Researchers have developed a new photonic neural network architecture that significantly improves task accuracy by leveraging the physical properties of light. Unlike traditional designs that mimic digital neural networks, this system uses physical transformations and multisynaptic optical paths to process information directly, avoiding errors from digital modeling and simulation. Tested on benchmark datasets including MNIST, Fashion-MNIST, and CIFAR-10, the network achieved classification accuracies of 99.79, 98.26, and 90.29 percent, respectively—outperforming both digital counterparts and existing hardware systems. The findings mark a major step toward more efficient and accurate AI hardware.

Controlling regioselectivity remains one of the major challenges in hydrocyanation. When an unsymmetrical alkyne undergoes this reaction, it typically yields a mixture of regioisomers. This complicates product separation, making it more difficult and costly. Now, scientists from the Institute of Physical Chemistry of the Polish Academy of Sciences have demonstrated a novel approach that addresses this problem. Their breakthrough enables full control over the reaction outcome — making hydrocyanation greener, more efficient, and more economical than ever before. Here is how it works.

Researchers at Southern University of Science and Technology developed an adaptive beam shaping method for laser micro-grooving to shape tiny grooves with sub-micron accuracy—even in hard-to-machine materials like silicon carbide.

By combining smart simulations and real-time adjustments, their system “teaches” lasers to self-compensation deviation between experimental and target results caused by diffraction and polarization, achieving 5× higher precision than traditional patterned laser ablation methods.

“Here, you can think of the laser as a shaped knife, and you can achieve the desired groove shape with a single stroke”, says Prof. Shaolin Xu.

To realize a sustainable low-carbon society, it is essential to establish a catalytic process that converts various concentrations of CO₂ in combustion exhaust gases from thermal power plants and other sources into useful chemicals using renewable hydrogen. However, due to the high oxygen (O₂) content (about 10%) in such exhaust gases, conventional catalytic methods face a major challenge in that H₂ reacts preferentially with O₂, making efficient CO₂ conversion technically impossible. A research team led by Hokkaido University has developed a tandem system that continuously captures and converts CO₂ in a wide concentration range, from atmospheric levels to exhaust gases. Their work is published in the journal Industrial Chemistry & Materials on June 13, 2025.

Researchers will study how ocean currents and river nutrients affect deep coral ecosystems on the West Florida Shelf – one of the Gulf’s largest and least-studied habitats. Funded by the Florida RESTORE Act Centers of Excellence Program, the project aims to support sustainable fisheries and conservation of these vital, little-explored habitats, which are home to economically important marine life. The research will guide science-based strategies for protecting the gulf coast’s long-term ecological and economic health.