Virtual fabrication modeling platforms may optimize designs that 2D design rule checks are no longer sufficient to achieve.
Author: Joseph Ervin Director, Semiconductor Process and Integration Lam Research
Advanced CMOS scaling and new memory technologies have introduced increasingly complex structures into the device manufacturing process. For example, the increase in NAND memory layers has achieved greater vertical NAND scaling and higher memory density, but has led to challenges in high aspect ratio etch patterning and foot print scaling issues. Unique integration and patterning schemes have been employed to solve these scaling challenges, but they create additional design rule challenges. Two-dimensional (2D) design rule checks (DRCs) are no longer sufficient to achieve performance and yield goals, due to the 3D nature of modern semiconductor devices. Design of Experiments (DOEs) for process characterization and optimization, traditionally used to save time and cost in developing process recipes, now require hundreds of physical experiments involving significant off-process time and substantial wafer testing. Moreover, non-intuitive interactions among process steps, as well as tightening process windows, have made it difficult to deliver concurrent performance and yield optimization using first principle modeling approaches. A 3D understanding of complex process sequences is required to solve these scaling challenges, and is provided by Coventor SEMulator3D, a virtual fabrication modeling platform.
Click here to continue reading.
While terms often are used interchangeably, they are very different technologies with different challenges.
Commercial chiplet marketplaces are still on the distant horizon, but companies are getting an early start with more limited partnerships.
The industry is gaining ground in understanding how aging affects reliability, but more variables make it harder to fix.
Key pivot and innovation points in semiconductor manufacturing.
Tools become more specific for Si/SiGe stacks, 3D NAND, and bonded wafer pairs.
Less precision equals lower power, but standards are required to make this work.
While terms often are used interchangeably, they are very different technologies with different challenges.
Open-source processor cores are beginning to show up in heterogeneous SoCs and packages.
Commercial chiplet marketplaces are still on the distant horizon, but companies are getting an early start with more limited partnerships.
New applications require a deep understanding of the tradeoffs for different types of DRAM.
Open source by itself doesn’t guarantee security. It still comes down to the fundamentals of design.
How customization, complexity, and geopolitical tensions are upending the global status quo.
127 startups raise $2.6B; data center connectivity, quantum computing, and batteries draw big funding.
Leave a Reply