Nucleation and Growth of Nano-Islands during Surface Reactions or Alloying with Increased Lattice Constant

Several processes lead to a self-organization with a regular structure on a surface. Many systems are well understood and even applied in industry to create samples with unique material, optical, and electronic properties. However, the behavior of some systems is still surprising and the underlying atomic processes are still a mystery. The repetitive formation and lifting of chemical reactions, during oxidation, nitridization, or sulfidization, as well as surface- and binary-alloy formation, and the exchange process in electrochemical atomic layer deposition, leads to ordered nano-islands growth, although the reason is unknown. Here we show that only two ingredients are required leading to such a behavior. Firstly, the surface reaction/alloying exhibits a larger lattice constant than the original, clean surface, resulting in surface stress and atoms that are pushed out on top of the terrace. Secondly, upon restoration/reduction, these expelled atoms have problems finding back their original positions resulting in a flux of adatoms and vacancies per cycle. The peculiar “nucleation & growth” in these systems differs significantly from standard, well-established models and theories. A precursor phase nucleates and grows in the early stages of the reaction to build up the critical surface stress leading to the expelled adatoms. The system is structurally fully reversible upon restoration before this critical stress is reached. In the irreversible nucleation stage adatoms are created in between the precursor structure leading to the self-organization. Using the oxidation-reduction cycles on Pt(111) as an example, we explain all peculiar nucleation & growth aspects. The precursors are the so-called “place-exchange” atoms that form rows or spokes on the surface. The combination of simultaneous adatom and vacancy growth nicely describes the surface evolution: applying our new model to the experimental data fits the entire evolution over 170 cycles with only three fit parameters. Finally, we present an overview of other systems, all showing similar behavior, indicating the generality of the above described process.