This review explores strategies for stabilizing the polymorphic phases of ammonium nitrate (AN), a promising chlorine-free oxidizer for green propellant systems. Despite its eco-friendly decomposition products, AN’s widespread use is limited by phase transitions near ambient temperature, which cause volumetric expansion, reduced density, and mechanical instability. To address these issues, researchers have investigated various additives such as potassium salts, metal oxides, organic compounds, polymers, and metal-organic frameworks (MOFs and ZIFs). These materials inhibit AN’s undesirable transitions and improve thermal performance. Recent innovations include hybrid systems and nanostructured additives that combine stability with enhanced energetic output.
Imatinib (IMA), a commonly used tyrosine kinase inhibitor, has become an increasing concern as a pharmaceutical contaminant because of its persistence in aquatic environments and possible mutagenic effects. This study introduces a rationally engineered MgV2O4@Cu2V2O7 (MVO@CVO) Z-scheme heterostructure, synthesized through an easy one-step hydrothermal process, for visible-light-driven photocatalytic degradation of IMA. The optimized MVO@CVO composite degraded over 90 % of IMA within 35 min under visible light, with an apparent rate constant roughly 8 times higher than pristine MVO and 3 times higher than CVO. UV-Vis DRS and Tauc analysis showed a bandgap reduction from 3.25 eV in MVO and 2.25 eV in CVO to 1.85 eV in the composite. Radical quenching experiments identified superoxide anions and surface-bound hydroxyl radicals as the primary reactive species. DFT calculations confirmed a bandgap around 2.0 eV, with a minimal interfacial lattice mismatch below 3.5 %, and strong electronic coupling at the interface, enabled by Cu-O-V linkages that facilitate charge transfer. The reconstructed Z-scheme mechanism enhances spatial charge separation, enabling CB electrons in MVO (-0.005 eV vs NHE) to reduce O-2 to center dot O-2(-) , and VB holes in CVO (+2.90 eV) to oxidize H2O/OH- to center dot OHads. This synergistic charge routing overcomes thermodynamic constraints usually seen in type-II systems. The composite maintained more than 90 % of its activity after five cycles, showing excellent reusability. These results position MVO@CVO as a promising, scalable, and durable photocatalyst for removing pharmaceutical pollutants in water purification.
In this paper, the impact of the inlet flow rate, operating pressure, stack configuration and inlet manifold width on the oxygen distribution among the channels of a fuel cell stack is studied. The standard deviation is used to calculate the severity of oxygen maldistribution. The oxygen flow in the U configuration is more uniformly distributed than in the Z configuration. The most even oxygen distribution is obtained for the UZ configuration. The standard deviation for the U, Z and UZ configurations at an inlet flow rate of 80 SLPM is calculated as 0.023, 0.038 and 0.009. The oxygen distribution for the U configuration is independent of the inlet flow rate. The standard deviation for the Z and UZ configurations reduces to 0.030 and 0.007 at an inlet flow rate of 40 SLPM. In addition, the operating pressure does not affect the oxygen distribution in three configurations. The standard deviation for the Z and UZ configurations increases from 0.38 and 0.009 to 0.4 and 0.012 as the inlet manifold width decreased from 15 mm to 12.2 mm. In contrast, the flow distribution in the U configuration stack improves and the standard deviation decreases from 0.023 to 0.019.
Achieving an efficient catalyst in the ATRP system with a simple preparation method and high recyclability is an important challenging issue. In this study, a dip-catalyst based on functional graphene oxide-coated bacterial cellulose as a green support was prepared. Graphene oxide (GO) nanosheets were first conjugated with ethylated-branched-polyethyleneimine (E-bPEI), showing as an effective ATRP ligand. Then, the ligand-modified GO was coated on the surface of bacterial cellulose (BC/GO-E-bPEI). In the next step, copper (I) was incorporated into the BC/GO-E-bPEI through strong chelation by E-bPEI (Cu@BC/GO-E-bPEI). The synthesized dip-catalyst, in the form of the strips, was used for the first time in the ATRP reaction of methyl methacrylate to investigate catalytic activity. The prepared dip catalysts showed high catalytic activity, reasonable control over molecular weight, and narrow molecular weight distribution. Importantly, the reaction was simply turned on/off at any time by insertion/removal of the strips. The recyclability of the catalyst was studied for 7 runs. Also, the amount of residual copper in the polymer was very low (1.5 ppm).
Object detection has achieved remarkable accuracy through deep learning, yet these improvements often come with increased computational cost, limiting deployment on resource-constrained devices. Knowledge Distillation (KD) provides an effective solution by enabling compact student models to learn from larger teacher models. However, adapting KD to object detection poses unique challenges due to its dual objectives—classification and localization—as well as foreground-background imbalance and multi-scale feature representation. While existing surveys have organized KD methods by task type or distillation signal, they lack architectural specificity and fail to address how knowledge flows through the modular internals of modern detectors. To address this critical gap and offer a unique perspective compared to existing literature, this review introduces a novel architecture-centric taxonomy for KD methods. We distinguish between CNN-based detectors (covering backbone-level, neck-level, head-level, and RPN/RoI-level distillation) and Transformer-based detectors (including query-level, feature-level, and logit-level distillation). We further evaluate representative methods using the MS COCO and PASCAL VOC datasets with mA@P0.5 as a performance metric, providing a comparative analysis of their effectiveness. The proposed taxonomy and analysis aim to clarify the evolving landscape of KD in object detection, highlight current component-level challenges, and guide strategic future research toward efficient and scalable detection systems.
