Towards sustainable structural materials for energy storage

Batteries are expected to play a pivotal role in the electrification of a range of sectors, including transport, aerospace and grid-scale storage. Of the next generation battery chemistries, Li-S batteries are a particularly attractive option due to their projected high energy density, low cost, and operating temperature range. However, the chemistry of such systems is complex, involving the electrochemical conversion between two insulating species (S and Li₂S), dissolution and shuttling of soluble intermediates, uncontrolled deposition and volume expansion; if not carefully regulated, these processes can lead to gradual capacity fade at best, explosive cell failure at worst. Here, I’ll discuss the development of free-standing carbon fibre electrodes that are conductive, lightweight and mechanically robust, and their role in addressing degradation mechanisms arising from volume expansion, polysulfide shuttling, dendrite formation and inventory loss. The fibres were prepared via electrospinning of biomass precursors followed by further heat treatment. The structural properties can be tailored by carefully tuning the treatment conditions, to enable free-standing cathodes with high sulfur loadings and enhanced polysulfide interactions to suppress shuttling and sulfur inventory loss. As anode supports, the carbon fibres provide a lithiophilic substrate for a homogeneous lithium-ion flux and low deposition overpotential, which favours large, uniform and low surface area lithium deposits. The assembled cell demonstrates greater capacity retention over long-term cycling than conventional Li-S cathode and anode substrates, making this process a promising route to achieving new electrode materials for Li‑S technologies.

Advancing Hematite Photoanodes for Photoelectrochemical Water Splitting: The Impact of g-C₃N₄ Supported Ni-CoP on Photogenerated Hole Dynamics

Solar-driven photoelectrochemical (PEC) water splitting using semiconductor materials represents a clean method for solar energy storage in clean fuels such as hydrogen. My PhD projects focus on the development of photoelectrode for PEC water oxidation and H2 production.

Thin Film Composite Membranes from Polymers of Intrinsic Microporosity for CO₂ Separation

Modern membrane technology provides substantial energy savings and economic and environmental benefits, serving as a viable alternative to energy-intensive industrial processes like distillation and adsorption. However, current commercial membranes, such as cellulose acetate and polysulfone, have limited gas permeability and selectivity. Polymers of intrinsic microporosity (PIMs) offer a promising solution due to their diverse chemical and structural properties. By incorporating bulky, rigid contortion centres, polymer chain packing can be disrupted, resulting in high excess free volume and thus, high gas permeabilities. Despite initially demonstrating excellent separation performance, PIMs often suffer from significant physical ageing and plasticization over time, which greatly limits their effectiveness in long-term applications. My PhD research has focused primarily on the synthesis of novel PIMs with improved gas-pair selectivity, long-term stability, and scalability. Notably, Thin Film Composite (TFC) membranes excel in handling high gas fluxes and are highly energy-efficient for CO2 capture, despite prior research on PIMs predominantly focusing on the fabrication of dense, thick membranes. Thus, my research also aims to develop reproducible PIM-based TFC membranes to fully leverage their potential for future commercial applications.

Environmental effects on the spatio-temporal evolution of plasma-wrinkled PDMS thin films

We investigate the surface plasma oxidation of polydimethylsiloxane (PDMS) elastomers and the impact of environmental conditions on the spatio-temporal evolution of the resulting wrinkling patterns. We employ a combination of small-angle light scattering (SALS), XPS, and optical and atomic force microscopy to examine the evolution of the crack density, and wrinkling profile following plasma exposure. Previous reports have rationalised the spontaneous isotropic wrinkling of PDMS-supported thin films following plasma exposure in terms of a thermally-driven differential deformation of the elastomer normal to the direction of incident exposure. Our findings challenge this widely accepted mechanism and provide evidence that, while temperature plays a role, other environmental factors largely govern this process and can account for the considerable variability of results reported in the literature using this method. These findings open new possibilities for exceptional control of PDMS plasma functionalisation, in terms of tunable surface patterning, adhesion, and wettability. 

Materials for a brighter future: designing photocatalysts for solar fuels production

As we strive to reach the net-zero target by 2050, it is clear that the production and use of sustainable fuels is essential. By using solar energy – a renewable source with immense energy output –, we can produce so-called solar fuels. A promising pathway towards the production of such fuels, which is both green and safe, is photocatalysis. In this process, some materials can convert molecules such as H₂O or CO₂ into value-added chemicals and fuels, upon light irradiation. These materials are called photocatalysts. H₂O splitting to produce H₂ is the most broadly studied photocatalytic reaction, partly due to its manageable thermodynamic barriers. Photocatalytic CO₂ reduction is another reaction of interest, combining utilisation of CO₂ emissions and production of sustainable fuels (such as methane or methanol). However, numerous challenges remain; major ones constitute the design, production, and deployment of an efficient photocatalyst. My research is focused on the development of photocatalysts for CO₂ photoreduction, and understanding of their chemical, sorptive and optoelectronic properties – which can regulate catalytic activity. 

