Defect Counts Research Articles

A new control chart based on proportional hazards and frailty regression models for monitoring high‐quality processes with censored observations

AbstractImproving quality and reducing defects in various processes have led to the design of numerous control charts for monitoring count data. These charts are characterized by a large number of zero defects and are modeled with zero‐inflated distributions. Besides, it is important to consider the influence of key factors on the number of defects in some processes. This paper presents three monitoring scenarios that utilize cumulative sum (CUSUM) control charts in Phase II. These scenarios are valuable for monitoring high‐quality processes, particularly with censored observations. The first scenario uses a zero‐inflated Poisson (ZIP) distribution while the second one employs a proportional hazards regression model to account for observed variables. The third scenario combines the proportional hazards and frailty regression models to consider both observed and unobserved variables. The performance of all three control charts is evaluated by applying shifts to both parameters of the ZIP distribution separately and simultaneously. Extensive simulation studies are conducted and the effects of low, moderate, and high censoring rates are thoroughly studied. The results highlight the importance of considering influential variables for detecting early shifts, as neglecting them significantly impairs the control chart performance. Finally, a real case study, involving defect counts in printed rolls at a label printing factory, has been provided and discussed to show the application of the monitoring procedures.

Low-Temperature Annealing of Nanoscale Defects in Polycrystalline Graphite

Polycrystalline graphite contains multi-scale defects, which are difficult to anneal thermally because of the extremely high temperatures involved in the manufacturing process. In this study, we demonstrate annealing of nuclear graphite NBG-18 at temperatures below 28 °C, exploiting the electron wind force, a non-thermal stimulus. High current density pulses were passed through the specimens with a very low-duty cycle so that the electron momentum could mobilize the defects without heating the specimen. The effectiveness of this technique is presented with a significant decrease in electrical resistivity, defect counts from X-ray computed tomography, Raman spectroscopy, and nanoindentation-based mechanical characterization. Such multi-modal evidence highlights the feasibility of nanoscale defect control at temperatures about two orders of magnitude below the graphitization temperature.

Similarity searching for wafer bin maps by measuring shape, location, and size similarities of defect patterns

A wafer bin map (WBM) is a visual representation of the spatial distribution of defective chips on a wafer. WBMs showing specific defect patterns are usually a result of process-assignable causes; thus, it is important to identify them to eliminate assignable causes. With advances in semiconductor manufacturing technology, identifying new defect patterns, and diagnosing their causes have become critical. However, most existing methods for WBM analysis use a supervised learning approach, which only detects previously known defect patterns. The similarity-search approach is a suitable alternative for defining new defect patterns. The proposed method uses an unsupervised approach to search for similar WBMs by measuring the similarities of three spatial features of defect patterns–shape, location, and size–which are useful for defining new defect patterns and diagnosing their causes. These three similarities are achieved using tensor voting and the mountain function for shape similarity, Euclidean distance for location similarity, and a combination of defect count and average radius for size similarity. The overall similarity was assessed using the weighted average of the three similarities. The weights are determined by quantifying the uncertainty of each similarity based on information entropy theory to better distinguish between similar patterns. The experimental results demonstrate the effectiveness of the proposed method compared to existing methods and highlight its capability to identify and describe the spatial features of defect patterns.

Optimizing defect rework for a software start-up's schedule variation: a Six-Sigma-based approach

PurposeManaging project completion within the stipulated time is significant to all firms' sustainability. Especially for software start-up firms, it is of utmost importance. For any schedule variation, these firms must spend 25 to 40 percent of the development cost reworking quality defects. Significantly, the existing literature does not support defect rework opportunities under quality aspects among Indian IT start-ups. The present study aims to fill this niche by proposing a unique mathematical model of the defect rework aligned with the Six Sigma quality approach.Design/methodology/approachAn optimization model was formulated, comprising the two objectives: rework “time” and rework “cost.” A case study was developed in relevance, and for the model solution, we used MATLAB and an elitist, Nondominated Sorting Genetic Algorithm (NSGA-II).FindingsThe output of the proposed approach reduced the “time” by 31 percent at a minimum “cost”. The derived “Pareto Optimal” front can be used to estimate the “cost” for a pre-determined rework “time” and vice versa, thus adding value to the existing literature.Research limitations/implicationsThis work has deployed a decision tree for defect prediction, but it is often criticized for overfitting. This is one of the limitations of this paper. Apart from this, comparing the predicted defect count with other prediction models hasn’t been attempted. NSGA-II has been applied to solve the optimization problem; however, the optimal results obtained have yet to be compared with other algorithms. Further study is envisaged.Practical implicationsThe Pareto front provides an effective visual aid for managers to compare multiple strategies to decide the best possible rework “cost” and “time” for their projects. It is beneficial for cost-sensitive start-ups to estimate the rework “cost” and “time” to negotiate with their customers effectively.Originality/valueThis paper proposes a novel quality management framework under the Six Sigma approach, which integrates optimization of critical metrics. As part of this study, a unique mathematical model of the software defect rework process was developed (combined with the proposed framework) to obtain the optimal solution for the perennial problem of schedule slippage in the rework process of software development.

