Artificial-Neural-Network-Driven Innovations in Time-Varying Process Diagnosis of Low-K Oxide Deposition.

To address the challenges in real-time process diagnosis within the semiconductor manufacturing industry, this paper presents a novel machine learning approach for analyzing the time-varying 10th harmonics during the deposition of low-k oxide (SiOF) on a 600 Å undoped silicate glass thin liner using a high-density plasma chemical vapor deposition system. The 10th harmonics, which are high-frequency components 10 times the fundamental frequency, are generated in the plasma sheath because of their nonlinear nature. An artificial neural network with a three-hidden-layer architecture was applied and optimized using k-fold cross-validation to analyze the harmonics generated in the plasma sheath during the deposition process. The model exhibited a binary cross-entropy loss of 0.1277 and achieved an accuracy of 0.9461. This approach enables the accurate prediction of process performance, resulting in significant cost reduction and enhancement of semiconductor manufacturing processes. This model has the potential to improve defect control and yield, thereby benefiting the semiconductor industry. Despite the limitations imposed by the limited dataset, the model demonstrated promising results, and further performance improvements are anticipated with the inclusion of additional data in future studies.

Diagnosing Time-Varying Harmonics in Low-k Oxide Thin Film (SiOF) Deposition by Using HDP CVD.

This study identified time-varying harmonic characteristics in a high-density plasma (HDP) chemical vapor deposition (CVD) chamber by depositing low-k oxide (SiOF). The characteristics of harmonics are caused by the nonlinear Lorentz force and the nonlinear nature of the sheath. In this study, a noninvasive directional coupler was used to collect harmonic power in the forward and reverse directions, which were low frequency (LF) and high bias radio frequency (RF). The intensity of the 2nd and 3rd harmonics responded to the LF power, pressure, and gas flow rate introduced for plasma generation. Meanwhile, the intensity of the 6th harmonic responded to the oxygen fraction in the transition step. The intensity of the 7th (forward) and 10th (in reverse) harmonic of the bias RF power depended on the underlying layers (silicon rich oxide (SRO) and undoped silicate glass (USG)) and the deposition of the SiOF layer. In particular, the 10th (reverse) harmonic of the bias RF power was identified using electrodynamics in a double capacitor model of the plasma sheath and the deposited dielectric material. The plasma-induced electronic charging effect on the deposited film resulted in the time-varying characteristic of the 10th harmonic (in reverse) of the bias RF power. The wafer-to-wafer consistency and stability of the time-varying characteristic were investigated. The findings of this study can be applied to in situ diagnosis of SiOF thin film deposition and optimization of the deposition process.

RT-CLAD: Artificial Intelligence-Based Real-Time Chironomid Larva Detection in Drinking Water Treatment Plants.

The presence of chironomid larvae in tap water has sparked public concern regarding the water supply system in South Korea. Despite ongoing efforts to establish a safe water supply system, entirely preventing larval occurrences remains a significant challenge. Therefore, we developed a real-time chironomid larva detection system (RT-CLAD) based on deep learning technology, which was implemented in drinking water treatment plants. The acquisition of larval images was facilitated by a multi-spectral camera with a wide spectral range, enabling the capture of unique wavelet bands associated with larvae. Three state-of-the-art deep learning algorithms, namely the convolutional neural network (CNN), you only look once (YOLO), and residual neural network (ResNet), renowned for their exceptional performance in object detection tasks, were employed. Following a comparative analysis of these algorithms, the most accurate and rapid model was selected for RT-CLAD. To achieve the efficient and accurate detection of larvae, the original images were transformed into a specific wavelet format, followed by preprocessing to minimize data size. Consequently, the CNN, YOLO, and ResNet algorithms successfully detected larvae with 100% accuracy. In comparison to YOLO and ResNet, the CNN algorithm demonstrated greater efficiency because of its faster processing and simpler architecture. We anticipate that our RT-CLAD will address larva detection challenges in water treatment plants, thereby enhancing water supply security.

Sequential Combination of Electro-Fenton and Electrochemical Chlorination Processes for the Treatment of Anaerobically-Digested Food Wastewater.

