The rapid progression of semiconductor device scaling toward the sub-nanometer regime has rendered particulate contamination a primary threat to manufacturing reliability. While conventional ex-situ inspection techniques offer high precision, they lack the real-time responsiveness required to trace dynamic sources of wafer defects within complex vacuum environments. This study introduces a monitoring framework that couples an in-situ real-time contaminant particle sensor (RTCPS) with artificial intelligence (AI) to forecast wafer-level defects during SiO2 plasma-enhanced chemical vapor deposition. By directly correlating real-time sensor signals with post-process wafer inspection data, we evaluated the linear relationship between in-situ monitoring and final contamination levels. This analysis revealed that particle generation during the cleaning process aligns much more closely with final wafer quality than during the deposition process. The cleaning process exhibited the coefficient of determination (R2) of 0.5149, which is over 2,500 times higher than the R2 of 0.0002 observed during deposition, providing a far more reliable indicator for wafer defect. Furthermore, a cross-validation-based AI model was developed to predict accumulated particle counts, achieving high precision with a mean deviation of approximately 73.3 particles. These results establish the RTCPS-AI integration as an essential diagnostic solution for linking in-situ process conditions directly to final wafer quality.
Precise control over defect structures in two-dimensional (2D) transition metal dichalcogenides (TMDCs) remains a critical challenge for high-performance gas sensors, as sensing characteristics are strongly governed by the density and distribution of adsorption-active sites. Among various defect types, grain boundary is of particular importance because they form interconnected networks of coordinatively unsaturated sites that enable both enhanced gas adsorption and efficient modulation of charge transport. However, practical strategies for controlling density of grain boundary in a simple and scalable manner remain limited, as it is primarily determined during nucleation and subsequent grain growth processes. Here, we report a pulsed laser-directed synthesis approach for controlling the density of grain boundary in MoS2 films. The laser pulse repetition frequency serves as a key parameter that dictates the energy distribution per pulse, thereby tuning the balance between nucleation and grain growth in a single processing step. As a result, the average grain size is reduced from about 27 to 8 nm with increasing pulse repetition frequency, accompanied by a significant increase in the density of grain boundary. Consistent with these structural changes, the NO2 gas sensing response is enhanced from about 10 to 48 %, demonstrating a clear correlation between the density of grain boundary and sensing performance. These results establish grain boundary as the dominant active sites governing gas adsorption and charge transfer in MoS2-based sensors and provide a scalable strategy for defect engineering in 2D TMDCs-based sensing platforms.
The development of techniques to sustain stable compact plasma discharge under low-pressure conditions is essential. Key design parameters for a coil-assisted compact plasma system were systematically swept, and variations in particle confinement characteristics were evaluated under an applied electric field environment. The investigated parameters included the number of coil turns, the inter-coil distance, and the thickness of the coil. An optimal design condition was identified, and the optimization results were interpreted through a physical analysis of electric field induced performance degradation. Practical design guidelines for efficient discharge systems are presented based on the results of three-dimensional electromagnetic simulations and particle simulation.
Second-order nonlinear optical (2NLO) materials are essential for practical photonic devices. Here, we demonstrate that external surface hydrophobic modification of silicalite-1 (SL1) films markedly enhances 2NLO activity by improving the degree of uniform orientation (DUO) of incorporated hemicyanine (HC-n) dyes. Attaching chlorodimethyloctadecylsilane to bare SL1 (bSL1) yielded strongly hydrophobic octadecylmodified SL1 (o-mSL1) (water contact angle: 75 degrees vs. 15 degrees), promoting unidirectional HC-n inclusion via hydrophobic-hydrophobic interactions. Although the number density in o-mSL1 is slightly reduced due to steric hindrance at channel entrances, the substantially higher DUO produces a significantly enhanced relative second harmonic generation intensity (rel-I2 omega). A maximum rel-I2 omega of 479.4% was achieved for HC-22@o-mSL1, surpassing 400.0 % for HC-15@bSL1, confirming that surface hydrophobicity directly governs DUO and 2NLO performance. This approach offers a practical route to high-performance inorganic-organic 2NLO materials.
