Exhaled biomarker detection, as a non-invasive and portable analytical method, holds great potential in the field of point-of-care testing. Acetone exhibits higher concentrations in the exhaled breath of diabetic patients compared to healthy individuals. Therefore, the development of highly sensitive, easy-to-operate, and portable acetone detection technology is desirable for the rapid screening of diabetes. A covalent organic framework, Eu-COF, was synthesized, and its fluorescence can be specifically quenched by acetone based on the mechanism of energy competition. A flexible Eu-COF-based fluorescent film, prepared using polyvinyl alcohol (PVA) as the substrate, was found to be suitable for use in a portable 3D printing device for on-site rapid detection of acetone. The linearity of concentration ranged from 0.2–8.0 mg m⁻³, with a detection limit of 0.1 mg m⁻³ (RSD = 8.6%). The accuracy of the device was validated by comparison with the results of GC-MS. Based on the F-test, there were no significant differences between the results of the portable device and GC-MS. The Eu-COF-based fluorescent film also exhibited a comparable response in humid environments. The detection device integrates sample collection and testing in one step, directly providing the concentration of acetone exhaled by the patient. It can serve as a rapid screening tool for point-of care, non-invasive diabetes detection.

This work explores the effect of a Zn_1−xSn_xO_y (ZTO) layer as a potential replacement for CdS in Sb₂(S,Se)₃ devices. Through the use of Afors-het software v2.5, it was determined that the ZTO/Sb₂(S,Se)₃ interface exhibits a lower conduction band offset (CBO) value of 0.34 eV compared to the CdS/Sb₂(S,Se)₃ interface. Lower photo-generated carrier recombination can be obtained at the interface of the ZTO/Sb₂(S,Se)₃ heterojunction. In addition, the valence band offset (VBO) value at the ZTO/Sb₂(S,Se)₃ interface increases to 1.55 eV. The ZTO layer increases the efficiency of the device from 7.56% to 11.45%. To further investigate the beneficial effect of the ZTO layer on the efficiency of the device, this goal has been achieved by five methods: changing the S content of the absorber, changing the thickness of the absorber, changing the carrier concentration of ZTO, using various Sn/(Zn+Sn) ratios in ZTO, and altering the thickness of the ZTO layer. When the S content in Sb₂(S,Se)₃ is around 60% and the carrier concentration is about 10¹⁸ cm⁻³, the efficiency is optimal. The optimal thickness of the Sb₂(S,Se)₃ absorber layer is 260 nm. A ZTO/Sb₂(S,Se)₃ interface with a Sn/(Zn+Sn) ratio of 0.18 exhibits a better CBO value. It is also found that a ZTO thickness of 20 nm is needed for the best efficiency.

Non-halogenated flame retardants are becoming the trend in the development of polymer flame retardant materials due to their high flame retardant efficiency and low generation of toxic smoke gases. Non-halogenated flame retardants achieve flame retardancy by forming a dense char layer and generating non-combustible gases, with the micro-porous structure of the char residue being crucial for studying the flame retardant mechanism. This study focuses on the segmentation of pores in scanning electron microscopy (SEM) images of the combustion char layer of non-halogenated flame retardant materials, which are cropped and labeled to form a unified dataset. We investigate the SEM image pore segmentation using data augmentation and transfer learning, addressing the challenge of limited sample size. We explore the impact of different data augmentation techniques and transfer learning on model performance. Additionally, we compare convolutional neural network (CNN) segmentation algorithms with traditional segmentation methods. Experimental results demonstrate that CNN segmentation algorithms outperform traditional methods in terms of segmentation accuracy. Offline data augmentation enhances model stability compared to online data augmentation, and adopting transfer learning significantly improves model performance metrics. Specifically, when training with VGG backbone weights through transfer learning, the average pixel accuracy and average intersection over union reach 94.49% and 89.88%, respectively.