• Surface nanocrystallization is an effective method to improve the high-temperature service performance of Super304H austenitic stainless steel. • Nanocrystalline materials lack thermal stability due to their excess energy associated with the high volume fraction of grain boundaries (GBs). • Near-surface ultrafine nanocrystals exhibit exceptional thermal stability since the GB mobilities are inhibited by highly stable local dislocation networks. • Formation of stable local dislocation networks is related to nano grain size, which occurs only when the grains be refined below a critical value. Nanocrystallization is an effective strategy for enhancing the service performance of high-temperature materials. However, the grain coarsening tendency of nanocrystals primarily constrains their applications at elevated temperatures. This study investigates the thermal stability, gradient microstructure characteristics and plastic deformation behaviour of nanocrystals produced by surface shot peening in an austenitic stainless steel. The shot-peened microstructures are studied experimentally through TEM observations, and numerically using atomistic computations. During shot peening, stored energy increases with grain refinement until it becomes critical ( d = d R ), triggering the recrystallisation of the microstructure. Near-surface ultrafine nanocrystals are much smaller than the critical recrystallisation grain size ( d R ), and exhibit exceptional thermal stability due to their boundary mobilities being inhibited by local dislocation networks. However, this stability occurs only when the grain size is much smaller than d R and below a critical value ( d S ). At this stage, the local dislocation networks are highly stable and no longer dominated by recrystallization processes that enable grain boundaries mobilities to increase and grains to coarsen. Therefore, an optimal balance between grain refinement strengthening and thermal stability can be achieved in regions where the mean grain size remains either above the recrystallization limit ( d > d R ), or below the dislocation network-dominated domain ( d < d S ).
Zhou et al. (Wed,) studied this question.