Mr. Wu YANG
Director, Technical Programs - Tessent Silicon Lifecycle Solutions, Siemens EDA
Wu Yang is the Director of product marketing at Siemens EDA. With 27 years of experiences in broader DFT, 3D IC testing, silicon learning, failure analysis and reliability, He has been a frequent speaker at many conferences and a regular contributor to papers, articles, and public speeches. He delivered multiple tutorials at ITC, ATS, ETS and VTS as well. He received a MS degree in electrical engineering from Portland State University.
Presentation Title
Industrial Trend: AI Empowered Silicon Lifecycle Management (SLM)
The semiconductor industry, driven by innovation and increasing complexity, necessitates a re-evaluation of traditional silicon lifecycle management (SLM) approaches. An integrated, intelligent, and AI-driven SLM strategy, leveraging platforms like Siemens AI FUSE, is critical for navigating the entire product journey, from initial design-for-test (DFT) and production, to in-field operation and post-mortem analysis. This talk explores key trends and advancements, highlighting a clear shift towards AI/ML-powered, data-driven methodologies that promise enhanced quality and reliability across the SLM cycle.
The foundational phase of SLM – design – is undergoing significant transformation through AI integration. The pursuit of near-zero defects per million (DPPM) at reduced test costs drives the adoption of sophisticated DFT techniques, including defect-oriented fault modeling. AI-intelligent solutions like Siemens EDA Tessent SSN (Streaming Scan Network) revolutionize the design phase by enabling highly efficient and comprehensive test at reduced cost. Furthermore, embedded Intellectual Properties (IPs) for comprehensive in-life monitoring are becoming standard practice. These embedded monitors provide invaluable real-time data throughout the chip's operational lifespan, which can be analyzed by generic AI engines for anomaly detection and proactive intervention.
As silicon moves from design to learning, encompassing bring-up, production ramp, and customer return analysis, the industry sees accelerated adoption of advanced scan and volume diagnosis-driven yield analysis. Root Cause Deconvolution (RCD) is a powerful ML engine embedded inside Siemens EDA YieldInsight that performs unsupervised learning on volume diagnosis data, identifying defect root causes with superior performance, resolution, and accuracy. This enables more effective failure and yield analysis. This AI-driven approach to yield learning is crucial for maintaining competitiveness in an industry characterized by tight margins and rapid technological cycles.
Finally, the in-field stage of SLM focuses on guaranteeing device safety, security, and reliability, especially for mission-critical applications. Robust Functional Safety (FuSa) implementations are paramount. In-system deterministic test achieves much higher test coverage than traditional pseudo-random built-in-self tests. Analytics of in-field operational data provide insights and root cause metrics for actions and decision-making. This also feeds back to the design stage, creating a continuous improvement loop. Together with function, health, and operation monitors, continuous in-system testing and monitoring reinforce reliability, validating device performance throughout its operational life. This comprehensive in-field management is essential for building trust and ensuring the long-term viability of semiconductor products.
In conclusion, the semiconductor industry is collectively moving towards a more proactive and AI-driven methodology for managing silicon throughout its entire lifespan. This paradigm shift, powered by advancements in DFT, silicon learning, embedded technology, and AI, promises to enhance product quality, accelerate learning cycles, and bolster the reliability and security of semiconductor devices. Embracing these integrated, AI-enabled SLM strategies is not merely an option but a necessity for sustained success in the modern semiconductor landscape.