Photodetectors that can operate over a wide range of temperatures, from cryogenic to elevated temperatures, are crucial for a variety of modern scientific fields, including aerospace, high-energy science, and astro-particle science. In this study, we investigate the temperature-dependent photodetection properties of titanium trisulfide (TiS₃)- in order to develop high-performance photodetectors that can operate across a wide range of temperatures (77 K–543 K). We fabricate a solid-state photodetector using the dielectrophoresis technique, which demonstrates a quick response (response/recovery time ~0.093 s) and high performance over a wide range of temperatures. Specifically, the photodetector exhibits a very high photocurrent (6.95 × 10⁻⁵ A), photoresponsivity (1.624 × 10⁸ A/W), quantum efficiency (3.3 × 10⁸ A/W·nm), and detectivity (4.328 × 10¹⁵ Jones) for a 617 nm wavelength of light with a very weak intensity (~1.0 × 10⁻⁵ W/cm²). The developed photodetector also shows a very high device ON/OFF ratio (~32). Prior to fabrication, the TiS₃ nanoribbons were synthesized using the chemical vapor technique and characterized according to their morphology, structure, stability, and electronic and optoelectronic properties; this was performed using scanning electron microscopy (SEM), transmission electron microscopy (TEM), Raman spectroscopy, X-ray diffraction (XRD), thermogravimetric analysis (TGA), and a UV–Visible–NIR spectrophotometer. We anticipate that this novel solid-state photodetector will have broad applications in modern optoelectronic devices.

This study investigated the localization accuracy of Grad-CAM as an Explainable AI (XAI) technique for industrial inspection, specifically evaluating it on the WM811K semiconductor wafer defect dataset - a large-scale, publicly available collection of 25,519 labeled wafer maps. Localization performance was measured using the Top-10% Intersection over Union (IoU) metric on a held-out test sample ([Formula: see text]), yielding a baseline accuracy of 0.129. A consistent radial attribution bias was observed across both Grad-CAM and XGrad-CAM, with both methods producing structurally similar bias profiles, suggesting the bias reflects a property of the model's learned representations rather than a gradient-weighting artifact. To address the radial attribution bias, the authors implemented a radial suppression method, which improved the average Top-10% IoU by 19.4% (from 0.129 to 0.155), verified through a two-tailed paired t-test ([Formula: see text]). Class-level analysis revealed significant improvements for edge-ring defects while other defect types remained stable. Finally, an oracle-based refinement demonstrated that geometric representation characteristics account for a substantial portion of residual localization errors, establishing an upper bound of 0.278 IoU.