Correlation of resistance switching and polarization rotation in copper doped zinc oxide (ZnO:Cu) thin films studied by Scanning Probe Microscopy

Journal of Materiomics xxx (xxxx) xxx

Contents lists available at ScienceDirect

Journal of Materiomics journal homepage: www.journals.elsevier.com/journal-of-materiomics/

Correlation of resistance switching and polarization rotation in copper doped zinc oxide (ZnO:Cu) thin films studied by Scanning Probe Microscopy Juanxiu Xiao a, b, Tun Seng Herng c, Yang Guo a, Jun Ding c, Ning Wang a, Kaiyang Zeng b, * a b c

State Key Laboratory of Marine Resource Utilization in South China Sea, Hainan University, No. 58, Renmin Avenue, Haikou, Hainan, 570228, China Department of Mechanical Engineering, National University of Singapore, 9 Engineering Drive 1, Singapore, 117576 Department of Materials Science and Engineering, National University of Singapore, 9 Engineering Drive 1, Singapore, 117576

a r t i c l e i n f o

a b s t r a c t

Article history: Received 5 May 2019 Received in revised form 27 August 2019 Accepted 3 September 2019 Available online xxx

This paper presents multiple-modes Scanning Probe Microscopy (SPM) studies on characterize the correlation of resistance switching (RS) and polarization rotation (PR) in copper doped ZnO (ZnO:Cu) thin films. Firstly, the bipolar RS behavior is confirmed by conductive Atomic Force Microscopy (c-AFM). The PR with almost 180
phase angle is confirmed by using the Piezoresponse Force Microscopy (PFM) on the same location. In addition, it elucidates that obvious PR behavior can be observed in the sample with increasing Cu concentration by combining Kelvin Probe Force Microscopy (KPFM). Furthermore, it is found that the region with downward polarization has low resistance state (LRS), whereas the region with upward polarization has high resistance state (HRS). Moreover, the Piezoresponse Force Spectroscopy (PFS) and Switching Spectroscopy PFM (SS-PFM) measurements further confirm that the existence of the built-in voltage, Vbuilt-in is largest in the ZnO:Cu (8 at.%) film deposited at the oxygen partial pressure of 2  104 Torr. The schematic diagrams of energy band diagram with varied built-in field, Ebuilt-in, polarization directions and redistributed charges are presented to explain the correlation between RS and PR behavior. © 2019 The Chinese Ceramic Society. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Resistive switching Polarization rotation Built-in voltage ZnO:Cu thin film Next generation memory

1. Introduction Resistance random access memories (RRAMs) based on transition metal oxide (TMO) have enormous potentials in the applications for next generation nonvolatile memory [1,2]. The II-VI compounds, such as ZnO, usually have similar structures to silicon and have advantages of easier fabrication [3]. Hence, the resistive switching (RS) behavior in ZnO-based materials have been extensively studied in recent years, especially due to their easier switching properties, high ROFF/RON ratio, low set and reset voltages, higher breakdown voltages, and good retention and endurance properties [4e8]. However, the mechanisms underlying the RS behavior remain unclear, as many internal and external factors may affect these mechanisms. Especially, the coupling between the RS behavior and the polarization rotation (PR) behavior was observed

* Corresponding author. E-mail address: [email protected] (K. Zeng). Peer review under responsibility of The Chinese Ceramic Society.

in pure ZnO thin films and nanostructures [9e12]. The PR was called polarization switching or polarization orientation in the earlier papers; we define it as polarization rotation (PR) in this paper, as ZnO based materials are well-known piezoelectric materials, but not conventional ferroelectric materials. Recent studies have found that, if ZnO thin films and nanostructures are switched to high resistant state (HRS), the materials can show PR phenomena under an external electric field [9,11,12]. It was also approved that such behavior was intrinsic properties of the ZnO films and not electrostatic effects at the sample surface during the measurements [11,12]. On the one hand, this coupling has potential to make multiple-states of data storage, such as low and high resistance states, and/or up and down polarization states. On the other hand, this PR behavior may also complicate the underlying electronic transport phenomena in those materials. It was reported that the PR behavior in some ferroelectric materials might change the conduction band profile in the contact of the electrode/ferroelectric film and result in RS phenomena [13e15]. Vice versa, a built-in field, Ebuilt-in varying during the RS phenomena has a complicate influence on PR behavior [16e21]. Firstly, due to the existence of

https://doi.org/10.1016/j.jmat.2019.09.001 2352-8478/© 2019 The Chinese Ceramic Society. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article as: Xiao J et al., Correlation of resistance switching and polarization rotation in copper doped zinc oxide (ZnO:Cu) thin films studied by Scanning Probe Microscopy, Journal of Materiomics, https://doi.org/10.1016/j.jmat.2019.09.001

