FULL P APER
Controlling Resistance Switching Polarities of Epitaxial BaTiO 3 Films by Mediation of Ferroelectricity and
Oxygen Vacancies
Ming Li , Jian Zhou , Xiaosai Jing , Min Zeng , Sujuan Wu , Jinwei Gao , Zhang Zhang , Xingsen Gao , Xubing Lu ,* J.-M. Liu , and Marin Alexe*
DOI: 10.1002/aelm.201500069
1. Introduction
BaTiO 3 (BTO) is an important ferroelec- tric, dielectric, and semiconducting per- ovskite oxide material. [ 1–3 ] It has been widely studied for various application potentials such as ferroelectric memo- ries, [ 4 ] multilayer capacitors, [ 5 ] and posi- tive temperature coeffi cient resistors, [ 6 ] etc. Recently, resistive random access memory (ReRAM) has been suggested as one of the promising candidates for next- generation nonvolatile memory applica- tions. [ 7,8 ] Attention has particularly been paid to BTO fi lms for the applications in ReRAM devices and encouraging results have been reported. [ 9–17 ] Correspond- ingly, various microscopic mechanisms were proposed to explain the resistance switching effects of BTO fi lms. One rep- resentative mechanism is the ferroelectric polarization induced interfacial charge coupling (variations of interfacial barrier height and width). The substantial roles of this mechanism in ferroelectric tunnel junctions, where the resistance switching originates mainly from a polarization modulation of the height and width of tunnel barrier, have been addressed. [ 9,10,12,14 ] Another mechanism is based on defects (ionic or electronic) mediated phenomena, in which formation/
rupture of conductive fi laments are believed to be responsible for the resistance switching in BTO fi lms, [ 11,13 ] and the forma- tion/rupture of conductive fi laments is closely related to the density and migration behaviors of oxygen vacancies (V O ) or other defects. [ 11,13 ] In addition to the above mentioned mecha- nisms, other mechanisms were proposed. For example, Wang et al. reported clear resistance switching behaviors for TiN/
BTO/TiN structure fabricated by RF sputtering and claimed that the thin TiO x as interfacial layer between BTO and TiN could cause the resistance switching. [ 15 ] Au et al. reported signifi cant resistive switching effects in Ag nanoparticles-embedded BTO fi lms and they claimed that the resistance switching is due to the Ag charge storage. [ 17 ]
Although excellent electrical properties such as high on/off resistance ratio, good retention, and high endurance etc., have been reported in the above mentioned literature, the resistive switching mechanisms in BTO-like ferroelectric oxide fi lms are In this work, the observations of different resistive switching polarities
of epitaxial BaTiO 3 (BTO) thin fi lms fabricated by pulsed laser deposi- tion are reported. The BTO fi lms with various ferroelectric states and oxygen vacancy (V O ) concentrations are achieved by carefully controlling the oxygen pressure during the depositions. For fi lms with no ferroelec- tricity and high V O concentrations, the resistance will change from a low resistance state (LRS) to a high resistance state (HRS) during a positive voltage cycle (0 → 3 → 0 V), and from a HRS to a LRS during a negative voltage cycle (0 →−3 → 0 V). However, completely opposite RS polarity is observed for the fi lms with weak ferroelectricity and intermediate V O concentrations. Such RS behaviors and polarity can be hardly observed or negligible for the fi lms with good ferroelectricity and nearly free of V O . It is proposed that the unique resistance switching polarities of BTO fi lms are attributed to the competition between the ferroelectricity and oxygen vacancy migration dynamics. Results clarify the complex RS mechanisms in the BTO fi lms, and address the competing ferroelectricity and V O migra- tion in modulating the RS behaviors of ferroelectric oxide-based resistive memory devices.
M. Li, J. Zhou, X. Jing, Prof. M. Zeng, Prof. S. Wu, Prof. J. Gao, Prof. Z. Zhang, Prof. X. Gao, Prof. X. Lu Institute for Advanced Materials and Guangdong Provincial Laboratory of Quantum Engineering and Quantum Materials
South China Normal University Guangzhou 510006 , China E-mail: [email protected] Prof. X. Lu, Prof. M. Alexe
Max Planck Institute of Microstructure Physics Weinberg 2 , Halle 06120 , Germany
E-mail: [email protected] Prof. J.-M. Liu
Laboratory of Solid State Microstructures and Innovation Center of Advanced Microstructures
Nanjing University Nanjing 210093 , China Prof. M. Alexe Department of Physics University of Warwick
Coventry CV4 7AL , West Midlands , UK
www.advelectronicmat.de
FULL P APER
still ambiguous and under dispute. Especially, for leaky ferro- electric semiconducting fi lms, the roles of the ferroelectricity and V O in modulating the RS behaviors are far from clarifi ed.
