Understanding the fundamental mechanisms of single-atom catalysis (SAC) is important to design systems with improved performance and stability. This is problematic, however, because singleatom active sites are extremely difficult to characterize with existing experimental techniques. Over the last 40 years, surface science has provided the fundamental insights to understand heterogeneous catalysis, but model systems in which metal atoms are stable on well-characterized metal-oxide substrates at reaction temperatures are scarce. In this perspective, I discuss what is already known about isolated metal atoms adsorbed on model metal-oxide surfaces, and how this information can be used to understand SAC. A key issue is that, although the highly-idealised model systems studied in surface science may not be representative of a real working catalyst, they do very much resemble what can be calculated using state-of-the-art theoretical modelling. Thus, surface science offers an opportunity to rigorously benchmark the theoretical approach to modelling SAC in future. Perhaps more excitingly, several groups have developed model systems where metal adatoms remain stable at elevated temperatures. To date however, there has been no clear demonstration of catalytic activity. The perspective closes with a brief discussion of the prospect for STM experiments under realistic reaction conditions.
Main TextThe rapidly emerging field of single-atom catalysis (SAC) aims to slash the precious metal loading in heterogeneous catalysts by replacing metal nanoparticles with so-called "single-atom" active sites 1 . While there are many reports of active SAC systems, the topic remains somewhat controversial because it is very difficult to characterize a system based on single atoms, and to distinguish between these and subnano particles 2 . In practice, most groups use aberration-corrected transmission electron microscopy (TEM) to demonstrate the atomic dispersion, sometimes supplemented by XANES, which can rule out significant metal-metal bonding 1 . The activity of the catalyst is then tested, and although some mechanistic information can be drawn by in-situ techniques such as IRAS, the catalytic mechanism is proposed largely on the basis of theoretical calculations. Such calculations are based on an idealised model of the system, in which both the support structure and active site geometry are assumed. Thus, a one-to-one correspondence between the experimental and theoretical results is difficult to prove.Traditionally, surface science has provided mechanistic information to understand heterogeneous catalysis. The idea is to strip away the complexity of a real catalyst and study well-defined singlecrystal samples in a highly-controlled ultrahigh vacuum (UHV) environment. This way, the adsorption of individual reactants can be studied in detail, and an understanding of the basic interactions can be built up. The downside of the approach is that the highly idealised model system may not be as representative of the real catalyst as o...