Browse Articles

Ion Diffusion‐Induced Multi‐Interface Reconstruction for High‐Resolution Perovskite X‐Ray Flat‐Panel Detectors

ABSTRACT A critical challenge with state‐of‐the‐art perovskite x‐ray flat‐panel detectors (FPDs) is their limited spatial resolution, primarily due to the presence of multiple poorly integrated interfaces. In this study, we report high‐resolution perovskite FPDs that achieve a record modulation transfer function (MTF) among polycrystalline perovskite direct‐conversion x‐ray FPDs of 6.2 line pairs per millimetre (lp mm −1 ) with a large imaging area of 8.5 × 8.5 cm 2 through an interface reconstruction strategy specifically tailored for perovskites. We reveal that Fick's law‐guided ion diffusion across hundreds of microns‐thick perovskites contributes to a reconstructed x‐ray sensing layer with highly integrated interfaces and a gradient energy band alignment. As such, we have realized an ultrasensitive x‐ray detection with a leading sensitivity‐to‐dark current ratio (2.61 × 10 11 µC Gy air −1 A −1 ) and outstanding stability under ambient conditions over 5760 h. The prototype perovskite FPDs exhibit a detective quantum efficiency (76.9%) and enable high‐resolution x‐ray imaging at a low dosage (0.98 µGy air ), substantially lower than previous polycrystalline perovskite FPDs. Our multi‐interface reconstruction strategy successfully addresses long‐standing issues in perovskite FPDs, advancing their progress from laboratory prototypes to commercial applications in digital radiography and industrial inspection.

A comparative evaluation of deep learning architectures, loss functions, and ensemble strategies for underwater fish species classification

Abstract The accurate classification of underwater fish species is important for biodiver-sity monitoring, sustainable fisheries management, and aquaculture. However, underwater fish classification remains challenging due to class imbalance, varying imaging conditions, and high visual similarity among species. This study presents a comprehensive evaluation of deep learning architectures, loss functions, and ensemble strategies for underwater fish species classification. Using the Mark Daniel Lampa Kaggle Fish Dataset comprising 13,304 images across 31 species, duplicate images were removed using a combination of MD5 hashing and perceptual hashing, resulting in a cleaned dataset for model training and evaluation. EfficientNet-B0, EfficientNetV2-S, Swin-Tiny Transformer, and ConvNeXt-Tiny were trained using both class-weighted cross-entropy loss and focal loss. The experimental results indicate that class-weighted cross-entropy loss consistently provided stronger performance for individual models, whereas focal loss produced a marginal improvement when used with the soft-voting ensemble. Among the individual models, EfficientNet-B0 achieved the highest accuracy of 94.81%. Ensemble learning further improved performance, with the four-model soft-voting ensemble achieving a test accuracy of 97.21% and a macro-F1 score of 0.9678 on the cleaned Kaggle dataset. An external dataset of 310 internet-sourced fish images was additionally used to assess model generalization under domain-shift conditions, where the ensemble achieved an accuracy of 76.39% and a macro-F1 score of 0.7335. While performance decreased substantially under domain shift, the results highlight the challenges of generalizing fish-species classification models beyond the training distribution. Fish species classification represents an important component of future end-to-end underwater monitoring systems incorporating fish detection and localization modules. The study provides a benchmark for future research on underwater fish-species classification, ensemble learning, and real-world deployment. The source code is openly available on GitHub and permanently archived on Zenodo (DOI: https://doi.org/10.5281/zenodo.19315919 ).