For example, in the works of Yan et al. [ 11 ] and Li et al., [ 13 ] only conductive fi laments formed by the V O migration are consid- ered, and the effect of ferroelectricity on the RS behaviors in BTO fi lms was ignored. In the work by Wang et al., [ 18 ] the RS effects in semiconducting ferroelectric BiFeO 3 (BFO) fi lms are attributed to the modulation of the Schottky-like barriers at both of the top/bottom interfaces due to the accumulation/depletion of V O in the BFO fi lms induced by the polarization switching.
For the same semiconducting ferroelectric BFO system, the RS effects are attributed only to the barrier variation of the bottom BFO/Nb-SrTiO 3 (STO) interface. [ 19 ] The barrier variation origi- nates from the accumulation/depletion of the free carriers in the Nb-doped STO substrates rather than the accumulation/
depletion of oxygen vacancies inside the BFO fi lms.
It is well known that oxygen pressure during the oxide fi lm deposition will greatly affect the fi lm crystal structure, ferroelec- tricity, conductivity, and charge transport properties. [ 20,21 ] In this work, we shall prepare a series of BTO fi lms with different fer- roelectric states and electrical conductions by carefully control- ling the oxygen pressure during the fi lm growth using pulsed laser deposition (PLD). Three typical types of resistive switching behaviors will be observed. Our major concern is to manipulate the ferroelectric polarization and V O concentration, so that the
polarization modulated consequence and V O -induced mecha- nism coexist and compete with each other. Along this line, one can expect a transition of the RS behavior from one to another, for example, the RS polarity reversal. Up to now, attention has been rarely paid to the RS polarity of the devices, while clari- fying the intrinsic resistance switching mechanisms remains critical and challenging. We will investigate the RS polarity variation in the BTO thin fi lms prepared under different condi- tions. Based on the characterizations of their crystal structures, ferroelectric polarizations, and electrical conductivities, the microscopic RS mechanisms responsible for the different RS polarities will be discussed. Additional energy band models will be proposed to explain these RS polarities.
2. Results
Figure 1 a shows the X-ray diffraction (XRD) 2 θ -Ω measure- ments of the six BTO fi lms deposited under different oxygen pressures ranging from 3.3 × 10 −6 mbar to 3.3 × 10 −1 mbar on the 20-nm thick La 0.7 Sr 0.3 MnO 3 (LSMO)-covered (001) STO sub- strates. For the convenience of expression, samples with BTO fi lms deposited in the oxygen pressures of 3.3 × 10 −1 , 3.3 × 10 −2 , 3.3 × 10 −3 , 3.3 × 10 −4 , 3.3 × 10 −5 , and 3.3 × 10 −6 mbar are nomi- nated as P-1, P-2, P-3, P-4, P-5, and P-6, respectively. The thick- ness of the BTO fi lms is ≈300 nm. The (00 m ) peaks from the
Figure 1. a) XRD 2 θ -Ω scans of BTO fi lms deposited in different oxygen partial pressures; b) the enlarged (003)/(300) refl ection peaks of the BTO fi lms;
c) RSM around the (103) pseudo-cubic refl ection of the BTO/LSMO/STO structures with BTO fi lm deposited in 3.3 × 10–3 mbar P O2 ; d) the impact of oxygen pressure on a/c lattice constant of the BTO fi lms.
FULL P APER
STO substrates and ( m 00)/(00 m ) peaks of the BTO fi lms reveal the good epitaxial BTO growth on the cube-on-cube orientation.
Since the LSMO layers are very thin (only ≈20 nm), the peak from the LSMO layers is not detectable. The peaks labeled with stars originate from the substrates due to the remaining Cu K β radiation. A more detailed analysis is carried out for the (003) peaks, as shown in Figure 1 b. For samples P-1 and P-2, the (300)/(003) peaks split into two groups of peaks of (003) and (300), implying that the a -domains and c -domains coexist in the two BTO fi lms. When the oxygen pressure decreases down to 3.3 × 10 −3 mbar, only the c -domains [(003) peak] can be iden- tifi ed and the a -domains [(300) peak] disappear. The (003) dif- fraction peaks shift toward the high angle with reduced oxygen pressure, indicating the out-of-plane lattice contraction. The reciprocal space mapping (RSM) measurements around the (103) pseudo-cubic refl ection of the BTO/LSMO/STO structure were also performed for all the samples so that their crystal structures and lattice constants can be evaluated. Figure 1 c is a typical RSM pattern for sample P-3. The calculated lattice parameters from the RSM results are shown in Figure 1 d. The large c / a ratios for the samples deposited in the intermediate oxygen pressures (P-3, P-4, and P-5) are evaluated, demon- strating the tetragonal crystal structures. The c / a ratios are close to 1.0 for the samples deposited in low oxygen pressures (P-6) or high oxygen pressures (P-1), demonstrating the cubic crystal structures. It is well known that a paraelectric BTO phase is cubic and tetragonal BTO corresponds to a ferroelectric phase.