2

J. Xiao et al. / Journal of Materiomics xxx (xxxx) xxx

Ebuilt-in in an as-grown ferroelectric film, a favored spontaneous polarization direction is observed in PZT and BTO thin film [17,22,23]. In addition, when put the sandwiched metal/ferroelectric film/metal system under external electric field, the Ebuilt-in will be varied and consisted with free charges redistributing at the interfaces of the electrode/the ferroelectric film. Also these charges can change depletion region at the interfaces and arouse the RS behavior, simultaneously affect the PR behavior by acting as nuclei for PR or defects to aggravate polarization relaxation [24e26]. Nevertheless, the effects of underlying electronic transport and RS behavior on the PR behavior of ZnO-based materials are not comprehensively studied. Our previous preliminary results revealed that the deteriorative endurance of the RS behavior in ZnO:Cu thin films was triggered by the built-in field determined from the PR behavior [27]. Boppidi, P. K. R. et al., proposed that high anneal temperature could arouse the PR behavior in their ZnObased device, and the invoked PR could partly lead to the enhancement of the device RS performance [28]. However, the investigation of the RS and PR behavior in their study were not from the same location. In addition, the influence of the RS behavior on PR behavior was not discussed. A detailed study of the underlying mechanism is not conducted yet. Hence, it is necessary to investigate the coupling between the RS behavior and the PR behavior in ZnO based thin films, so as the underlying mechanism. The inevitable existence of intrinsic defects and intentional doping may contribute to the underlying mechanisms. Recent studies have indicated that the electrochemical migration of oxygen vacancies, VO, in the vicinity of the interface may result in RS and also favorable to the PR behavior [9,27,29]. Oxygen vacancies are inevitable to form under oxygen deficient conditions during the film deposition, especially by Pulsed Laser Deposition (PLD) [9,30]. Therefore, in this study, to investigate the VO effects, the thin films were deposited at different deposition oxygen partial pressure (PO2 ) by PLD technique. On the other hand, dopants can also tune the band structure of the ZnO samples [31]. Among doped ZnO compounds, the study shows that copper doped ZnO films might generate more defect states resulting in a better RS behavior [28,32]. It was reported that the PR behavior in copper doped ZnO materials could be also enhanced to some extend [26,33]. Our group's previous work also revealed that ZnO:Cu thin films with a Cu content of 0e10 at.% were wurtzite with the predominated caxis (002) texture. However, when the content of Cu content was above 11%, a secondary phase of CuO was detected which indicating an exceeded solid solubility of Cu in ZnO film [33]. Hence, ZnO:Cu thin films can be a good system for comprehensive understanding of the coupling of the local RS, the PR and the underlying electronic transport. To study the effects of Cu concentration, ZnO:Cu sapmles with 0 at.%, 2 at.%, 8 at.% and 12 at.% Cu content were deposited in this work. To understand the atomistic mechanisms of RS and PR behaviors in the ZnO:Cu thin films, a study at a single defect level on the same location is required. The applications of various Scanning Probe Microscopy (SPM) techniques have offered a pathway to address structure, electrical and electromechanical properties on the nanometer and atomic scales in the last three decades [34e38]. For example, conductive Atomic Force Microscopy (c-AFM) and Piezoresponse Force Microscopy (PFM) have been used extensively to study RS and PR behaviors in various piezo/ferroelectric materials [11,39,40]. It is also noticed that, by investigating on the surface potential with Kelvin Probe Force Microscopy (KPFM) technique, the generation/migration of charge carriers and polarization bound charges integrated with the underlying RS and PR behaviors can also be distinguished [41,42]. Therefore, herein we will apply the multiple-modes SPM techniques on the same location to have detailed study of the relationships between the RS and PR of the

