The Digital Process Innovation Center

The Digital Process Innovation Center (DIC) researches cutting-edge digital technologies such as machine learning to establish system platforms and promote their deployment at factories. The DIC aims to realize smart factories, i.e., fully autonomous factories incorporating cloud, big-data, IoT, and other advanced technologies. To fabricate 3D BiCS FLASH™ and other cutting-edge flash memories, it is necessary to centralize information management and control manufacturing machines accordingly. To meet the global demand for flash memories, the DIC is also devoted to establishing sophisticated automatic production systems and ensuring in-process quality control while maintaining high productivity.

 

The Digital Process Innovation Center

Technology Highlights

Automatic Production System

Automatic Production System

Automatic Production System

The manufacturing process of the flash memory is extremely complicated. Unlike automobile production that proceeds sequentially, the fabrication of flash memories consists of a complicated combination of processes supported by roughly 5,000 machines of 200 types that are in constant operation. Flash memories being fabricated on silicon wafers are transferred between machines in a clean enclosure called a FOUP (front-opening unified pod). Once a FOUP is loaded into a machine, wafers are processed or measured, and upon completion, the FOUP is transferred to the next machine. To maintain the quality of flash memories, all the machines are subjected to control, inspection, and maintenance. To obtain the maximum performance from and ensure high-accuracy control of all the machines, the DIC deploys an automatic production system using cutting-edge digital technologies.

Quality Control in Manufacturing Processes

Quality Control in Manufacturing Processes

Quality Control in Manufacturing Processes

In the course of manufacturing of flash memories, each machine records the run-time, operating conditions, temperature, humidity, and other parameters as logs. In addition, the results of in-process inspection and measurement and the results of the final electrical characteristics evaluation are saved as digital data. The amount of these data reaches 40 TB per day. The DIC possesses data infrastructure to systematically analyze such big data using machine learning. This makes it possible to predict the quality and characteristics of final products before they are completed and to perform quality control while adjusting and controlling manufacturing processes.

 

Aiming to Realize Digital Innovation

A flash memory foundry generates not only a huge amount of data but also various types of data, including text, numeric data, still images, and video. Since different technologies are necessary to analyze different types of data, the DIC introduces state-of-the-art analysis technologies from Silicon Valley while establishing open innovation programs with Japanese universities and enterprises.

PAGETOP