Controlled Peroxide Treatment Enhances Low‑Power Switching in Zinc Peroxide‑Based Programmable Metallization Cells

Abstract

This study investigates how peroxide surface treatment modifies the resistive switching behavior of zinc peroxide (ZnO₂)-based programmable metallization cells (PMCs). Peroxide exposure converts the hexagonal ZnO lattice into a cubic ZnO₂ phase, but excessive exposure leads to crystalline breakdown. Devices incorporating ZnO₂ (Cu/ZnO₂/ZnO/ITO) switch at markedly lower currents than control Cu/ZnO/ITO devices. However, prolonged peroxide treatment degrades switching stability, likely due to microstructural changes in ZnO₂. Fine tuning of peroxide processing is therefore essential for optimal PMC performance.

Background

Resistive random‑access memory (RRAM) is emerging as a key technology for next‑generation data storage, overcoming the physical limits of volatile DRAM and flash memory. The programmable metallization cell (PMC), a subclass of RRAM, offers high scalability and straightforward fabrication. Zinc oxide (ZnO) is a popular candidate for PMC stacks because of its low cost, chemical stability, and wide bandgap (~3.3 eV). Yet, ZnO‑based PMCs typically require high operation currents due to their intrinsic n‑type conductivity. Various strategies—dopant incorporation, growth control, buffer layers, nanorod insertion, and material stacking—have been explored to reduce the switching current, but these often involve complex processing steps.

Recent work demonstrated that inserting a zinc peroxide (ZnO₂) layer yields both volatile and non‑volatile switching. Peroxide treatment converts the ZnO surface from a hexagonal to a cubic phase, increasing resistivity and enabling Schottky contacts or photodiode functions. The impact of peroxide‑induced phase transformation on memory switching, however, remains underexplored. This paper addresses that gap by systematically studying the effect of peroxide treatment time on the structure and switching characteristics of ZnO₂-based PMCs.

Methods

ZnO thin films were deposited on commercial ITO/glass substrates (Uni‑Onward Corp.). The films were immersed in 30 % H₂O₂ at 100 °C for 1, 3, and 9 minutes, then rinsed with deionized water and dried under N₂. Surface‑treated samples are denoted ST_x (x = 1, 3, 9). A non‑treated film (NT) served as control. Cu top electrodes (150 µm diameter) were sputtered through a shadow mask to form Cu/ZnO/ITO sandwiches. Transmission electron microscopy (TEM, JEOL 2100FX) assessed crystal structure and morphology, while an Agilent B1500 semiconductor analyzer measured electrical characteristics.

Results and Discussion

Cross‑sectional TEM of the NT film (Fig. 1a) revealed a perpendicular ZnO growth with hexagonal wurtzite structure (JCPDS#36‑1451). Peroxide treatment for 1 minute (ST1) produced a bilayer: an upper polycrystalline cubic ZnO₂ (JCPDS#77‑2414) and a remaining ZnO layer (Fig. 1d–f). Extending treatment to 3 minutes (ST3) oxidized deeper regions, yielding a mixture of crystalline and amorphous ZnO₂ (Fig. 1g–i). At 9 minutes (ST9), the entire resistive layer became amorphous ZnO₂ with dispersed nanocrystals (Fig. 1j–l). The progression from hexagonal to cubic and eventually to amorphous phases indicates that excess peroxide destabilizes the lattice, likely due to high‑energy oxygen radicals.

Electrical measurements (Fig. 2c–f) showed that pristine resistance increased by 5–6 orders of magnitude when ZnO₂ was present. The ST9 device exhibited a slight resistance drop relative to ST3, consistent with the formation of an amorphous, lower‑resistivity ZnO₂ matrix. Notably, the forming current compliance dropped from 100 mA (control) to 35 µA (ST9), demonstrating the high‑resistivity layer’s ability to trigger switching at low current.

All devices displayed bipolar, counter‑clockwise I–V hysteresis. The reset voltage remained fixed at –2 V, while the set voltage varied slightly with treatment time (Fig. 3a–d). The coefficient of variation in V_set increased with longer peroxide exposure, indicating that microstructural irregularities broaden the set‑voltage distribution. Endurance tests (Fig. 3f–i) revealed that the control device maintained an ON/OFF ratio of ~13 over 10⁴ cycles, whereas ST1 achieved a ~46× window at only 200 µA compliance. ST3 and ST9 operated at 100 µA and 35 µA respectively, but exhibited progressively degraded uniformity due to branching conducting filaments caused by random grain orientations and amorphous regions.

Retention measurements (Fig. 3j) confirmed non‑volatility for all samples over 7000 s at room temperature. The observed instability in ST3 and ST9 is attributed to multi‑branch filament formation, driven by the high density of grain boundaries or the random distribution of ZnO₂ nanocrystals.

Conclusion

Peroxide‑treated ZnO films convert to high‑resistivity ZnO₂, enabling low‑current forming and switching in PMC devices. A modest 1‑minute treatment (ST1) balances low operation current (200 µA) with reliable endurance and large memory window. Excessive peroxide exposure, however, introduces structural irregularities that compromise switching stability. Thus, precise control of peroxide processing offers a simple, scalable route to low‑power, high‑density RRAM architectures suitable for future non‑volatile memory applications.