The XRD results reveal that the samples P-3, P-4, and P-5 have good ferroelectric properties and the others may not.
Atomic force microscope surface morphology measurements were also done to check the fi lm quality, as seen in Figure S1 (Supporting Information). Except for sample P-1, all other
samples exhibit smooth surfaces. Terrace growth patterns in the samples P-3 and P-4 can be identifi ed by careful observa- tion. Usually, it is diffi cult to persevere the initial vicinal struc- ture on the top surface of such thick fi lms; therefore, this obser- vation confi rms the high quality of the BTO fi lms used in the present study.
The resistive switching properties of the Au/BTO/LSMO device structures were investigated. The schematic diagram of the measurement setup is shown in the inset of Figure 2 a. The small devices with area of 60 × 60 µm 2 were used for electrical characterization to avoid the effect of the fi lm nonuniformity.
Positive and negative biases were applied to the top electrode while the LSMO bottom electrodes were always grounded. To reduce the effects from the transient current as small as pos- sible, a small voltage step of 5 mV and long current integration time (PLC 20 in the current integration time of B1500A) were adopted during all the current–voltage ( I – V ) measurements (The experimental results for transient current measurements on the sample P-3 are shown in Figure S2 (Supporting Infor- mation). It should be noted that the resistances of the LSMO bottom electrodes in all these samples should be less infl uenced by the different oxygen ambient pressures for the BTO thin- fi lm deposition, as identifi ed by earlier work, [ 22 ] because the conductivity of LSMO is dominated by the Mn 3+ –Mn 4+ double- exchange mechanism. This allows a safe exclusion of the vari- ation in electrical conductivity for LSMO bottom electrodes.
Figure 2 shows the typical I – V characteristics of all the samples.
The observed I – V hysteresis loops were measured by the four continuous voltage scanning cycles of 0 →+3 → 0 → –3 → 0 V without an initial electrical forming process. Three types of RS behaviors can be observed. For the sample P-6, clear bipolar RS effects are exhibited. During the positive voltage scanning www.advelectronicmat.de
Figure 2. The resistive switching behaviors of BTO fi lms deposited in different oxygen pressures: a) P-6 (3.3 × 10–6 mbar); b) P-5 (3.3 × 10–5 mbar);
c) P-4 (3.3 × 10–4 mbar); d) P-3 (3.3 × 10–3 mbar); e) P-2 (3.3 × 10–2 mbar); f) P-1 (3.3 × 10–1 mbar). The inset of Figure 2 a shows the schematic diagram of the Au/BTO/LSMO/STO sample structure and measurement set up.
FULL P APER
cycle (0 →+3 → 0 V), the sample changes from a LRS state to a HRS state, whereas it changes from a HRS to a LRS during the negative voltage scanning cycle (0 → –3 → 0 V), as shown in Figure 2 a. We assign this kind of RS behavior as type “A” RS behavior. Similarly, the samples P-5 and P-4 also exhibit clear bipolar RS effects, as shown in Figure 2 b,c. However, the resist- ance changes from a HRS to a LRS during the 0 →+3 → 0 V cycle, and from a LRS to a HRS during the 0 →−3 → 0 V cycle, whose hysteresis direction is completely opposite to that of the type “A” RS behavior. We assign it as type “B” RS behavior. The two types of RS modes exhibit opposite RS polarities, implying that the mechanisms for them should be different. The samples P-3, P-2, and P-1 show almost identical I – V features, as shown in. Figure 2 d–f. Comparing the RS behaviors of type “A” and type “B” modes, they suggest the following different features:
(1) Unlike the large resistance ratio of more than 10 3 shown in type “A” and type “B” samples, only weak RS effects can be observed. (2) The weak RS effect exhibits a unipolar behavior, while the bipolar RS behaviors are observed in the type “A” and
“B” samples. (3) The I – V hysteresis characteristics are almost symmetrical and the magnitude of current at ±3 V decreases several orders of magnitude down to only pico-amphere scale, implying that the BTO fi lm changes from a good conductor to a good insulator. The RS characteristics observed in the samples P-3, P-2, and P-1 are classifi ed into the type “C” RS behavior.
The huge differences of the RS behaviors in these samples suggest different underlying microscopic mechanisms, which are assumed to be closely related to the differences in their
crystal structures and electrical properties due to different growth oxygen pressures. To elucidate the origin for the huge RS differences, the ferroelectricity of all the samples was fi rst investigated by macroscopic polarization-voltage (P-V) meas- urements. At room temperature, saturated hysteresis loop can be observed only for the samples P-3 and P-2. For the sample P-1, the P-V curve is linear-like, demonstrating very weak or nearly no ferroelectricity. The samples grown in oxygen pres- sure lower than 3.3 × 10 −3 mbar are too leaky to obtain macro- scopic P-V characteristics at room temperature. Nevertheless, the P-V measurements performed at 80 K revealed well satu- rated ferroelectric hysteresis for the sample P-4. However, we still cannot observe normal macroscopic P-V hysteresis for the samples P-5 and P-6 even at low temperatures due to very leaky fi lms. Figure 3 a,b shows the macroscopic P-V measurements results (The leaky P-V curves for the samples P-5 and P-6 are not shown here). To further check the ferroelectricity of the leaky BTO fi lms, piezoresponse force microscopy (PFM) meas- urements were carried out for the samples P-5 and P-6. Under the same polarizing and reading conditions, the PFM phase exhibits clear ferroelectric switching features for the sample P-5, but no evidence of ferroelectricity can be observed for the sample P-6, as shown in Figure 3 c,d. Based on the macroscopic P-V and nanoscale PFM measurement results, it can be con- cluded that the samples, P-6 and P-1 have no ferroelectricity;
the samples P-3 and P-2 have comparatively good ferroelec- tricity, and only weak ferroelectric performances for the sam- ples P-4 and P-5 can be revealed. These results are consistent Figure 3. a) P-V characteristics measured at 10k Hz for samples of P-3, P-2, and P-1 at room temperature; b) P-V characteristics of sample P-4 meas- ured at 10k Hz at room temperature and 80 K; c) PFM phase of sample P-5; d) PFM phase of sample P-6. Both of them are poled with same writing voltages of –8V and +8V.
FULL P APER
with the crystal structure analysis by XRD results shown in Figure 1 .
In parallel to the ferroelectricity characterizations, the elec- trical conductivity of the BTO fi lms was also measured. It is known that the conduction of the nondoped BTO fi lms is con- tributed from the free electrons provided by V O . [ 21 ] Experimen- tally, using aberration-corrected annular-bright-fi eld scanning transmission electron microscopy, Xu et al. proved that the V O concentration increases with reduced oxygen pressure during the deposition. [ 23 ] In Zhao’s work, [ 21 ] Rutherford backscattering spectrometry measurements on the BTO fi lms deposited in the oxygen pressure ranging from 2.2 × 10 −7 to 2.2 × 10 −4 mbar also confi rmed the same relationship between the V O concentra- tion and deposition oxygen pressure. In the present work, the V O concentration in the BTO fi lms is higher at lower oxygen pressure during the PLD. Since it is very diffi cult to accurately measure the concentration of light elements, such as oxygen, we have a rough comparison on the V O concentrations of the samples P-6, P-4, and P-3 from the point of view of conductivity.
The temperature dependent resistances of the Au/BTO/LSMO structures were measured at a constant voltage of 2 V with the same capacitor size. As shown in Figure S3 (Supporting Infor- mation), the three samples exhibit huge resistance differences, which indirectly imply the signifi cant V O density difference inside them. The sample P-3 exhibits the largest resistance and the sample P-6 the smallest resistance. The negative Hall coef- fi cient obtained by the Hall measurement (results not shown here) shows that the major carrier is electron. The present resistance and Hall measurements indicated that the free car- riers (electrons) provided by the doping of V O have been greatly reduced in the samples P-4 and P-3. The signifi cant increase of the resistance indirectly proved that the V O concentration has been greatly reduced in the samples P-4 and P-3. The sample P-3 can be nearly dealt as a good insulator, and most probably there are no oxygen vacancies in the BTO fi lm.
3. Discussion
Based on the above analysis on the ferroelectricity and electrical conductivity of the BTO fi lms, the big differences of the RS behaviors of the BTO samples should be attributed to the dif- ferent ferroelectric properties and V O concentrations. To inter- pret the three different types of RS behaviors, we propose the following microscopic RS mechanisms.
