Thickness‐Defined Transport Regime Toward Near‐Ideal Area‐Type Switching
ABSTRACT Area‐type memristors are attractive for analog in‐memory computing, but their use is often limited by parasitic charge storage, causing readout drift, non‐pinched hysteresis, and degraded readout fidelity. We show that near‐ideal area‐type switching emerges within a tunneling‐governed transport regime in ultrathin layers that suppresses charging‐related readout artifacts. By varying the thickness in TiN/Ta 2 O 5 /Pt in a range of 3–10 nm, we reveal the transition from filamentary switching to near‐short‐circuit transport with two gradual‐switching modes with different readout stability. The 5–6 nm thickness window suppresses non‐pinched hysteresis and short‐time drift while preserving analog switching, yielding nearly polarity‐symmetric hysteresis and readout, clear area scaling, and continuous near‐linear conductance modulation. Our analysis indicates that trap‐assisted tunneling dominates the stabilized‐state response, whereas ionic redistribution governs pristine‐state reconstruction and the stability boundaries. A compact model is derived for the 5 nm device which reproduces the quasi‐static I–V behavior and the dynamic resistance evolution based on pulse‐width‐dependent characteristics. The verified model is provided as Verilog‐A code enabling simulation in circuit design tools. Our results demonstrate that ultrathin scaling is a regime‐defining design rule for near‐ideal area‐type operation in oxide memristors, implying a fundamental approach to fabricating robust multi‐state memristors for in‐memory computing applications.
Authors
- Thomas Mikolajick (ORCID: https://orcid.org/0000-0003-3814-0378)
- Ronald Tetzlaff (ORCID: https://orcid.org/0000-0001-7436-0103)
- Gianaurelio Cuniberti (ORCID: https://orcid.org/0000-0002-6574-7848)
- Leif Riemenschneider (ORCID: https://orcid.org/0000-0002-0254-9174)
- Richard Schroedter (ORCID: https://orcid.org/0000-0003-3259-4571)
- Stefan Slesazeck (ORCID: https://orcid.org/0000-0002-0414-0321)
- Benjamin Max (ORCID: https://orcid.org/0000-0002-6468-6596)
- Cuo Wu (ORCID: https://orcid.org/0000-0001-9198-9173)
- Hao Shen
- Abdallh Herbawi (ORCID: https://orcid.org/0009-0000-8695-3272)
- Jason Vincent
- Junjie Zhao (ORCID: https://orcid.org/0009-0001-7081-9105)
- Hampus Hoffman
Institutions
- NaMLab (Germany) (DE)
- Technische Universität Dresden (DE)
- RWTH Aachen University (DE)
- FH Aachen (DE)
Publication Details
- Journal
- Advanced Materials
- Published
- 2026-10-14
- DOI
- https://doi.org/10.1002/adma.75393
- Primary Topic
- Advanced Memory and Neural Computing
- Type
- article
- Field-Weighted Citation Impact
- 0.00