A two-stage sequential electro-Fenton (E-Fenton) oxidation followed by electrochemical chlorination (EC) was demonstrated to concomitantly treat high concentrations of organic carbon and ammonium nitrogen (NH4+-N) in real anaerobically digested food wastewater (ADFW). The anodic Fenton process caused the rapid mineralization of phenol as a model substrate through the production of hydroxyl radical as the main oxidant. The electrochemical oxidation of NH4+ by a dimensionally stable anode (DSA) resulted in temporal concentration profiles of combined and free chlorine species that were analogous to those during the conventional breakpoint chlorination of NH4+. Together with the minimal production of nitrate, this confirmed that the conversion of NH4+ to nitrogen gas was electrochemically achievable. The monitoring of treatment performance with varying key parameters (e.g., current density, H2O2 feeding rate, pH, NaCl loading, and DSA type) led to the optimization of two component systems. The comparative evaluation of two sequentially combined systems (i.e., the E-Fenton-EC system versus the EC-E-Fenton system) using the mixture of phenol and NH4+ under the predetermined optimal conditions suggested the superiority of the E-Fenton-EC system in terms of treatment efficiency and energy consumption. Finally, the sequential E-Fenton-EC process effectively mineralized organic carbon and decomposed NH4+-N in the real ADFW without external supply of NaCl.

Ultrasonication study for suspending single-walled carbon nanotubes in water.

A systematic calorimetry-based technique was developed to standardize single-walled carbon nanotube (SWNT) dispersion protocol. Simple calorimetric experiments were performed to benchmark the performance of the ultra-dismembrator. Temperature profiles for the sonication period were utilized to estimate energy input to the system. Energy loss profile was generated for the ultradismembrator in use and a calibration relationship was formulated that could standardize the sonication process. The standardized protocol was used to prepare aqueous SWNT suspensions-sonicating SWNTs in a varied range of input energy (18-100 kJ) in water. SWNT mass fractions suspended for each energy input was accurately measured and the suspended SWNT samples were characterized for morphology, surface potential, cluster size and structure, and chemical functionality using high resolution transmission electron microscopy (HRTEM), electrophoresis, dynamic and static light scattering (DLS/SLS), and Raman spectroscopy. The study demonstrated that suspended mass of SWNTs increased up to 18 kJ of energy input with no further increase upon continued energy input. The physicochemical properties showed similar trend for energy input. The aggregate cluster size, surface potential behavior, as well as the Raman defect properties plateaued after the initial energy input. The significant changes observed were limited to morphological properties, i.e., shorter length, debundled, and sharp edged SWNTs and fractal cluster formation (lower D(f)) with increased input energy.

Removal of micropollutants and NOM in carbon nanotube-UF membrane system from seawater.

One of the main problems for seawater reverse osmosis desalination is membrane fouling associated with natural organic matter. Bisphenol-A (BPA) and 17alpha-ethinylestradiol (EE2) are well-known endocrine-disrupting compounds that have been detected in wastewater and seawater. In this study, the contribution of carbon nanotubes (CNTs, single-walled carbon nanotubes) to membrane fouling control and the potential adsorption mechanisms of BPA and EE2 were investigated using artificial seawater (ASW) in a bench scale ultrafiltration (UF) membrane coupled with CNTs. For high ionic strength ASW, UVA254 nm is a good alternative for highly aromatic dissolved organic carbon (DOC) determination, with a very strong linear relationship (R2 > or = 0.99) with increasing DOC concentrations. Approximately 80% of DOC in ASW was rejected by the CNT-UF system where 31% of DOC was removed due to adsorption by CNTs. The presence of CNTs shows a 20% increase in membrane flux in ASW. A strong linear correlation between retention and adsorption of BPA and EE2 was obtained. The percentage of adsorption/retention of BPA and EE2 in UF-CNTs follows the order: 94.0/96.6 (DI + CNTs, EE2) > 86.2/90.0 (ASW + CNTs, EE2) > 73.6/78.9 (DI + CNTS, BPA) > or = 74.1/77.3 (ASW + CNTS, BPA) > 29.8/29.8 (ASW, EE2) approximately equal to 27.3/27.3 (ASW, BPA) > or = 25.3/25.3 (DI, EE2) approximately equal to 24.8/24.8 (DI, BPA). This indicates that retention by the UF-CNT system is mainly due to adsorption. Overall, EE2 adsorption was greater than BPA during the UF-CNT experiments, presumably due to the higher hydrophobicity of EE2 than BPA.

