Theses - Chemical Engineering and Biotechnology

Permanent URI for this collection

https://www.repository.cam.ac.uk/handle/1810/219484

Browse

Now showing 1 - 20 of 240

Controlled Access
Computational Techniques for Studying Nanoporous Materials
Rampal, Nakul
The central goal of this thesis has been the development of new computational techniques to accelerate the discovery and characterization of nanoporous materials. Chaper 1 introduces the field of nanoporous materials/reticular chemistry, provides a brief history of molecular simulation in reticular chemistry, and finally, discusses some key challenges that need to be addressed from a computational perspective. Chapter 2 introduces the objectives of this thesis including the organization of this thesis, and some of the important questions this thesis aims to answer. Chapter 3 begins with a detailed discussion on the theory behind GCMC simulations, including the partition function and the different moves in a GCMC simulation and their associated probabilities. Next, the chapter goes into detail of how the potential energy function *U* is calculated, i.e. force fields, including it’s two main contributing terms, (i) the bonded potential, *U_bonded*, and (ii) the nonbonded potential, *U_nonbonded*. Finally, the theory behind the calculation of different geometric properties like the accessible surface area, largest cavity diameter (LCD), and the pore limiting diameter (PLD) are discussed. Chapter 4 introduces our recent advances in HTS to rapidly screen *in silico* the adsorption properties of hundreds of MOFs for CO/N₂ separations. Our approach involves the use of a multi-scale toolbox combining high-throughput molecular simulations, data mining and advanced visualization, as well as process system modeling, backed up by experimental validation. Chapter 5 extends the high-throughput screening approach introduced in the previous chapter to rapidly screen the properties of not hundreds, but thousands of MOFs for H₂ storage. We also discuss how principal component analysis (PCA) can be used to extract meaningful insights from the vast amount of data generated from such screening studies. We validate our screening approach by synthesizing and evaluating the performance of the selected MOF (HKUST-1) in its monolithic form. Chapter 6 begins with an introduction to Small Angle X-ray Scattering (SAXS) and lattice gas models. Next, we introduce the concept of a monolith, and show experimentally the existence of interparticle mesopores - inaccessible from powders - that push final adsorption capacities above levels expected for single crystals. Finally, we show how lattice-gas models in combination with GCMC simulations can be used to accurately capture the monolithic structure across both the microporous and mesoporous range enabling the robust future predictions of the adsorption characteristics of monolithic materials. Chapter 7 begins with a derivation of the BET equation from first principles. We follow this up with a discussion on how the BET equation can be used to calculate the BET area, i.e. BET method, and some state-of-the-art problems with this method. Finally, we introduce an algorithmic approach called BETSI that addresses some of these problems. Chapter 8 summarizes the key results of this thesis and provides some context on the future outlook and challenges in this field.
Embargo
Halide Perovskite and Metal-Organic Framework Composites for X-ray Detection
Salway, Hayden
Our ability to detect X-ray radiation and generate detailed images on internal structures in a non-destructive manner, has had profound impacts on all of our lives. Whether that is through medical imaging, allowing vital diagnosis of tumours and visualisation of broken bones; through security imaging, a vital resource for international security; manufacturing and food processing control, ensuring the food we eat is safe to eat, free from harmful objects; to scientific research, unearthing foundational principles, contributing to the progression of society. X-ray detectors are omnipresent in all our lives, contributing to our safety, health, and the progress of science and technology. To expand the potential of this crucial technology, a new generation of X-ray detector materials is required to overcome current limitations and expand performance beyond that currently achievable. In this thesis, two families of materials with unique properties, halide perovskites and metal-organic frameworks are brought together for the first time, in a sol-gel processable, monolithic manner and utilised to advance the development of novel X-ray detector materials. Utilising the exceptional optoelectronic properties of halide perovskites, and benefitting from the enhanced stability, processability, and potential for functionalisation provided by metal-organic frameworks, a new class of X-ray detector materials are formed. Concurrently, these materials are comprehensively characterised using a suite of structural and photophysical techniques to provide mechanistic insights into their formation and photophysical processes. These crucial insights enabled the refinement and optimisation of synthesis protocols and choice of building-blocks, to enhance stability beyond previously achievable in comparable composites, and develop optimised monolithic perovskite@MOF composites tailored towards X-ray detector applications. Overall, this thesis utilises the unique properties of perovskites and metal-organic frameworks to develop, robust, stable, and scalable X-ray detectors with outstanding promise to overcome limitations of stand-alone perovskites and current detector materials. By synergistically combining two materials, detectors with new multifunctionality are possible. We show monolithic perovskite@MOFs can play a key role in future X-ray detector devices, beyond their encapsulation and stabilisation properties, contributing to the efficient transport of X-ray stimulated charges and limiting ion migration. This work opens an array of applications and contributes to bringing perovskite-based X-ray detectors closer to commercialisation.