Carbon Nanotube Production from Molten Li₂CO₃ via High Temperature Electrolysis

My PhD research focuses on tackling global warming and the utilisation of carbon dioxide in carbon nanotube production with electrochemical methods. I'm currently working on investigating the mechanism of carbon nanotubes growth in molten carbonate salts and optimising the high temperature electrochemical cell. 

Olivine dissolution in CO₂-saturated water under geological carbon storage conditions

Geological carbon storage (GCS) is one of the most promising and reliable techniques to avoid releasing CO₂ into the atmosphere. Among the four trapping mechanisms of GCS, mineralization provides the most permanent way for fixation of CO₂ by conversion into carbonate rocks. Olivine is one of the most promising minerals for this purpose because it is abundant in Earth's upper mantle. 

My PhD aims to study three stages or aspects of carbon mineralization:

1. the solubility of the gas impurities, for example CO and N₂, of the injected CO₂ stream in brine under reservoir conditions;
2. the dissolution kinetics of Olivine under reservoir conditions;
3. Olivine carbonation under reservoir conditions. Experiments were conducted first, and models have been being developed to correlate the experimental data or rationalize the results. 

The results from the PhD project extend the database of Olivine dissolution rates to a higher temperature, help understand the process of Olivine carbonation, and provide the first data for CO solubility in brine with high molality of salt. This project helps understand the long-term fate of the injected CO₂ and the geochemical reaction under reservoir conditions, and helps optimize the design of a GCS project.

Data-driven modelling and prediction of complex systems

Complex systems (CSs) are ubiquitous in nature, technology, and society, encompassing seemingly unrelated domains such as fluids, climate, social networks, and financial markets. They are characterised by intricate interactions among their constituents at the microscopic scale, resulting in emergent behaviour that manifests in macroscopic observables. Understanding their evolution poses major challenges, often addressed by coarse-graining techniques and phenomenological models. Equilibrium-based models, which assume a time-invariant dynamics of the CS at hand, prevail due to their simplicity, serving as zeroth- or first-order approximations of the CS behaviour. However, they are not capable of describing phenomena like phase transitions or dynamical regime changes. To address these limitations, we employ machine learning (ML) techniques, in particular physics-informed ML and neural-differential equations, together with statistical mechanics techniques. The overarching goal is to develop a comprehensive theoretical-computational framework applicable across different scientific and application domains. Our aim is to enhance understanding of CS behaviour under both equilibrium and non-equilibrium conditions, while also bridging the gap between the time evolution of macroscopic observables and the underlying microscopic dynamics, and uncovering similarities between different CS classes.

From white- to black-box modelling in chromatographic separation processes

High-fidelity (white-box) process models of chromatographic separation systems are described by complex Partial Differential and Algebraic Equations (PDAEs). This often results in high computational costs that can complicate further online applications, such as optimisation and control. To tackle such challenges, hybrid (grey-box) models, that integrate first-principles knowledge with empirical data, and data-driven (black-box) models, that rely primarily on process data, can be employed to reduce the computational burden. In this work, we are proposing a novel methodology for the development of hybrid and data-driven models for separation processes.
We focus on the twin-column Multicolumn Countercurrent Solvent Gradient Purification (MCSGP) process, used for the purification of monoclonal antibodies. The process model assumes lumped kinetics, and after spatial discretisation, using 50 collocation points, comprises 3309 variables and 4119 equations. To decrease the computational expense, we train Artificial Neural Networks (ANNs) to directly predict the periodic elution profile of the chromatographic system at Cyclic Steady State (CSS). We investigate various model structures with different degrees of physicochemical process knowledge and compare their performance to the high-fidelity model formulation, using accuracy and simulation time amongst other performance metrics. 

Humans, Algorithms, and Chemical Engineering: Collaborative tools for Reactor Design & Beyond

My PhD focuses on the use of machine learning to enhance various aspects of Chemical Engineering. Specifically, speeding up the design of expensive experiments such as CFD simulations of chemical reactors with different geometries. In doing so, I identify novel reactor geometries that are more efficient, resulting in more sustainable chemical processes. These new reactors are then 3D printed and experimentally validated using reacting flows. Another key aspect is considering how humans interact with these new algorithms, particularly in settings where experts such as chemists are present. I have designed methodologies where these experts can have a say in the experimental design process, resulting in finding better solutions in fewer experiments. My PhD combines aspects of machine learning, experimental design, and chemical engineering, with a strong consideration as to how humans interact and enhance the links between these fields. 

Kinetics of Enzymatic Reactions

My PhD project centers on optimizing glycosylation reactions within the SUGAR-TARGET platform, a system designed for precise sequential glycosylation using immobilized enzymes. By refining these processes, we aim to produce bespoke glycoproteins with high specificity and efficiency. These custom glycoproteins have significant potential in the treatment of cancer and autoimmune diseases, offering targeted therapeutic options with improved efficacy. 

