Wafer‐Scale 3D Integration for High‐Density Multi‐Valued Neuromorphic Logic
Advanced Science · 2026-11 · 13 (63)
DOI: 10.1002/advs.77437 ↗At a glance
Journal article in Advanced Science, published online on 2026-08-25. Written by Chang-Hyeon Kim and 5 co-authors. Institution: School of Electrical and Electronics Engineering, Chung-Ang University, Seoul, South Korea. 48 references. The full text is freely available (open access, CC BY 4.0). Funded by: National Research Foundation of Korea, Ministry of Trade Industry & Energy, Hansung University.
Abstract (original language)
As conventional transistor scaling approaches fundamental limits, multi‐valued logic (MVL) has emerged as a promising strategy to enhance information density and reduce circuit complexity beyond binary complementary metal‐oxide‐semiconductor (CMOS) technology. Here, we present a monolithic three‐dimensional (3D) vertically stacked MVL architecture based on a tellurium (Te)/indium‐gallium‐zinc‐oxide (IGZO) heterojunction field‐effect transistor (H‐FET) integrated with a crystallinity‐enhanced Te field‐effect transistor (CE‐Te FET). Engineered interfacial band alignment in the H‐FET induces carrier confinement and gate‐tunable electron‐dominated transport, enabling intrinsic ternary switching through controlled current modulation. Complementary channel engineering of the CE‐Te FET via oxidation and crystallinity recovery suppresses off‐state leakage and optimizes transconductance matching, stabilizing the intermediate logic state. Using low‐temperature CMOS‐compatible processes, we realize wafer‐scale vertically stacked ternary circuits exhibiting robust three‐level operation, high uniformity, and long‐term stability. System‐level analysis reveals substantial gains in logic density and area efficiency enabled by H‐FET‐assisted 3D integration. Furthermore, the ternary voltage characteristics are directly mapped to ternary weight neural networks, enabling high‐accuracy handwritten digit classification. This scalable inorganic 3D MVL platform establishes a practical pathway toward high‐density logic and multi‐valued neuromorphic computing architectures beyond conventional binary planar scaling.
Subjects
Author keywords
Authors and institutions
- School of Electrical and Electronics Engineering, Chung-Ang University, Seoul, South Korea.
- Department of Electronics and Information Engineering, Hansung University, Seoul, South Korea.
- Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA.
- School of Electrical and Electronic Engineering, Program in Semiconductor Convergence, Inha University, Incheon, South Korea.
ORCID profiles: Tae In Kim, Hyuck-In Kwon, Ick-Joon Park
Funding
- National Research Foundation of Korea
- Ministry of Trade Industry & Energy
- Hansung University
Cite
APA
Kim, C. H., Kim, M. S., Rhim, J. B., Kim, T. I., Kwon, H. I., & Park, I. J. (2026). Wafer‐Scale 3D Integration for High‐Density Multi‐Valued Neuromorphic Logic. Advanced Science, 13(63). https://doi.org/10.1002/advs.77437
Vancouver
Kim CH, Kim MS, Rhim JB, Kim TI, Kwon HI, Park IJ. Wafer‐Scale 3D Integration for High‐Density Multi‐Valued Neuromorphic Logic. Advanced Science. 2026;13(63). doi:10.1002/advs.77437
BibTeX
@article{Kim2026,
author = {Kim, Chang-Hyeon and Kim, Min Seok and Rhim, Joong Bum and Kim, Tae In and Kwon, Hyuck-In and Park, Ick-Joon},
title = {Wafer‐Scale 3D Integration for High‐Density Multi‐Valued Neuromorphic Logic},
journal = {Advanced Science},
year = {2026},
volume = {13},
number = {63},
doi = {10.1002/advs.77437}
}
EUVIMED