Open Access
Development and application of optical microscopy tools for the study of axon guidance
Wunderlich, Lucia; Wunderlich, Lucia [0000-0001-7200-1713]
During their development, neurons extend axons towards their target cells, where they branch to establish connections. The navigation of axons relies on the presence of chemical cues sensed through guidance cue receptors. Upon receptor activation, intracellular signalling pathways are initiated, one of which induces local protein synthesis (LPS), a key process to enable rapid navigational responses to guidance cues. Impairments of axon guidance and LPS are associated with several neurological disorders. In this thesis, state-of-the-art optical microscopy-based tools were developed to improve the efficiency and versatility of commonly used methods for studying axon guidance both *in vitro* and *in vivo*. Furthermore, imaging-based studies were performed on selected guidance cue receptors to investigate their regulation of cue-induced LPS. For the experimental investigations, *Xenopus laevis* retinal ganglion cells (RGCs) were utilised as a model system, enabling comprehensive studies on isolated outgrowing axons both *in vitro* and *in vivo*. Initially, an imaging method was established to examine intricate axonal structures within the highly complex physiological environment. To achieve this, expansion microscopy (ExM) was combined with light sheet fluorescence microscopy (LSFM) to visualise RGC branching *in vivo*. By tracing individual axons, this technique offers a valuable tool for studying cue-dependent arborisation within the brain. Then, ExM and structured illumination microscopy (SIM) were applied in a study aimed at investigating how guidance cue receptors facilitate cue-dependent responses through changes in specific mRNA translation. A mechanism was explored that involves the direct interaction of the guidance cue receptors deleted in colorectal cancer (DCC) and neuropilin-1 (Nrp1) with the translation machinery. This interaction was found to be mediated through RNA binding proteins (RBPs), enabling a receptor-specific mRNA subset to be rapidly and locally translated in response to cue stimulation. Further investigations focused on the cue-induced intracellular transport dynamics of DCC within the endosomal system. During these studies, DCC was observed to colocalise with ribonucleoprotein (RNP) granules on endosomes, suggesting a model in which DCC facilitates the association of RNP granules with endosomes through its affinity to RBPs. Finally, an optical system was developed to enhance the throughput of commonly employed assays for stimulating axons and enabling directed axonal outgrowth *in vitro*. This system employs surface-immobilisation of guidance cues and adhesion proteins. Protocols based on the principle of light-induced molecular adsorption of proteins (LIMAP) were established to guide outgrowing RGC axons towards their physiological target tissue. In summary, this work describes the development of highly sophisticated tools designed to facilitate the study of axon guidance both *in vitro* and *in vivo*. Additionally, valuable insights were gained into the cue-induced mechanisms that initiate LPS through guidance cue receptors. These advancements hold great potential for enhancing our comprehension of axon guidance and its implications for neurological disorders.
Controlled Access
Question Answering on Dynamic Knowledge Graph for Chemistry
Zhou, Xiaochi; Zhou, Xiaochi [0000-0002-4008-9965]
The field of chemistry relies heavily on accessing diverse types of data and information, both for human users seeking information and for implementing applications. While the chemistry Knowledge Graph provides a solution for representing this data and information, it poses challenges for human users to access it efficiently. A Knowledge Graph Question Answering system is one of the solutions. However, due to the specific nature of the chemistry Knowledge Graph, off-the-shelf solutions for Knowledge Graph Question Answering might be less effective. As a result, this thesis explores implementing Knowledge Graph Question Answering on the chemistry Knowledge Graph, addressing challenges querying it. The challenges addressed in this thesis include the non-shallow structure, the semantic heterogeneity, the embedding of numerical values, and the large scale of the chemistry Knowledge Graph. The thesis studied the two main methods for Knowledge Graph Question Answering: the Semantic Parsing and Knowledge Graph embedding methods against the chemistry Knowledge Graph. The thesis also investigated models including LDA-based topic modelling, StarSpace-based text classification, CRF-based Named Entity Recognition, Knowledge Graph embedding models (TransE, Complex, TransR, and TransRA), relation prediction, and score alignment. The first study implements a Semantic Parsing-based Question Answering system using CRF-based Named Entity Recognition to extract key components from questions. The system also applies an ontology lookup service to ground components to semantic representations. StarSpace-based text classification locates suitable SPARQL query templates. SPARQL queries are formed by filling semantic representations into templates, enabling data retrieval. Evaluation results show that the system outperforms Wolfram Alpha and Google Search Engine baselines in some question types. The second study explores integrating semantic agents to expand system coverage. Modifications to OntoAgent ontology enable agent discovery, matching, and invocation. Semantic agent descriptions include question templates for automated training question generation. Evaluation results show a 1.0 F1 score for StarSpace question classification and a 0.95 F1 score for CRF-based Named Entity classification. For agent-related questions, 83% have correct requests and 81% have correct answers. The final study investigates Knowledge Graph embedding-based Question Answering methods and various embedding techniques. Knowledge Graph embedding represents information, while BERT-based relation prediction predicts question embeddings. Answer candidates are ranked based on triple likelihood calculated from the topic entity, predicted relation, and candidate embeddings. A novel Knowledge Graph embedding algorithm, TransRA, has numerically higher filtered mean reciprocal rank than other embedding methods. A BERT-based score alignment model integrates and re-ranks answers, increasing mean reciprocal rank by 0.41. Evaluation results show filtered mean reciprocal rank ranging from 0.53 to 0.88 across domains.
