Ultra-low Energy Consuming Memtransistors by Nanoparticle Decoration on Metal Oxide Thin Films
- 주제(키워드) neuromorphic device , synaptic device , synaptic plasticity , low-energy consumption
- 발행기관 아주대학교 일반대학원
- 지도교수 박성준
- 발행년도 2025
- 학위수여년월 2025. 8
- 학위명 석사
- 학과 및 전공 일반대학원 지능형반도체공학과
- 실제URI http://www.dcollection.net/handler/ajou/000000035040
- 본문언어 영어
- 저작권 아주대학교 논문은 저작권에 의해 보호받습니다.
초록/요약
A nanoparticle (NP)-engineered electrolyte-gated metal oxide synaptic transistor (EGMT) is demonstrated to simultaneously achieve ultralow-energy consumption and wide dynamic range (DR), addressing a key trade-off in neuromorphic hardware. The device integrates a solution- processed IGZO channel with a polymer electrolyte (PEO/LiAsF6), where interfacial aluminum NPs function as ion trap sites to stabilize multilevel conductance states. Operated at a drain voltage of 1 mV, the EGMT exhibits a wide DR over 78 across 50 conductance states with an energy consumption of 0.62 pJ per spike. Neural network simulations using a 784–60–10 architecture, based on experimentally measured conductance trajectories, reveal approximately 99.7% and 91.4% reductions in training and inference energy, respectively, compared to conventional CMOS systems. These results establish the proposed EGMT platform as a highly efficient and scalable building block for future neuromorphic systems requiring precise analog modulation under strict energy constraints.
more목차
1. Background 1
1.1 Neuromorphic electronics 1
1.2 Research trend of synaptic devices 2
1.3 From single devices to large-scale arrays 4
1.4 Selection of electrolyte and channel materials 6
1.5 Type of artificial neural network 8
2. Introduction 10
2.1 Importance of energy-efficiency and dynamic range (DR) in neuromorphic systems 10
2.2 Promise of electrolyte-gated memtransistors 12
2.3 Recent advances and limitations in EGMTs 15
2.4 Introducing energy-efficient EGMTs with wide DR 16
3. Results and discussion 19
3.1 Design and electrical characteristics of synaptic transistors with Al NPs 19
3.2 Surface characterizations of Al NP-decorated IGZO films 23
3.3 Synaptic functionalities of NP-engineered EGMT with low energy consumption 27
3.4 Long-term reliable operation of scalable EGMTs 30
3.5 Neuromorphic computing 33
4. Conclusions 39
5. Experimental section 40
5.1 Materials 40
5.2 IGZO precursor solution synthesis 40
5.3 Polymer-electrolyte synthesis 40
5.4 Fabrication of EGMTs with Al NPs 40
5.5 Electrical characterization 41
5.6 Simulation for neuromorphic computing 41
6. References 47
