Research

We develop computational and knowledge-driven approaches to understand and design molecular and materials systems. Our work connects chemical knowledge, artificial intelligence, molecular modelling, and molecular assembly, with a growing interest in how these methods can support chemistry education.

Chemical Knowledge & Artificial Intelligence

We develop methods that allow chemical knowledge to be represented, connected, and used computationally. This includes molecular representations, ontologies, knowledge graphs, machine learning, and AI-based workflows.

Chemical knowledge is distributed across molecular structures, experimental data, calculations, publications, and databases. We investigate how this information can be organised in machine-readable form while retaining the chemical relationships that give it meaning.

Knowledge graphs are central to this work. They provide a structured way to connect compounds, properties, reactions, methods, and scientific evidence. We use these representations to search chemical information, identify relationships, and support hypothesis generation and chemical reasoning.

We are also interested in AI agents for chemistry. Our work explores how language models can interact with structured chemical knowledge, molecular modelling tools, and scientific databases. The aim is to build workflows that remain chemically interpretable and keep the chemist involved in the reasoning process.

Molecular Assembly & Materials Design

We study how molecular building blocks assemble into complex inorganic and hybrid structures. A central question is how composition, connectivity, geometry, and electronic structure determine the architecture and properties of the resulting system.

Computational models allow us to explore structural spaces that are too large to investigate experimentally. We use these models to identify recurring assembly principles and to connect local building-block geometry with the structure of complete molecular systems.

Programmable Molecular Assembly

Polyoxometalates provide an important platform for this research because they combine structural diversity with rich electronic and redox chemistry. We have studied heteropolyoxovanadates and late-transition-metal-containing POMs, including systems based on V, Pd, Au, Pt, and Cu.

Our aim is to go beyond explaining known structures and use computational models to identify plausible new ones. Structural enumeration, molecular modelling, quantum-chemical calculations, chemical knowledge, and experimental observations are combined to understand how molecular architectures emerge and how they might be designed.

The same concepts can be extended beyond polyoxometalates to other molecular and reticular systems, including metal-organic polyhedra, molecular cages, and framework materials.

POMtronics and Molecular Function

Polyoxometalates show complex relationships between redox chemistry, protonation, supramolecular interactions, electronic structure, and metal-metal bonding. These features make them useful molecular systems for studying dynamic chemical and electronic behaviour.

Within the concept of POMtronics, we investigate how changes in bonding and electronic structure can produce molecular functionality. Current interests include reversible metal-metal bond formation, electron storage, molecular switching, photoredox processes, and catalytic reactivity.

The aim is to understand these processes at the molecular level and use that understanding to guide the design of responsive inorganic molecular systems.

Digital Chemistry Education

Chemistry relies on several different ways of representing the same system. Students move between formulas, molecular structures, graphs, numerical models, and physical concepts. We are interested in how digital and physical tools can help learners connect these representations.

Our work includes molecular modelling, digital molecular representations, interactive tools, physical molecular models, and the use of artificial intelligence in chemistry education. A particular interest is how students construct, modify, and reason with chemical representations.

Physical Models and Programmable Structures

Large nanoscale and reticular structures are often difficult to represent with conventional molecular modelling kits. We have investigated low-cost interlocking disks as modular building elements for constructing polyhedral, reticular, and chemistry-inspired structures.

Their geometry allows learners to explore connectivity, symmetry, curvature, strain, and structural stability directly through construction. These models are useful for teaching, workshops, and outreach activities where spatial reasoning and molecular assembly are central.

This work complements our interest in digital molecular editors and computational environments where learners can construct and manipulate molecular and materials structures.