Kinetics of vacancy-assisted reversible phase transition in monolayer MoTe2

Phase transitions between the 2H and $${1{{{\rm{T}}}}}^{{\prime} }$$ phases in monolayer MoTe2 show promise for many semiconductor applications, yet the atomistic pathways for nucleation and growth remain unclear. Here, we investigate these kinetic mechanisms using atomistic simulations based on a machine learning interatomic potential trained on SCAN-DFT data, combined with mean field kinetic theory. The transition is found to involve both diffusive and diffusionless mechanisms. Nucleation of $${1{{{\rm{T}}}}}^{{\prime} }$$ is initiated by the coalescence of neighboring Te monovacancies into mobile divacancies, which interact with other Te vacancies to form small triangular $${1{{{\rm{T}}}}}^{{\prime} }$$ islands. Islands then grow by consuming pre-existing vacancies, when depleted, by absorbing migrating divacancies; beyond a critical size they can advance without vacancies but with a higher barrier. Removing the stimulus triggers a rapid, diffusionless $${1{{{\rm{T}}}}}^{{\prime} }$$ to 2H transition that reorganizes Te vacancies into three spoke-like lines at the island center. Subsequent cyclic and reversible transitions between $${1{{{\rm{T}}}}}^{{\prime} }$$ and 2H are diffusionless, requiring only mild stimuli and no additional vacancies. Phase transitions between the 2H and 1T’ phases in monolayer MoTe2 show promise for semiconductor applications, yet the atomistic pathways for nucleation and growth remain unclear. This paper investigates these kinetic mechanisms using atomistic simulations based on a machine learning interatomic potential trained on SCAN-DFT data.