Understanding the working state of the surface structure is essential for practical rational design of catalysts. Yet recent studies reveal that under reactive conditions, catalysts can undergo substantial restructuring -- often dynamic, geometrically and compositionally complex -- that escapes the reach of the conventional static view of catalysis. These phenomena often involve an ensemble of transient, irregular, and heterogeneous surface structures and require sampling a broader set of accessible configurations. Still, even this expanded view can fall short when the system is far from the thermodynamic equilibrium picture. This perspective reviews the growing body of experimental and computational evidence for such non-equilibrium restructuring phenomena. We categorize the scenarios where the system can be stranded, kinetically trapped in metastable regimes and never thermalize, as well as their physical origins. The restructuring arises not only from the intrinsic dynamics of the catalyst but also from the complex interplay with interfacial species, spawning a zoo of restructuring pathways of various chemical nature, sizes, and time scales. To meet the growing complexity, we outline promising directions in computational chemistry, machine learning, and integration with experiments. We call for a shift in perspective: to embrace complexity as a defining feature of catalysis, to not shy away from its inherent messiness, and to revisit deactivated or failed catalysts not as dead ends, but as rich, underexplored gold mines of mechanistic insight.
Zhang et al. (2025) studied this question.