Currently, 3D NAND research focuses on the Bit-Cost Scalable (BiCS) flash memory technology to integrate the high density memory cells into the 3D NAND chip. However, the charge trap nitride (CTN) based 3D NAND, the main-stream flash memory technology, faces physical limits in improving the cell density. The cell density improvement of the CTN-based 3D NAND is limited largely by two main factors: i) threshold voltage(Vth) shift due to the cell-to-cell interference caused by charge migration and dispersion and ii) spacer oxide thickness limitation due to the high voltage operation.

Ferroelectric 3D NAND can be considered one of the possible alternatives to improve the cell density of the 3D NAND by overcoming the weakness of the CTN-based 3D NAND mentioned above. In particular, the HfO2-based ferroelectric cell is known for maintaining the polarity in the ferroelectric cell numerically down to 3 Å and 2 nm, minimizing the interference between the cells and improving the cell density. As the cells in Ferroelectric 3D NAND changes between program and erase (P/E) states at lower voltages than those in the CTN-based 3D NAND, the required spacer oxide thickness reduces. Thus, it can address the challenges of stack up in conventional CTN-based 3D NAND.

The various electrical properties of the HfO2 complicate its integration into the 3D NAND fabrication processes due to the co-existence of the multiple phases of the HfO2 layer. The composition of the multiple phases of the HfO2 layer is determined by the process conditions. The HfO2 has three major electrical properties: dielectric, anti-ferroelectric, and ferroelectric properties. The property depends on the phases of which formation is influenced by the fabrication conditions, dopant type, and dopant concentration. Research on HfO2-based ferroelectrics is mostly performed based not on 3D NAND but on 2D ferroelectric capacitor or ferroelectric field effect transistor (Fe-FET). Thus, the properties and mass-production possibilities of HfO2 based 3D ferroelectric NAND(3D Fe-NAND) have not yet been verified. Few existing papers propose the 3D Fe-NAND structure but they utilize 3D NAND mimicking vehicles which have lower process integration complexities than the current product, making it difficult to accurately prove the possibility of mass production of the 3D Fe-NAND.

In 2022 IMW, we share the results of 3D FE-NAND on the conventional 3D CTN NAND test vehicle for mass production by substituting the ONO layer with the HfO2 ferroelectric layer. The present research shows that the 3D Fe-NAND is applicable to the next generation 3D NAND by experimentally verifying the cell characteristics with the same level of fabrication complexity and cell density as those of the 3D CTN NAND product. The 3D Fe-NAND MLC operation was verified experimentally for the first time to our knowledge with the P/E window of 3.4 ± 0.29 V and data retention time of over 365 days. We also demonstrated TLC Vth placement, proving the uniform characteristics of the cells and subsequently suggesting the viability of potential mass production capability. Yet, the process still requires further improvement in the P/E window and the data retention for the TLC operation at a product level. Nevertheless, it is a strong candidate for next generation 3D NAND memory.

The publication: Link