As mentioned above, the sample P-6 has nearly no ferro- electricity. Thus, the RS effect is unlikely due to ferroelectric switching. In addition, the features of RS behaviors shown in Figure 2 a are also not likely due to the conductive fi lament mechanism, whose RS behaviors are usually unipolar. We pro- pose that the RS effects are ascribed to the local migration of V O driven by external electric fi eld. To construct the energy band model, the charge transport characteristics of the interfaces between the Au/BTO and BTO/LSMO have been investigated.
As shown in Figure S4a (Supporting Information), the linear I – V characteristics of the in-plane LSMO/BTO (P-6)/LSMO device demonstrate the ohmic contact of the BTO/LSMO inter- face. On the contrary, the I – V characteristics of the in-plane Au/
BTO (P-6)/Au device exhibit the typical Schottky diode behavior,
as shown in Figure S4b (Supporting Information), indicating clearly the Schottky contact at the Au/BTO interface. Therefore, the charge transport in the sample P-6 is dominated only by the Schottky barrier at the Au/BTO top interface. When a posi- tive voltage is applied to the top electrode, the oxygen vacan- cies will be repelled away from the top Au/BTO interface. [ 24–26 ] Then the width and height of the Schottky barrier will increase, as shown in Figure 4 a (point “a” in the I – V curve). After scan- ning the voltage from point “a” to point “b”, the V O concentra- tion on the top interface will be continuously reduced with the continuous positive voltage stress. Consequently, the width and height of the Schottky barrier at the top interface will be further increased, as shown in Figure 4 b. Comparing points “a” and “b”
at the same bias voltage, the BTO fi lm will change from a LRS in point “a” to a HRS in point “b” due to the increased height and width of the Schottky barrier at the Au/BTO top interface.
When a negative bias voltage is applied to the Au electrode, the oxygen vacancies will move toward the Au/BTO top interface.
In point “c”, a small number of V O accumulated at the Au/BTO interface lowers the Schottky barrier height in comparison with its virgin state (see Figure 4 c). After the negative bias voltage sweeps from c →−3 V → d, the oxygen vacancies are inten- sively accumulated at the top interface. Then the barrier (height and width) at the top interface will be further weakened (see Figure 4 d). Therefore, the BTO fi lm switches from the HRS in point “c” to the LRS in point “d”, due to the decrease of the barrier in height and width at the Au/BTO interface. To further confi rm this V O migration model, we carried out current stress measurements under different temperatures. When the sample P-6 is biased at a constant voltage of ±2V, we observe that: (1) The stress current decreases/increases with the increase of the stress time, which further confi rms the model that the posi- tive/negative voltages will repel/attract oxygen vacancies from the interface and result in the increase/decrease of the resist- ance. (2) The stress current decreases/increases faster at a higher temperature, also confi rming the model in which the oxygen vacancies have higher mobility at higher temperatures.
The corresponding results are shown in Figure S5 (Supporting Information).
For the samples grown with oxygen pressures higher than 3.3 × 10 −6 mbar, their RS behaviors (type “B” and type “C”) are completely different from that of type “A”. In this case, the V O migration model cannot be used to explain the type “B” and type
“C” RS behaviors. A number of theoretical [ 27 ] and experimental works [ 28–30 ] showed that the charge transport across the metal/
ferroelectrics/conductive oxide device is controlled by the ferro- electric polarization switching induced Schottky barrier variation at the top/bottom interface. The ferroelectric polarization controls the accumulation or depletion of electron charges at the inter- face, and the associated bending of the n -BTO conduction band determines the transport regime across the interface. According to the investigation on ferroelectricity, the samples with type “B”
RS behaviors (samples P-5 and P-4) have strong or weak ferro- electricity at room temperature. Therefore, the ferroelectricity should play an important role in inducing the type “B” RS behav- iors. To confi rm this assumption, we calculate the ferroelectric polarization-induced interface barrier changes of the sample P-4 using the theoretical model proposed by Pintilie et al. [ 31,32 ] , as shown in Table 1 . It shows that the ferroelectric switching does
www.advelectronicmat.de
FULL P APER
modulate the barrier height at the top/bottom interface. The bottom BTO/LSMO interface is no longer of the ohmic contact.
