Research

In our group we currently conduct state-of-the-art computational research in four main areas (see below). Most of this research is carried out in close collaboration with leading experimental groups in the field and from worldwide academic institutions. Whenever possible, we try to guide the design of novel catalysts with an enhanced performance and ask our experimental collaborators to synthesize and test them to support our theoretical predictions.

Matching the global energy demand in a clean, reliable and economically affordable way is one of biggest challenges of this century. While solar or wind power shows great promise to meet the forthcoming energy gap, they are limited by the fact that they cannot generate energy on a constant basis; basically, because the sun (wind) does not shine (blow) at all times and with the same intensity everywhere. A practical solution to this problem is to store the excess electricity from renewable sources in the form of high-energy-density chemical bonds such as molecular hydrogen, H₂. For example, one could dissociate water into H₂ and O₂ in an electrolysis cell, and store the former gas in a tank. Then, when the energy demand increases, the stored H₂ could be combined with O₂ present in the air to produce water and electricity in a fuel cell, thus closing a sustainable energy cycle without any carbon footprint.

What is then preventing us from using water and sunlight to power our homes, cars and electronic devices? The answer is very simple. Dissociating water into H₂ and O₂ is a two-reaction process that requires a large amount of energy, specifically the formation of the latter. Thus, we need chemical substances (i.e. electrocatalysts) that reduce this energetic cost to make the overall process feasible. While at present there are some materials that can catalyse this reaction efficiently, they are made of very rare elements like platinum or iridium, which prevents their global commercialization. Hence, it is essential to find alternative catalysts based on Earth-abundant elements if we want this technology to compete with fossil fuels and meet the global energy demand.

One of the main research aims of the CCEM Group is to use supercomputers and machine learning models to speed up the discovery of highly active, stable, and selective catalysts based on Earth-abundant elements to be integrated into commercial energy conversion devices to supply clean, reliable and inexpensive energy to the world. Some of the electrochemical reactions that we are most interested in include the hydrogen and oxygen evolution reactions, as well as the reduction of N₂ and CO₂ for the sustainable production of chemical fuels and feedstocks such as ammonia, methanol and formaldehyde.

In the current search of new cost-effective catalysts, graphene is arguably one of the most attractive materials due to its intriguing physical, chemical, and mechanical properties. The discovery of graphene as single-atom-thick crystallites in 2004 by Andre Geim and Konstantin Novoselov led them to win the Nobel Prize in Physics and opened the era of the so-called 2D-materials. Since then, hundreds of other inorganic layered materials have been reported including the families of transition metal dichalcogenides (e.g. MoS₂, WS₂), transition metal oxides (e.g. Mo₃, V₂O₅), transition metal trichalcogenides (e.g. NbS₃), and transition metal halides (e.g. MgBr₂, MoCl₂). Because of their different physical and chemical properties compared to their bulk counterparts, as well as their high tunability, 2D-materials are regarded as promising cost-effective catalysts for future technological applications.

Some of the members in our group are using advanced computational methods to explore various families of 2D materials as high-performing catalysts for the sustainable and scalable production of organic building blocks. Interestingly, some of our most recent research in this area has led to the prediction of a promising 2D catalyst that our experimental collaborators are currently synthesizing and testing in the laboratory.

Transition metal oxides (TMOs) are known to effectively promote – as a catalyst or support – a wide variety of chemical reactions for different applications including photocatalysis, fuel cells, biomass conversion, and synthesis of fine chemicals and drugs. This versatility can be attributed, in many cases, to the presence of different types of structural (e.g. kinks, steps, terraces, vacancy sites, and dopants) and electronic defects (e.g. electrons and holes), which can drastically modify their properties, and therefore, their catalytic performance. In other cases, the intrinsic properties and chemical reactivity of TMOs can be rationalised with the Lewis acidity and basicity of their constituent elements. This remarkable tunability, together with the high thermal and chemical stability of many TMOs, make these materials suitable candidates for a broad range of technological applications.

Metal alloys offer unique catalytic advantages through their tunable electronic structures and synergistic interactions between constituent metals, often outperforming monometallic systems. Our group investigates bimetallic alloys using first-principles calculations and data-driven strategies to design efficient catalysts for energy and industrial applications. To accelerate this process, we apply machine learning, combining classification and regression algorithms, to predict key adsorption properties such as binding energies as well as to explore ordered and disordered alloy surfaces. For instance, focusing on CO adsorption on Cu-based alloys, we developed a two-step ML framework that accurately identifies stable adsorption sites and predicts binding strengths, enabling rapid and cost-effective screening of alloy surfaces for CO₂RR, for details see our pulication page.

Besides the application of TMOs in electrocatalysis, in our group we are very interested in the modelling of TMOs for catalytic reactions under thermal conditions, including (but not limited to) CO oxidation, solid oxide fuel cells, and the activation of non-polar molecules such as H₂ and CH₄. To model TMOs we mostly employ periodic density functional theory (DFT) methods except in the case of highly correlated TMOs, for which we use DFT+U following the rotationally invariant approach of Dudarev et al.

Our key objectives include: (i) elucidating the underlying reaction mechanisms; (ii) understanding the factors that govern catalytic activity in order to establish reliable reaction descriptors, through the use of electronic and bonding analysis as well as microkinetic models; and (iii) applying these descriptors for the rational design of catalysts or in machine learning-based screening using existing molecular or ligand databases—leveraging our newly developed package, DART—for the discovery of improved catalysts in industry relevant processes.

We are particularly interested in main group catalysis and organometallic transformations that involve the activation of typically inert bonds, such as O–O, C–H, and C–C bonds, as well as the conversion of small molecules like CO₂, NOₓ, and methane.

Our group is also making significant contributions to the discovery of novel oxygen evolution reaction (OER) catalysts, enabling green hydrogen production through water splitting. We actively employ data-driven approaches, using machine learning-based surrogate models to identify promising candidate complexes. Our goal is to develop algorithms that can guide future calculations via active learning (AL).These and other related reactions are being explored in close collaboration with our experimental partners.