Researchers at the Institute of Industrial Science, The University of Tokyo have developed MatterGen-e^-, a machine learning model for the inverse design of inorganic crystals. The model generates crystal structures with electronic-structure descriptors to match a target property, producing more valid, stable, unique, and novel structures than when the crystal structure is considered alone. Control experiments showed that only physically meaningful electronic descriptors improve stability and validity, whereas diversity increased regardless of the descriptors used.

This work develops an asymmetric polarization strategy for covalent triazine frameworks (CTFs) by introducing naphthalene units to strengthen the internal electric field and promote charge separation. The resulting CTF-P1 achieves an H₂O₂ production rate of 2762.0 μmol g⁻¹ h⁻¹ in pure water without sacrificial agents. Mechanistic studies reveal that enhanced polarization facilitates both oxygen reduction and water oxidation, enabling efficient photocatalytic H₂O₂ synthesis.

Through atomic dragging, HEAH (high-entropy alloy hydride) imparts high-distortion characteristics to the MgH₂ surface, weakening interfacial Mg–H bonds and enhancing de-/hydrogenation kinetics. Meanwhile, dehydrogenated HEA enables spontaneous H₂ dissociation without energy barrier, leading to rapid hydrogenation at 200 °C.

 An international collaborative research group led by the National Institute for Fusion Science (NIFS) of the National Institutes for Natural Sciences, the University of Electro-Communications (UEC), Clemson University, and Grenoble Alpes University has demonstrated a new method for evaluating nuclear charge radii using atomic spectroscopy with the Tokyo-EBIT experimental facility at UEC. The research group performed high-precision extreme ultraviolet spectroscopy of “highly charged ions (HCIs)”— atoms that have been ionized many times. By comparing the precise atomic energy levels obtained from the experiments with the latest atomic structure calculations, the nuclear size effect on the atomic energy-levels could be successfully evaluated, leading to the determination of the nuclear charge-radii.

 In the present study, the research targets were the rare-earth elements lutetium (Lu) and ytterbium (Yb). These are neighboring elements on the periodic table. The research group used a new method to evaluate the difference in the sizes of these atomic nuclei. As a result, it was found that the anomalous trend—suggested by conventional recommended databases—that “the lutetium nucleus is larger than that of its adjacent element” is incorrect, and clarified that the odd-even staggering observed in nuclear radii is restored to its expected form. However, this experiment revealed a degree of odd-even staggering so unexpectedly significant that it cannot be explained by currently widely used theoretical calculations, raising the discrepancy between experiment and theory as a new challenge.

 This research not only demonstrates the potential of HCI spectroscopy for new nuclear structure studies but also raises new challenges for understanding the structure of heavy nuclides.

Nanocoating technology reduces contactinterfacial resistance between electrodes and electrolytes, maximizing efficiency in both hydrogen production and power generation

Fingerprint Authentication Without Storing Fingerprints | POSTECH

A new study reveals how manganese-modified biochar helps beneficial fungi break down agricultural residues and accelerate the formation of valuable humic substances, offering a promising approach to sustainable organic waste recycling and soil improvement.

Soils are among the planet's largest carbon reservoirs, yet salinity and land degradation threaten their ability to store carbon and support agriculture. Improving carbon storage in salt-affected farmland could therefore offer a dual benefit: restoring soil health while supporting climate change mitigation.

A team at the University of Pittsburgh has found a way to make computer predictions about hard-to-model metals more accurate. In tests, their approach better predicted how pairs of metal atoms behave, including the notoriously difficult case of two chromium atoms. The improved calculations could give scientists more reliable data as they work to design new materials and train machine-learning tools.