Laser‐Induced Synthesis and Passivation of Intrinsically Antioxidative Nano‐Copper for Durable Electronics

ABSTRACT The ever‐increasing costs, along with rigorous storage requirements, are the primary obstacles to the sustainable development of pure metallic inks for advanced electronics. The chemically reductive conversion of copper (Cu) from its much cheaper oxides offers a comparable alternative, whereas suffering from the intrinsic oxidation issues that lead to conductivity failures. Here, we report laser‐induced synthesis and passivation of robust nano‐Cu, achieving superior durability under harsh humid‐thermal conditions (190°C and 90°C, 90% RH) in the absence of additional encapsulations. Through selective and controllable metallization, the nascent Cu is equipped with well‐defined dual‐ligand barriers at interface, which are functionalized by formate crystallographic coordination and oleylamine hydrophobization. Such surface modification imparts the as‐formed Cu with limited resistance increase over prolonged humid‐thermal fluctuations. As demonstrations, we create conformal and wearable sensor systems with passivated Cu interconnects that withstand humid‐thermal erosion and provide signal alerts. This laser technology bridges the gap between high‐performance Cu nanomaterials and cost‐effectiveness, empowering endurable electronics for real‐life extremes.

A preliminary evaluation of SPEEK nanofibers as multi-drug carrier delivering diclofenac, paracetamol and alcoholic extract of Strobilanthes ciliata

Selective Neural Responses to Conspecific Vocalizations in Marmoset Area 32

Understanding how primate frontal circuits encode socially informative vocalizations is central to elucidating the neural basis of communication. In common marmosets, functional MRI has identified robust selectivity for conspecific vocalizations in the pregenual anterior cingulate cortex (pgACC; area 32), but the underlying neural dynamics driving this response remain unknown. Here, we investigated this by recording single-neuron activity in area 32 of five awake marmosets (three female and two male) using acute and chronic Neuropixels probes and Utah microelectrode arrays during presentation of a standardized cross-species auditory stimulus set comprising marmoset, macaque, and human vocalizations as well as a suite of nonvocal sounds. Of 1,713 neurons recorded across 17 sessions, 945 (55%) showed significant auditory responsiveness, and 362 (21%) were categorically selective. Selective neurons exhibited their strongest responses to marmoset calls, and population preference indices revealed a robust bias toward conspecific vocalizations. Temporal population analyses showed rapid encoding: decoding accuracy for detection of marmoset calls exceeded chance within ∼50–100 ms, and representational similarity analysis demonstrated that marmoset calls formed a distinct cluster separate from macaque, human, and nonvocal stimuli. These results provide the first direct evidence that ACC area 32 contains a specialized neural representation for conspecific vocal signals. By linking fMRI-defined selectivity to single-neuron population dynamics, our findings identify area 32 as a key node in primate vocal communication networks and illuminate how frontal circuits transform auditory input into socially meaningful categorical information.

Cu‐Pd Dual Single Atoms Promoting Selective CO ₂ Photoreduction to C ₂ Products in Seawater

ABSTRACT The solar‐powered CO 2 conversion via the photocatalysis route offers a sustainable pathway toward carbon neutrality while mitigating energy/environmental pressure. Nevertheless, the selective and efficient conversion of CO 2 via photoreduction to C 2 products remains a formidable challenge. Here, we engineered a dual‐single‐atom photocatalyst by controllably embedding Pd and Cu single atoms into a TiO 2 matrix. The optimized catalyst (Cu 0.5 Pd 0.5 /TiO 2 ) exhibits the outstanding yield (119.2 µmol/g cat ) and selectivity (84.8%) for acetic acid production from CO 2 photoreduction, performed in seawater and in a photothermal‐aided reactor. Various in situ/ex situ characterizations were employed to investigate atomic‐level structure‐performance correlation and reaction mechanism in practical condition. In situ x‐ray photoelectron spectroscopy, in situ atomic force microscopy‐Kelvin probe force microscopy, transient‐state surface photovoltage, and in situ electron paramagnetic resonance (EPR) collectively indicate that loading Pd and Cu single atoms onto TiO 2 apparently accelerates charge kinetics. This modification results in increased photogenerated electrons for CO 2 reduction, facilitating C─C coupling and hydrogenation reactions. Additionally, in situ infrared (IR) spectroscopy and theoretical computations affirm the pivotal function of Pd single atoms for lowering the energy barrier to form the * OCCO intermediate, apparently improving selectivity for acetic acid production. Overall, our work presents an innovative approach to tackle kinetic and thermodynamic challenges for light‐induced CO 2 ‐to‐C 2 conversion.

