The electrode is one of the key factors that influences and controls the resistive switching characteristic of a resistive switching device. In this work, we investigated the write-once-read-many-times (WORM)-resistive switching behavior of BiFeO3 (BFO)-based devices with different top electrodes, including Pt, Ag, Cu, and Al. The WORM-resistive switching behavior has been observed in Pt/BFO/LaNiO3 (LNO), Ag/BFO/LNO, and Cu/BFO/LNO devices. In the initial high resistance state, the Pt/BFO/LNO device shows space-charge-limited current conduction due to the large Schottky barrier height at the Pt/BFO interface, while the Ag/BFO/LNO and Cu/BFO/LNO devices exhibit Schottky emission conduction due to the small barrier height at both top electrode/BFO and BFO/LNO interfaces. In the low resistance state, the metallic conduction of the Pt/BFO/LNO device is a result of the formation of conduction filaments composed of oxygen vacancies, and yet the metallic conduction of Ag/BFO/LNO and Cu/BFO/LNO devices is due to the formation of oxygen vacancies-incorporated metal conduction filaments (Ag and Cu, respectively). The observed hysteresis I-V curve of the Al/BFO/LNO device may be attributed to oxygen vacancies and defects caused by the formation of Al–O bond near the Al/BFO interface. Our results indicate that controlling an electrode is a prominent and feasible way to modulate the performance of resistive switching devices.
BiFeO3 (BFO), a room-temperature multiferroic material, has great application potential in multiple areas such as multiferroic devices,1,2 photovoltaic devices,3,4 magnetoelectric coupling devices,5,6 resistive switching devices,7,8 and photocatalysis applications.9,10 The advantages of BFO-based resistive switching devices have been reported in recent works, and various factors that affect the resistive switching behaviors have been investigated. Due to ferroelectric properties, the poling process of BFO increases the activated trap levels, resulting in a high resistance switching ratio and the increased transition voltage from Ohmic to space-charge-limited behavior.11 Besides, the charged domain wall plays an essential role in forming the conductive paths and modifies the resistive switching characteristic of BFO.12,13 Other external parameters, such as light illumination, magnetic field, and doping, can also influence the resistive switching behavior of BFO. Light illumination affects the ON/OFF ratio or set voltage (Vset) of BFO, which results from photon-induced charge carriers.14,15 Due to the magnetoelectric effect and Lorentz force when applying a magnetic field, the Vset and resistance of BFO are modulated by the magnetic field.16 As a commonly used way, elemental doping can control the oxygen vacancies and lattice symmetry of BFO; hence, the resistive switching characteristic is affected effectively by the doping element such as Mn, La, etc.17,18
In addition, the electrode is a significant factor that impacts the resistive switching characteristic of a resistive switching device. Recent works reported that the exploited electrode materials, such as Ag, Pt, indium tin oxide, TiInSnO, Nb–SrTiO3, La2/3Sr1/3MnO3, SrRuO3, etc., strongly influence ion migration, conduction filament, and the Schottky barrier, resulting in the optimization of the bipolar resistive switching behavior of BFO-based devices.19–21 The write-once-read-many-times (WORM) resistive switching behavior of BFO has been investigated recently.22,23 However, the effect of the electrode on the WORM resistive switching behavior of BFO has not been reported yet. In this work, we investigated the WORM resistive switching behavior of BFO-based devices with different top electrodes. The BFO-based devices with Pt, Ag, Cu, and Al as top electrodes were prepared. The Pt/BFO/LaNiO3 (LNO), Ag/BFO/LNO, and Cu/BFO/LNO devices show WORM-resistive switching behavior, while the Al/BFO/LNO device exhibits a hysteresis behavior of the current-voltage (I-V) curve. The different electrical behaviors of the devices with different top electrodes are attributed to the variation in the Schottky barrier at the top electrode/BFO interfaces and the formation of different conduction filaments.
In this work, we use an LNO thin film as the bottom electrode. The LNO thin films were prepared on SrTiO3 (STO) single crystal substrates by combining room temperature magnetron sputtering and a 600 °C O2 annealing process. To obtain stoichiometric BFO thin films, a Bi1.1Fe1.0O3 ceramic target was used for RF magnetron sputtering. The obtained thin films were further treated by an annealing process under O2 atmosphere at 650 °C for 30 min. The thickness of BFO thin films is 150 nm. The top electrodes, including Pt, Ag, Cu, and Al, were deposited by electron-beam evaporation. The prepared top electrodes on BFO are circular, where the diameter is 200 μm and the interval space is 800 μm.