Microscopic Evaluation of Dentinal Defects after Root Canal Preparation with Different Hand and Nickel-Titanium Files; An In Vitro Study

ABSTRACT Background: Effective endodontic instrumentation aims to remove microorganisms, debris, and tissue from the root canal while maintaining dentinal integrity. This study compares dentinal defect incidence following canal preparation with different hand files, nickel-titanium rotary files, and reciprocating files Materials and Methods: Eighty single-rooted mandibular premolars with mature apices were collected. Four groups (n = 20) were established based on canal patency establishment, canal preparation technique, irrigation solution, and final irrigation. After root sectioning at 3 mm, 6 mm, and 9 mm from the apex, slices were examined under a stereomicroscope, and dentinal defects were recorded. A second examiner reviewed the images. Results: The results showed significant difference of P = 0.031 among Hand files vs. Rotary ProTaper files: In Hand files vs. Reciprocating WaveOne files significant difference was P = 0.048, and for rotary ProTaper files vs. Reciprocating WaveOne files: No significant difference (P = 0.643). Dentinal defect counts were similar at 3 mm, 6 mm, and 9 mm. Statistically significant variation was observed between hand files and rotary files, as well as hand files and reciprocating files Conclusion: Both rotary and reciprocating files showed a statistically significant increase in dentinal defect incidence compared to hand files. However, clinical implications should be considered cautiously. Instrumentation technique selection should be based on clinical context, operator experience, and patient factors. Further clinical studies are needed for validation.

Enhancing Software Quality through Early-Phase of Software Verification and Validation Techniques

Purpose: The core objectives of this conducted research were threefold: first, to assess whether early-phase verification and validation (V&V)techniques, encompassing software design review, code review, and inspection, significantly contribute to defect prevention throughout the software development process; second, to evaluate the impact of these techniques on the efficiency of software development in terms of time, cost, and resource allocation; and third, to conduct a comprehensive comparative analysis between early-phase verification and validation (V&V)techniques and traditional post-development testing regarding their effectiveness in defect prevention and software quality improvement.
 Methodology: Employing an experimental approach, this study conducted case studies within Tanzanian software development organizations. Data were meticulously gathered through surveys and interviews involving software professionals. Both quantitative and qualitative data were systematically analyzed to ensure data reliability and ethical considerations.
 Findings: The research findings indicate that early-phase verification and validation (V&V) techniques shine as powerful tools for defect prevention, leading to a substantial reduction in defect counts and elevating software quality metrics, such as reliability and user satisfaction. Notably, these techniques demonstrated a favorable impact on software development efficiency by shortening development cycles and curtailing costs. When compared to traditional post-development testing, early-phase verification and validation (V&V) techniques consistently outperformed in the realm of defect prevention and software quality enhancement.
 Unique Contribution to Theory, Practice and Policy: The study's outcomes underscore the transformative potential of early-phase verification &validation techniques, urging the software development industry to pivot from defect detection to defect prevention. The insights garnered here offer practical guidance for organizations aiming to streamline their software development processes, economize costs, and deliver top-notch software products that closely align with user expectations. While conducted in a Tanzanian context, these findings reverberate across the global software development landscape, underscoring the universal relevance of early-phase verification and validation (V&V) techniques as pivotal enablers of software quality assurance and operational efficiency.

Polymer Nanoparticles Applied in the CMP (Chemical Mechanical Polishing) Process of Chip Wafers for Defect Improvement and Polishing Removal Rate Response.