Acceptance of semi-automated likelihood ratio systems for pattern evidence among chinese forensic and legal professionals

Abstract Traditional methods of pattern evidence evaluation in forensic science have long relied on subjective expert judgment, raising concerns about transparency, validity, and reliability. In response, the international forensic community has increasingly advocated the use of the likelihood ratio (LR) framework to provide a more scientific approach to evidence evaluation. Semi-automated LR systems have emerged as essential tools to operationalize this framework in practice. However, the successful adoption of these systems ultimately depends on practitioners’ willingness to integrate them into their work. This study applied the Technology Acceptance Model to investigate the factors influencing Chinese forensic practitioners’ acceptance of semi-automated LR systems. Data from 236 valid responses were analyzed using structural equation modeling. This study takes perceived usefulness and perceived ease of use as the core variables. The results reveal that practitioners attach more importance to a system’s practical value than to its operability. They prioritize system accuracy and transparency over ease of use. Perceived usefulness exerts an indirect effect on behavioral intention via users’ attitude toward the system, rather than a direct one. In addition, profession exerted a significant moderating effect on three paths in the model.

Stromal Cells Recruitment by Sono‐Piezoelectric/Pyroelectric Peptide Hydrogels Promotes Bone Regeneration After Postsurgical Osteosarcoma Recession: A Real‐World‐Matched Study

ABSTRACT Currently, the elimination of postsurgical osteosarcoma (OS) and defect repair are still separately explored, which is unlike real‐world clinical scenarios. Inadequate bone marrow stromal cells (BMSCs) compromise their repair and antitumor efficiencies. Here, PFSSTKT (PFS)‐functionalized KLD‐12 peptide hydrogels (KLD‐PFS) have been engineered and integrated with 2D BiOIO 3 nanosheets to obtain the injectable sono‐piezoelectric/pyroelectric peptide hydrogels (KLD‐PFS@BiOIO 3 ) featuring a supramolecular peptide nanofiber (SMPNF) structure. Differing from the dominant BaTiO 3 in sono‐piezoelectric dynamic therapy (SPDT) of cancer, BiOIO 3 nanosheets can produce ROS through sono‐piezoelectric and pyroelectric catalytic processes under ultrasound irradiation, thus enabling the combination of SPDT with pyroelectric dynamic therapy against residual OS. More significantly, PFS as a bone marrow homing peptide enables KLD‐PFS to capture and recruit more BMSCs, and the inherent SMPNF structure, direct ultrasound‐induced current stimuli, sono‐piezoelectricity/pyroelectricity‐induced current stimuli and ROS birth expedite BMSCs differentiation and bone regeneration. These multifaceted actions follow the signaling pathways associated with calcium flux and cancer–neuron communication disruptions and metabolic dysfunction rectification. They have been successfully validated to repress residual OS and favor bone regeneration in a clinical scenarios‐matched postsurgical osteosarcoma and bone defect model. This study offers a promising strategy for comprehensive osteosarcoma management.

Strength enhancement and microstructural evolutions of dredged slurry stabilized with waste-modified composite binder

Anisotropic Masked Mycobacterium Potentiates Amplified Antitumor Trained Immunity via Spleen Targeting and Myelopoiesis Conversion

