In this study, we propose a non-destructive method combining optical microscopy images and convolutional neural networks (CNNs) to estimate component mixing ratios in pharmaceutical powder blends. Two CNN architectures, VGG16 and Xception, were trained as regression models on binary mixtures (CEOLUS PH-101 and PEARLITOL 200SD) and ternary mixtures (Ibuprofen, Pharmatose 100M, and CEOLUS UF-702). The Xception model achieved high predictive accuracy, with a coefficient of determination of 0.821 for the binary mixture and 0.935 for the ternary mixture, along with a low Root Mean Square Error of Prediction (RMSEP) and a low Mean Absolute Error (MAE). Feature representations from the intermediate CNN layers were visualized using UMAP (uniform manifold approximation and projection), revealing distinct clustering according to mixing ratios, indicating that the model effectively captured subtle morphological differences in particle structure. This approach offers a promising tool for real-time, non-destructive quality control in pharmaceutical manufacturing, potentially reducing reliance on sampling-based or chemical analytical methods.

This study proposes a simple gas-phase coating technique for synthesizing core–shell particles with metallic cores via a one-step aerosol process based on the control of gas–liquid–solid interfaces. By adjusting the precursor concentration and hydrogen gas flow rate, the particle morphology, size, and crystal structure could be precisely tuned. During synthesis, the balance among the formation and reduction behaviors of Fe, FeO, and SiO₂ phases was optimized to achieve the formation of a stable silica shell. As the precursor concentration increased, the core size increased while the SiO₂ shell became thinner, in good agreement with the predicted trend. Powder X-ray diffraction analysis revealed that a sufficient hydrogen supply is essential to suppress the formation of oxide impurities and to achieve complete reduction to metallic iron. The obtained core–shell particles exhibited a uniform spherical morphology and well-defined core–shell structures, demonstrating the effectiveness of interface control in gas-phase synthesis.

In this study, the feasibility of a CO₂ separation process based on ion-induced nucleation was investigated. Ion-induced nucleation is a heterogeneous nucleation process in which gas-phase molecules aggregate around an ion, allowing cluster formation under milder supersaturation conditions compared to homogeneous nucleation. First, the adsorption–desorption dynamics of vapor molecules on an ion surface were modeled using queueing theory, and the theoretical predictions were compared with molecular dynamics (MD) simulation results. The arrival-interval and binding-time distributions obtained from MD simulations showed good agreement with exponential distributions under low vapor pressure conditions. Next, ion-induced nucleation of CO₂ was investigated using MD simulations of CO₂–N₂ gas mixtures containing a single ammonium ion, representing the dilute ion limit. Under conditions where homogeneous nucleation was suppressed, the formation and growth of CO₂ nanoclusters around the ion were clearly observed. These results demonstrate that ion-induced nucleation can effectively lower the nucleation barrier and promote particle formation. The findings of this study suggest the potential of combining ion-induced nucleation with sonic nozzle separation to realize a low-energy and efficient CO₂ separation process.

This study aimed to clarify the effects of nanoplastics on microorganisms in aquatic environments and establish guidelines for controlling their toxicity. Positively charged polystyrene nanoparticles were used as model nanoplastics, and Escherichia coli at different growth stages (exponential and stationary phases) were employed as model microorganisms. Particle exposure experiments were conducted under various salt concentrations. Bacterial size, particle adhesion, cell membrane integrity, and colony-forming ability were evaluated using flow cytometry (FCM) and confocal laser scanning microscopy (CLSM). The results showed that under low-salinity conditions, nanoparticle adhesion increased, leading to significant cell membrane damage and reduced colony-forming ability. In contrast, under high-salinity conditions, particle adhesion and cytotoxicity were suppressed due to electrostatic screening effects. Cytotoxic effects caused by particle adhesion were more pronounced in exponentially growing cells than in stationary-phase cells. These findings demonstrate that not only exposure conditions but also the microbial growth state strongly influence the toxicity of nanoplastics toward microorganisms.

Porous Al particles with etching pits on their surfaces were produced by anodic etching of Al particles placed inside a rotating barrel. In this method, electrical current flows through the Al particles as they contact a Pt plate at the bottom of the barrel, causing anodic etching of their surfaces. The structure of the etching pits formed on the Al particle surfaces could be controlled by varying the current and electrolysis time. Furthermore, using an electrolyte containing a surfactant prevented particle flotation and enabled the formation of etching pits on Al surfaces with particle sizes smaller than 5 μm. Porous Mg particles can also be produced by performing barrel anodic etching using Mg particles as the starting material. The porous particles obtained using this method are expected to have various applications, such as in sensors, catalyst supports, and batteries.