DTCO breaks through the limitations of traditional one-way processes, not only covering the joint optimization of unit layout, but also developing a variety of methods and tools including design for manufacturing (DFM) and design for test (DFT), realizing the deep integration of the design end and the manufacturing end. Today, DTCO has become one of the core technologies in chip manufacturing and is widely used by advanced chip manufacturers such as TSMC and Samsung to accelerate process upgrades and improve the yield of new products.
Adhering to the original intention of the HPO concept, Oriental Jingyuan has been working hard to deepen its roots in the fields of computational lithography and chip manufacturing volume detection over the past 11 years since its establishment, and has gradually built a HPO strategic product series.
DMC (Design Manufacturability Check): It enables rapid manufacturability checks of design layout lithography results and builds a bridge between back-end physical design tools and computational lithography tools.
· PHD (Patterning Hotspot Detection): realizes comprehensive simulation detection of bad spots in the patterning process of the mask plate and improves the accuracy of mask manufacturing;
ODAS (Offiline Data Analysis System): enables the direct creation of CD-SEM recipes using design layouts, improving CD-SEM measurement efficiency;
PME (Process Margin Explorer): enables classification and grouping of DRSEM results based on the design layout, significantly improving the effectiveness of DRSEM results;
YieldBook : Integrates design data and chip manufacturing yield-related data on the same platform, breaking down the strict barriers between design data and manufacturing data, and providing a data foundation for HPO overall yield optimization.
HPO2.0 Strategic Planning: AI-driven Intelligent Upgrade
In recent years, with the advancement of AI technology, especially the rapid development of domestic AI models, it has become possible to further realize more efficient and intelligent HPO solutions. At the same time, due to the company's forward-looking layout in the underlying architecture, starting from Day One, Oriental Jingyuan's computational lithography product PanGen has been based on the hybrid supercomputing architecture of CPU+GPU and CUDA development since the early stage of research and development. It can run in the same computing environment as the AI framework, and can naturally be highly compatible with the current technological dividends of AI technology's rapid development.
Against this background, Oriental Jingyuan officially launched the HPO2.0 product planning and gradually carried out related development work. The HPO2.0 plan will integrate AI functions into Oriental Jingyuan's existing products, and use AI to create new products and applications, covering computational lithography, yield equipment and yield management software products, aiming to promote the progress of China's integrated circuit design and manufacturing through integrated solutions.
In the field of computational lithography: We will use the introduction of AI-related capabilities to improve the accuracy of the current optical proximity correction (OPC) model, and carry out etching modeling, complete the curve mask reverse lithography optimization process, and also need to deploy the layout of the whole chip fuzzy matching. With the assistance of AI, the improvement of layout-related basic capabilities provides technical support for the in-depth development of Oriental Jingyuan's computational lithography products, which will help further consolidate Oriental Jingyuan's leading position in domestic computational lithography products.
In terms of yield equipment products: We will introduce AI capabilities for all of Oriental Jingyuan's yield equipment products, and use AI to assist Oriental Jingyuan's yield equipment products in achieving control optimization, fault analysis, image processing, data mining and other functions, so that Oriental Jingyuan's yield equipment products will gain unique advantages in AI applications and promote the upgrade of yield equipment from "data collection and analysis" to "intelligent decision-making."
Fab yield data analysis products: We will use AI technology to create a series of yield analysis AI tools, such as an integrated process modeling tool that combines software and hardware, and establish a full-chain data platform from design layout to chip yield. On this data platform, we will establish a variety of AI process models and AI yield models to achieve an intelligent detection solution that combines software and hardware.