ZnO:Cu thin films. 2. Materials and methods 2.1. Samples preparation The ZnO:Cu samples (with 0 at.%, 2 at.%, 8 at.% and 12 at.% Cu content) were deposited by using pulse laser deposition (PLD) technique under oxygen partial pressure of 1  106 Torr and 2  104 Torr. The thickness of the films was about 240 nm. The used substrates were commercial Pt-coated Si wafer (Addition Engineering Ltd, CA, USA), in which the Pt coating layer acted as bottom electrode for all of the SPM measurements. The details of preparation procedure was described in the previous work [33]. 2.2. Characterization The crystallinity of the films was characterized by using X-ray diffraction (XRD, Briker AXS D8 Advance) [33,43]. It has confirmed that the films are fully crystalline nature with a preferentially [0001] orientation texture. The Cu percentage of the films was investigated by energy dispersive spectroscopy (EDS) and X-ray photoemission spectroscopy (XPS) [33]; and the oxygen vacancy concentration was characterized by performing the Raman spectra and X-ray Absorption Spectroscopy (XAS) analysis [27]. The details of those measurements will not be repeated here. To study the correlations between the RS, PR and surface potential in ZnO:Cu samples, two sequence poling processes were firstly conducted by c-AFM and followed by KPFM and PFM measurements. It should be noticed that, in the c-AFM measurements, the bias was applied through the bottom electrode and the tip was grounded. Herein it is referred as “sample bias” (Fig. S1(a), Supplementary Information); whereas in the PFM and KPFM measurements, the bias was applied to the conductive tip, and the bottom Pt electrode of the sample was grounded, and this is referred as “tip bias” (Fig. S1(b), Supplementary Information). The two sequence poling processes were defined as “box-in-box” and “up-down” poling processes in this paper (Figs. S1(c) and S1(d): Supplementary Information). In the “box-in-box” poling processes, a sample bias of 10 V was firstly applied on a 5  5 mm2 area and followed by applying a sample bias of 10 V on a 2.5  2.5 mm2 area in the middle of the area poled by 10 V (Fig. S1(c), Supplementary Information). In the “up-down” poling process, sample biases of 10 V, 0 V, 10 V were applied to three adjacent 3  1 mm2 areas (Fig. S1(d), Supplementary Information). In these poling processes, the scan rate was set as 0.3 Hz. After these poling processes, current image of a larger area (10  10 mm2) that included these biased regions was obtained by c-AFM with a small DC sample bias of 1 V. To study the correlations between the resistance state, polarization direction and surface potential, the c-AFM biased region was then scanned under KPFM as well as PFM modes. In the PFM measurements, 1 V AC tip bias was used; the parameters for the KPFM measurements were the same as these used in the previous studies [43e45]. These correlations were further investigated by Piezoresponse Force Spectroscopy (PFS) technique. In the PFS measurements, the tip was fixed at an arbitrary location. A triangular waveform, which was composed by a sequence of square waves supposed with an AC bias (Fig. S2 Supplementary Information), was applied to the tip at a frequency of 200 mHz with alternative of DC bias-on and off. In this waveform, t1 is the time of bias-off, and t2 is the time of bias-on. Both times were set to be 25 ms. To avoid the contribution of electrostatic effect, the bias-induced remnant piezoresponse was acquired at every pulse between the adjacent voltage steps (i.e., bias-off state). The piezoelectric response hysteresis loop, Pr, is

Please cite this article as: Xiao J et al., Correlation of resistance switching and polarization rotation in copper doped zinc oxide (ZnO:Cu) thin films studied by Scanning Probe Microscopy, Journal of Materiomics, https://doi.org/10.1016/j.jmat.2019.09.001