Open Access
Monitoring Tissue Function Dynamics in vitro with Bioelectronics: Towards Understanding Barrett’s Oesophagus Pathogenesis
Van Niekerk, Douglas Carl
Barrett’s oesophagus (BE) is a non-malignant, specialised intestinal metaplasia of the oesophageal mucosa and a precursor of oesophageal adenocarcinoma (EAC). EAC is a particularly aggressive cancer with a high mortality rate, and as such, early intervention and prevention is considered to be imperative. In order to entertain such therapeutic strategies, the pathogenesis of BE needs to be better understood and models thereof developed to act as testing platforms for prophylactic candidates. BE pathogenesis is driven by persistent gastro-oesophageal reflux, which impairs mucosal barrier function, leading to epithelial ulceration, and initiates a sustained inflammatory response. This complex microenvironment reprogrammes both local cell types as well as the BE progenitor *cell of origin* (the identity of which remains under much debate). The pathogenic microenvironment presents selective pressures which act upon the reprogrammed cells, which express differing relative finesses – competitive exclusion of the local keratinocytes results in re-epithelialization of the wound site by the BE progenitor cell type. Early-stage pathogenesis, characterised by tissue function dysregulation and epithelial erosion prior to ulceration, is considered a potential point of prophylactic application. Moreover, the mucosa is a dynamic system and the evolution of tissue function in response to the periodic reflux insult is a complex, transient process. The objective of this work is thus to develop a platform capable of simulating the oesophageal mucosa and pathogenic microenvironment, while non-destructively monitoring the tissue function. The oesophageal mucosa is a barrier tissue, responsible for modulating flux between lumen and stroma and as such, oesophageal tissue function is defined as its barrier function, or impedance to trans-epithelial flux. One means of probing barrier function is to measure the flux of solvated, ionic species across the epithelium, in response to an applied electrical potential difference. Poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS), a conjugated polymer blended with a poly-electrolyte, is identified as a suitable material upon which to base the platform, as it both allows for highly efficient ionic-to-electronic flux coupling as well as possessing mechanical properties similar to the extracellular material in native tissue. Using an ice-templating technique, the conjugated polymer can be fabricated so as to take the form of a macroporous scaffold, which can then be used as a cell-culture substrate, which enforces the three-dimensional architecture of native tissue. This scaffold was embedded within a suspended cell culture well-insert, with separate apical and basolateral compartments, to yield the electronic transmembrane (e-transmembrane) device. Colonising the scaffold with fibroblasts and culturing epithelium on the apical surface was shown to produce epithelial models which histologically recapitulated native tissue. Further, the device was shown to be capable of monitoring the linearized, ionic impedance of intestinal, endothelial and renal tubule epithelial models, with the correlation between measurement and barrier function confirmed through the use of epithelial tissue phantoms. While the initial iteration of the e-transmembrane adheres to the conventional, two-electrode configuration, I have demonstrated a significant improvement in measurement sensitivity by extending the design to a three-electrode configuration, which incorporates a reference electrode. In particular, the utility of a non-conventional reference electrode design is demonstrated, wherein a polarizable material, shaped so as to minimize field and impedance distortion, is shown to improve sensor performance within the specific use context of the e-transmembrane device. The measurement error due to the reference electrode inclusion has been derived and verified by way of epithelial tissue phantoms, with the performance improvement illustrated in measurements of renal tubule epithelial models. The e-transmembrane form-factor, which is an extension of conventional suspended cell-culture inserts, allows for the adaptation of existing three-dimensional tissue culture protocols; a commonly used oesophageal model, comprising of the co-culture of fibroblasts and immortalised oesophageal epithelial primary cells (EPC2-hTERT), is shown here to translate into the e-transmembrane, with the barrier function of the model measurable by the device. Periodic exposure to a simplified model of gastro-oesophageal refluxate simulates the early-stage pathogenic microenvironment in the e-transmembrane oesophageal model. I show that the model of early-stage pathogenesis displays erosion of the stratified epithelium and the activation of repair mechanisms in accommodation to the persistent insult. Ascorbic acid (vitamin C) is an anti-oxidant which has been shown to improve barrier function and wound healing in stratified squamous epithelia. I further demonstrate that concomitant exposure of the mucosal model to ascorbic acid and the refluxate results in reduced epithelial erosion and expedited barrier function repair, indicating the prophylactic utility of the dietary micronutrient.
Open Access
Graphene Brain on a Chip Platform for the Study of Neurodegeneration
Hui, Ernestine
A complex network of interconnected neurons forms the brain. This network is disrupted in neurodegenerative diseases---such as Parkinson's disease (PD), and there are limited tools available to study the molecular mechanisms behind the degradation of the neuronal network. This indicates a gap in the understanding of how neurodegenerative diseases and PD initiate and propagate. Therefore, graphene microelectrode arrays (MEAs) were explored to bridge this gap. However, there are currently no protocols or methods available to fabricate reliably and apply the graphene MEAs for the study of neurodegeneration. This work demonstrates a new reproducible protocol for the fabrication of graphene MEAs, a new method for using the graphene MEAs---including analysing the simultaneous calcium imaging, electrophysiology, and super-resolution imaging data obtained, and a physiologically relevant application for the graphene MEAs in neurodegeneration research. The fabricated devices demonstrated a higher spatiotemporal resolution and lower impedance than commercially available MEAs, due to the properties of the graphene---such as the high transparency---as well as the fabrication techniques. Due to the higher spatiotemporal resolution provided by the graphene, the imaging of spontaneous neuronal activity, electrophysiology, and correlative imaging and electrophysiology recordings of neurons on a network level and a single-cell level could be obtained. In addition, the transparency allowed for the sub-cellular investigations of the degenerating neurons using super-resolution microscopy. A machine-learning model was used to classify the neuronal spikes and analyse the imaging and electrophysiology data. Using U18666A to induce Niemann-Pick disease type C, the graphene MEAs and methodology for simultaneous imaging and electrophysiology were validated. It was found that U18666A had a significant degenerative impact on the synchronicity, activity, connectivity, and morphology of the neurons---which was verified by electrophysiology, microscopy, correlative imaging and electrophysiology, and super-resolution microscopy. In addition, further investigations into PD were carried out and preliminary degenerative effects of cholesterol and alpha-synuclein combined were observed using the graphene MEAs. A reliable, reproducible, and valuable tool for the study of neurodegeneration has been developed. This protocol and methodology immediately allows researchers to investigate neurodegenerative diseases in depth, which expands and adds to the scientific knowledge pool for PD. A deeper understanding of neuronal pathology opens the doors for developing therapeutic interventions for other neurodegenerative diseases as well. Therefore, this work unleashes a new tool for the neuroscience and drug development communities.
Open Access
Film Formation via Drying of Polymer Solutions
Othman, Ahmed
Drying of skin and personal care products, to create thin films, is a process employed to deliver product-specific functions to targeted areas. Uniform, defect-free films are essential, as defects and instabilities can adversely affect product performance. To optimise product functionality, understanding the drying process and resulting final film morphology is essential. This project investigated film formation by drying droplets and thin films of a polymer solution, specifically polyvinylpyrrolidone (PVP) and ethanol, a formulation often utilised in skin and personal care products. An imaging rig was designed and optimised to study the drying and spreading dynamics, as well as dried morphology of PVP droplets at ambient conditions. Drying PVP–ethanol droplets on glass substrates resulted in two distinct edge morphologies: a ringlike pattern, similar to the well known "coffee-ring", and a scalloped pattern. This scalloped pattern formed due to a surface tension-driven flow, initiated by the presence of water within PVP. This instability was suppressed by the removal of water from PVP. These findings were supported by a lubrication approximation analysis, which predicts the instability wavelength and exhibits good agreement with experimental observations. Octamethyltrisiloxane was added to PVP–ethanol mixtures to reverse the Marangoni flow and facilitate uniform deposition. However, evaporation from this mixture resulted in liquid–liquid phase separation. This was found to be caused by the transfer of water vapour into the droplet, altering the mixture’s miscibility. The surrounding relative humidity significantly influenced droplet evaporation dynamics, with lower humidity levels preventing phase separation. A confocal Raman microscope was used to characterise PVP distribution during drying. In 1D PVP–ethanol film, the initial film thickness was found to determine the drying uniformity, PVP distribution, and solvent retention. Films of 30–40 μm thickness exhibited a uniform PVP distribution and minimal solvent retention, whereas films of 500 μm developed a concentrated polymer skin layer that hindered drying. In typical thickness ranges found in the application of personal care products (30–40 μm), PVP–ethanol displays limitations in solvent retention and slow release.
