Deterministic and efficient switching of sliding ferroelectrics

Recent studies highlight the scientific importance and broad application prospects of two-dimensional sliding ferroelectrics, which prevalently exhibit vertical polarization with suitable stackings. It is crucial to understand the mechanisms of sliding ferroelectricity and to deterministically and efficiently switch the polarization with optimized electric fields. Here, applying our dream-Allegro multitask equivariant neural network, which simultaneously predicts structure-dependent interatomic potentials and Born effective charges, we construct a comprehensive model for the boron nitride $(\mathrm{BN})$ bilayer. The molecular dynamics simulations reveal a remarkably high Curie temperature of up to 1400 K, facilitated by robust intralayer chemical bonds and delicate interlayer van der Waals interactions. More importantly, it is found that, compared to the out-of-plane electric field, the inclined field not only leads to deterministic switching of electric polarization, but also largely lowers the critical strength of the field, due to the presence of the in-plane polarization in the transition state. This strategy of an inclined field is demonstrated to be universal for other sliding ferroelectric systems with monolayer structures belonging to the symmetry group $p\overline{6}m2$, such as transition metal dichalcogenides.