Research

Knowledge Engineering

Knowledge Engineering (KE) is a field of artificial intelligence that deals with developing and maintaining intelligent systems capable of solving complex problems using symbolic methods. These systems mimic the decision-making process of a human expert, thereby enhancing the quality and efficiency of problem-solving in various domains.

The research in KE focuses on the chemical domain, where we use ontologies to represent knowledge. Ontologies are digital frameworks that enable the instantiation of heterogeneous chemical information in a machine-readable format. This process involves structuring the information in a manner that facilitates efficient and accurate reasoning.

We establish a repository of valuable chemical information and knowledge by populating the knowledge graph with instantiated knowledge. This knowledge graph serves as the backbone for our problem-solving systems, allowing us to build software agents that replicate the decision-making processes of domain experts. These agents use evidence-based reasoning to access the knowledge graph, query information and generate new knowledge.

Our work results in a system capable of solving complex chemical problems more efficiently and effectively than traditional methods. By mimicking the decision-making processes of a human expert, our software agents enable us to automate complex decision-making tasks, improving the quality of decision-making while reducing the time required to solve complex problems.

Inorganic Materiomics

Materiomics is a holistic approach to the study of material systems, focusing on their functionality and behaviour rather than just a collection of individual properties. Our research is dedicated to exploring the less-explored inorganic frontier of materiomics.

Our work focuses on the study of molecular metal oxides, the most abundant materials in the Earth’s crust. Metal oxides are believed to have played a catalytic role in the formation of important biological molecules and have contributed to technological and societal advances throughout human history. We are particularly interested in the self-organization of polyoxometalates (POMs), the most explored class of molecular metal oxides. POMs are negatively charged and made of early transition metals, and we are interested in understanding how their electronic structure and properties can be tuned to achieve advanced functionality, such as “POMtronics.”

Adaptive Self-Organization in POMs

While millions of reported POM compounds exist, only a few classical archetypes have dominated POM-related applications. It is crucial to understand how different synthetic factors influence the adaptive self-organization of POMs and how we can tailor it towards the discovery of new, broadly useful compounds. Experimentally, studying the whole reactive space is impractical due to the unlimited number of synthetic combinations that can influence POM self-organization. However, the virtual structural POM space can be computationally explored, defined by overall connectivity, nuclearity, or building block type/ratio. We have successfully exploited this in silico approach to understand how structural and electronic factors interact to determine the magnetically important heteropolyoxovanadates (heteroPOVs) and catalytically relevant late transition metal-based POMs. Through this project, we are developing theories and concepts that can rationalize structural trends in POM chemistry and enable predictive structure discovery. Using a computer-aided approach, we attempt to develop new POM materials (V, Pd, Au, Pt, and Cu-based) and explore their self-organization using a combination of theoretical, spectroscopic, and spectrometric techniques.

POMtronics

POMs exhibit a complex interplay of supramolecular, photo-electronic, redox, and protonation/cation interaction properties. Obtaining insights into this complexity often requires a combined theoretical and experimental approach. Once well-understood, we can make smart and advanced applications of POMs. In our project, we are looking at dynamic functionalities that impact changes in covalent bonding, particularly the metal-metal bonds in POMs and the dynamics of the photo-redox activation of hydroxo moieties. Both topics strongly impact the catalytic conversion of fine organics, such as hydrogen atom transfer catalysis. The metal-metal bonding formation in POMs is also important for their application as molecular switches and electron storage materials. Our research aims to understand these complex interactions better and explore their potential for innovative, dynamic POM-based applications.

Educational Tools

Nanoscale materials are a challenging subject in chemistry education, with their complex structures requiring hands-on manipulatives for effective teaching. However, traditional modelling kits such as ball-and-stick kits or zometools are not always practical or cost-effective, particularly for constructing large multi-atom models like polyoxometalates, nanotubes, and mesoporous materials. Similarly, linearly binding legoidal toys are not always suitable for creating the spatial connectivity and flexibility required for constructing chemistry-relevant assemblies.

To overcome these limitations, we have found that plastic interlocking disks (ILDs) can be an excellent alternative. Originally designed for training dexterity in young children, ILDs are inexpensive and readily available. We have demonstrated that they can be used to construct chemistry-relevant polyhedral and reticular assemblies that are typically several decimeters in size, making them ideal for classroom demonstrations.

What’s more, the 8- and 6-fold symmetrically grooved ILDs offer unique opportunities for teaching spatial thinking and structure programming abilities. The hands-on experience of working with semi-robust ILDs provides a tactile understanding of the emerging strain effects and anticipation of the construction stability. As a result, the combination of visual and sensational components makes ILDs a discovery-based learning tool that is increasingly popular in outreach workshop activities.