A5090160531- Bayesian Modeling and Causal Inference
- Advancements in Photolithography Techniques
- Integrated Circuits and Semiconductor Failure Analysis
- Electron and X-Ray Spectroscopy Techniques
- Optical Coatings and Gratings
- Industrial Vision Systems and Defect Detection
- Data Management and Algorithms
- Semiconductor materials and devices
- Advanced Surface Polishing Techniques
- Rough Sets and Fuzzy Logic
- Data Quality and Management
- Data Mining Algorithms and Applications
- Semantic Web and Ontologies
- AI-based Problem Solving and Planning
- Surface Roughness and Optical Measurements
- Gene expression and cancer classification
- Bayesian Methods and Mixture Models
- Copper Interconnects and Reliability
- Anomaly Detection Techniques and Applications
- Network Security and Intrusion Detection
- Fault Detection and Control Systems
- Genetic Associations and Epidemiology
- Advanced Measurement and Metrology Techniques
- Advancements in Semiconductor Devices and Circuit Design
- Software Reliability and Analysis Research
Nantes Université
2015-2024
Laboratoire des Sciences du Numérique de Nantes
2017-2024
IMEC
2015-2024
IMT Atlantique
2022-2024
Laboratoire de Physique Subatomique et des Technologies Associées
2022-2024
Centre National de la Recherche Scientifique
2010-2023
École Centrale de Nantes
2023
Institut National de Physique Nucléaire et de Physique des Particules
2022
École Polytechnique
2007-2021
Sorbonne Université
1996-2021
In data analysis, latent variables play a central role because they help provide powerful insights into wide variety of phenomena, ranging from biological to human sciences. The tree model, particular type probabilistic graphical models, deserves attention. Its simple structure - allows and efficient inference, while its capture complex relationships. the past decade, model has been subject significant theoretical methodological developments. this review, we propose comprehensive study...
Discovering the genetic basis of common diseases in human genome represents a public health issue. However, dimensionality data (up to 1 million markers) and its complexity make statistical analysis challenging task. We present an accurate modeling dependences between markers, based on forest hierarchical latent class models which is particular probabilistic graphical models. This model offers adapted framework deal with fuzzy nature linkage disequilibrium blocks. In addition, can be reduced...
At 7nm and beyond, designers need to support scaling by identifying the most optimal patterning schemes for their designs. Moreover, can actively help exploring options that do not necessarily require aggressive pitch scaling. In this paper we will illustrate how MOL scheme be optimized achieve a dense SRAM cell; optimizing device performance lead smaller standard cells; metal interconnect stack needs adjusted unidirectional metals vertical transistor shift design paradigms. This...
This proposes a novel ensemble deep learning-based model to accurately classify, detect and localize different defect categories for aggressive pitches thin resists (High NA applications).In particular, we train RetinaNet models using ResNet, VGGNet architectures as backbone present the comparison between accuracies of these their performance analysis on SEM images with types patterns such bridge, break line collapses. Finally, propose preference-based strategy combine output predictions...
Recommender systems are applications to retrieve useful information from large amount of online data assist users in discovering interesting items/products the system. Collaborative filtering, content-based demographics-based filtering and hybrid approach main approaches realize recommendation systems. Most existing algorithms use a single deal with problems. Besides, traditional mainly dyadic relationships between items whereas real world generally conceptualized terms objects relations...
CD-SEM images inherently contain a significant level of noise. This is because limited number frames are used for averaging, which critical to ensure throughput and minimize resist shrinkage. noise SEM may lead false defect detections erroneous metrology. Therefore, reducing in utmost importance. Both conventional filtering techniques recent most discriminative deep-learning based denoising algorithms restricted with certain limitations. The first enables the risk loss information content...
For many years traditional 193i lithography has been extended to the next technology node by means of multi-patterning techniques. However recently such a became challenging and expensive push beyond for complex features that can be tackled in simpler manner Extreme UltraViolet Lithography (EUVL) technology. Nowadays, EUVL is part high-volume manufacturing device landscape it reached critical decision point where one further single print on 0.33NA full field scanner or move EUV double...
Design-Technology co-optimization becomes a key knob to enable CMOS scaling. In this work we evaluate the technology options including lithography as well device that are considered N10 scaling by exploring their impact on representative designs such standard cells, SRAM and analog contexts. This paper illustrates design angle needs be early in development of node. assessment decisions start from constraints power/performance, area cost, all which create Co-Optimization space.
The imec N7 (iN7) platform has been developed to evaluate EUV patterning of advanced logic BEOL layers. Its design is based on a 42 nm first-level metal (M1) pitch, and 32 pitch for the subsequent M2 layer. With these pitches, iN7 node an 'aggressive' full-scaled N7, corresponding IDM or foundry N5. Even in 1D style, single exposure 16 half-pitch layer very challenging lithography, because its tight tip-to-tip configurations. Therefore, industry considering hybrid use ArFi-based SAQP...
Recently, artificial intelligence technologies and machine learning methods have offered attractive prospects to design manage crisis response processes, especially in suicide management. In other domains, most algorithms are based on big data help diagnose suggest rational treatment options medicine. But psychiatry related behavior clinical evaluation. They more heterogeneous, less objective, incomplete compared fields of Consequently, the use psychiatric may lead accurate sometimes...
One of the many constrains High Numerical Aperture Extreme Ultraviolet Lithography (High NA EUVL) is related to resist thickness. In fact, one consequences moving from current 0.33NA 0.55NA (high NA) Depth Focus (DOF) reduction. addition, as feature lines shrink down 8nm half pitch, it essential limit aspect ratio avoid pattern collapse. The direct consequence such a situation that thickness 30nm, usually used for 32nm pitch dense line/space (LS), will not be suitable 16nm where target...
With the planned introduction of double patterning techniques, focus attention has been on tool overlay performance and whether or not this meets required for patterning. However, as we require tighter performance, impact selected integration strategy plays a key part in determining achievable performance. Very little given at time to example deposition steps, oxidation CMP steps that they have wafer deformation therefore degraded which directly reduces available budget. Also, selecting...
The overlay metrology budget is typically 1/10 of the control resulting in total measurement uncertainty requirements 0.57 nm for most challenging use cases 32nm technology generation. Theoretical considerations show that based on differential signal scatterometry (SCOL<sup>TM</sup>) has inherent advantages, which will allow it to achieve generation and go beyond it. In this work we present results an experimental theoretical study SCOL. We results, comparing with standard imaging metrology....
This paper summarizes findings on the iN7 platform (foundry N5 equivalent) for single exposure EUV (SE EUV) of M1 and M2 BEOL layers. Logic structures within these layers have been measured after litho etch, variability was characterized both with conventional CD-SEM measurements as well Hitachi contouring method. After analyzing patterning layers, impact potential interconnect reliability studied by using MonteCarlo process emulation simulations to determine if current litho/etch...
Customer Relationship Management is an every day task for companies, even the ones dealing with Small Data.We are more interested here by Lead Scoring that refers to practice of calculating and assigning a score leads (business contacts or qualified prospects) company.In this paper, we present one way building scoring model Bayesian network using small amount data. In addition its ability handling uncertainty, networks knowledge representation models can be built from expert knowledge. our...
The integration of a three-layer BEOL process which includes an intermediate 21 nm pitch level, relevant for the 3 technology node, is demonstrated. A full barrier-less Ruthenium (Ru) dual-damascene (DD) metallization allowed to test different dimensions minimum island, via extension and tip-to-tip (T2T). Five-track place route (PNR) SRAM constructions were realized with self-aligned block (SAB) technique. Stacked vias showed resistance modulation size island due change in chamfer. High...