents. To correctly identify the correct crystal structure it was necessary to obtain and refine a powder neutron dataset. This presented before obscured information on the relative positions of hydrogen atoms and inevitably led to the successful elucidation of the crystal structure of 2,4-dichloro- 5-sulfamoylbenzoic acid. With reference to oxamic acid two conformations, namely 'cis' and 'trans' were identified from the refinement of laboratory X-ray diffraction. Infrared analysis and lattice energy calculations were also used to distinguish between the two conformations with some success. With respect to computational crystal structure prediction, presented here is a new computational strategy for crystal structure prediction from molecular structure only. The traditional lattice energy output from a polymorph prediction sequence is reranked in terms of hydrogen bonding and graph set merit points. My research here has to a certain extent managed to combine these attributes and enabled the successful prediction of 8 out of the initial 11 chosen test structures obtained from the Cambridge Structural Database (CSD).

The reaction of specially prepared salicylaldehyde benzyl ethers with the activated alkenes, methyl acrylate or acrylonitrile, in the presence of the catalyst, DABCO, has afforded Baylis-Hillman products, which have been subjected to conjugate addition with either piperidine or benzylamine. Hydrogenolysis of these conjugate addition products in the presence of a palladium-on-carbon catalyst has been shown to afford the corresponding 3-substituted coumarins, while treatment of O-benzylated Baylis-Hillman adducts with HCl or HI afforded the corresponding 3-(halomethyl)coumarins directly, in up to 94%. The 3-(halomethyl)coumarins have also been obtained in excellent yields (up to 98%) and even more conveniently, by treating the unprotected Baylis-Hillman products with HCl in a mixture of AcOH and Ac2O, obtained from tert-butyl acrylate and various salicylaldehydes. The generality of an established route to the synthesis of coumarins via an intramolecular Baylis-Hillman reaction, involving the use of salicylaldehyde acrylate esters in the presence of DABCO, has also been demonstrated. Reactions between the 3-(halomethyl)coumarins and various nitrogen and carbon nucleophiles have been shown to proceed with a high degree of regioselectivity at the exocyclic allylic centre to afford 3-substituted coumarin products. The electronimpact mass spectra of selected coumarin derivatives have been investigated using high-resolution and B/E linked scan data. Fragmentation pathways have been proposed and fragmentation modes associated with different coumarin-containing analogues have been compared. A series of coumarin-containing analogues of ritonavir (a clinically useful HIV-1 protease inhibitor) have been prepared and characterized. The synthetic approach has involved the coupling of coumarin derivatives with a hydroxyethylene dipeptide isostere to afford ritonavir analogues containing coumarin termini. An interactive docking procedure has been used to explore the docking of ritonavir and a coumarincontaining analogue into the enzyme active site.

The properties of superparamagnetic iron oxide nanoparticles (SPION) have led to them being a major area of research within the ‘nano-revolution’. A number of SPION species have been used in disease imaging, including multimodal contrast agents active in positron emission tomography (PET) and magnetic resonance imaging (MRI), as well as bioconjugate species where biomolecules have been immobilised on the nanoparticle’s surface. The western world faces an epidemic of conditions for which monoclonal antibodies (mAbs)have become seen as a ‘magic bullet’. However, the expense of mAb therapy, possible side effects and the desire to maximise treatment success require improvements in patient stratification and selection. Chapter 1 introduces the field of biomedical imaging and immunotherapy and describes how research into immunoPET and PET/MRI imaging overlap in the following chapters. Chapter 2 describes the development of reliable, reproducible methods of synthesising SPION and introducing a number of biocompatible coatings with useful functionalities. These techniques underlie the chemistry to be discussed in Chapter 3- the immobilisation of biomolecules on the surface of SPION. Typical linker chemistry is discussed, with the relative merits of different approaches expanded upon. The effect of initial reaction stoichiometry on enzymatic activity is explored as a model for later experiment design. Chapter 4 incorporates several of the aspects examined in earlier chapters to describe the synthesis of radiolabelled SPION bioconjugated to the clinically approved antibody trastuzumab to give a novel immunoPET contrast agent. Fluorescence activated cell sorting (FACS) analysis and fluorescence microscopy confirm the in vitro validation of these agents. In vivo experiments show how these agents require further development before reaching a human clinical context. Chapter 5 relates the effort to synthesise novel coordination systems based on the hypoxiaselective imaging agent 64Cu-ATSM. The co-ordination chemistry of these systems with several metals is described.

The work presented in this thesis focused on the synthesis of gold nanoparticles, exploring new ways to synthesise them and also using new tools to improve the study and discovery of new nanomaterials. Whilst there has been a special concern in understanding how they are organised and which are the best conditions to achieve specific shapes, there is gap in finding new approaches that can allow fast synthesis of nanoparticles and fast screening of the chemical space and real time observation of the reaction under different reaction conditions. In the first chapter of this thesis we are going to present a new synthesis method to prepare gold nanoparticles-POM hybrids. Also, we will discuss how POMs can influence the aggregation of nanoparticles, depending on the size and charge of the POM. For example, gold nanoparticles will aggregate more easily if they are surrounded by small and less negatively charged POMs. In the next chapters of the thesis, we will aim to demonstrate that an automated system is able to evolve gold nanomaterials, this means that an automated system will use raw materials (simple chemicals, in this specific case HAuCl4, CTAB and NaBH4) to synthesise very simple nanostructures, such as spheres, then reuse those spheres and other chemicals to produce even more complex structures. In chapter 2 we will go through the process of building an automated system, in this case, the system will be designed to synthesise gold nanoparticles. We will start by designing the automated system and testing it, we will see the flaws that those different systems had and how we overcame them by doing some improvements on them, such as more control over the temperature of the reaction, keeping a constant temperature of the reagents, improving the cleaning process, trying different concentrations of the reagents, trying different algorithms and different ways to calculate the fitness factor, etc. until we found a system that was very reliable, which was able to provide reproducible results. In the last chapter of this thesis, we will focus on the results obtained in the different versions of the automated system. First, we will explain our first objective, which was obtaining gold nanorods of very short aspect ratio. We will analyse the results we obtained for that objective, trying different fitness factor calculations and how the different ways to calculate the fitness factor affected the process to obtain the desired product. We will discuss why we changed the fitness factor calculation and how this helped to achieve our objective. Then, we will jump to the next level of the project, which was to synthesise very simple gold nanoparticles from raw chemicals and reuse these simple structures to obtain more complex structures. This demonstrates that we have an automated system able to evolve gold nanomaterials; the first of this kind. In this chapter, we will also talk about the different techniques used to characterise the product, where we used in-line analytical techniques such as UV-Vis or image analysis (which are the ones used to calculate the fitness factor that the algorithm is going to use) and other techniques to fully characterise the final product such as TEM.