Open Access
A dynamic knowledge graph approach to self-driving chemical laboratories
Bai, Jiaru; Bai, Jiaru [0000-0002-1246-1993]
The contemporary design of self-driving laboratories faces difficulties in scalability and interoperability when it comes to the vision of a globally connected research network. This is due to heterogeneous data formats and resources as an obstacle to holistic integration. This thesis investigates a potential solution to the interoperability problem in chemical experiments by utilising a dynamic knowledge graph to unify the representation of data, software, hardware, and workflow. The developed approach is applied to a few selected case studies. To realise a self-driving chemical laboratory, the design-make-test-analyse cycle is first reformulated as the process of propagating information through a dynamic knowledge graph by means of a chain of actions. Our approach utilises ontologies to capture the data and material flows involved in the experimentation, and employs autonomous agents as executable knowledge components to carry out both computational and physical tasks. The iterative workflow is automatically managed by a derived information framework with data provenance semantically preserved following the FAIR principles – Findable, Accessible, Interoperable and Reusable. The derived information framework is also applied to an automated flood impact assessment in the smart cities domain to demonstrate its generalisability. On the computational front, the dynamic knowledge graph approach is applied to automate the calibration of a kinetic reaction mechanism, which demonstrated a reduction of calibration time from months when done manually to days while an increase in accuracy measured as a 79% decrease in objective function value. In the wet lab, we demonstrate the practical application by linking two robots in Cambridge and Singapore to achieve a collaborative closed-loop optimisation for an aldol condensation reaction in real time. The two robots effectively produced a Pareto front for the cost-yield optimisation problem over the course of three days of operation. The dynamic knowledge graph approach is also applied to optimise two Suzuki coupling reactions for efficient synthesis of challenging molecules, obtaining 82 mg of the final product from a 40 mL scale-up that would be otherwise difficult to access without extensive process redesign and manual synthesis efforts.
Open Access
Developing bioelectronic sensors for 3D in vitro models at the air liquid interface: Innovative technology for biomedical and respiratory research.
Barron, Sarah; Barron, Sarah [0000-0002-9922-2388]
Respiratory diseases and airborne infections are significant global health issues and a major cause of morbidity and mortality worldwide. According to the World Health Organization (WHO), respiratory diseases rank as the third leading cause of death globally, with over 10 million deaths per year attributed to respiratory illnesses. Chronic respiratory conditions are characterized by aberrant immune and epithelial barrier function, leading to respiratory inflammation and tissue damage. Despite substantial progress in identifying therapeutic targets and drug candidates, the efficacy of novel treatments in clinical trials remains low. One of the reasons for this, is that the current gold standard models of human disease used for drug screening rely on two-dimensional (2D) cell culture and/or animal testing. However, neither of these models accurately reflects the native microenvironment of human tissues. Recent advancements in bioengineering have resulted in the development of *in vitro* tissue models that more closely mimic the native tissue microenvironment. These frameworks include the development of extracellular matrix (ECM) materials, microfluidics, and three dimensional (3D) organoids. However, the clinical translatability of these models is limited by the lack of sensor technologies capable of monitoring them non-invasively and in real time. Gold standard analysis techniques rely on sample labelling or destructive end-point analysis, while electronic methods require large, rigid electrode materials that are ill adapted to soft 3D microarchitecture. Additionally, these electronics require submersion in an electrolyte for operation, which is inadequate for native air-liquid interfaced tissue such as the lungs. In this context, the goal of this thesis project was to address these challenges from multiple angles and in a multidisciplinary manner. The first goal was biology-focused and aimed at transitioning from 2D monocultures to a multicellular 3D model. The development of the biological respiratory model was based on a co-culture of primary human bronchial epithelial cells and lung fibroblasts in collagen coated cell culture inserts. This model highlighted the impact that multicellular cross-talk can have on the development/differentiation of *in vitro* models and their differing response to drug stimulation. The second goal was materials-focused and aimed at creating a biomimetic ECM environment of the lung. A composite electroactive scaffold was developed using an organic conducting polymer blended with human Fibrinogen protein, which replicated the mechanical and morphological properties of the human airway. The platform was capable of monitoring barrier formation, in real-time over 32 days, and, permits continuous, simultaneous electrical readouts. The third goal was technology-focused and resulted in the design and fabrication of a novel microelectronic device that was capable of monitoring epithelial barrier function and perturbations of primary *in vitro* models at the air liquid interface (ALI). By developing integrated bioelectronic sensors, this work demonstrates the ability to monitor advanced 3D tissue-like systems non-invasively and in real time. Additionally, and for the first time, patient-specific electrical signatures of respiratory epithelial dysfunction were measured at ALI. This technology will also have the potential for use in other air interfaced model systems, including the gut, skin, eye and brain, for diagnostic, drug screening and personalised medicine applications. Overall, it is hoped the versatile biotechnology platforms developed here will contribute to the field of biomedical and respiratory research.
