Abstract
The growing demand for fast, energy-efficient, and scalable non-volatile memories (NVMs) is driven by data-intensive applications and edge computing. Among emerging solutions, ferroelectric memories based on hafnium-zirconium-oxide (HZO) materials are gaining attention due to low switching power, high speed, and CMOS compatibility. Notably, ferroelectric diodes stand out for their simple two-electrode structure, which eases integration into high-density 3D architectures. However, device scaling is increasingly limited by copper interconnect challenges. As the pitch and diameter of copper interconnects shrink below 10 nm, parasitic resistance rises, degrading performance. To address this, we present monolithic 3D integration of a vertical ferroelectric diode using highly aligned single-walled carbon nanotubes (SWCNTs) as nanoscale edge electrodes and interconnects. The unique interface between crystalline carbon and the Hf0.5Zr0.5O2 ferroelectric layer enables polarization-controlled Schottky-to-ohmic transition, yielding intrinsic diode-like behavior with self-selectivity, eliminating the need for external selector devices. This ferroelectric diode shows exceptional performance with ultra-low programming energy of 0.86 fJ, a high nonlinearity factor of approximately 2070, retention exceeding 104 s, and endurance lasting over 106 cycles. Its large memory window supports varying output current levels and reliable multilevel switching. Crucially, it maintains stable operation at extreme nanoscale, proving HZO-based ferroelectric devices can retain high performance with nanoscale edge electrodes. Given the superior scalability of single-crystalline carbon nanotube interconnects compared to conventional polycrystalline metal ones, this unique configuration paves the way for ultra-scaled, low-power 3D memory architectures meeting modern data-intensive computing demands.
| Original language | English |
|---|---|
| Article number | 121108 |
| Journal | Carbon |
| Volume | 248 |
| DOIs | |
| Publication status | Published - 5 Feb 2026 |
Bibliographical note
Publisher Copyright:© 2025 Elsevier Ltd.
Keywords
- Energy-efficient electronics
- Ferroelectric diode
- Low power device
- Non-volatile memory
- Self-selective
- Single-walled carbon nanotubes
Fingerprint
Dive into the research topics of '3D vertical ferroelectric memory with wafer-scale aligned carbon nanotube edge-electrodes'. Together they form a unique fingerprint.Press/Media
Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver