Skip to content
DraconiticResearch
Search Sources API
203,203 studies31 sourcesUpdated 17 min ago
PreprintCC BYChemRxiv

Decoupling Thermal Kinetics and Structural Evolution in HfO2 RRAM via Broadband Impedance Spectroscopy: A SPICE Model for Analog Applications

Abstract

The integration of HfO2-based resistive random-access memory (RRAM) into analog circuits and neuromorphic hardware requires accurate, physics-based small-signal models. In this work, we employ broadband impedance spectroscopy to physically isolate the distinct structural and kinetic components governing the alternating current (AC) response of TiN/Ti/HfO2/W memristive devices, thereby providing such a model. Measurements were performed over a wide temperature range (77 K to 350 K), across multiple device areas, and at various resistance states controlled by compliance current. The AC response reveals distinct relaxation processes, which are accurately described by an electrical equivalent circuit that isolates the individual electrical signatures of the filament constriction, the interfacial boundary, and the bulk conductive path. The low-frequency relaxation, attributed to the filament constriction, exhibits a strong exponential temperature dependence, allowing for the extraction of the thermal activation energy associated with trap-assisted charge transport within the oxygen vacancy-rich gap. In contrast, the bulk-related response is temperature-independent but strongly depends on the resistive state. The model is further validated by the scaling of the high‑frequency pole with device area, confirming its geometric origin. As a result, our model decouples the thermal kinetics from the structural evolution of conductive filaments in emerging non-volatile memories. Finally, to demonstrate the practical utility of this model, it is implemented in LTSpice to design all-memristive analog filters and a neuromorphic signal mixer. This approach takes advantage of the properties of both pristine and programmed devices to eliminate the need for conventional passive components.

Related

  1. Orthogonal phase modulation and Lissajous mode decoupling in light-induced thermoelastic spectroscopy for real-time multi-component gas sensing
    Hanxu Ma, Shunda Qiao, Ying He +2
    Reports on Progress in PhysicsPeer-reviewedLicence not statedFull text28 citedPhysical chemistryOpenAlex
  2. Conformational Analysis of Methoxy- and Methylthio-Substituted Thiophenes by Broadband Rotational Spectroscopy and Computational Chemistry.
    Timothy J Barnum, Nathan Riordon, Lukas Baker +1
    The journal of physical chemistry. APeer-reviewedCC BYFull textPhysical chemistryEurope PMCPMC
  3. Model-free Reconstruction of Molecular Energy Levels by Broadband Kilohertz-accurate Cavity-enhanced Spectroscopy
    S. Vasilchenko, A. -W. Liu, C. -X. Zuo +6
    arXivPreprintarXiv licenceFull textPhysical chemistryAtomic, molecular & optical physicsarXiv
  4. Hydrothermally synthesised Gd2Te4O11 nanorods for broadband visible light photodetection and NLO applications
    Manoranjan Sahoo, Prabhukrupa Chinmay Kumar, Ramakanta Naik
    Nanoscale AdvancesPeer-reviewedCC BY-NC 3.0Full textMaterials & polymersPMC
  5. Phase evolution and broadband electromagnetic wave absorption mechanisms of electrospun polymer-derived SiC-based fibrous ceramic membranes
    Gaoang Huang, Defang Zu, Xuewen Jiang +9
    Journal of Advanced CeramicsPeer-reviewedCC BYFull text8 citedMaterials & polymersOpenAlex
BibTeX
@misc{duenas2026decoupling,
  title = {{Decoupling Thermal Kinetics and Structural Evolution in HfO2 RRAM via Broadband Impedance Spectroscopy: A SPICE Model for Analog Applications}},
  author = {Salvador Dueñas and Guillermo Vinuesa and Héctor García and Helena Castán},
  howpublished = {ChemRxiv},
  year = {2026},
  month = oct,
  doi = {10.26434/chemrxiv.15009980/v1},
  url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15009980/v1},
  note = {CC BY},
}
RIS
TY  - UNPB
TI  - Decoupling Thermal Kinetics and Structural Evolution in HfO2 RRAM via Broadband Impedance Spectroscopy: A SPICE Model for Analog Applications
AU  - Salvador Dueñas
AU  - Guillermo Vinuesa
AU  - Héctor García
AU  - Helena Castán
PY  - 2026
DA  - 2026/10/06
PB  - ChemRxiv
DO  - 10.26434/chemrxiv.15009980/v1
UR  - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15009980/v1
AB  - The integration of HfO2-based resistive random-access memory (RRAM) into analog circuits and neuromorphic hardware requires accurate, physics-based small-signal models. In this work, we employ broadband impedance spectroscopy to physically isolate the distinct structural and kinetic components governing the alternating current (AC) response of TiN/Ti/HfO2/W memristive devices, thereby providing such a model. Measurements were performed over a wide temperature range (77 K to 350 K), across multiple device areas, and at various resistance states controlled by compliance current. The AC response reveals distinct relaxation processes, which are accurately described by an electrical equivalent circuit that isolates the individual electrical signatures of the filament constriction, the interfacial boundary, and the bulk conductive path. The low-frequency relaxation, attributed to the filament constriction, exhibits a strong exponential temperature dependence, allowing for the extraction of the thermal activation energy associated with trap-assisted charge transport within the oxygen vacancy-rich gap. In contrast, the bulk-related response is temperature-independent but strongly depends on the resistive state. The model is further validated by the scaling of the high‑frequency pole with device area, confirming its geometric origin. As a result, our model decouples the thermal kinetics from the structural evolution of conductive filaments in emerging non-volatile memories. Finally, to demonstrate the practical utility of this model, it is implemented in LTSpice to design all-memristive analog filters and a neuromorphic signal mixer. This approach takes advantage of the properties of both pristine and programmed devices to eliminate the need for conventional passive components.
ER  -