The Ag/ZnMn2O4/p+-Si resistive-switching device reported by Zhang et al. (2018) can be operated in both unipolar and bipolar switching modes within the same cell, but only the bipolar mode is stable under endurance cycling, while the unipolar mode shows large cycle-to-cycle variation of the set and reset processes and quickly degrades. The physical origin of this unipolar-mode non-uniformity — why one switching pathway in a single device is reproducible while the other is not — was left unresolved. Solving it requires identifying the microscopic mechanism (filament chemistry, geometry, and the stochastic thermal/ionic/electrochemical process governing unipolar set and reset) that produces the asymmetry and that accounts for the reported conduction-mechanism assignments.

Background

Resistive random-access memory (RRAM) stores information by reversibly switching a thin oxide between a high-resistance state (HRS) and a low-resistance state (LRS) with voltage pulses. In unipolar resistive switching (URS), the set (transition to LRS) and reset (transition back to HRS) both occur at the same voltage polarity and are distinguished only by the current level allowed during the operation (the compliance current); in bipolar resistive switching (BRS), set and reset require opposite voltage polarities. In transition-metal oxides both modes are usually filamentary: after a one-time high-voltage electroforming step, a nanoscale conductive filament — typically oxygen-vacancy-rich oxide, or deposited metal when an electrochemically active electrode such as Ag is used — bridges the electrodes. URS is commonly attributed to a thermochemical fuse–antifuse mechanism, in which Joule heating ruptures the filament (reset) and a current-limited burst reforms it (set), a process whose outcome depends on the stochastic filament morphology and local temperature; BRS is commonly attributed to interface-type valence-change (migration of oxygen vacancies that alters the resistance of a tip or interface region of the filament) or electrochemical-metallization processes (dissolution and redeposition of a metal such as Ag) that only partially modulate a filament. ZnMn2O4 is a spinel-structured ternary oxide; the source device was made by a sol-gel (chemical-solution) deposition of ZnMn2O4 on a heavily p-doped silicon substrate (p+-Si) with a Ag top electrode, forming a simple metal/oxide/semiconductor stack in which the n-type oxide and p+-Si also form a p–n junction. Prior and adjacent results frame the question. Peng and Wu (2009) showed that ZnMn2O4 itself supports stable unipolar switching with ON/OFF ratios above 10^4 in a different electrode configuration, so the material is not intrinsically incapable of uniform URS. The same group as the source paper had earlier reported BRS in sol-gel ZnMn2O4/p+-Si stacks (2014). Coexistence of both switching modes in a single cell is a documented phenomenon in ZnO (Lee et al. 2010, in an Ag/ZnO/Pt stack very similar in electrode choice), NiO (Choi and Kim 2014), and the spinel ferrite NiFe2O4 (Li et al. 2022); in these systems the electroforming conditions — notably the compliance current — select the mode (high forming current favoring the strong filament of URS, low current favoring BRS), and the two modes can follow distinct carrier-transport mechanisms that can each be stable over many cycles. That selection rule is specific to those systems and should not be assumed to describe how the two modes coexist in the Ag/ZnMn2O4/p+-Si device at issue, where both are reported in the same device. Against this backdrop, Zhang et al. (2018) reported an asymmetry: in their Ag/ZnMn2O4/p+-Si cell the bipolar mode was stable under endurance cycling, whereas the unipolar mode exhibited significant variation of the set and reset processes over only a small number of cycles. The reported electrical specifics (measured under a 30 mA compliance current) were: the sol-gel film was spinel after annealing at 650 °C; the bipolar mode switched at a set of about +8.25 V and a reset of about −9.75 V and held an ON/OFF ratio of roughly 10–10^2 stably over 100 cycles, whereas the unipolar mode switched at a set of about 7.0 V and a reset of about 1.8 V, held a larger ON/OFF ratio of roughly 10^3–10^7, but varied from cycle to cycle and was characterized over only about six cycles. Their electrical analysis assigned, in the URS mode, both HRS and LRS conduction to the filament mechanism, while in the BRS mode the HRS was dominated by space-charge-limited conduction (SCLC — a trap-controlled injection regime in which current grows roughly as voltage squared beyond a trap-filled limit), the LRS was filamentary, and the overall bipolar behavior was ascribed to the space-charge region of the oxide/p+-Si p–n junction. The paper did not resolve why the unipolar pathway in the same cell is non-reproducible.