Open Access
Renewables Integration Challenges of the Energy Transition
Atherton, John
Throughout the world a large-scale energy transition is unfolding. This fundamentally includes a vast expansion of variable renewable energy (VRE) generation. In Britain this primarily consists of onshore and offshore wind, as well as solar. The non-dispatchability of these technologies presents a unique range of challenges for the energy network. This thesis investigates the key difficulties of renewable energy integration in Britain, as well as some potential solutions. Renewables integration is investigated using a multidomain methodology to process and model live data. Using a knowledge graph supported infrastructure, the British energy network is modelled. This consists of a representation of Britain’s transmission network, power plants, and energy storage capabilities (both existing, and potential future expansions). These research investigations are expanded upon with respect to energy policy, and energy market modelling. A north/south divide is identified as a fundamental transmission limitation using both economic dispatch and curtailment analysis methodologies. From a network perspective, this inflexibility is particularly significant with respect to wind power from Scotland (both onshore and offshore), resulting in increased rates of curtailment. From a market perspective, a lower percentage rate of dispatchable energy penetration coincides with increased market volatility to the point where even coal has been strategically dispatched at higher prices, at a rate not seen in recent years prior. Co-located energy storage systems (ESSs) were modelled with the ability to both arbitrage price, and charge using otherwise curtailed energy to simulate their potential as a solution to both these challenges. ESSs co-located with Scottish wind farms were found to reduce curtailment volumes significantly compared to southern placements, while the latter achieved faster payback due to a greater of unconstrained opportunities to arbitrage prices. All ESSs, however, achieved payback due to significant energy price volatility experienced in recent years. Given these ongoing challenges to the electrical grid and markets, the role and placement of storage will remain topics of key significance, upon which these findings shed novel insight.
Open Access
Development and characterisation of a human in vitro model of the Neurovascular Unit using a biomimetic tissue engineered system
Barberio, Chiara
The neurovascular unit (NVU) is a complex structure comprising endothelial cells, pericytes, astrocytes, neurons, and extracellular matrix (ECM) elements. It plays an essential role in regulating cerebral blood flow and maintaining the integrity of the blood-brain barrier (BBB). Over the years, several tissue engineering approaches modelling the NVU *in vitro* have gained popularity as they offer a platform to investigate the cellular cross-talk, unravel disease mechanisms as well as brain targeted therapeutics effects. While *in vivo* models have been invaluable in advancing our understanding of the brain functions, they also have significant limitations when it comes to tissue engineering purposes. Besides being expensive to establish and maintain, *in vivo* models are often species-specific and may not accurately reflect human pathophysiology. Additionally, ethical concerns around the use of animals in research can limit the types of studies that can be performed. These limitations highlight the need for alternative *in vitro* models that can more accurately mimic the NVU structure and functions in health and disease. Two-dimensional (2D) cell culture models, such as Transwell systems, have been widely used for *in vitro* NVU modelling allowing the investigation of cell-cell interactions and cellular influence on the BBB properties and integrity. However, one setback of such 2D models is that they do not fully recapitulate the complex 3D microenvironment of the NVU, which can negatively impact cell behaviour and signalling pathways. Additionally, in 2D models, shear stress and other mechanical forces experienced by endothelial cells are not representative of physiological conditions *in vivo*. Organ-on-chips technology, on the other hand, has emerged as a promising and more robust alternative platform to study the dynamic interactions between neuronal and vascular components under controlled conditions. Nevertheless, some challenges, such as difficulty in accurately replicating the complex 3D architecture and physiology of the native NVU still remain. Tissue engineering integrating bioelectronics has emerged as a promising strategy for *in vitro* modelling and monitoring of complex biological systems such as the NVU. This approach entails combining tissue-engineered equivalents (e.g., scaffolds, hydrogels) with integrated electronic sensors and devices, enabling real-time monitoring of cellular and molecular processes *in situ*. In addition, the seamless incorporation of bioelectronics within tissue-engineered systems enables the real-time and continuous recording of experimental conditions, hence avoiding end-point assay experiments. Furthermore, tissue engineering integrating bioelectronics has the remarkable potential to revolutionize disease modelling and drug screening by allowing the development of high-throughput platforms for screening drug candidates and personalized medicine. While there are still challenges to overcome, such as the biocompatibility of the electronic components, device design optimisation and the need for better integration methods, this approach holds great promise for advancing our understanding of complex biological systems and improving the design/development of therapeutics. The objective of this dissertation was to create a novel bioelectronic model of the neurovascular unit (NVU) that could continuously monitor the blood-brain barrier (BBB) *in vitro*. To achieve this goal, the research combined fundamental principles from 3D cell biology, material science, and tissue engineering. The first approach was to establish a 2D Transwell system as a reductionist and informative support for multiple NVU model configurations. The 2D system was used to determine the optimal cell culture conditions for endothelial cells, astrocytes, and neuronal cells. The resulting NVU *in vitro* platform was then characterised using optical and electrical methods to evaluate BBB integrity under different conditions. To improve and advance such biological model, the next step was to create a 3D biomimetic scaffold that resembled the brain extracellular matrix (ECM). The scaffold was designed to be a more physiologically relevant substrate than the 2D Transwell systems. A composite electroactive scaffold integrating ECM elements was generated to test its suitability to support neuronal cell culture and differentiation *in situ*. The biomimetic 3D scaffold was then used to host the multicellular NVU model, which demonstrated the formation of an apical endothelial barrier. Finally, the 3D NVU system was successfully incorporated into an in-house engineered bioelectronic platform called "e-Transmembrane." This platform enabled in-line, non-invasive, and dynamic monitoring of the NVU cell types and BBB development. The integration of the 3D NVU system into the e-Transmembrane platform represented a significant leap towards a more sophisticated and realistic model of the NVU and its BBB. In conclusion, this study described above has significant potential for advancing our understanding of the blood-brain barrier (BBB) and the neurovascular unit (NVU) *in vitro*. Furthermore, such bioelectronic NVU model could represent a reliable test bed for drug discovery and screening by providing a more reliable and predictive platform for testing the permeability of drugs across the BBB.
