One of the largest barriers to commercialization of novel heterogeneous catalysts is their long-term stability and deactivation under reaction conditions. As the field of heterogeneous catalysis continues to mature, significant attention has been spent on studying deactivation mechanisms to understand how to design catalysts with long-term activity. Here, we use the reverse water-gas shift (RWGS) reaction as a model reaction to characterize the deactivation mechanism of rhodium nanoparticles (RhNPs) on TiO2 supports. These RhNPs are synthesized via an aqueous, one-pot, colloidal synthesis method that results in a narrow size distribution and can be tuned to produce particles of varying size and crystallinity. A TiO2 support was functionalized with various loadings of RhNPs and tested under reaction conditions, (3:1 H2:CO2, 300 °C to 600 °C) for 16+ hours. Notably, these catalysts shift in selectivity from methanation to RWGS after a few hours before showing overall deactivation to a steady-state. The rate of change in selectivity and overall rate of deactivation are strongly tied to the size and structure of the RhNPs. This structural sensitivity is investigated through a variety of characterization techniques, including E-TEM, XAS, AP-XPS, in-situ DRIFTS, and H2-Chemisorption. Additionally, we have shown that the typical pretreatments used on heterogeneous catalysts have a pronounced effect on the structure and performance of these RhNPs. We report on how our results align with the literature’s understanding of how single atoms of Rh preferentially perform RWGS while small clusters preferentially perform methanation, and go further to investigate how various deactivation mechanisms, such as sintering, strong metal-support interactions, and particle decomposition, play a role in the selectivity and long-term activity of the catalysts. These studies enhance the field’s understanding of selective catalyst deactivation and provide valuable insights into characterizing and designing catalysts to minimize long-term deactivation that can be applied to future systems.