Furthermore, the potential change across the BTO/LSMO inter- face is more signifi cant than that of the Au/BTO interface, which may be due to the much larger screen length of LSMO (2.0 Å) than that of Au (0.3 Å). Therefore, charge transport behaviors across the two back to back metal-semiconductor Schottky con- tacts are dominated by the bottom BTO/LSMO interface. Based on the above analysis, we propose the following model to explain the RS behaviors shown in the type “B” samples (P-5 and P-4), as shown in Figure 5 . When a positive voltage is applied onto the top Au electrode (Figure 5 a), the depletion layer width and barrier
height at the top interface will increase due to the downward polarization, while they are decreased at the bottom BTO/LSMO interface. When the positive voltage scans from voltage point “a”
to point “b” (Figure 5 b), the downward polarization will continu- ously increase, therefore the energy band at the bottom interface will keep bending-down, leading to the decrease of the resistance in point “b”. On the contrary, a negative voltage applied onto the top electrode will produce an upward polarizarion, which will result in the decrease/increase of the barrier height and width at the top interface/bottom interface (Figure 5 c). When the negative voltage scans from point “c” to point “d” (Figure 5 d), the upward polarization will continuously increase. The energy band at the bottom interface will keep bending-up, leading to the increase of the resistance in point “d”.
For the sample P-3, we also estimate the barrier changes by ferroelectric polarization switching using the same methods for the sample P-4, where only the parameters used in the calcula- tion are somewhat different. Since the sample P-3 has better fer- roelectricity, a higher P r value of 4.5 µC cm –2 is adopted, and dif- ferent virgin values of the barrier height at the interface are used considering the better insulating properties (different positions for the conduction band). The estimated barrier height of the top Au/BTO interface will change from 1.84 to 1.92 eV (down- ward polarization) or 1.76 eV (upward polarization). The barrier height at the bottom BTO/LSMO interface will change from 1.64 to 1.22 eV (downward polarization) or 2.06 eV (upward polarization). Although signifi cant barrier height changes can be observed, the RS effects of sample P-3 are rather weak, as
Table 1. The barrier height changes with and without polarization switching of sample P-4.
Barrier height Virgin value
[eV]
Polarization down [eV]
Polarization up [eV] a)
Top interface (Au/BTO) 1.65 1.71 1.59
Bottom interface (BTO/LSMO) 1.45 1.17 1.73
a) The following parameters are used for the calculation of the barrier height: P = 3µC/cm 2 , ε static = 60, l 1 = 0.3Å, ε M1 = 1.5, l 2 = 2.0 Å, ε M2 = 2. The P and ε static values are taken from the data adopted by Daniel et al. in Ref. [32]; The l 1 and ε M1 values are estimated from the Cu values used in Ref. [32], since they are rarely reported.
The l 2 and ε M2 values are from the experimental results by X. Hong et al. (Appl.
Phys. Lett. 86, 142501(2005) and by Mistrik et al. (J. Appl. Phys. 99, 08Q317(2006).
The real parameters may have some deviation to the presently used for calculation.
The calculated results are enough to demonstrate the change of the barrier height with ferroelectric switching.
Figure 4. The schematic energy band diagrams for sample P-6 (3.3 × 10–6 mbar) under different bias voltage conditions, which indicate that the bipolar RS polarities are dominated by V O migration under external electric fi eld. a) positive voltage in position “a”, V O slightly depleted at the top interface;
b) positive voltage in position “b”, V O deeply depleted at the top interface; c) negative voltage in position “c”, V O slightly accumulated at the top interface; d) negative voltage in position “d”, V O deeply accumulated at the top interface.
FULL P APER
www.advelectronicmat.de
shown in Figure 2 d. As already known from the results in Figure 3 a, the sample P-1 doesn’t have ferroelectricity and the sample P-2 has weaker ferroelectricity than the sample P-3. They have different ferroelectricity. However, all of them exhibit nearly the same current–voltage behaviors, as shown in Figure 2 d–f. The present results suggest that the ferroelectric polarization cannot necessarily induce the RS effect in insulating ferroelectric fi lm without suffi cient free carriers (electrons cre- ated by V O in the present case). It should be pointed out that remarkable RS effect is possible for ferroelectric tunnel diode based on very thin BTO fi lm, in which the charge transport is dominated by the direct tunneling process of electrons. [ 9 ] For the present metal (Au)-insulator (300 nm BTO)-metal (LSMO) (MIM) structures, the weak I – V hysteresis loops as observed in the samples P-3, P-2, and P-1 are believed to be mostly due to the dielectric relaxation current. As also shown in Figure S2 (Supporting Information), the current–voltage behaviors of the sample P-3 are closely related to the voltage sweeping speeds, which implies that the I – V hysteresis shown in Figure S2 (Sup- porting Information) is not from the steady transport current.
Our present results demonstrate that ferroelectric switching can induce signifi cant RS effects for thick BTO fi lms, but it only works on the semiconductor-type BTO fi lms (P-5 and P-4), which further demonstrates that the free carriers (electrons) is critical for the thick ferroelectric fi lm to obtain remarkable RS effects.
