Quanhong Ou

Lower-background photoelectrochemical (PEC) bioanalysis is crucial for high-sensitivity biomolecule detection and early clinical disease screening by effectively eliminating matrix interference and reducing the background photocurrent noise. In this work, a novel, simple, label-free split-type PEC aptasensor was successfully constructed for interleukin-6 (IL-6) analysis based on electrospun Bi2O3 nanofibers (NFs) coupled with in situ grown iodine vacancy (IV) BiOI nanosheets (IV-BiOI/Bi2O3 NFs). The introduction of heterojunction interfaces and IV in BiOI/Bi2O3 NFs can synergistically regulate light absorption capacity, optimize electronic band structure, efficiently promote the separation and migration of photogenerated electron–hole pairs, and inhibit charge recombination, thus greatly enhancing PEC photocurrent response. Aptamer-capped mesoporous silica nanospheres encapsulate ascorbic acid to specifically recognize IL-6 (10.0 ng/mL), triggering large amounts of AA release. This boost-on photocurrent response yields a 30.9-fold enhancement versus the control group. By separating biorecognition reactions from photocurrent readout, the split-type strategy avoids electrode surface biomodification interference and achieves an ultralow background. The sensor presents a wide linear range, low detection limit (0.6 pg/mL), favorable selectivity, and stability and achieves reliable detection in real serum and sweat samples. This work provides a new paradigm for defective heterojunction-based PEC biosensing with label-free operation, simple fabrication, and a high signal-to-noise ratio.

Lipstick is a common forensic trace, yet its rapid and non‐destructive preliminary classification remains challenging due to complex formulations, spectral redundancy, and limited model interpretability. Although FTIR spectroscopy has demonstrated potential for discrimination, current studies still inadequately address robust independent sample evaluation, variable reduction, and interpretable analysis. To address these methodological limitations, this study established an FTIR spectroscopy based machine learning workflow for the preliminary classification of 36 lipstick samples in a forensic screening context. Multiple spectral preprocessing strategies, classification models, and feature selection methods were systematically evaluated within a grouped nested cross validation framework to reduce optimistic bias and better reflect practical generalization. In the full spectrum model, Linear Discriminant Analysis (LDA) based on the original spectrum achieved the best overall classification performance, with an accuracy of 93.75%. To further improve the usability of the model, various feature selection methods were compared, and ReliefF‐SPA‐LDA was identified as the primary workflow. This method preserves over 98% of the class discriminative information while reducing the dimensionality of spectral data by 95.5%. SHAP analysis further revealed the key discriminative wavenumbers at 1247 cm⁻¹ and 1760 cm⁻¹, enhancing the interpretability of the classification results. This study, by establishing a compact and interpretable FTIR framework, helps bridge the gap between spectral classification and practically applicable forensic screening, enabling rapid preliminary screening and source narrowing of lipstick traces.

Traditional processing significantly alters the phytochemical composition and therapeutic properties of Polygonatum sibiricum (P. sibiricum), making reliable authentication of differently processed products essential for quality control. However, existing spectroscopic-chemometric approaches often face three key limitations: loss of chemically meaningful spectral information during dimensionality reduction, insufficient integration between feature-selection strategies and classification models, and limited interpretability of model outputs. In this study, a robust and interpretable analytical framework integrating Fourier transform infrared (FTIR) spectroscopy with machine learning was developed to classify four processed P. sibiricum products, including raw product (RPP), nine-steamed-and-nine-sun-dried product (NSND), honey-processed product (HPP), and wine-processed product (WPP). A systematic comparison of three complementary feature-extraction strategies, principal component analysis (PCA), simulated annealing (SA), and successive projections algorithm (SPA) was conducted to balance spectral compression and chemical interpretability. To improve the reliability and generalization of the model, all data processing and validation procedures were performed at the physical-sample level, where replicate spectra were treated as a single unit. The results demonstrate that classification performance is closely associated with spectral regions related to O-H stretching, C-H stretching, and glycosidic C-O/C-O-C vibrations, reflecting processing-induced variations in polysaccharide-related structures. Overall, this study establishes a physically consistent and chemically interpretable FTIR–machine learning framework, providing a more reliable strategy for authentication and quality evaluation of processed herbal materials.

Antimony sulfide (Sb₂S₃) faces challenges in optoelectronics due to sulfur vacancy-induced recombination and crystallization heterogeneity. A two-step defect engineering strategy combining hydrothermal growth and postselenization treatment has been developed, enabling the fabrication of single-crystalline antimony sulfoselenide (Sb₂(S,Se)₃) microrods with suppressed defects and enhanced optoelectronic properties. Theoretical calculations confirm selenium passivation of sulfur vacancies is energetically favorable (-0.66 eV vs vacancy formation energy of 0.75 eV), promoting selective Se substitution at defect sites. The selenization temperature (300 ∼ 400 °C) has been controlled to achieve tunable Se gradient incorporation, which passivates vacancies while minimizing interfacial disorder. The optimized Sb₂(S,Se)₃ microrod photodetector exhibits a high responsivity of 9.68 A/W at 550 nm, an external quantum efficiency of 2185%, fast response times (4.0 ms rise and 3.8 ms decay), and a detectivity of 7.2 × 10¹⁰ Jones. Systematic characterizations and ab initio molecular dynamics simulations demonstrate that the optoelectronic performance enhancement arises from improved crystallinity, reduced defect density, and optimized carrier dynamics. This work presents a generalizable approach for defect engineering in low-dimensional metal chalcogenides through spatially controlled heteroatom doping, providing insights for the design of high-performance optoelectronic devices.

A low-temperature chemical vapor deposition method was developed to synthesize high-quality, single-crystalline Bi ₂ S ₃ microrods with superior visible photodetection properties.