Chemical mechanical planarization (CMP) is a wafer-surface-polishing planarization technique based on a wet procedure that combines chemical and mechanical forces to fully flatten materials for semiconductors to be mounted on the wafer surface. The achievement of devices of a small nano-size with few defects and good wafer yields is essential in enabling IC chip manufacturers to enhance their profits and become more competitive. The CMP process is applied to produce many IC generations of nanometer node, or those of even narrower line widths, for a better performance and manufacturing feasibility. Slurry is a necessary supply for CMP. The most critical component in slurry is an abrasive particle which affects the removal rates, uniformity, defects, and removal selectivity for the materials on the wafer surface. The polishing abrasive is the source of mechanical force. Conventional CMP abrasives consist of colloidal silica, fume silica or other inorganic polishing particles in the slurries. We were the first to systematically study nanoparticles of the polymer type applied in CMP, and to compare traditional inorganic and polymer nanoparticles in terms of polishing performance. In particular, the polymer nanoparticle size, shape, solid content dosing ratio, and molecular types were examined. The polishing performance was measured for the polishing removal rates, total defect counts, and uniformity. We found that the polymer nanoparticles significantly improved the total defect counts and uniformity, although the removal rates were lower than the rates obtained using inorganic nanoparticles. However, the lower removal rates of the polymer nanoparticles are acceptable due to the thinner film materials used for smaller IC device nodes, which may be below 10 nm. We also found that the physical properties of polymer nanoparticles, in terms of their size, shape, and different types of copolymer molecules, cause differences in the polishing performance. Meanwhile, we used statistical analysis software to analyze the data on the polishing removal rates and defect counts. This method helps to determine the most suitable polymer nanoparticle for use as a slurry abrasive, and improves the reliability trends for defect counts.

A design methodology based on two dimensional fuzzy linguistic variables for attribute control charts with real case applications

Attribute control charts (ACCs) effectively evaluate process’ situation based on nonconforming items (NCIs) and defects of items in a process. The traditional control charts (CCs) that analyze process control situations need a major adjustment to take into account uncertainties into eal case problems that can occur due to human evaluations and hesitancies, measurement systems errors, and process situations. The fuzzy set theory (FST) has been used successfully to reach this aim. Recently, some fuzzy set extensions have been integrated into traditional CCs to more extensively and sensitively model the uncertainty by involving other uncertainty parameters, such as non-membership value and hesitancy function. In this paper, one of these extensions, Pythagorean fuzzy sets (PFSs), has been adjusted with CCs to model the vagueness of human evaluations, their hesitancies, and uncertainties regarding the process. For this aim, the well-known ACCs based on the total count of defects and NCIs and called c, u, p, and np CCs, respectively, have been re-designed by using PFSs. Additionally, the control procedure of these CCs has been deeply analyzed with respect to fuzzy limit values and fuzzy measurements. Also, a rule-based approach that checks in-control and out-of-control situations regarding distance measurements is suggested to check the sample’s status more accurately. The proposed CCs based on PFSs have also been applied on a real case study from a manufacturing factory. The obtained results confirm that the CCs based on PFSs have an ability to create accurate, sensitive, and flexible results by using the input data while analyzing the process’ stability.

Secondary defects of as-grown oxygen precipitates in nitrogen doped Czochralski single crystal silicon

In this work, as-grown defects and their secondary defects in 300 mm nitrogen-doped Czochralski-grown single-crystal silicon (NCZ-Si) are investigated. Secondary defects are revealed by homogeneous epitaxial layer growth, and as-grown defects are decorated via gaseous hydrogen chloride (HCl) etching. Localized light scattering (LLS) technique is utilized to inspect defect distribution and provide the latex sphere equivalent (LSE) size to differentiate defect types. With the locations of LLS, the morphology of defect is observed by scanning electron microscopy (SEM). According to the results, it indicated that secondary defects are extrinsic dislocation loops and stacking fault (SF) loops, which are induced by as-grown oxygen precipitates during the growth and cooling process. Furthermore, the correlation between the counts of secondary defects and nitrogen concentration is obtained. It contributes to the optimization of nitrogen doping to avoid the generation of detrimental defects for device operation.