Adopting and Adapting: Development of an artificial virus for efficient RNA delivery and improved stability to enhance immunity against viral diseases

RNA-based technologies offer promising solutions for combatting infectious diseases, but their application is hindered by RNA degradation before reaching their site of action and the challenges associated with existing delivery vectors. To address these issues, we developed an artificial virus that resembles the structure and functionalities of natural viruses, yet avoids their inherent drawbacks such as toxicity, pre-existing immunity, and production challenges. 

This novel system employs a shell of biodegradable amphiphilic polymers, simulating the viral envelope, and a core composed of lipid nanoclusters that mimic the viral nucleocapsid. The structure, morphology, size, charge and functionality of the artificial virus were characterized using cryo-electron microscopy (cryo-EM), small angle neutron scattering (SANS), zeta sizing and cell transfection, providing detailed insights into this novel system. This synthetic vector has demonstrated enhanced stability at room temperature and effective delivery capabilities. In vivo testing using mouse models as an influenza vaccine confirmed its potential, successfully protecting mice against influenza challenges. This approach not only broadens the scope of RNA technology in therapeutic applications but also leverages biomimicry to circumvent limitations faced by current delivery systems.

Stimuli-responsive nanoparticles by self-assembly of sequence-controlled multifunctional biodegradable polymers for enhanced RNA delivery and thermostability

Synthetic polymers have shown significant promise in drug delivery applications owning to their structural and functional versatility. However, their further clinical translation is impeded by potential toxicity concerns. In this project, a library of novel biocompatible multifunctional polyesters were synthesised via a Quantitative One-pot Iterative Living Ring-Opening Polymerisation (QOIL-ROP) method. This innovative strategy enables the precise control of polyester chemical structure and functionality. The self-assembly of multifunctional polyesters in aqueous environment was explored, highlighting the effects of polyester chemical strucrture on the self-assembled nanoparticle morphology. The multifunctional polyesters were then evaluated for their efficacy in intracellular delivery of self-amplifying RNA (saRNA). By tuning key parameters such as polyester molecular weight, charge density, and block sequence, enhanced saRNA delivery and negligible cytotoxicity was achieved, as compared to the benchmark system based on cationic polyethylenimine (PEI). Notably, the optimised formulations exhibited improved long-term saRNA thermostability at refrigerator temperatures, outperforming current approved mRNA vaccines that require ultra-cold chain storage. Taken together, the multifunctional polyesters synthesised via the QOIL-ROP method are considered as promising therapeutic delivery vehicles owning to their simple and scalable production, excellent biocompatibility, efficient delivery and improved long-term formulation stability.

Diglycine crystallisation in the presence of impurity

Crystallisation is a widely used separation technique when purifying the final product or intermediate product in biomolecule production. In the synthesis of peptides, amino acids are employed as reactants and the unreacted residues are mixed with the product stream. Our study investigate how downstream separation process is affected by different impurity levels. Diglycine and glycine were selected as the model compounds for their simple structures, good aqueous solubility, and availability. 
With the common peptide synthesis techniques, the coupling efficiency for joining two amino acid molecule ranges from 90 to 95%, so we tested multiple impurity levels below 10%. The concentration curves measured with time indicated that 2.5% impurity reduced the nucleation rated of diglycine by over 60%, delayed the induction time from 1.6h to more than 5 h. 10% glycine further increased the induction time to over 15 hours. Though low impurity levels were found only affected crystallisation kinetics, at high impurity level where both solutes were supersaturated, the thermodynamics of the process are also changed, illustrated by the changed in crystal morphology, detected by PXRD. The structure of the new crystal form was determined from the spectrum, attributed to cocrystal formation by glycine, diglycine, and water in 1:1:1 ratio.

Development of multi-scale model of thrombolytic therapy for ischemic stroke

Stroke stands as the second leading cause of death globally. For decades, alteplase - a recombinant tissue plasminogen activator (tPA) - has been the sole FDA-approved thrombolytic agent for acute ischemic stroke (AIS) treatment. However, clinical trials also demonstrate its serious ICH risk and low recanalization rate for large clot occlusion. The therapeutic benefits of tPA are further compromised by its short half-life and the body's inhibitor PAI-1. Furthermore, efficacy of thrombolytic therapy is also affected by the biological conditions of the patients and thrombus properties.  New Clinical trials focus on testing novel thrombolytics drugs and improving their dose regimens, but it is very time consuming to design such trials for AIS therapy. Short therapeutic window of thrombolytic therapy makes it difficult to conduct any pharmacokinetic-pharmacodynamic (PK-PD) clinical study, which is very important for dose regimen design. Therefore, conducting in-silico studies of the PK-PD of thrombolytic agents can be very useful in understanding and evaluating their performance. The overall aim of my PhD project is to develop multiphysics thrombolysis models which can simulate the physical and biochemical processes such as protein kinetic, blood flow and drugs transport mimicking laboratory and in vivo experiments of thrombolysis.