The crystalline phase of the samples was measured by x-ray diffraction (XRD, DMAX 1400), where the radiation source is Cu Kα (λ = 0.154 nm). The surface morphology of BFO films was measured by an atomic force microscope (AFM, SPA-300HV) and a scanning electron microscope (SEM, UItra55). We measured the electrical properties of the devices in a Lakeshore probe station, and a Keithley 6517B electrometer was used to apply and collect the electrical signal.
III. RESULTS AND DISCUSSION
A. Characterization of the BFO thin film
The XRD θ–2θ pattern of the BFO/LNO structure is shown in Fig. 1(a). Due to the effect of the STO (100) substrate, BFO and LNO thin films are (100) preferred orientation. No diffraction peaks belonging to other crystalline phases were observed, indicating the pure phase of the BFO films. Figures 1(b) and 1(c) show the SEM and AFM images of the BFO thin film. The grains are close-packed and no pin holes exist, indicating the high quality of the BFO thin films.
B. I-V characteristic of the devices
To understand the effect of the top electrodes on the resistive switching behavior of BFO, we measured the I-V characteristic of the devices in the voltage range of ±7 V, as shown in Fig. 2. The Pt/BFO/LNO, Ag/BFO/LNO, and Cu/BFO/LNO devices exhibit pronounced WORM-resistive switching behavior, as shown in Figs. 2(a)–2(c). When applying small voltages to the top electrodes, the measured current is small and increases slowly with the increase in voltage, indicating the high resistance state (HRS) of the as-prepared BFO thin films. When the applied voltages exceed the set voltages (Vset), the leaps of the currents have been observed, indicating the switch of the devices from the HRS to the low resistance state (LRS). The set voltages of Pt/BFO/LNO, Ag/BFO/LNO, and Cu/BFO/LNO devices are 2.4, 4.2, and 4.6 V, respectively. The ON/OFF ratios, defined as the ratio of HRS resistance to LRS resistance that is read at the voltage of 0.1 V, are 3 × 10−4, 3 × 10−3, and 2.8 × 10−3 for Pt/BFO/LNO, Ag/BFO/LNO, and Cu/BFO/LNO devices, respectively. The devices remain in the LRS after being switched and show good repeatability for more than 100 cycles (in the voltage range of ±7 V). However, the I-V curve of the Al/BFO/LNO device exhibits a hysteresis characteristic rather than the resistive switching characteristic, as shown in Fig. 2(d). The observed different characteristics of the devices with different top electrodes indicate that the interface of the top electrode/BFO plays an important role in the electrical properties of the BFO-based devices.
After the devices were switched to the LRS, all the devices, including Pt/BFO/LNO, Ag/BFO/LNO, and Cu/BFO/LNO, show linear I-V characteristics in the voltage range of ±7 V, as shown in Figs. 3(d)–3(f). The linear I-V characteristic indicates the Ohmic conduction of the devices in the LRS. Although the devices in the HRS show Ohmic conduction too at small voltages (<1 V), the current is several orders of magnitude smaller than which in the LRS, implying the formation of conduction filaments in the LRS.
Figure 4(a) shows the schematic equilibrium band structure diagrams of the devices with different top electrodes. The electron affinity, work function, and energy bandgap of BFO are 3.3, 4.7, and 2.8 eV, respectively.20,28 LNO is a metal with perovskite-type structure, and its work function is 4.5 eV.29,30 The work functions of Pt, Ag, Cu, and Al are 5.65, 4.26, 4.42, and 4.28 eV, respectively.31–34 Accordingly, the Schottky barrier heights and built-in potentials (Vbi) at the electrode/BFO interfaces can be calculated by considering the electron affinity and work functions, as marked in Fig. 4(a). The Schottky barrier height at the BFO/LNO interface is 1.2 eV, and band bending (bend downward toward the interface) of BFO is 0.2 eV. For the Pt/BFO/LNO device, the Pt/BFO interface has a barrier height of 2.35 eV and a Vbi of 0.95 eV. The high barrier height at the Pt/BFO interface and large band bending (bend upward toward the interface) of BFO at the interface inhibit the Schottky emission current, resulting in SCLC conduction at higher voltages (1 V < V < Vset) in the HRS. For Ag/BFO/LNO, Cu/BFO/LNO, and Al/BFO/LNO devices, the barrier height at the top electrode/BFO interfaces are 0.96, 1.12, and 0.98 eV, respectively, which are close to the barrier height at BFO/LNO interfaces and are conducive to the Schottky emission of electrons. As a result, the Ag/BFO/LNO and Cu/BFO/LNO devices at higher voltages (1 V < V < Vset) is dominated by Schottky emission until the formation of conduction filaments.
