Recent application of neural networks (NNs) to modeling interatomic interactions has shown the learning machines' encouragingly accurate performance for select elemental and multicomponent systems. In this study, we explore the possibility of building a library of NN-based models by introducing a hierarchical NN training. In such a stratified procedure NNs for multicomponent systems are obtained by sequential training from the bottom up: first unaries, then binaries, and so on. Advantages of constructing NN sets with shared parameters include acceleration of the training process and intact description of the constituent systems. We use an automated generation of diverse structure sets for NN training on density functional theory-level reference energies. In the test case of Cu, Pd, Ag, Cu-Pd, Cu-Ag, Pd-Ag, and Cu-Pd-Ag systems, NNs trained in the traditional and stratified fashions are found to have essentially identical accuracy for defect energies, phonon dispersions, formation energies, etc. The models' robustness is further illustrated via unconstrained evolutionary structure searches in which the NN is used for the local optimization of crystal unit cells.

We present an approach based on two bio-inspired algorithms to accelerate the identification of nanoparticle ground states.

No abstract

Prescreening candidate structures with reliable classical potentials is an effective way to accelerate ab initio ground state searches. Given the growing popularity of machine learning force fields, surprisingly little work has been dedicated to quantifying their advantages over traditional potentials in global structure optimizations. In this study, we have developed a neural network (NN) model and systematically benchmarked it against a commonly used Gupta potential and an embedded atom model in the search for stable Au N clusters (30 ≤ N ≤ 80). An efficient simultaneous optimization of clusters in the full size range was achieved with our recently introduced multitribe evolutionary algorithm. Density functional theory (DFT) evaluations of candidate configurations identified with the three classical models revealed that the NN structures were lower in energy by at least 10 meV/atom for 30 of the 51 sizes. We also demonstrated that DFT evaluation of all NN-relaxed structures during evolutionary searches resulted in finding even more stable configurations, which highlights the need for further improvement of the NN accuracy to avoid excessive DFT calculations. Overall, the global searches produced putative ground states with matching or lower DFT energies compared to all previously reported Au clusters with 30–80 atoms.

We have discovered a stoichiometric LiB with hexagonal boron layers by compressing known LiB0.9 with linear boron chains. The sp to sp 2 rebonding occurred at room temperature and moderate pressures above 21 GPa. The study was motivated by a long-standing prediction that LiB in the stable layered configuration could be a close analog to the MgB2 superconductor. Apparent stacking disorder in LiB and an evidenced stoichiometry shift in LiBy (down to y ≈ 0.75) made the materials' characterization a challenge. Ab initio modeling allowed us to establish LiBy's pressure-dependent composition and predict overlooked related stable structures. The synchrotron powder diffraction data indicates that synthesized LiB remains metastable under ambient pressure.

Two-dimensional (2D) BNSi2 has been recently proposed to be a viable candidate material with a graphene-like structure. We have carried out ab initio evolutionary ground state searches and uncovered a number of 2D structures in the (BN)xSi1−x pseudobinary system with considerably lower energies. Nevertheless, our formation energy analysis shows that none of these or previously proposed polymorphs are stable with respect to phase separation into 2D Si and BN. Examination of the related CSi, BNC2, and BNSi2 2D compounds indicates that the lack of miscibility in the last two systems is due to limited involvement of nitrogen electronic states in the covalent bonding.

We have combined a neural network formalism with metaheuristic structural global search algorithms to systematically screen the Mg–Ca binary system for new (meta)stable alloys.

Magnetic fluctuations induced by geometric frustration of local Ir-spins disturb the formation of long-range magnetic order in the family of pyrochlore iridates. As a consequence, Pr2Ir2O7 lies at a tuning-free antiferromagnetic-to-paramagnetic quantum critical point and exhibits an array of complex phenomena including the Kondo effect, biquadratic band structure, and metallic spin liquid. Using spectroscopic imaging with the scanning tunneling microscope, complemented with machine learning, density functional theory and theoretical modeling, we probe the local electronic states in Pr2Ir2O7 and find an electronic phase separation. Nanoscale regions with a well-defined Kondo resonance are interweaved with a non-magnetic metallic phase with Kondo-destruction. These spatial nanoscale patterns display a fractal geometry with power-law behavior extended over two decades, consistent with being in proximity to a critical point. Our discovery reveals a nanoscale tuning route, viz. using a spatial variation of the electronic potential as a means of adjusting the balance between Kondo entanglement and geometric frustration.