Reconfigurable Logic-in-Memory Oxide Transistors Enabled by Transferable Ferroelectric HZO.

Area of Science:

• Materials Science
• Electrical Engineering
• Nanotechnology

Background:

• Indium oxide (In2O3) shows promise for logic-in-memory (LOM) applications due to high electron mobility and low-temperature processing.
• Direct integration of high-κ ferroelectrics like hafnium zirconium oxide (HZO) with In2O3 degrades interfaces, causing device instability.
• Existing methods face challenges with chemical incompatibilities and interfacial defects.

Purpose of the Study:

• To develop an interface-layer-free integration method for ferroelectric HZO and In2O3 channels.
• To demonstrate the feasibility of using advanced membrane transfer techniques for LOM devices.
• To create stable and high-performance dual-gate ferroelectric In2O3 transistors and reconfigurable logic circuits.

Main Methods:

• Utilized advanced membrane transfer techniques to deposit a ferroelectric HZO layer onto In2O3 channels.
• Formed a van der Waals-like junction with an approximate 0.8 nm interfacial gap, preserving In2O3 stoichiometry.
• Fabricated dual-gate ferroelectric In2O3 transistors and integrated them into inverter circuits.

Main Results:

• Achieved interface-layer-free integration, avoiding chemical incompatibilities and preserving In2O3 channel quality.
• The transferred HZO exhibited a dielectric constant of 26 and low leakage current (<10^-7 A cm^-2 at 1 MV cm^-1).
• Demonstrated dual-gate ferroelectric In2O3 transistors with a large memory window, stable endurance (>10^9 cycles), and reconfigurable NOR/NAND logic functions.

Conclusions:

• The membrane transfer method provides a viable route for integrating ferroelectric HZO with In2O3 for LOM applications.
• This approach overcomes interfacial degradation issues, leading to enhanced device performance and stability.
• The process is compatible with silicon back-end-of-line thermal budgets and scalable for wafer-level integration, paving the way for high-density, multifunctional LOM architectures.