4. Conclusion
In conclusion, we have carried out a comprehensive study on the resistive switching behaviors of epitaxial BTO thin fi lms deposited on LSMO-covered STO substrates. We have demon- strated that the RS behaviors of BTO fi lms are due to the com- petition between ferroelectric polarization and V O migration.
It has been revealed that the free carriers are crucial to obtain the polarization switching induced large RS effects in thick fer- roelectric fi lms. Especially, we have observed two completely opposite RS switching polarities for the BTO fi lms deposited in different oxygen pressures, and have clarifi ed the microscopic mechanisms responsible for the different RS polarities from the viewpoints of the ferroelectricity and electrical conductivity.
The present work clarifi es the intrinsic relationship between the complex RS behaviors and their embedded mechanisms for BTO-like ferroelectric semiconducting oxide fi lms, and provides benefi cial insights in the accurate design and control of the RS behaviors in the high performance ReRAM devices.
5. Experimental Section
The BTO/LSMO oxide heterostructures were epitaxially grown on (001)-oriented SrTiO 3 single crystal substrate by PLD. The LSMO bottom electrodes were grown at a substrate temperature of 600 °C and 0.21 mbar oxygen pressure. After the growth of ≈20 nm LSMO layers, the BTO fi lms with thickness of ≈300 nm were deposited at a substrate temperature of 650 °C. The laser fl uency and laser frequency were fi xed to be 0.6 J cm –2 and 3 Hz for the deposition of all the BTO fi lms. The BTO fi lms were deposited by only varying the deposition oxygen pressure while keeping other processing parameters completely identical. The gold top electrodes were evaporated to fabricate the Au/
BTO/LSMO/STO RS devices. The crystal structure was examined by X-ray diffraction measurement using a Philips X’pert MRD meter. The 2 θ -Ω scan was carried out to check the crystal phase and preferred orientation. The crystal structure and the strain status of the BTO fi lms on STO substrates were further investigated by RSM, from which the lattice constants were calculated. The surface morphologies of substrates and deposited fi lms were studied by atomic force microscope (Digital Instruments 5000) working in tapping mode. The ferroelectricity of the leaky BTO fi lms was investigated by piezoresponse force microscopy (Cypher, Asylum Research). The resistance-temperature properties Figure 5. The schematic energy band diagrams for sample P-4 (3.3 × 10–4 mbar) under different bias voltage conditions a) positive voltage in position
“a”; b) positive voltage in position “b”; c) negative voltage in position “c”; d) negative voltage in position “d”. “ δ ” means the dead layer width between the metal/ferroelectric; “W1”and “W2” mean the depletion layer width in the top and bottom interface. The inset of the fi gure shows the positions
“a”, “b”, “c”, and “d” in the IV hysteresis loop.
FULL P APER
of the BTO fi lms were measured by a Keithley 6430 sub-femtoampere sourcemeter with a variable temperature vacuum probe station (Jannis).
Ferroelectric and resistive switching characteristics were investigated by a Radiant ferroelectric tester Premier II and an Agilent B1500A high precision semiconductor product analyzer, respectively.
Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.
Acknowledgement
M.L. and J.Z. contributed equally to this work. This work was supported by the National Natural Science Foundation of China (Contract Nos.
61271127, 51431006, and 51472093). M.A. acknowledges the support of Royal Society via Wolfson Award. X. B. Lu acknowledges the support from the Alexander von Humboldt foundation. This work was also supported by the Program for ChangJiang Scholars and Innovative Research Team in University (Grant No. IRT1243), the Program for International Innovation Cooperation Platform of Guangzhou (Grant No.
2014J4500016), and the National Undergraduate Innovation Program (2013, 2014). We acknowledge Richard Morris for the critical reading of the manuscript.
Received: February 17, 2015 Revised: March 18, 2015 Published online:
[1] J. Junquera , P. Ghosez , Nature 2003 , 422 , 506 .
[2] K. J. Choi , M. Biegalski , Y. L. Li , A. Sharan , J. Schubert , R. Uecker , P. Reiche , Y. B. Chen , X. Q. Pan , Science 2004 , 306 , 1005 .
[3] S. Valencia1 , A. Crassous , L. Bocher , V. Garcia , X. Moya , R. O. Cherifi , C. Deranlot , Nat. Mater. 2011 , 10 , 753 .
[4] J. F. Scott , Ferroelectrics Rev. 1998 , 1 , 1 .
[5] a) C. L. Jia , K. Urban , Science 2004 , 303 , 2001 ; b) L. M. Levinson , S. I. Hirano , in Grain Boundaries and Interfacial Phenomena in Elec- tronic Ceramics , American Ceramic Society Westerville, OH 1981 , p. 215 .