Open Access
Functionalization of phage endolysins to treat infections of spore-forming Clostridia bacteria in the era of antibiotic resistance
Mills, Gabrielle
Certain *Clostridia* species are notable pathogens, not only because they form highly resistant endospores but because they produce toxins and exhibit antibiotic resistance. *Clostridioides difficile* is a nosocomial pathogen responsible for thousands of deaths every year, whereas *Clostridium perfringens* is a common cause of food poisoning and a notable veterinary pathogen. Research on *Clostridia* spores and phage therapy has been limited to a selection of species, partially because of their anaerobic nature. While recent work has endeavoured to bridge this gap, there is still a pressing need to understand the resistance mechanisms of *Clostridia* spore-formers—especially *Clostridioides difficile*—and develop improved treatments for these difficult pathogens. The aim of this thesis was to augment current knowledge of *Clostridia* and their phage-based therapies, specifically through analysis of *Clostridium perfringens* SM101 as well as *Clostridioides difficile* strains 630, R20291, and SH1. The first portion of the thesis specifically looks at the *Clostridia* spore coat: the composition through proteomics and genetic engineering and the structure via phase contrast and TEM imaging. Over 3,000 coat and exosporium proteins were identified in *Clostridia* species, and TEM showed that *Clostridioides difficile* 630 and *Clostridium perfringens* SM101 both possess an exosporium like that of *Clostridium sporogenes*. Then, phage were induced and isolated to assess their specificity and efficacy against *Clostridia*, yet endolysin domains were the ultimate focus for host range analyses and subsequent experimentation. Finally, the feasibility of combining endolysin binding domains and antibacterial proteins was investigated for improvements in lytic and binding activity. To do so, a novel methodology was formulated to synthesize these proteins in susceptible *Escherichia coli*. Fusions of hen-egg-white- lysozyme and a phage endolysin domain of *Clostridium perfringens* show enhanced lytic activity against *Clostridia* bacteria, including outgrown vegetative cells.
Open Access
Rheology of bubbly liquids containing a non-Newtonian continuous phase
Gibson, Douglas
Bubbly liquids are two-phase materials where gas is dispersed in a liquid with bubble volume fractions ranging between 0 and 0.5. Bubbly liquids are found in many real-world contexts and their bulk rheological behaviour is determined by the continuous phase rheology, the bubble phase properties and the interaction between the two phases. The majority of research concerning bubbly liquids has focussed on materials with a Newtonian continuous phase. However, many of the bubbly liquids found in industrial and consumer settings have a viscoplastic continuous phase. This project has investigated the rheology of bubbly liquids with a viscoplastic continuous phase, namely a commercial polymer solution called ClearGlide^TM that belongs to a family of yield stress fluids known as Carbopols. Bubbly liquids with air volume fractions in the range 𝜙_a = 0 − 0.5 were produced by whisking the continuous phase in a planetary mixer. Aeration profiles describing how 𝜙_a evolved were found to have the same general shape regardless of continuous phase dilution or mixer speed. When plotted using normalised aeration parameters the majority of data sets followed a common curve which was quantified using an exponential relationship. The bubbly liquids had polydisperse log-normal bubble size distributions. The variation in bubble size with continuous phase dilution and mixing time was considered with respect to bubble break-up. Original models accounting for the process geometry were proposed and found to provide a good description of the aeration process. These models provided insight into the mechanisms taking place during whisking, specifically entrainment and disengagement of bubbles. Rheological behaviour was investigated using both shear and oscillatory rheometry. Unaerated ClearGlide^TM exhibited typical yield stress fluid behaviour comprising three distinct regions: solid-like linear elastic deformation at low stress, followed by a yielding transition, and subsequent liquid-like shear-thinning at large stress that was described using the Herschel-Bulkley model. Suitable rheological parameters were identified to characterise each region. Bubbly ClearGlide^TM exhibited similar behaviour to that of its continuous phase. Aeration generally resulted in a reduction in the rheological parameters with increasing air volume fraction. Correlations quantifying the impact of air volume fraction were determined. The combined effect of the continuous and bubble phase material properties was considered and found to be complex, with the continuous phase having a greater impact on the bulk rheology.
Restricted
A generalized catalyst design and optimization framework for gas-phase heterogeneous catalysis
Shambhawi, Shambhawi
[Restricted]
Embargo
Inverse Methods for Fast Field Cycling and NMR Correlation Experiments
Beckmann, Julian
This thesis focuses on the development of inverse methods for nuclear magnetic res- onance (NMR) correlation or exchange experiments and new fast field cycling (FFC) modelling approaches. Moreover, the theoretical framework to consider fast field cycling as an inverse problem is provided and the possibility to employ established regularization methods for the inversion of NMR dispersion profiles is investigated. It is shown that the relaxation behaviour of systems with sufficiently understood molecular dynamics such as ionic liquids can be accurately described by simplifying relaxation models. This is further supported by findings from molecular dynamics simulations and physicochemical measurements. Further, the concept of generalized cross validation was applied to modified to- tal generalized variation (MTGV) regularization and it was shown that deep learning can be employed for inversion of NMR signals from exchange or correlation ex- periments. In addition, the inversion performance of deep learning, Tikhonov and MTGV regularization was compared on a vast data set of simulated NMR signals providing strong evidence of deep learning achieving the best reconstruction results in a clear majority of instances. Finally, fast field cycling was treated as an inverse problem and MTGV regularization was employed to reconstruct correlation time distributions from NMR dispersion profiles. From this, detailed insights into the molecular dynamics of several cata- lyst samples were obtained previously not accessible with established modelling approaches.