Transformer‐optimized generation, detection, and tracking network for images with drainage pipeline defects

AbstractRegular detection of defects in drainage pipelines is crucial. However, some problems associated with pipeline defect detection, such as data scarcity and defect counting difficulty, need to be addressed. Therefore, a Transformer‐optimized generation, detection, and counting method for drainage‐pipeline defects was established in this paper. First, a generation network called Trans‐GAN‐Cla was developed for data augmentation. A classification network was trained to improve the quality of the generated images. Second, a detection and tracking model called Trans‐Det‐Tra was developed to track and count the number of defects. Third, the feature extraction capability of the proposed method was improved by leveraging Transformers. Compared with some well‐known convolutional neural network‐based methods, the proposed network achieved the best classification and detection accuracies of 87.2% and 87.57%, respectively. Furthermore, the F1 scores were 87.7% and 91.9%. Finally, two pieces of onsite videos were detected and tracked, and the numbers of misalignments and obstacles were accurately counted. The results indicate that the established Transformer‐optimized method can generate high‐quality images and realize the high‐accuracy detection and counting of drainage pipeline defects.

Quantitative and technological quality indicators of the new “Nasib” cotton variety

The paper presents the results of the analysis of cotton weight per boll, yield, fiber output, fiber length and fiber quality of the newly created medium-fiber “Nasib” cotton variety, which is studied as the selection test. A three-year analysis of fiber microneur, hardness, length, uniformity, index of short fibers, elongation at break, soiling, defect count, reflectivity and yellowness is presented.

Multi-surface defect detection for universal joint bearings via multimodal feature and deep transfer learning

ABSTRACT Surface defect detection by machine vision has received increased attention concerning the quality control of universal joint bearings (UJB). The defect distribution and counting information are important for product quality optimisation. However, vision defect detection for UJB remains a challenging task due to the diversity of background textures and defect characters on multiple surfaces. In this study, a multi-surface defect detection (MSDD) method consisting of region segmentation, feature extraction and detection is proposed. First, defect regions are accurately localised by the proposed adaptive defect region segmentation algorithm, which suppresses the inference of background variety. Then, a novel defect feature named multimodal fusion shape descriptor that integrates the global information and local information of defects is constructed to generate the discriminative defect representation. Finally, a defect feature extraction ability transfer strategy based on the transfer learning mechanism is proposed to address the problem of insufficient defect samples. The experimental results show that our method achieves the best accuracy of 94.8% macro-F1 and processes 28 defect images per second. Besides, the application effect in the practical production line indicates that our method meets the accuracy and real-time requirements of MSDD for UJB.

Automatic Detection and Counting System for Pavement Cracks Based on PCGAN and YOLO-MF

The regular detection of pavement cracks is critical for life and property security. However, existing deep learning-based methods of crack detection face difficulties in terms of data acquisition and defect counting. An automatic intelligent detection and tracking system for pavement cracks is proposed. Our system is formed of a pavement crack generative adversarial network (PCGAN) and a crack detection and tracking network called YOLO-MF. First, PCGAN is used to generate realistic crack images, to address the problem of the small number of available images. Next, YOLO-MF is developed based on an improved YOLO v3 modified by an acceleration algorithm and median flow (MF) algorithm to count the number of cracks. In a counting loop, our improved YOLO v3 detects cracks and the MF algorithm tracks the cracks detected in a video. This improved algorithm achieves the best accuracy of 98.47% and F1 score of 0.958 among other algorithms, and the precision-recall curve was close to the top right. A tiny model was developed and an acceleration algorithm was applied, which improved the detection speed by factors of five and six, respectively. In on-site measurement, three cracks were detected and tracked, and the total count was correct. Finally, the system was embedded in an intelligent device consisting of a calculating module, an automated unmanned aerial vehicle, and other components.

PMON252 Δ-9-tetrahydrocannabinol does not affect motility, morphology or viability, but alters the transcriptome of cryopreserved sperm

