Search

Atomistic simulations hold significant value in clarifying crucial phenomena such as phase transitions and energy transport in materials science. Their success stems from the presence of potential energy functions capable of accurately depicting the relationship between system energy and lattice changes. In magnetic materials, two atomic scale degrees of freedom come into play: the lattice and the spin. However, accurately tracing the simultaneous evolution of both lattice and spin in magnetic materials at an atomic scale is a substantial challenge. This is largely due to the complexity involved in depicting the interaction energy precisely, and its influence on lattice and spin-driving forces, such as atomic force and magnetic torque, which continues to be a daunting task in computational science. Addressing this deficit, we present DeepSPIN, a versatile approach that generates high-precision predictive models of energy, atomic forces, and magnetic torque in magnetic systems. This is achieved by integrating first-principles calculations of magnetic excited states with deep learning techniques via active learning. We thoroughly explore the methodology, accuracy, and scalability of our proposed model in this paper. Our technique adeptly connects first-principles computations and atomic-scale simulations of magnetic materials. This synergy presents opportunities to utilize these calculations in devising and tackling theoretical and practical obstacles concerning magnetic materials.

Atomic simulations hold significant value in clarifying crucial matters such as phase transitions and energy transport in materials science. Their success stems from the presence of potential energy functions capable of accurately depicting the relationship between system energy and lattice changes. In magnetic materials, two atomic scale degrees of freedom come into play: the lattice and the magnetic moment. Nonetheless, precisely portraying the interaction energy and its impact on lattice and spin-driving forces, such as atomic force and magnetic torque, remains a formidable task in the computational domain. Consequently, there is no atomic-scale approach capable of elucidating the evolution of lattice and spin at the same time in magnetic materials. Addressing this knowledge deficit, we present DeepSPIN, a versatile approach that generates high-precision predictive models of energy, atomic forces, and magnetic torque in magnetic systems. This is achieved by integrating first-principles calculations of magnetic excited states with advanced deep learning techniques via active learning. We thoroughly explore the methodology, accuracy, and scalability of our proposed model in this paper. Our technique adeptly connects first-principles computations and atomic-scale simulations of magnetic materials. This synergy presents opportunities to utilize these calculations in devising and tackling theoretical and practical obstacles concerning magnetic materials., 8 pages, 4 figures

X-ray Laue diffraction is an important method for characterizing the local crystallographic orientation and elastic strain in polycrystalline materials. Existing analysis methods are designed mainly to index a single or a few Laue diffraction pattern(s) recorded in a detector image. In this work, a novel method called dictionary–branch–bound (DBB) is presented to determine the crystallographic orientations of multiple crystals simultaneously illuminated by a parallel X-ray incident beam, using only the spot positions in a detector image. DBB is validated for simulated X-ray Laue diffraction data. In the simulation, up to 100 crystals with random crystallographic orientations are simultaneously illuminated. Fake spots are randomly added to the detector image to test the robustness of DBB. Additionally, spots are randomly removed to test the resilience of DBB against true spots that are undetected due to background noise and/or spot overlap. Poisson noise is also added to test the sensitivity of DBB to less accurate positions of detected spots. In all cases except the most challenging one, a perfect indexing with a mean angular error below 0.08° is obtained. To demonstrate the potential of DBB further, it is applied to synchrotron microdiffraction data. Finally, guidelines for using DBB in experimental data are provided.

Plastic deformation of metals refines the microstructure and increases the strength through work hardening, but this effect of deformation is counterbalanced by dynamic recovery. After deformation to large strains, the microstructure typically shows a lamellar morphology, with finely spaced lamellar boundaries connected by triple junctions. Here we report that mechanically assisted triple junction motion is an important mechanism of dynamic recovery, and it replaces two boundaries by one, while maintaining the structural morphology. The observation rationalizes many features of the evolution of microstructure and strength in metals deformed to large strains, including an approach to a steady state.

The migration of a recrystallization boundary in pure aluminum was followed during in situ annealing in a scanning electron microscope. The microstructure was characterized using the electron channeling contrast technique, and a typical stop-go grain boundary motion was observed during annealing. Thermal grooving associated with boundary migration on the inspected free surface was characterized after the in-situ experiment using atomic force microscopy. The results show that new thermal grooves develop at places where the recrystallization boundary segments remain stationary for a relatively long time. The kinetics of thermal grooving are determined. Effects of the surface oxidation layer on the formation of thermal grooving as well as the overall influence of grooves on boundary migration are discussed. © (2013) Trans Tech Publications, Switzerland.