Perovskites are promising materials candidates for optoelectronics, but their commercialization is hindered by toxicity and materials instability. While compositional engineering can mitigate these problems by tuning perovskite properties, the enormous complexity of the perovskite materials space aggravates the search for an optimal optoelectronic material. We conducted compositional space exploration through Monte Carlo (MC) convex hull sampling, which we made tractable with machine learning (ML). The ML model learns from density functional theory calculations of perovskite atomic structures, and can be used for quick predictions of energies, atomic forces, and stresses. We employed it in structural relaxations combined with MC sampling to gain access to low-energy structures and compute the convex hull for CsPb(Br 1−x Cl x ) 3 . The trained ML model achieves an energy prediction accuracy of 0.1 meV per atom. The resulting convex hull exhibits two stable mixing concentrations at 1/6 and 1/3 Cl contents. Our data-driven approach offers a pathway towards studies of more complex perovskites and other alloy materials with quantum mechanical accuracy.

Recently the possibility to use ion beam mixing combined with suitable annealing has been suggested as a possible means to synthesize individual silicon quantum dots in a silica layer, with the possibility to function as single-electron transistors. For this to work, it is necessary to have a careful control of the ion beam mixing in Si/SiO 2 /Si heterostructures, as well as understand the nature of not only the composition, but also the chemical modication of the SiO 2 layer by the mixing with Si. We describe here a procedure to synthesize Si/SiO 2 /Si heterostructures in molecular dynamics, with an energy minimization scheme to create strong and stable interfaces. The created heterostructures are irradiated at energies and uences matching corresponding experiments. The results show a considerable degree of interface mixing, as expected. They also show some densication of the silica layer due to recoil implantation, and formation of a considerable number of coordination defects. Due to the strong covalent bonding in silicon and silica, the densication is not fully elastically relaxed even in the presence of a nearby surface.

We present an update of the DScribe package, a Python library for atomistic descriptors. The update extends DScribe’s descriptor selection with the Valle–Oganov materials fingerprint and provides descriptor derivatives to enable more advanced machine learning tasks, such as force prediction and structure optimization. For all descriptors, numeric derivatives are now available in DScribe. For the many-body tensor representation (MBTR) and the Smooth Overlap of Atomic Positions (SOAP), we have also implemented analytic derivatives. We demonstrate the effectiveness of the descriptor derivatives for machine learning models of Cu clusters and perovskite alloys.