Abstract Cannabis is one of the most widely consumed recreational drugs among reproductive aged adults around the globe. It is known that Δ-9-tetrahydrocannabinol (THC), the psychoactive component of cannabis, can interact with the endocannabinoid system (ECS) via cannabinoid receptor 1. Components of the ECS are highly involved in regulating mammalian reproductive processes such as oocyte, sperm, and embryo development by modulating gene expression, apoptosis, and several intracellular pathways. Previous research in our lab demonstrated that THC exposure to bovine oocytes, the ideal translational model for human, was detrimental to in vitro maturation, increased apoptosis in blastomeres and altered gene expression (Misner et al., 2021). Since cannabis consumption is more prevalent among males, we investigated THC effects on male fertility and sperm function. We hypothesized that in vitro THC treatment decreases motility, increases morphological defects, and alters gene expression in sperm. To test this hypothesis, cryopreserved sperm from 5 bulls of known fertility (n=5) was incubated for 6- or 12-hours with concentrations of THC mimicking the plasma levels of THC following therapeutic (0.032μM) and low and high recreational usage (0.32 μM and 3.2 μM) in humans. Bull sperm was used as a translational model since it is highly morphologically similar to human sperm. Sperm motility was assessed manually on a minimum of 100 sperm per treatment group in each bull using a Makler counting chamber and morphology was assessed using Giemsa staining and manual counting of head, neck, midpiece and tail defects of 300 sperm per treatment in each bull. Apoptosis was evaluated using Annexin V-FITC and PI fluorescent staining coupled with flow cytometry. To evaluate gene expression, RNA from THC-treated sperm was subjected to transcriptome analysis using a GeneChip Bovine Gene 1.0 ST Array. Our results show that THC-treatment at both 6-hours and 12-hours had no influence on sperm motility, contrary to findings in the literature. Furthermore, we also observed no significant effects on sperm morphology and apoptosis following THC-treatment. However, transcriptome analysis identified 39 significantly differentially expressed genes (DEG) following sperm exposure to the lowest concentration of THC, while 196 and 33 genes were differentially expressed following exposure to the mid and high doses correlated with recreational cannabis use, respectively. Upon further analysis of these DEG using DAVID 6.8, several gene ontology and functionally related gene groups were enriched including genes related to nucleosome assembly and function, spermatogenesis, apoptosis and ribosomes and translation. Additional validation of a subset of DEG will be performed using digital droplet PCR. These findings suggest that in vitro THC treatment may cause intracellular and molecular effects, ultimately altering sperm competence, despite not affecting phenotypic characteristics such as motility, morphology or viability (apoptosis). Presentation: Monday, June 13, 2022 12:30 p.m. - 2:30 p.m.

Enhanced Lot Acceptance Testing Based on Defect Counts and Posterior Odds Ratios

Optimal defects-per-unit test plans based on posterior odds ratios are developed for the disposition of product lots. The number of nonconformities per unit is modeled by the Conway–Maxwell–Poisson distribution rather than the typical Poisson model. In essence, a submitted batch is conforming if its posterior acceptability is sufficiently large. First, a useful approximation of the optimal test plan is derived in closed form using the asymptotic normality of the log ratio. A mixed-integer nonlinear programming problem is then solved via Monte Carlo simulation to find the smallest number of inspected items per lot and the maximum tolerable posterior odds ratio. The methodology is applied to the manufacturing of paper and glass. The suggested sampling plan for lot sentencing provides the specified protections to both manufacturers and customers and minimizes the needed sample size. In terms of inspection effort and accuracy, the proposed approach is virtually an advantageous extension of the classical frequentist perspective. In many practical cases, it yields more precise assessments of the current consumer and producer risks, as well as more realistic decision rules.

Effects of morphology and graphitization of carbon nanomaterials on the rheology, emulsion stability, and filtration control ability of drilling fluids

Nanomaterials are good potential additives in drilling fluid to improve the thermal and physical-mechanical properties of the fluids, but their use in this field is still scarce. Ester-based drilling fluid (EBDF) is an environmentally friendly drilling fluid. However, its use is still restricted to low to moderate drilling temperatures due to its poor thermal and hydrolytic stability. This study investigates the roles of morphology and graphitization of carbon nanomaterials on the rheology, filtration, and emulsion stability of EBDF at high pressure high temperature (HPHT) drilling conditions. Various carbon nanomaterials were investigated, including graphene nanopowder, commercial graphene nanoplatelets, in-house graphene nanoplatelets, graphene oxide, and carbon nanotubes cotton. The results showed that the behaviour of EBDF depended on the morphology and graphitization of carbon nanomaterials. Graphene nanopowder demonstrated the best performance due to its large size, low graphitic defect count, and most hydrophobic surface, which improved the EBDF emulsion stability and HPHT filtration properties. Hydrophobic surface of graphene nanopowder sheets provides superior protection against thermal degradation of the emulsifier films in drilling fluid. Filtration tests revealed that graphene nanopowder reduced the filtrate and filter cake thickness by 20% and 25%, respectively, at a graphene concentration of 0.007 wt%.

