KAIST Develops Low-Voltage High-Density Memory Device Applying Ferroelectric Material

A high-performance, highly integrated next-generation memory device using ferroelectric materials has been developed domestically. Ferroelectric materials are highlighted as key materials in the development of next-generation semiconductor technologies because they facilitate charge storage in memory devices. The next-generation memory device developed by a domestic research team is evaluated to have overcome the limitations of DRAM and NAND Flash memory, the two major pillars of the current memory semiconductor industry, by applying ferroelectric materials.

KAIST announced on the 6th that Professor Jeon Sang-hoon’s research team successfully developed next-generation memory and storage memory technologies using hafnia ferroelectric materials.

Group photo of Professor Jeon Sang-hoon's team. Provided by KAIST

Hafnia ferroelectric is a non-volatile insulating film that exhibits physical properties such as CMOS process compatibility, operating speed, and durability, making it a material actively researched for use as a core material in next-generation semiconductors.

DRAM memory is volatile memory used to store data in everyday devices such as smartphones, computers, and USBs, and loses stored data when external power is cut off. Conversely, it has been widely used as main memory due to its low process cost and high integration density.

However, as the size of DRAM memory devices decreases, the capacitance of the storage capacitor that stores information in the DRAM device also decreases, making memory operation difficult.

To overcome these limitations, the research team developed an ultra-thin high-permittivity material based on hafnia ferroelectric. The focus was on enabling the storage capacitor to achieve high storage capacity even in physically small areas. The developed ultra-thin high-permittivity material allows storage in a layer as thin as 2.4 ? (one ten-thousandth the thickness of a human hair) of SiO2 (silicon oxide) effective thickness, which is the thinnest among previously reported DRAM capacitors.

During the research process, the team also developed ‘FRAM memory,’ a ferroelectric memory considered a potential candidate to replace DRAM memory technology. This memory can store and erase non-volatile information at low voltages below 1V, comparable to current DRAM levels, making it an essential element for next-generation memories that require significantly improved energy efficiency.

The research team also succeeded in developing next-generation memory technology based on hafnia ferroelectric to overcome the limitations of NAND Flash memory. NAND Flash memory is also widely used in smartphones, computers, and USBs and is classified as non-volatile memory for data storage.

The semiconductor industry has advanced NAND Flash memory technology by stacking multiple layers to increase storage capacity. However, recently, physical limitations have made it difficult to stack more than 500 or 1000 layers.

In response, the research team studied applying ferroelectric materials as NAND Flash materials and added a thin TiO2 layer at the material interface, enabling vertical stacking of over 1000 layers while maintaining data stability against external environmental interference.

Additionally, noting that oxide channel-based memory devices in existing NAND Flash technology cannot completely erase data, the team developed a new structure of high-performance oxide channel-based NAND Flash device. This device can store more data than existing devices and maintain data stability for over 10 years.

Professor Jeon Sang-hoon of KAIST said, “We expect this research result to serve as a breakthrough in memory semiconductor technology development, which has been stalled due to scaling issues,” adding, “In the future, it can also contribute to the commercialization of various artificial intelligence computing and edge computing technologies.”

Meanwhile, this research was conducted in collaboration with Samsung Electronics and Hanyang University, supported by the Korea Evaluation Institute of Industrial Technology (KEIT) public-private joint investment project for advanced semiconductor workforce training, the Ministry of Science and ICT’s Innovation Research Center (IRC) support project, and Samsung Electronics.

This content was produced with the assistance of AI translation services.