Fan-Out Panel-Level Packaging (FO-PLP) enables high area utilization and manufacturing efficiency, but process-induced warpage caused by the coefficient of thermal expansion (CTE) mismatch and polymer shrinkage remains a major challenge. This study presents a classifier-gated hybrid machine-learning framework for the rapid and accurate FO-PLP warpage prediction using a database generated from a validated three-dimensional finite element process model. A Random Forest classifier first estimates the probability of each global warpage mode, while cluster analysis reduces the spatial training dataset. Two mode-specific artificial neural networks are then combined through probability-weighted fusion to predict the full warpage field and enable warpage prediction for previously unseen geometry layouts. The framework was evaluated on 16 independent finite element designs spanning both warpage modes. Compared with an equivalent single-network model, the proposed approach consistently achieved lower mean and maximum prediction errors across all designs, with the greatest improvements at the panel edges and corners where the prediction is most challenging. In addition, the clustering strategy reduced the training-set size and computational cost. These results demonstrate that integrating warpage-mode classification with mode-specific learning improves both the prediction accuracy and training efficiency, providing a practical tool for the fast warpage assessment of new FO-PLP layout designs.
Rain gardens are widely used for stormwater treatment; however, the performance of alternative sorbent materials under varying rainfall conditions remains insufficiently understood. In this study, the removal of nutrients and heavy metals was evaluated in laboratory-scale rain gardens amended with halloysite of two grain size fractions (1-2 mm and 2-4 mm) under rainfall events lasting 30 and 120 min. Three column systems were tested: a reference column containing dolomite, sand, and gravel (C1), and two halloysite-amended columns (C2 and C3). Synthetic stormwater containing N-NH4+, N-NO3-, P-PO43-, Cu, Zn, and Pb was applied. Halloysite improved nutrient removal in all rainfall scenarios compared with the reference column. Dissolved inorganic nitrogen (DIN) removal reached 84% in column C3, whereas soluble reactive phosphorus (SRP) removal reached 85% in column C2. Complete removal of Cu, Zn, and Pb was observed in all columns, highlighting the dominant role of dolomite in heavy metal retention. These findings demonstrate the potential of halloysite as a nutrient-removing amendment in rain garden.
This study investigated the resistance of recycled aggregate concrete (RAC) to chloride penetration under freeze-thaw cycles using both numerical simulation and experimental validation. A five-phase mesoscale model of RAC was developed. The salt freeze-thaw test was conducted to validate the simulated values and explore the effects of different recycled coarse aggregate (RCA) substitution ratios, as well as varying dosages of fly ash and superfine fly ash, on chloride concentration in RAC. The experimental data showed satisfactory agreement with the simulated values, confirming the model's validity. Additionally, the RCA volume fraction, interfacial transition zone (ITZ) thickness, and adhesive ratio of old mortar were analyzed using simulation. The variation patterns of compressive strength, relative dynamic elastic modulus, and mass loss of RAC also revealed the mechanisms by which RCA substitution ratio and the dosages of fly ash and superfine fly ash act under salt freeze-thaw conditions. This study provides guidance for enhancing the resistance of RAC to chloride penetration and its durability under freeze-thaw conditions.
The complexity and heterogeneity of tumors make monotherapy inadequate for effective tumor elimination, highlighting the urgent need for multifunctional synergistic therapeutic strategies. In this study, a near-infrared (NIR)-responsive multimodal therapeutic nanoplatform (FSINPs) was constructed by simple adsorption of indocyanine green (ICG) and Fe3+ onto silk fibroin nanoparticles. Molecular docking showed that ICG binds stably to silk fibroin mainly via hydrogen bonds. Density functional theory (DFT) calculations predicted that Fe3+ strongly coordinates with the sulfonate groups of ICG and quenches ICG fluorescence via intermolecular charge transfer. Under conditions mimicking the acidic and high-glutathione tumor microenvironment, the ICG-Fe3+ coordination is disrupted, leading to fluorescence recovery of FSINPs. Fe3+ catalyzes the Fenton reaction to generate hydroxyl radicals (·OH), thereby achieving chemodynamic therapy (CDT). Upon 808 nm laser irradiation, ICG acts as a dual photosensitizer capable of both photothermal therapy (PTT) and photodynamic therapy (PDT), generating local hyperthermia with a photothermal conversion efficiency as high as 48.7% and producing singlet oxygen (1O2). The photothermal effect facilitates ·OH production, and CDT enhances photodynamic efficacy. The synergistic action of the three therapeutic modalities results in potent light-activated cytotoxicity toward tumor cells. In vivo experiments demonstrated that FSINPs enable tumor microenvironment-responsive fluorescence imaging for over 48 h and achieve complete tumor eradication in a 4T1 tumor model without obvious systemic toxicity. This study provides a new strategy for constructing activatable imaging and highly synergistic PTT/CDT/PDT-integrated silk fibroin-based nanomedicines and offers computational chemistry references for their rational design and development.
