Fast SET and Slow RESET: Asymmetric Resistive Switching in BaTiO3/Nb:SrTiO3 Epitaxial Heterojunctions

Background

Ferroelectric resistive switching—driven solely by polarization reversal—has attracted intense interest due to its inherent speed and reversibility. It has been demonstrated in a variety of ferroelectric/semiconductor heterojunctions, such as BaTiO₃/Nb:SrTiO₃ and MoS₂/BaTiO₃/SrRuO₃, where the barrier height and width are modulated by the ferroelectric field effect. Reported phenomena include enhanced tunneling electroresistance, bipolar resistive switching combined with negative differential resistance, and optically or electrically controlled electroresistance. Despite these advances, the switching dynamics in BaTiO₃/Nb:SrTiO₃ heterojunctions have not been explored. Here we report an asymmetric resistive switching: a ~10 ns SET under +8 V versus a ~10⁵ ns RESET under –8 V. We attribute this behavior to the interplay between ferroelectric polarization screening by electrons and oxygen vacancies at the interface.

Methods

Commercial (100) 0.7 wt% NSTO substrates were sequentially cleaned in ethanol, acetone, and de‑ionized water, then dried with air. BaTiO₃ (BTO) films (~100 nm) were deposited by pulsed laser deposition (KrF excimer, 248 nm, 25 ns, 300 mJ, 5 Hz) at 700 °C under 1 Pa O₂ for 15 min, followed by 10 min at 1 Pa O₂ and a controlled cool‑down to room temperature at 10 °C/min in vacuum. Au top electrodes (0.04 mm²) were sputtered through a shadow mask; the bottom electrode was indium pressed onto the NSTO substrate. Transport measurements employed a Keithley 2400 sourcemeter, while voltage pulses (10 ns–1 s) were generated by an Agilent 33250A arbitrary waveform generator. Surface morphology, ferroelectricity, and electrostatic potential were characterized by atomic force microscopy (AFM), piezoresponse force microscopy (PFM), and scanning Kelvin probe microscopy (SKPM) using an Oxford AR instrument. PFM phase, amplitude, current, and SKPM images were acquired over a 2 × 2 µm² area after writing with +8 V and –8 V, under a biased conductive tip of 0.5 V. All measurements were performed at room temperature.

Results and Discussion

Figure 1 illustrates the current–voltage (I–V) characteristics of the Au/BTO/NSTO/In stack at small bias (–0.5 to 0.5 V) following pulses of varying amplitude and width. Positive pulses set the device to a low‑resistance state (LRS; ~3 × 10⁴ Ω), while negative pulses return it to a high‑resistance state (HRS; ~3 × 10⁶ Ω), yielding an OFF/ON ratio of 100. The transition is gradual, enabling multi‑level resistance tuning. Time‑resolved measurements reveal that the SET transition saturates with 10 ns pulses above 4 V, whereas the RESET transition requires millisecond‑scale pulses to fully recover the HRS. This four‑order‑of‑magnitude asymmetry underscores the distinct underlying mechanisms.

Figure 2 confirms the ferroelectric nature of the BTO films: AFM shows an atomically flat surface, and PFM hysteresis loops exhibit coercive voltages of ±3.1 V. PFM and SKPM imaging after domain writing demonstrate that the local current is higher when the polarization points toward the Nb:SrTiO₃ substrate (positive bias) and lower when it points away (negative bias). No filamentary conduction is observed, indicating a purely field‑effect mechanism. The observed switching speeds cannot be explained by pure polarization reversal (which would be ~10 ns for both SET and RESET) nor by oxygen‑vacancy drift alone, suggesting a combined electron–vacancy screening process.

We propose that during SET (positive bias), electrons rapidly screen the positive bound charge at the BTO/NSTO interface, lowering the Schottky barrier and enabling a fast transition. During RESET (negative bias), oxygen vacancies must migrate to screen the negative bound charge; their slower mobility imposes a longer timescale. This model accounts for the observed asymmetry and aligns with prior observations of electron‑vacancy interfacial screening in related heterostructures.

Conclusions

Asymmetric resistive switching has been demonstrated in BaTiO₃/Nb:SrTiO₃ heterojunctions, with SET occurring in ~10 ns and RESET in ~10⁵ ns. The disparity originates from electron‑mediated screening during SET and oxygen‑vacancy‑mediated screening during RESET at the BTO/NSTO interface. This fast‑write/slow‑erase behavior is promising for applications requiring rapid programming and long‑term retention.

Abbreviations

BTO:

BaTiO₃

HRS:

High‑resistance state

LRS:

Low‑resistance state

NSTO:

Nb:SrTiO₃

PFM:

Piezoresponse force microscopy

SKPM:

Scanning Kelvin probe microscopy