The effects of inserted layers (HfO2, ZrO2, Y2O3, La2O3) on the ferroelectric and synaptic properties of Zr-doped HfO2 sandwich structure
- Authors
- Chung, Chulwon; Park, Kyungsoo; Yun, Seunghyeon; Park, Junhyeok; Jeong, Hyeon Cheol; Choi, Changhwan
- Issue Date
- Nov-2025
- Publisher
- Elsevier BV
- Keywords
- Ferroelectric thin films; Insertion layer materials; Artificial synapse device; FeFETs
- Citation
- Applied Surface Science, v.710, pp 1 - 11
- Pages
- 11
- Indexed
- SCIE
SCOPUS
- Journal Title
- Applied Surface Science
- Volume
- 710
- Start Page
- 1
- End Page
- 11
- URI
- https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/208610
- DOI
- 10.1016/j.apsusc.2025.163918
- ISSN
- 0169-4332
1873-5584
- Abstract
- HfO2-based ferroelectrics (FEs) have gained significant attention as synaptic devices due to multistate capability enabled by multi-domain FEs and nonvolatile characteristics. However, when these FE thin films are applied to synaptic devices, the polarization switching often results in non-linear behavior during the potentiation and depression of synaptic weights. In this study, we introduced insertion layer structures to induce a depolarization field, effectively suppressing abrupt polarization switching and increasing the coercive field (Ec) to enhance synaptic device performance. The study investigated the behavior of Zr-doped HfO2 (HZO) FE properties using different insertion layers (HfO2, ZrO2, Y2O3, La2O3). The FE performance was further evaluated for synaptic devices, including the linearity of remnant polarization (Pr) under varying electric field, endurance characteristics, and cycle-to-cycle variation. Among the materials tested, HfO2 as the insertion layer yielded the most promising results, showing superior endurance (2x106 cycle) and minimal cycle-to-cycle variation (14.3%) due to its lower oxygen vacancy levels (6.6%). The HfO2-inserted FeFET device demonstrated improved synaptic weight control, providing a more states and enhanced linearity in potentiation and depression compared to FeFET without insertion layer. Finally, pattern recognition simulations using the MNIST dataset revealed an accuracy of 87.76%, which represents an improvement of 3.75% over reference.
- Files in This Item
-
Go to Link
- Appears in
Collections - 서울 공과대학 > 서울 신소재공학부 > 1. Journal Articles

Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.