Search for: All records

Abstract Automated, reliable, and objective microstructure inference from micrographs is essential for a comprehensive understanding of process-microstructure-property relations and tailored materials development. However, such inference, with the increasing complexity of microstructures, requires advanced segmentation methodologies. While deep learning offers new opportunities, an intuition about the required data quality/quantity and a methodological guideline for microstructure quantification is still missing. This, along with deep learning’s seemingly intransparent decision-making process, hampers its breakthrough in this field. We apply a multidisciplinary deep learning approach, devoting equal attention to specimen preparation and imaging, and train distinct U-Net architectures with 30–50 micrographs of different imaging modalities and electron backscatter diffraction-informed annotations. On the challenging task of lath-bainite segmentation in complex-phase steel, we achieve accuracies of 90% rivaling expert segmentations. Further, we discuss the impact of image context, pre-training with domain-extrinsic data, and data augmentation. Network visualization techniques demonstrate plausible model decisions based on grain boundary morphology. 

The ability to adapt to changing internal and external conditions is a key feature of biological systems. Homeostasis refers to a regulatory process that stabilizes dynamic systems to counteract perturbations. In the nervous system, homeostatic mechanisms control neuronal excitability, neurotransmitter release, neurotransmitter receptors, and neural circuit function. The neuromuscular junction (NMJ) ofDrosophila melanogasterhas provided a wealth of molecular information about how synapses implement homeostatic forms of synaptic plasticity, with a focus on the transsynaptic, homeostatic modulation of neurotransmitter release. This review examines some of the recent findings from theDrosophilaNMJ and highlights questions the field will ponder in coming years.

 

To translate the exceptional properties of colloidal nanoparticles (NPs) to macroscale geometries, assembly techniques must bridge a 10⁶‐fold range of length. Moreover, for successfully attaining a final mechanically robust nanocomposite macroscale material, some of the intrinsic NPs’ properties have to be maintained while minimizing the density of strength‐limiting defects. However, the assembly of nanoscale building blocks into macroscopic dimensions, and their effective macroscale properties, are inherently affected by the precision of the conditions required for assembly and emergent flaws including point defects, dislocations, grain boundaries, and cracks. Herein, a direct‐write self‐assembly technique is used to construct free‐standing, millimeter‐scale columns comprising spherical iron oxide NPs (15 nm diameter) surface functionalized with oleic acid (OA), which self‐assemble into face‐centered cubic (FCC) supercrystals in minutes during the direct‐writing process. The subsequent crosslinking of OA molecules results in nanocomposites with a maximum strength of 110 MPa and elastic modulus up to 58 GPa. These mechanical properties are interpreted according to the flaw size distribution and are as high as newly engineered platelet‐based nanocomposites. The findings indicate a broad potential to create mechanically robust, multifunctional 3D structures by combining additive manufacturing with colloidal assembly.

 

Water electrolysis is the key to a decarbonized energy system, as it enables the conversion and storage of renewably generated intermittent electricity in the form of hydrogen. However, reliability challenges arising from titanium‐based porous transport layers (PTLs) have hitherto restricted the deployment of next‐generation water‐splitting devices. Here, it is shown for the first time how PTLs can be adapted so that their interface remains well protected and resistant to corrosion across ≈4000 h under real electrolysis conditions. It is also demonstrated that the malfunctioning of unprotected PTLs is a result triggered by additional fatal degradation mechanisms over the anodic catalyst layer beyond the impacts expected from iridium oxide stability. Now, superior durability and efficiency in water electrolyzers can be achieved over extended periods of operation with less‐expensive PTLs with proper protection, which can be explained by the detailed reconstruction of the interface between the different elements, materials, layers, and components presented in this work.