ABSTRACT Cancer progression involves systemic immune suppression and tumor recurrence facilitated by increased splenic protumoral myelopoiesis. To address this clinical challenge, we developed an engineered Trojan Mycobacterium as a trained immunity (TI)‐mediated splenic myelopoiesis converter (t‐SMC). This design provides an intravenously injectable anisotropic TI inducer with an aspect ratio of 3.6 for preferential spleen targeting and cellular uptake via a Mycobacterium “peeling‐off and masking‐up” strategy with human serum albumin incorporating an enzyme‑activatable Toll‑like receptor 7/8 agonist that boosts TI activation. t‐SMC induces antitumoral myelopoiesis with metabolic and epigenetic reprogramming in splenic macrophages. This shift toward antitumoral myelopoiesis markedly enriches the tumor microenvironment with inflammatory myeloid effectors in an E.G7‐OVA model, and the adoptive transfer of splenic myeloid cells confirmed the TI‐driven antitumor effect. Furthermore, comparative analysis revealed that the bacterial scaffold (alb‐Mycobac) initiates myeloid reprogramming, but only the integrated TLR7/8 agonist (t‐SMC) crosses the therapeutic threshold for tumor regression. In the MC38 surgical recurrence model, preoperative t‐SMC combined with standard adjuvant αPD‐L1 plus oxaliplatin achieved complete inhibition of recurrence in 50% of treated mice (3 of 6) with 62.5% long‐term survival, establishing TI priming as a translation‐ready preoperative adjuvant that converts immune‐checkpoint non‐responders into responders.

Protein hydrolysate from olive oil waste as an effective biostimulant to enhance strawberry resilience against water stress

Direct Growth of 2D Bilayers on Au(111) for Low‐Coercive Sliding Ferroelectricity

ABSTRACT Two‐dimensional (2D) semiconductors directly grown on metal substrates can enable pristine and low‐defect interfaces that are difficult to achieve through transfer‐based fabrication. Although monolayer transition metal dichalcogenides (TMDs) have been widely synthesized on metals, controllable bilayer growth remains challenging owing to distinct thermodynamics and kinetics of second‐layer growth on inert first‐layer surfaces—a significant hurdle given their importance for interlayer‐coupled functionalities like sliding ferroelectricity. Low‐symmetry rhenium dichalcogenides (ReX 2 , X = S, Se) are especially attractive because their weak interlayer coupling and multiple thermodynamically stable stacking configurations support low sliding barriers. Here, by optimizing precursor flux and growth kinetics, we achieved controlled chemical vapor deposition (CVD) growth of bilayer ReS 2 with well‐defined parallel and antiparallel stacking directly on Au(111). As‐grown parallel‐stacked bilayers exhibit a large piezoelectric response (effective d 33, eff = 9.84 pm V −1 ) and sliding ferroelectricity with an ultralow coercive voltage of approximately 2 V. Tip‐defined ferroelectric tunnel junction and ferroelectric field‐effect transistor measurements further support the ferroelectric interpretation. Comparative studies of as‐grown, transferred, and exfoliated bilayers reveal that the cleaner, more conformal interface of the as‐grown bilayers is associated with a lower coercive bias. This work establishes a scalable route for stacking‐specific bilayer TMD growth on metal toward low‐coercive sliding ferroelectricity.

Assessment of phase uncertainty in nonlinear nearshore wave records

Abstract Given a time-series record of ocean surface elevation, spectral analysis is often used to determine characteristic statistics of the record, such as wave height and mean period. A simple linear analysis shows that the phases of the Fourier components in a time-series have some effect on the value of the resulting characteristic statistics; two records with identical wave components, but different phases, will yield two different sets of bulk statistics ( $$H_{m0}$$ , $$T_p$$ , etc.). In the linear sense, this variance from different phase realizations is due to “spectral leakage”, or the existence of Fourier components with non-integer number of periods in the record. It is also expected that variance due to nonlinearity should exist in nearshore wave records, as different phase realizations may lead to different low-frequency motions, which may be poorly resolved in a typical time-series record. This phase-driven variance is a type of aleatory uncertainty, and must be quantified statistically. Here, in the context of a high energy wave event in Duck, North Carolina, USA, the behavior of phase uncertainty is examined through numerical simulation. The analysis indicates that nonlinearity does play a role in uncertainty. The primary outcome of this study are empirical relations for the expected value of phase-driven variance, which can be used to understand the baseline precision of spectral statistics derived from time-series records.