Embargo
Packed bed reactors at the pellet scale: magnetic resonance studies of hydrodynamics, mass transport and reaction
Elgersma, Scott
Packed bed reactors are widely used in the chemical and petrochemical industries to facilitate heterogeneous catalytic reactions. Despite the widespread use of packed bed reactors in industry, the flow, transport, and reaction phenomena occurring at the scale of individual catalyst pellets is poorly understood due to a lack of experimental techniques capable of non-invasively probing these phenomena at the pellet scale. This thesis presents the development and implementation of magnetic resonance (MR) techniques to investigate the hydrodynamics, mass transport and reaction occurring at the pellet scale within packed bed reactors. The aim of this work is to gain physical insight into the processes occurring at the pellet scale, which can subsequently be applied to optimize the catalyst and reactor technology used in industry. A novel MR method is developed and implemented to quantify the liquid-solid mass transfer coefficient in packed beds. The method, which utilizes MR relaxation exchange and a magnetization transport model, enables the true mass transfer process at the pellet scale to be probed without the confounding effect of molecular dispersion. The MR method is applied to measure the mass transfer coefficient for single phase liquid flow and two-phase gas-liquid flow in beds of SiO₂ and TiO₂ pellets. The limiting mass transfer coefficient at zero flow is directly measured for the first time, helping to resolve a debate in literature regarding its value. An MR velocity imaging methodology is developed to acquire 3D images of the time-averaged velocity and turbulent kinetic energy for turbulent flows in packed beds of commercially relevant α-Al₂O₃ pellets. The method is subsequently used to investigate the effect of pellet shape on hydrodynamics, revealing that the hydrodynamics at both the bed and pellet scale are influenced by pellet shape. Further, MR velocity imaging is applied to investigate the effect of the tube-to-pellet diameter ratio on the hydrodynamics. For the beds studied, the pellet scale hydrodynamics are found to be independent of the tube-to-pellet diameter ratio, but some differences in the near wall hydrodynamics are observed. Finally, MR based chemical shift imaging (CSI) is implemented to map the intra-pellet chemical composition at *operando* conditions during the hydrogenation of styrene to ethylbenzene over a Pd/Al₂O₃ catalyst. Inhomogeneous partial wetting of the catalyst, representative of that known to occur in commercial reactors, is found to cause substantial heterogeneity in chemical composition across a single catalyst pellet. Commonly used 1D reaction-diffusion models are inadequate to describe the observed reaction heterogeneity. The pellet-scale composition maps provide novel insight regarding the coupling between transport and reaction at the pellet scale and have important implications for pellet scale catalyst models.
Open Access
Monitoring Biological Events on Native Cell Membranes with Organic Bioelectronics
Lu, Zixuan
Traditional drug screening technologies rely heavily on animal testing, which is not only expensive and time-consuming, but also raises many ethical issues. Also, most drug candidates fail during the different phases of drug testing. Therefore, developing a low-cost, rapid, and animal-free system to allow mass-screening of a drugs’ specificity and efficacy is the promising way to increase testing efficiency for novel drugs. Since more than half of currently approved drugs target cell membranes, studying drug interactions with cell membranes is one of the key factors in understanding drug cytotoxicity and efficacy. The recently developed membrane-on-a-chip system, involving cell-derived supported lipid bilayers (SLBs) integrated with bioelectronic devices, is able to rapidly sense drug responses from cell membrane. This system eliminates the complexities and challenges of traditional testing systems, such as sterile cell-culture environments or physiological monitoring of animals. Also, quantification of drug responses with this system can be easily achieved by membrane quality analysis (i.e. electrochemical impedance spectroscopy (EIS)). Therefore, the membrane-on-a-chip system is a promising technology to advance next-generation drug discovery. Colleagues from our group previously demonstrated the detection of drug response with overexpressed transmembrane proteins with this systems and quantified the response with a model circuit. The key progress of this dissertation is sensing bioevents associated with naturally-expressed transmembrane proteins. This dissertation also discusses on multiple aspects of the membrane-on-a-chip system, such as strategies for device design and fabrication, electrochemical characteristics of SLBs, noise reduction for accurate measurements, and applications for biomedical research and drug screening. First, the two issues (1. gold-ring marks; 2. delamination of sacrificial layer induced by bubble formation) caused by photolithography during microfabrication are optimised and discussed for achieving high productivity and stable performance of these organic devices. Then, the experimental data was combined with numerical simulations to explore the relationship between changes in SLB quality and impedance output, delving into a deeper understanding of the impedance profiles of devices with and without SLBs, as well as extracted parameters such as membrane resistance (R_m). This approach was employed to investigate the relationship between microelectrode area and sensor sensitivity with changes in SLB state, towards rational device design. We highlight the trend of electrode size (polymer volume) required for sensing bilayer presence as well as the dependence of electrode sensitivity on SLB capacitance and resistance. Lastly, I illustrate how the flexible approach of including electrode and transistor measurements to amalgamate characteristic impedance spectra of transistors overcoming the problem of low-frequency noise and errors seen with traditional EIS. With a deeper understanding of the membrane-on-a-chip system, we further apply this technology to: detect SARS-CoV-2 viral entry and screen antibody efficacy to inhibit viral invasion, which is discussed in chapter 4; Study drug-induced blockage of voltage-gated calcium (CaV) ion channels on SLBs derived from neuroblastoma and primary cortical cells, which is discussed in chapter 5. In chapter 4, I demonstrate characteristic electrochemical impedance spectra of the early and late pathways of SARS-CoV-2 entry on SLBs derived from two cell lines: human lung epithelial Calu-3 cells and human embryonic kidney cells which overexpress the ACE2 receptor (HEK293-ACE2), respectively. The sensitivity of virus detection is further assessed by testing a range of viral particle concentrations; the detectable concentration reaching as low as ~10² virus pseudo particles (VPPs) per ml. We successfully apply this platform not only as a sensitive viral particle detector, but also as a drug-screening platform for screening antibodies that specifically target either ACE2 proteins on the host membrane or SARS-CoV-2 Spike proteins. For the neuronal membrane application demonstrated in chapter 5, I first performed differentiation on SH-SY5Y neuroblastoma cells, and then form native SLBs from the SH-SY5Y cells before and after differentiation on microelectrode arrays. An upregulated dose response of blocking of naturally expressed CaV ion-channels after SH-SY5Y differentiation was successfully detected via EIS. Further, the same setup was adapted to the native SLBs derived from rat primary cortical cells for the first time. The response of CaV ion-channel blocking on rat cortical neuron SLBs is comparable to the response from SH-SY5Y SLBs with and without differentiation. This dissertation demonstrates that this membrane-on-a-chip system is a rapid, ultra-sensitive, cell-free, and high-throughput platform to sense biological events, including specific interactions with endogenously expressed levels of membrane proteins. This could be an important contribution to the development of next-generation diagnostics and antibody/drug screening technologies for healthcare and biomedical research.
