Advanced Coatings (Belgium)

The growing demand for natural and new food products has prompted researchers to look into alternative food innovations that provide nutritional value while addressing challenges such as food wastage and poverty alleviation. In response, the development and assessment of chocolate-coated banana candies was the main focus of this study. In particular, it looked at the product's appearance, taste, texture, and aroma as well as general customer acceptability, differences in sensory qualities among treatments, and shelf life under both ambient and cold storage conditions. The formulation with the best performance was subjected to microbiological and proximate tests. The study employed an experimental-developmental method using a completely randomized design. Evaluation samples were appropriately coded, and the experiment was replicated three times with ten semi-trained panelists and 100 customer respondents. Sensory evaluation was carried out using a 9-point Hedonic Scale, and the gathered data were analyzed using descriptive statistics and analysis of variance. Treatment C, with Green Cavendish bananas, had the highest mean scores across all sensory measures, followed by Treatment B (Red Banana) and Treatment A (Española). There were no apparent differences among treatments in terms of texture, taste, aroma, or appearance, according to statistical analysis. However, Treatment A (Española) was chosen for microbiological and proximate testing as well as shelf-life evaluation due to its overall consumer acceptance. The results validated its potential nutritional benefit and appropriateness for prolonged use. The product remained stable for up to 20 days at room temperature and 30 days under chilled conditions, according to a shelf-life examination. The Aerobic Plate Count was analyzed at 60 cfu/g, showing good microbiological quality and storage stability, and no Fecal Coliforms, Salmonella, Escherichia coli, yeast, or mold were found, further supporting product safety.

High-haze materials have been widely used in optical and photonic applications, yet most exhibit fixed internal structures with static haze levels, limiting their adaptability across different usage scenarios. Here, we report a facile strategy to fabricate strain-dependent tunable high-haze films by blending silica nanofibers (NFs) into a polystyrene-block-poly(ethylene-ran-butylene)-block-polystyrene (SEBS) matrix. The composite SEBS/NF films exhibit increasing light scattering and haze with the NF content, reaching up to 94%. Upon mechanical stretching, the absence of chemical bonding between the rigid NFs and flexible SEBS allows interfacial sliding, generating nanoscale cavities. These cavities significantly amplify the internal refractive index mismatch, leading to a reversible and strain-dependent enhancement of haze. In addition, the inherent hydrophobicity of SEBS promises environmental stability and anti-fouling properties, enabling robust performance in underwater conditions. Together, these features position the SEBS/NF films as promising candidates for multi-functional optical systems, including underwater light diffusers and adaptive light management devices.

The Strong Lottery Ticket Hypothesis (SLTH) posits that large, randomly initialized neural networks contain sparse subnetworks capable of approximating a target function at initialization without training, suggesting that pruning alone is sufficient. Pruning methods are typically classified as unstructured, where individual weights can be removed from the network, and structured, where parameters are removed according to specific patterns, as in neuron pruning. Existing theoretical results supporting the SLTH rely almost exclusively on unstructured pruning, showing that logarithmic overparameterization suffices to approximate simple target networks. In contrast, neuron pruning has received limited theoretical attention. In this work, we consider the problem of approximating a single bias-free ReLU neuron using a randomly initialized bias-free two-layer ReLU network, thereby isolating the intrinsic limitations of neuron pruning. We show that neuron pruning requires a starting network with $Ω(d/\varepsilon)$ hidden neurons to $\varepsilon$-approximate a target ReLU neuron. In contrast, weight pruning achieves $\varepsilon$-approximation with only $O(d\log(1/\varepsilon))$ neurons, establishing an exponential separation between the two pruning paradigms.

ABSTRACT Perovskite quantum dots (PQDs) offer exceptional optoelectronic properties but suffer from poor stability, limiting their practical use. Silica (SiO 2 ) encapsulation can improve the thermal and photostability of CsPbBr 3 PQDs, yet conventional methods relying on slow tetraethyl orthosilicate (TEOS) condensation under ambient humidity yield low‐density shells that permit moisture penetration and rapid degradation under harsh conditions. Here, we present a ligand‐assisted reprecipitation (LARP) strategy in which decylphosphonic acid (DPA) replaces oleic acid (OA) as the surface ligand to enable in situ SiO 2 encapsulation. The intrinsic acidity of DPA self‐catalyzes TEOS hydrolysis, driving the formation of a cross‐linked Si‐O‐Si network and producing dense, uniform SiO 2 shells directly on the PQD surface. The resulting DPA‐CsPbBr 3 QDs@SiO 2 retain 91.8% of their initial photoluminescence intensity after 18 min of ultrasonic treatment in water, far exceeding the 12.7% retention of OA‐CsPbBr 3 QDs@SiO 2 . They also exhibit excellent photostability and X‐ray stability. Embedding these PQDs in hydroxyl‐terminated polysiloxane enables the fabrication of flexible scintillator films with high stability and spatial resolution for X‐ray imaging. This simple, low‐cost, and scalable approach offers a versatile route to robust PQDs for advanced optoelectronic applications.

The growing demand for distributed sustainable energy solutions has driven innovations in atmospheric moisture and droplet-enabled electricity generation. This study introduces a dual-mode moisture-droplet energy-harvesting coating (MDEC) that integrates a moisture electricity generator (MEG) and a triboelectric droplet electricity generator (DEG) into a single scalable coating system. By employing hybrid MXene-bridged graphene oxide (GO) microspheres as the hybrid ink electrode and a fluorocarbon resin dielectric layer, the developed MDEC overcomes the limitations of traditional metal-based electrodes that cannot be scaled for manufacturing and the inefficiency of single-energy harvesting schemes in moisture environments. The MEG component achieves a voltage output of 0.85V at 25% relative humidity through ion concentration gradient diffusion, whereas the DEG component has a peak power density of 36W m-2 with a short-circuit current of 301µA and an open-circuit voltage of 36.5V. Modular integration of 360 units enables linear voltage scaling up to 301V, successfully powering commercial LEDs and charging capacitor devices. This design offers a promising pathway for scalable low-power electronics and Internet of Things (IoT) applications.