Does Stimulus-Preceding Negativity Reflect Predictions in a Somatosensory Roving Paradigm?

The mismatch negativity (MMN) has since decades become one of the most prominent phenomena in cognitive neuroscience, recognized as a brain response indexing automatic change detection. While substantial work has characterized poststimulus mismatch responses (MMRs), comparatively little is known about how prior information is represented before stimulus onset. Thirty-four healthy human volunteers (12 females) underwent a somatosensory roving paradigm with electroencephalography to investigate whether anticipatory brain activity reflects stimulus history accumulation. Poststimulus analyses revealed canonical somatosensory MMRs, corresponding to a modulation of the N140 component (MMN), source localized in the secondary somatosensory cortex, and a P300 MMR. Both were parametrically modulated by train length, with opposing direction for standard and deviant trials, decreasing over the course of the experiment. Prestimulus analyses revealed a stimulus-preceding negativity (SPN) that systematically varied as a function of stimulus history; longer trains of stimulus repetitions were associated with reduced anticipatory negativity. The SPN did not vary throughout the experiment, but its relationship with poststimulus responses evolved, showing an initially negative correlation with the N140 that diminished over time and an opposite pattern for the P300. Source reconstruction localized the SPN to a network involving inferior frontal, frontopolar, and somatosensory cortices, suggesting a possible role in top–down gain modulation of upcoming sensory stimuli. These findings indicate that learning of statistical regularities is not only observable in the poststimulus phase following violation of statistical regularities (i.e., prediction error signals), but also in neural activity preceding sensory inputs, thereby providing new avenues for investigating endogenous predictive processes.

An On‐Skin Visually Monitorable Gel‐Matrix Zinc–Ion Battery for Electrically Stimulated Wound Healing

ABSTRACT Aqueous zinc‐ion batteries (ZIBs) feature intrinsic safety, low cost, and environmental compatibility, making them attractive for untapped cutting‐edge applications, such as the biomedicine area. Particularly, wearable medical power sources require stable electrical output, mechanical flexibility, and high biocompatibility. Herein, a stable and flexible gel‐matrix aqueous ZIB was designed as an on‐skin power source for electrically stimulated wound healing. The elaborately designed graded porous framework hydrogel electrolyte (GPF‐HGE) obtains high ionic conductivity (17.19 mS cm −1 ) and improved mechanical deformation tolerance (46 kPa and 205 J m −2 ) due to the porous framework formed in the solvent exchange process. Its transparency property, combined with an annular ring‐shaped electrode that generates an electric field aligned with the endogenous wound electric field, enables direct application to wound sites and real‐time visual monitoring. The assembled Zn|GPF‐HGE|I 2 @AC full cells show superior long‐term cycling stability accompanied by slight capacity fading, ensuring a steady energy supply for electrical stimulation (ES). Benefiting from its robust electrochemical stability and excellent biocompatibility, the obtained ES device can provide reliable electrical signals and promote wound healing through synergistic antibacterial and anti‐inflammatory effects. This work highlights the potential of aqueous ZIBs as advanced biomedical power sources beyond conventional energy storage applications.

Military terrain-conditioned decoy-aware mask refinement network for open-vocabulary camouflaged object segmentation

Submillimeter‑Scale Untethered Magnetic Actuators Enabling on‑Demand Manipulation and In Situ Viscosity Sensing