Removal of bisphenol A and 17α-ethinyl estradiol from landfill leachate using single-walled carbon nanotubes.

In this study, the adsorption of bisphenol A (BPA) and 17α-ethinyl estradiol (EE2) from landfill leachate onto single-walled carbon nanotubes (SWCNTs) was investigated. Different leachate solutions were prepared by altering the pH, ionic strength, and dissolved organic carbon (DOC) in the solutions to mimic the varying water conditions that occur in leachate during the various stages of waste decomposition. The youngest and oldest leachate solutions contained varying DOC and background chemistry and were represented by leachate Type A (pH = 5.0; DOC = 2500 mg/L; conductivity = 12,500 µS/cm; [Ca(2+)] = 1200 mg/L; [Mg(2+)] = 470 mg/L) and Type E (pH = 7.5; DOC = 250 mg/L; conductivity = 3250 µS/cm; [Ca(2+)] = 60 mg/L; [Mg(2+)] = 180 mg/L). These solutions were subsequently combined in different ratios to produce intermediate solutions, labeled B-D, to replicate time-dependent changes in leachate composition. Overall, a larger fraction of EE2 was removed as compared to BPA, consistent with its higher log K(OW) value. The total removal of BPA and EE2 decreased in older leachate solutions, with the adsorptive capacity of SWCNTs decreasing in the order of leachate Type A > Type B > Type C > Type D > Type E. An increase in the pH from 3.5 to 11 decreased the adsorption of BPA by 22% in young leachate and by 10% in old leachate. The changes in pH did not affect the adsorption of EE2 in the young leachate, but did reduce adsorption by 32% in the old leachate. Adjusting the ionic strength using Na(+) did not significantly impact adsorption, while increasing the concentration of Ca(2+) resulted in a 12% increase in the adsorption of BPA and a 19% increase in the adsorption of EE2. DOC was revealed to be the most influential parameter in this study. In the presence of hydrophilic DOC, represented by glucose in this study, adsorption of the endocrine disrupting compounds (EDCs) onto the SWCNTs was not affected. In the absence of SWCNTs, hydrophobic DOC (i.e., humic acid) adsorbed 15-20% of BPA and EE2. However, when the humic acid and SWCNTs were both present, the overall adsorptive capacity of the SWCNTs was reduced. Hydrophobic (π-π electron donor-acceptor) interactions between the EDCs and the constituents in the leachate, as well as interactions between the SWCNTs and the EDCs, are proposed as potential adsorption mechanisms for BPA and EE2 onto SWCNTs.

Removal of phenol and substituted phenols by newly developed emulsion liquid membrane process.

Emulsion liquid membrane (ELM) stabilized by non-Newtonian conversion of the liquid membrane phase and dispersed in a Taylor-Couette flow was applied to treat model industrial wastewaters containing phenols and selected substituted phenols (hydroquinone, three chlorophenols and two nitrophenols) at relatively high concentrations. Under optimized operating conditions, all the compounds except hydroquinone were extracted with the maximum extraction efficiency of over 96% and the time to reach maximum extraction ranged from 2 to 30 min. For 2-chlorophenol, 2,4-dichlorophenol, and 2-nitrophenol, maximum extraction efficiency of over 99% was reached in less than 2 min of contact time. Extraction of these compounds from the mixture was also highly efficient. Factors that affected the overall extraction efficiency such as acid-base speciation of the target compounds and the affinity between the target compounds and the membrane phase were determined. This study suggested that the newly developed ELM process might provide a promising alternative technology to treat industrial wastewaters containing high concentrations of phenolic compounds.