Atomic simulations of primary irradiation damage in U–Mo–Xe system

To shed a light on Xe bubble nucleation in U–Mo fuel from the view of primary irradiation damage, a reported U–Mo–Xe potential under the framework of embedded atom method has been modified within the range of short and intermediate atomic distance. The modified potential can better describe the interactions between energetic particles, and can accurately reproduce the threshold displacement energy surface calculated by the first-principles method. Then, molecular dynamics simulations of primary irradiation damage in U–Mo–Xe system have been conducted under different contents. The raise of Xe concentration brings about a remarkable promotion in residual defect quantity and generates bubbles in more over-pressured state, which suggests an acceleration of irradiation damage under the accumulation of the fission gas. Meanwhile, the addition of Mo considerably reduces the residual defect count and hinders irradiation-induced Xe diffusion especially at high contents of Xe, corroborating the importance of high Mo content in mitigation of irradiation damage and swelling behavior in U–Mo fuel. In particular, the variation of irradiation damage with respect to contents suggests a necessity of taking into account the influence of local components on defect evolution in mesoscale simulations.

İntratekal Gadolinyumlu Manyetik Rezonans Myelografide THRIVE Sekansının Katkısı

İntratekal Gadolinyumlu Manyetik Rezonans Myelografi (İG-MRM) spontan intrakraniyal hipotansiyonun (SİH) etyolojisini araştırmak için yapılan bir inceleme yöntemidir. İncelemede genellikle yağ baskılı 2 boyutlu T1A turbo spin eko (2B-T1A TSE) sekansı tercih edilmektedir. “T1-weighted high-resolution isotropic volume examination” (THRIVE) sekansı; yüksek çözünürlüklü yağ baskılı T1 görüntüleme imkanı sunan, yeni nesil 3 boyutlu gradient eko (GRE) görüntüleme tekniğidir. Bu çalışmada; THRIVE sekansının intrakraniyal hipotansiyon ile ilişkili İG-MRM bulgularını saptamadaki tanısal başarısını 2B-T1A TSE ile karşılaştırarak değerlendirmeyi amaçladık. Mart 2018 – Kasım 2019 tarihleri arasında klinik ve radyolojik bulgularına göre SİH tanısı konan ve İG-MRM yapılan 16 hasta dahil edildi. 2B-T1A TSE ve THRIVE sekansları; servikal, torakal ve lomber düzeydeki beyin omurilik sıvısı (BOS) opasifikasyon düzeyleri, tekniklerin her birinde saptanan epidural BOS kolleksiyonu, dural defekt ve meningeal divertikül bulgularının sayısı açısından karşılaştırıldı. Servikal ve torakal düzeyde THRIVE sekansındaki BOS opasifikasyon düzeyi 2B-T1A TSE’ye göre anlamlı düzeyde daha düşüktü (p0,05). SİH olgularının İG-MRM’sinde THRIVE sekansının kullanımı ile küçük dural defektler ve meningeal divertiküller rutin sekanslara göre daha kolay saptanabilir. Ancak sekansın gradient eko tabanlı olması nedeni ile özellikle servikal ve torakal düzeyde BOS alanları yeterince opasifiye olmayabilir.

(Digital Presentation) Research on Single Wafer RCA Clean in High Aspect Ratio Trenches

