Dynamics of cluster catalysts under reaction conditions

Catalysts are substances that make it easier for a particular chemical reaction to happen, without being themselves consumed in the reaction. This may mean that some industrial processes require lower temperatures, and therefore less energy. In other cases, some chemicals cannot be obtained at all without using a catalyst. Designing new catalysts holds enormous potential to reduce the worlds energy consumption and enable new technological developments. However, the problem is that we often do not understand well enough how catalysts actually work to make progress in improving them. We now know that a catalyst typically looks very different from its default state while it is being used in a reaction: It may constantly change its shape and chemical composition, and this is an important part of its function. Understanding this active state is crucial in order to design new and better catalysts. In my project, I will investigate a special class of catalysts consisting of small clusters of just a few metal atoms, which sit on some cheap material. Interestingly, it turns out that this support material also plays a big role in the activity of the catalyst, and part of the project is to investigate this interaction. Because real catalysts are so complex and hard to understand, the plan is to use model systems: Starting from a highly ordered support (a single crystal), it is possible to know exactly where every single atom is. I will then use a size-selective cluster source, which again allows obtaining extremely high control over what exactly is put onto this surface for example, lots of clusters that all consist of exactly seven platinum atoms. Creating these very well-defined systems will make it easier to later interpret what happens during a reaction. To look at the clusters at the conditions that interest us, I will use a special type of scanning tunnelling microscope, which allows looking at single atoms while adding some low pressure of a reaction gas mixture and to heat up the sample. These conditions are still more mild than what is typical for a real catalyst, but are enough to already see some dynamics. With these tools, I will then first determine what types of gas mixtures and temperatures are most interesting to explore. Once this becomes clear, I can take the same samples and expose them to much higher pressures in a microreactor, where it is also possible to detect reaction products. Going back and forth between microscopy, the reactor, and additional spectroscopic experiments, I will be able to explore what exactly happens to the clusters while they catalyse a reaction and how, exactly, the catalysis works.

Whether platinum catalyst particles stay active on a surface, become covered, or even end up buried can be decided by the oxide support they sit on, especially by how oxygen-rich or oxygen-poor that support is. Catalysts underpin cleaner, more energy-efficient routes to fuels and chemicals, including reactions that could upgrade carbon dioxide. A major challenge is that a catalyst rarely keeps its "as-prepared" structure once it meets hot, reactive gases: metal particles can change size or chemistry, and reducible oxides such as titanium dioxide (TiO) can rearrange as they gain or lose oxygen. To design better catalysts, we need to understand these working states, not just the starting materials. In this project we worked on simplified "model catalysts" with unusually high control. We prepared TiO surfaces whose oxygen content can be set reproducibly, deposited platinum as both very small clusters and larger nanoparticles, and then exposed the same samples to relevant gases at elevated temperature and low (near-ambient) pressures while observing the surface with an atom-resolving microscope and complementary spectroscopy. Two publications summarize the main outcomes. First, in Nanoscale (2024) we established preparation protocols that let the TiO crystal return to a consistent bulk reduction level over many cycles, and we showed that this bulk state strongly changes surface reactivity in O, H, CO and CO. Notably, an oxygen-poor TiO surface could be re-oxidized by CO under our conditions, while a nearly oxygen-balanced surface did not show this reaction. With platinum present, the support state and the surrounding gas together decided whether particles remain exposed, become covered by a thin TiO layer, or are buried deeper inside the oxide. Second, in JACS (2025) we disentangled the effects of particle size, oxygen pressure, and support stoichiometry by following platinum from ultra-high vacuum up to 0.1 mbar oxygen. Sub-nanometer clusters oxidized easily (even at room temperature) and were less thermally stable under oxidizing conditions than well-crystallized nanoparticles. Crucially, on a reduced, defective TiO support, platinum particles were buried quickly by newly formed TiO layers during support re-oxidation, completely deactivating the catalyst. A side-by-side comparison with a common TiO powder catalyst showed that matching and maintaining the support's oxygen content is essential if model studies are to reflect real catalysts. Overall, the project shows that catalyst behaviour depends on multiple factors, but that controlling the oxide support offers a practical lever to keep precious-metal particles accessible and stable in realistic reaction environments. Ongoing work in CO hydrogenation atmospheres supports the same conclusion, where re-oxidation of reduced surfaces by excess CO can be detrimental.