ABSTRACT Untethered magnetic actuators offer a promising platform for minimally invasive surgery due to their compact size, mechanical compliance, multifunctionality, and remote controllability. However, the structural simplicity imposed by current construction strategies limits their operational versatility to leverage established surgical paradigms, and systematic studies of locomotion across the variable viscosities of bodily fluids remain scarce. Herein, we introduce an interference‐fit assembly strategy that couples stereolithographic 3D‐printed microstructures with customized magnetic pixels to fabricate submillimeter‐scale untethered magnetic actuators. Theoretical and experimental analyses in model‐fluid environments spanning physiologically relevant viscosity ranges demonstrate that, once above the critical magnetic flux density required for stable motion, the locomotion velocity scales linearly with the driving frequency. The velocity‐frequency slope exhibits a hyperbolic dependence on ambient viscosity, enabling semi‐quantitative viscosity sensing concurrently with motion and functional execution. Furthermore, we showcase multifunctional on‐demand manipulation capabilities: enhanced diffusion and directional transport with qualitative viscosity sensing; in situ quantitative viscosity measurement inspired by a rotational viscometer; mechanical fragmentation via continuous rotation; and a rigid‐soft synergistic magnetic gripper for targeted cargo capture and delivery. Collectively, this strategy establishes a versatile and scalable paradigm for designing multifunctional, untethered, submillimeter‑scale magnetic actuators with prospective potential for biomedical applications.

8-aminoguanine as a therapeutic strategy to reverse hallmarks of urothelial cell aging

Emergence of Behavioral Tinnitus in Gerbils Is Associated with Reduced Spontaneous Rates in Single Auditory Nerve Fibers

Tinnitus is often initiated by damage to the peripheral auditory system, for example, by acoustic overexposure. Animal studies have shown that such noise-induced tinnitus is related to increased spontaneous activity in the dorsal cochlear nucleus as well as further along the central auditory pathway. However, the role of spontaneous activity of the auditory nerve, connecting the peripheral and central auditory systems, in tinnitus emergence remains unknown. In the current study, tinnitus was induced by exposing anesthetized Mongolian gerbils of either sex to a 115 dB SPL narrowband noise. After one day of recovery, animals were behaviorally tested for gap detection deficits using a gap-prepulse inhibition of the acoustic startle reflex (GPIAS) paradigm, indicative of tinnitus. Noise-induced threshold shifts did not differ between animals with and without signs of tinnitus. Interestingly, single auditory nerve fibers recorded from animals with signs of tinnitus had significantly reduced spontaneous rates compared with both noise-exposed animals without signs of tinnitus and sham-exposed animals. Furthermore, spontaneous rate reduction was specific to fibers tuned to frequencies within the frequency bands that showed gap detection deficits. On the other hand, interspike interval variability and bursting behavior increased in fibers from noise-exposed compared with sham-exposed animals but did not differ with gap detection deficits. These findings suggest that tinnitus-related central hyperactivity may be initiated by reduced spontaneous rates of its innervating auditory nerve fibers. This is consistent with current theoretical models explaining the central manifestation of tinnitus and offers a more detailed definition of tinnitus-related deafferentation.

Suppressing Nonradiative Losses via Computation‐Guided Molecular Design of Self‐Assembled Interfacial Layers for High‐Performance Organic Solar Cells

ABSTRACT The rational design of efficient self‐assembled interlayers (SAIs) is pivotal for overcoming the efficiency bottleneck in organic solar cells (OSCs), in which substantial nonradiative energy losses limit further performance improvements. In this study, guided by theoretical calculations, we report a series of new SAI materials ( BT‐nBZCz ; n = 2, 3, and 4) based on a benzocarbazole core functionalized with a benzo[b]thiophene (BT) unit, and systematically investigate their impact on nonradiative losses and OSC performance. Notably, binary OSCs incorporating BT‐4BZCz achieve a significantly higher power conversion efficiency (PCE) of 20.46% compared to 18.08% for BT‐free SAIs. Combined theoretical and experimental analyses reveal that BT functionalization improves energy level alignment, enhances dipole moment, and optimizes molecular packing, crystallization kinetics, and fibrillar network formation relative to BT‐free SAIs. These improvements promote more efficient exciton dissociation, reduce charge recombination, and in particular, suppress nonradiative energy losses (0.284 vs. 0.217 eV) in the corresponding OSCs. Furthermore, BT‐4BZCz exhibits excellent universality, maintaining high performance across various binary photoactive blends. This work presents a viable molecular engineering strategy for SAIs to suppress nonradiative losses and advance the development of high‐performance OSCs.