I ntroduction. High aspect ratio (AR) processes are widely used in front end of line (FEOL) IC manufacturing. Figure 1 depicts high (AR) trench cleaning process. Polymerization residues produced during the dry etch process will cause many problems such as STI voids in the follow-up process if the residues are not removed. This can lead to lower yield and reduction of chip life[1]. It is difficult to completely clean the residue inside trench with aspect ratio greater than 10.As shown in Figure.2, poor wetting of the liquid in the high AR trench, prevents the cleaning of the whole trench. Typically, the wafer surface can be easily cleaned, however residues remain inside of the trench. The cleaning of polymerization residues inside the trench and eliminating watermarks are the main issues in high (AR) cleaning processes. In this paper, we design an infiltration experiment, polymerization residues cleaning experiment and drying experiment to verify single wafer RCA clean performance on high (AR) trench processes. The verification of the single wafer RCA wafer clean is through yield data. Experimental. In the infiltration experiment, a layer of 500A SiO2 was grown on the high aspect ratios trench wafer. As shown in Table I, This was followed by treating the wafers with different process conditions on ACM single wafer tool. The remaining oxide thickness inside trench was measured by transmission electron microscope (TEM). Figure 3 shows experimental steps.In the polymerization residue cleaning experiment, 7 wafers with trenches after dry etching were treated with different chemical and process mode shown in Table II on ACM single wafer tool. The defect count was measured by the inline defect detection equipment and the scanning electron microscope.(SEM)In the drying experiment, high AR wafers were processed on an ACM single wafer tool, under different drying conditions. The trench wafer were then transferred into a vacuum bag and the air was evacuated. If there is no watermark on wafer surface within 15 minutes, we consider that the wafer has been dried in the macro level.Wafers were split between being cleaned by single wafer RCA clean and by bench RCA clean. The wafers are scanned for defect count on the wafer surface after the cleaning process for both the base line (bench) and the single wafer tool. Finally, we review yield on device completion to check results of single wafer and bench clean. Results. Table I shows the result of dHF infiltration experiment. When the pre rinse DIW time exceeds 30 seconds, the whole trench is soaked by DIW, allowing the dHF to permeate the whole trench, with oxide loss being similar from the top to the bottom of the trench.Table II shows the result of cleaning experiment. ACM Single wafer dHF+SC1 (mega) process can effectively clean the whole trench. Figure 4 and 5 shows defect map and SEM picture after single DHF+SC1(mega).Figure 6 shows the effect of chuck rotation speed and N2 flow rate on wafer drying. Figure 10 shows particle performance under different drying process conditions. Figure 7 shows the inline defect map and SEM after dry process.Figure 8 & 9 show the particle performance after the clean process. No special particle map appears with the single wafer clean process. Fig 10 shows the yield performance after all of the processes are completed. Average yield of single clean process is better than the baseline bench clean process. Table 3 shows detailed yield data from the wafer splits . Conclusion. Infiltration and cleaning experiment result shows that single wafer RCA clean can effectively clean the polymerization residues inside of high aspect ratio trenches. Dry experiment result shows that single wafer process can reduce watermark and particles in the dry step. Inline particle performance and final yield performance shows that single wafer RCA clean is better than baseline bench. Single wafer clean has lower standard deviation than bench clean and SW RCA clean in high (AR) trench process is feasible and effective. References. Xue. Research on High Aspect Ratio TSV Cleaning Technology enhanced by Megasonic[D]. Beijing: University of Chinese Academy of Sciences, 2016.JiHoon Cha, ChangSup Mun, etal. Low Si recess on cleaning process by dilute HF/SC-1 with megasonic[J]. Solid State Phenomena, 2008, 134:209-212.H Wang, Yue Ma, Fuping.Chen, etal. Removal of Fine Particle using SAPS Technology and Functional Water[J]. Solid State Phenomena, 2013, 195:185-190.Dae-Hong EOM, Geun-Bae LIM, etal. Reaction of Ozone and H2O2 in NH4OH Solutionsand Their Reaction with Silicon Wafers[J]. Japanese Journal of Applied Physics, 2004, 43:3335-3339. Figure 1

Deep learning characterization of surface defects in the selective laser melting process

Surface defects in the selective laser melting (SLM) process adversely affect the surface quality of additive manufacturing workpieces. Although some studies have been conducted to classify or detect defects, the investigation of the distribution of defects has received relatively little attention. Currently, there are many studies which focus on the counting of objects and they typically adopt a backbone convolutional neural network to obtain an initial feature map, which contains more semantic information but loses some geometric details. In this paper, the distribution of surface defects and defect count estimation of additive manufactured components are first studied as well as presentation of a developed deep learning characterization method (DLCM) based on a detail-aware dilated convolutional neural network (DDCNN) incorporated with a fine details feature map extractor designed to obtain final fine semantic features. It features two dilated convolutional layer combination blocks (DCBs), which are proposed to fuse low-level features with semantic features. Data are acquired from the surface of a workpiece fabricated by SLM. A series of experiments have been conducted to validate the performance of the DLCM. Compared with the other main state-of-the-art methods, the proposed DLCM yields better results.