Open Access
Transfer Learning for Accelerated Process Development
Felton, Kobi; Felton, Kobi [0000-0002-3616-4766]
One of the greatest challenges in process development is the limited amount of data that can be collected. Techniques that can draw insights from this limited data have the possibility to accelerate process development. This thesis presents a collection of studies on using transfer learning to accelerate various aspects of process development. Part I focuses on reaction optimization, where I propose a benchmarking framework for comparing machine learning strategies for reaction optimization and demonstrate the benefits of using multi-task learning to accelerate chemical reaction optimization. In Part II, I explore the use of reinforcement learning and multi-fidelity Bayesian optimization for accelerating feedback controller tuning, specifically for distillation control systems. Finally, in Part III, I take two perspectives on using machine learning for predictive thermodynamics, a key aspect of process modelling. I introduce *DeepGamma* for predicting activity coefficients and ML-SAFT for predicting PCP-SAFT parameters, showing steps towards improving thermodynamic predictions using transfer learning. Together, all of these studies demonstrate the potential of transfer learning to accelerate process development, providing valuable insights for future research and practical applications.
Embargo
Optimal control reformulation for the solution of decision-making problems in chemical engineering
Mappas, Vasileios
This thesis focuses on the challenges that arise from the solution of decision-making problems in terms of convergence, efficiency and robustness. State-of-the-art solvers could fail to find the optimal solution or solve small instances of these problems. The original contribution of the current research consists of the development of a new methodology, based on optimal control theory, to solve different types of decision-making problems. The goal behind developing optimal control reformulations is to effectively solve these problems, while overcoming the drawbacks of the widely used commercial solvers, and three different types of problems are examined. First, the proposed solution scheme is used for finding the best pairings between control and manipulated variables and utilising the optimal tuning parameters, simultaneously, while satisfying path and end-point constraints. The proposed methodology results in the same pairings and gives better control actions compared to well-established methods. Furthermore, a novel multistage feedback controller is presented, which is proven to be able to handle disturbances with and without uncertainty and switching between operating points. Second, the Reverse Osmosis (RO) cleaning maintenance scheduling problem is examined. It is shown that the proposed framework can solve successfully this type of problem, even for large-scale configurations, long time horizons and arbitrary realistic model complexity of the underlying dynamic model of the RO process and produce an automated solution for the membrane cleaning scheduling, obviating the need for any form of combinatorial optimisation. Two-Point Boundary Value Problems (TPBVPs) are examined and solved utilising a double shooting approach, where the initial TPBVP is reformulated into a set of Ordinary Differential Equations (ODEs) and is solved as a least-squares optimisation problem in order to obtain the solution trajectory. Furthermore, a TPBVP formulation is extended and applied to solve dynamic optimisation problems (OCP), based on Pontryagin's Minimum Principle. Three different algorithms are used to solve the OCPs: the forward, backward and double shooting methods. In the case of OCPs, the proposed solution scheme obviates the need for discretising the control actions through the time horizon. The proposed algorithm successfully solved the TPBVPs and found the optimal solution of the OCP, showing a robust performance.
Embargo
Data-driven Dynamic Control Scheme for Antibody Producing CHO Cell Cultures in Fed-batch
Gangadharan, Nishanthi
Effective process control is a basic requirement for biopharmaceutical manufacturing to achieve high efficiency and enhanced quality control. High non-linearity and uncertainties associated with bioprocesses challenge the ability of traditional controllers to deliver satisfactory performance, thereby creating a need for advanced model based control strategies for efficient bioprocess control. Unlike the chemical or traditional pharmaceutical sector, bioprocess control is a complex process that involves controlling the behaviour of billions of cells in a bioreactor that evolves with non-linear dynamics over time. Variability in cell culture behaviour arising from heterogeneous culture conditions, product types, and cell types make developing a generalised control action challenging. Historic bioprocess data can provide valuable insights into the underlying dynamics of cell culture. Process Analytical Technology (PAT) initiative has highlighted the importance of identifying critical process parameters (CPP) of a bioprocess that influence critical quality attributes (CQA), to achieve seamless integration of analytical data with real-time monitoring and control for enhanced process understanding and to overcome manufacturing challenges. This thesis explores the different stages of development of a novel data-driven dynamic control scheme for bioprocesses in the context of antibody producing CHO cell cultures in fed-batch bioreactors. Multivariate monitoring techniques in biopharmaceuticals has resulted in the generation of large amounts of data comprising real-time measurements of critical quality and performance attributes, and if exploited efficiently can provide opportunity for developing superior control action. This work employs a novel methodology for condensing online data into an offline data matrix to achieve a comprehensive view of the critical process parameter landscape. The methodology was found superior to traditionally employed averaging and helped increase the number of variables available for representing the design space of the process. The augmented data set was then used to extract novel parameter relationships by applying concepts from network theory. The ability of these newly identified parameter relationship in describing the process efficiently was tested by constructing symbolic regression models. The performance of newly identified variables in predicting process behaviour was found superior to variable relationships found in literature. Following this, new models were generated using support vector machines (SVM) to predict the performance of these cultures at a future time point. These models helped decide the expected trajectory of a culture, based on process knowledge derived from historical bioprocess data. Customised control strategies were developed, to acknowledge the process dynamics of different days of the culture, by employing an optimisation algorithm along with the different models developed throughout the course of this study. The control scheme dynamically recalculates the expected trajectory and proposes customised reactive control action when encountered with a deviation from the expected trajectory. The proposed control scheme was able to recommend sensible control actions during the different test cases designed for control. This closed-loop model-based multi-attribute control scheme, that employs concepts from data science, network theory and control theory, ensure that the cultures remain on a pre-defined well established trajectory thereby minimising variability.

Browse

Recent Submissions