Problem Statement

Determine the physical origin of the cycle-to-cycle non-uniformity of the unipolar resistive switching mode in the Ag/ZnMn2O4/p+-Si device — the same cell in which the bipolar mode switches stably. Specifically, identify at the microscopic level the filament chemistry (oxygen-vacancy-based versus Ag-related), geometry, and stochastic process — thermal (Joule-heat-driven fuse–antifuse rupture), ionic, or electrochemical — that makes the unipolar set and reset irreproducible over cycles while the bipolar pathway in the identical stack remains reproducible, and do so consistently with the reported conduction-mechanism assignments (both unipolar states filamentary; bipolar HRS dominated by space-charge-limited conduction with a p–n-junction space-charge-region contribution). The explanation should account for the roles of the Ag active electrode, the sol-gel polycrystalline spinel film, and the p+-Si substrate, and should be validated against the observed electrical phenomenology (variability of set/reset voltages and currents versus stable bipolar endurance). What is required is the joint explanation rather than any one of its parts: a resolving answer must commit to a filament chemistry and to the stochastic process governing the unipolar set and reset, state the filament geometry qualitatively, remain consistent with the reported conduction-mechanism assignments, account for the roles of the stated stack components, and explain why the bipolar pathway in the identical stack stays reproducible. Addressing the individual chemistry, geometry, or process questions alone does not resolve the task, and the required level is qualitative — no quantitative structural fitting or new measurement is required.

References

  1. Zhang, Yupei; Wang, Hua; Xu, Jiwen; Li, Zhida; Yang, Ling. Coexistence of Bipolar and Unipolar Resistive Switching Behavior in Ag/ZnMn2O4/p+-Si Device. Journal of Wuhan University of Technology-Mater. Sci. Ed. 33(6), 1433-1436 (2018). DOI: 10.1007/s11595-018-1987-5. https://doi.org/10.1007/s11595-018-1987-5
  2. Peng, Haiyang; Wu, Tom. Nonvolatile resistive switching in spinel ZnMn2O4 and ilmenite ZnMnO3. Applied Physics Letters 95, 152106 (2009). DOI: 10.1063/1.3249630. https://doi.org/10.1063/1.3249630
  3. Xu, Jiwen; Yang, Zupei; Zhang, Yupei; Zhang, Xiaowen; Wang, Hua. Bipolar resistive switching behaviours in ZnMn2O4 film deposited on p+-Si substrate by chemical solution deposition. Bulletin of Materials Science 37, 1657-1661 (2014). DOI: 10.1007/s12034-014-0731-9. https://doi.org/10.1007/s12034-014-0731-9
  4. Lee, Seunghyup; Kim, Heejin; Park, Jinjoo; Yong, Kijung. Coexistence of unipolar and bipolar resistive switching characteristics in ZnO thin films. Journal of Applied Physics 108, 076101 (2010). DOI: 10.1063/1.3489882. https://doi.org/10.1063/1.3489882
  5. Choi, Dooho; Kim, Chang Soo. Coexistence of unipolar and bipolar resistive switching in Pt/NiO/Pt. Applied Physics Letters 104, 193507 (2014). DOI: 10.1063/1.4875918. https://doi.org/10.1063/1.4875918
  6. Li, Jiacheng; Yao, Chuangye; Ke, Yifu; Huang, Wenhua; Thatikonda, Santhosh Kumar; Qin, Ni; Bao, Dinghua. Understanding the coexistence of unipolar and bipolar resistive switching in NiFe2O4 resistive memory devices. Applied Physics Letters 120, 133501 (2022). DOI: 10.1063/5.0086415. https://doi.org/10.1063/5.0086415
  7. Carlos, Emanuel; Branquinho, Rita; Martins, Rodrigo; Kiazadeh, Asal; Fortunato, Elvira. Recent Progress in Solution-Based Metal Oxide Resistive Switching Devices. Advanced Materials 33, 2004328 (2021; published online December 2020). DOI: 10.1002/adma.202004328. https://doi.org/10.1002/adma.202004328
  8. Min, Kyunghwan; Jung, Dongmyung; Kwon, Yongwoo. Investigation of switching uniformity in resistive memory via finite element simulation of conductive-filament formation. Scientific Reports 11, 2447 (2021). DOI: 10.1038/s41598-021-81896-z. https://doi.org/10.1038/s41598-021-81896-z
  9. Liu, Wei-Cheng; Chen, Yibing; Zheng, Ziping; Bao, Xin-Tong; Ma, Zelin; Zhang, Hao-Ran; Pan, Shusheng; Yang, Huiping; Ge, Jun. Electrode-Interface Effect on Low-Resistance Unipolar Ag-Electrochemical Metallization Switching in ZnO Memristors. Physica Status Solidi Rapid Research Letters (2025). DOI: 10.1002/pssr.70250. https://doi.org/10.1002/pssr.70250