[6] H. I. Yoo , C. R. Song , D. K. Lee , J. Electroceramics 2002 , 8 , 5 . [7] R. Waser , M. Aono , Nat. Mater. 2007 , 6 , 833 .
[8] A. Sawa , Mater. Today 2008 , 11 , 28 .
[9] Z. Wen , C. Li , D. Wu , A. D. Li , N. B. Ming , Nat. Mater. 2013 , 12 , 617 .
[10] A. Chanthbouala , V. Garcia , R. O. Cherifi , K. Bouzehouane , S. Fusil , A. Barthelemy , Nat. Mater. 2012 , 11 , 860 .
[11] Z. B. Yan , Y. Y. Guo , G. Q. Zhang , J.-M. Liu , Adv. Mater. 2011 , 11 , 1351 .
[12] K. C. Sekhar , J. P. B. Silva , K. Kamakshi , M. Pereira , M. J. Gomes , Appl. Phys. Lett. 2013 , 102 , 212903 .
[13] S. Li , H. Z. Zeng , S. Y. Zhang , X. H. Wei , Appl. Phys. Lett. 2013 , 102 , 153506 .
[14] X. S. Gao , J. M. Liu , K. Au , J. Y. Dai , Appl. Phys. Lett. 2012 , 101 , 142905 .
[15] C. Wang , M. H. Kryder , J. Phys. D: Appl. Phys. 2008 , 41 , 245301 . [16] R. K. Pan , T. J. Zhang , J. Y. Wang , J. Z. Wang , D. F. Wang ,
M. G. Duan , Thin Solid Films 2012 , 520 , 4016 .
[17] K. Au , X. S. Gao , J. Wang , Z. Y. Bao , J. M. Liu , J. Y. Dai , J. Appl. Phys.
2013 , 114 , 027019 .
[18] C. Wang , K. J. Jin , Z. T. Xu , L. Wang , C. Ge , H. B. Lu , H. Z. Guo , M. He , G. Z. Yang , Appl. Phys. Lett. 2011 , 98 , 192901 .
[19] Z. Q. Hu , Q. Li , M. Y. Li , Q. W. Wang , Y. D. Zhu , X. L. Liu , Z. Z. Zhao , Y. Liu , S. X. Dong , Appl. Phys. Lett. 2013 , 102 , 102901 . [20] T. Zhao , F. Chen , H. B. Lu , G. Z. Yang , Z. H. Chen , J. Appl. Phys.
2000 , 87 , 7442 .
[21] T. Zhao , Z. H. Chen , F. Chen , H. B. Lu , G. Z. Yang , H. S. Cheng , Appl. Phys. Lett. 2000 , 77 , 4338 .
[22] R. Cauro , A. Gilabert , J. P. Contour , R. Lyonnet , M.-G. Medici , J.-C. Grenet , C. Leighton , I. K. Schuller , Phys. Rev. B. 2001 , 63 , 174423 .
[23] Z. T. Xu , K. J. Jin , L. Gu , Y. L. Jin , C. Ge , C. Wang , H. Z. Guo , H. B. Lu , R. Q. Zhao , G. Z. Yang , Small 2012 , 8 , 1279 .
[24] K. Sibuya , R. Dittmann , S. Mi , R. Waser , Adv. Mater. 2010 , 22 , 411 . [25] X. B. Yan , Y. D. Xia , H. N. Xu , X. Gao , H. T. Li , R. Li , J. Yin , Z. G. Liu ,
Appl. Phys. Lett. 2010 , 97 , 112101 .
[26] T. J. Zhang , R. K. Pan , Z. J. Ma , M. G. Duan , D. F. Wang , M. He , Appl. Phys. Lett. 2011 , 99 , 182106 .
[27] X. H. Liu , Y. Wang , J. D. Burton , E. Y. Tsymbal , Phys. Rev. B 2013 , 88 , 165139 .
[28] A. Q. Jiang , C. Wang , K. J. Jin , X. B. Liu , J. F. Scott , C. S. Hwang , T. A. Tang , H. B. Lu , G. Z. Yang , Adv. Mater. 2011 , 23 , 1277 .
[29] A. Tsurumaki , H. Yamada , A. Sawa , Adv. Funct. Mater. 2012 , 22 , 1040 .
[30] H. T. Yi , T. Choi , S. G. Choi , Y. S. Oh , S.-W. Cheong , Adv. Mater.
2011 , 23 , 3403 .
[31] L. Pintilie , M. Alexe , J. Appl. Phys. 2005 , 98 , 124103 . [32] D. Pantel , M. Alexe , Phys. Rev. B 2010 , 82 , 134105 .