Figure 4(b) shows the leakage currents of the devices in the HRS. The Pt/BFO/LNO device has the lowest leakage current due to the high barrier height and large band bending at the Pt/BFO interface. The currents of Ag/BFO/LNO and Cu/BFO/LNO devices are greater than that of Pt/BFO/LNO device, which are attributed to the small barrier height at both top electrode/BFO and BFO/LNO interfaces. However, the Al/BFO/LNO device shows current that is two times larger than Ag/BFO/LNO and Cu/BFO/LNO devices, even though the barrier height at the top electrode/BFO interfaces is very close. This phenomenon indicates that the Al/BFO interface may be different from other three top electrode/BFO interfaces. The Al that diffuses into BFO may combine with the oxygen atoms to form the Al–O bond due to the large Gibbs energy (−1582.3 kJ/mol),35 which results in high density of oxygen vacancies near the Al/BFO interface. The oxygen vacancies may reduce the resistance of BFO, i.e., increase the current of the device. On the other hand, the oxygen vacancies and caused defects near the interface may act as trap centers of carriers, resulting in the hysteresis characteristic of the I-V curve by carrier trapping/detrapping.
C. Temperature-dependent resistance in the LRS and HRS
We further measured temperature-dependent resistance of the LRS and HRS of the devices, as shown in Fig. 5. With the decrease in temperature from 300 to 100 K, the resistances of the devices in the HRS increase gradually, indicating the insulating behavior of BFO. The insulating behavior is due to the semiconductor band structure of BFO as well as the Schottky barrier at the electrode/BFO interfaces. By contrast, the resistances in the LRS decrease with the drop of temperature, which indicates the metallic properties of the LRS of the devices. The temperature dependence of metallic resistance can be described by R(T) = R0[1 + α(T − T0)], where R0 is the resistance at temperature T0 and α is the resistance temperature coefficient.36,37 By linearly fitting the temperature-dependent LRS resistances of the devices, we obtained the resistance temperature coefficients of the devices, as shown in the insets in Figs. 5(a)–5(c). The resistance temperature coefficient of the Pt/BFO/LNO device at room temperature is 1.3 × 10−3 K−1, which is close to the previously reported values of the devices with conduction filaments that relate to oxygen vacancies.38–40 Thus, the metallic behavior of the Pt/BFO/LNO device in the LRS may be due to the formation of conduction filaments composed of oxygen vacancies in BFO. The calculated resistance temperature coefficients of Ag/BFO/LNO and Cu/BFO/LNO devices are 2.1 × 10−3 and 3.6 × 10−3 K−1, which are in agreement with the values of Ag and Cu (approximately 3.8 × 10−3 and 3.9 × 10−3 K−1 for Ag and Cu bulks, respectively, within ∼2.5 × 10−3 and ∼4 × 10−3 K−1 for Ag and Cu high-purity nanowires).41,42 This metallic behavior implies that the conduction filaments in Ag/BFO/LNO and Cu/BFO/LNO devices are composed of Ag and Cu atoms, respectively, due to the diffusion of top electrodes under a bias voltage. In addition, the oxygen vacancies may be incorporated into metal filaments, which may decrease the mean free path of electrons in the metal filaments.36,42 As a result, the resistance temperature coefficients are slightly smaller than the pure metal but higher than the oxygen filaments.
In summary, we prepared BFO-based devices with different top electrodes, including Pt, Ag, Cu, and Al, and investigated the effect of top electrodes on the WORM switching behavior. The Pt/BFO/LNO, Ag/BFO/LNO, and Cu/BFO/LNO devices show pronounced WORM-resistive switching behavior. The Pt/BFO/LNO device shows SCLC conduction in the HRS due to the large Schottky barrier height at the Pt/BFO interface and switches to the LRS due to the formation of conduction filaments composed of oxygen vacancies. The Ag/BFO/LNO and Cu/BFO/LNO devices exhibit Schottky emission conduction in the HRS owing to the small barrier height at both top electrode/BFO and BFO/LNO interfaces and show metallic conduction behavior in the LRS due to the formation of metal conduction filaments, i.e., Ag and Cu filaments (which may contain oxygen vacancies). The Al/BFO/LNO device exhibits a hysteresis I-V curve, which may be attributed to the oxygen vacancies and defects caused by the formation of Al–O bond near the Al/BFO interface.
This work was financially supported by the Natural Science Foundation of Sichuan Province, China (No. 2022NSFSC2014) and the Doctoral Fund of Southwest University of Science and Technology, China (Nos. 18zx7132 and 19zx7131).
Conflict of Interest
The authors have no conflicts to disclose.
Yajun Fu and Wei Tang contributed equally to this work.
Yajun Fu: Conceptualization (lead); Supervision (equal); Writing – original draft (lead); Writing – review & editing (lead). Wei Tang: Investigation (lead); Visualization (lead). Jin Wang: Resources (equal); Supervision (equal). Linhong Cao: Project administration (equal); Resources (equal).
The data that support the findings of this study are available within the article or from the corresponding author upon reasonable request.