J. Xiao et al. / Journal of Materiomics xxx (xxxx) xxx

calculated using the equation Pr ¼ A*cos(f) and usually plotted as a function of applied biases in the shape of the hysteresis loop, where A is the amplitude, and f is the phase angle, both are obtained at bias off state [46]. With the use of the high-voltage mode, PFS measurement can be performed at the bias higher than ±10 V. Hence, in this study, the pulse voltage was gradually increased until the saturated hysteresis loops were obtained, i.e., the phase angle change is ~180
. Eventually, the biases of ±10 V and ±15 V was applied to the ZnO:Cu (2 at.% and 8 at.%, deposited at PO2 ¼ 1  106 Torr) samples, and ±25 V was applied to the ZnO:Cu (0 at.%, 2% at.%, 8 at.% and 12 at.%, deposited at PO2 ¼ 2  104 Torr) samples, respectively. Furthermore, the SS-PFM (Switching Spectroscopy PFM) measurements were conducted in a grid of 64  64 points within a scanning area of 2  2 mm2. Both amplitude and phase loops were acquired, and then the piezoresponse hysteresis loops were calculated on all of the 64  64 points. The imprint was then determined from each hysteresis loop and then a twodimensional imprint map was generated for this 2  2 mm2 area. All of the SPM measurements are conducted with a commercial system (MFP-3D, Asylum Research, CA, USA), with commercially available Pt-coated Si tips (AC240TM, Olympus, Japan). The tip has the nominal spring constant of 2 N/m and nominal tip radius of approximately 28 nm and average resonance frequency of 65 kHz (specifications from manufacturer). 3. Results and discussion 3.1. Characteristics by multiple-modes SPM techniques Firstly, “set” and “reset” experiments at a fixed area are conducted with c-AFM technique. For the ZnO:Cu samples deposited under oxygen partial pressure of PO2 ¼ 2  104 Torr, the current is negligible from the c-AFM measurements (within applied bias of ±10 V) due to the high resistances of the ZnO:Cu samples. However, the ZnO:Cu samples deposited at oxygen partial pressure of PO2 ¼ 1  106 Torr have shown moderate conductivity. Fig. 1(a) and Fig. 2(a) are the current images after the “box-in-box” poling processes (defined in Fig. S1(c), Supplementary Information] on ZnO:Cu samples (2 at.% and 8 at.%, PO2 ¼ 1  106 Torr), respectively. In addition, it is found that 10 V can “set” samples from high resistance state (HRS) to low resistance state (LRS), and 10 V can “reset” samples from LRS to HRS. It is also noticed that the magnitude of the current at LRS decreases as the copper concentration increases from 2 at.% to 8 at.%, indicating the sample with 8 at.% Cu has higher resistance. Afterward, the relationships among the RS and the surface potential distribution, as well as the changes of the PR after the “set” and “reset” experiments in the ZnO:Cu samples are further studied by the KPFM and PFM measurements on the same location. Fig. 1(b)-(d) and 2(b)e(d) show the corresponding KPFM and PFM images for ZnO:Cu (2 at.%) and ZnO:Cu (8 at.%) sample respectively. The surface potential value of the biased regions is flattened by the potential of the unbiased region as in the previous studies [43e45]. Figs. 1(b) and 2(b) are the surface potential images after two times grounded tip scanning. It is found that, comparing with that of the unbiased region, the area biased by 10 V sample bias shows higher surface potential whereas the area biased by 10 V sample bias shows lower surface potential. It is also noted that the surface potential in the 10 V poled area in ZnO:Cu (8 at.%) sample is actually higher than that of the ZnO:Cu (2 at.%) sample after the same poling processes. These results suggest that the ZnO:Cu (8 at.%) sample (deposited atPO2 ¼ 1  106 Torr) may have a better charge storage behavior, which is consistent with that reported in the previous study [45]. Figs. 1(c) and 2(c) are the PFM phase images of the ZnO:Cu (2 at.% and 8 at.%) samples. From the histogram of phase images in Fig. 2(e),

3

polarizations with phase angle contrast of around 180
can be observed. It is found that numbers of polarizations are at the phase angle of about 45
, and other polarizations are around at the phase angle of about 135
. It is obvious that most of polarizations in the region poled by 10 V (green square: 2.5  2.5 mm2) are switched to upward direction (toward to the sample surface), and at the same time, this region is in HRS. On the other hand, it is also noticed that for the region biased with 10 V (red square: 5  5 mm2), most of polarizations are in downward direction (toward to the bottom electrode), and the region is in LRS. This suggests that the LRS region is associated with the polarizations in the downward direction. There are more polarizations at the phase angle of about 45
in the ZnO:Cu (8 at.%) sample, indicating more polarizations are in the downward direction in this sample. Furthermore, the corresponding responses can be also observed from the PFM amplitude images (Figs. 1(d) and 2(d)). Comparing the multiple-modes SPM results, it can be concluded that, when the sample is subjected to negative sample bias, the emerging of current is observed, which can compensate the downward polarization (Fig. 2(e)), and this may favor the incurring current. Therefore, in this negative sample biased region, a surface potential increase is also observed, and this corresponds to the presence of additional positive charges attributing to hole injections. On the other hand, when sample is subjected to positive sample bias, there is no observable current and the polarization is re-rotated to upward direction (Fig. 2(e)), the decrease in surface potential should due to negative charge ions diffusion to the sample surface. It should be noticed that as the doped copper concentration increasing, the polarization is rotated to downward direction within the region with LRS, and this is significantly different from that in the undoped ZnO film, in which the polarization rotation was only observed in the region with HRS [9]. In addition, PR retention or time dependent study is also conducted to confirm the phase change is an intrinsic property of the samples (Fig. S3). The 10 V/-10 V (“box-in-box”) tip bias poling processes are implemented on the ZnO:Cu (2 at.%) thin film sample deposited at PO2 ¼ 1  106 Torr. The PFM phase images obtained just after the poling processes/after 19 h are shown in Figs. S3(a) and S3(b). It is obvious that even after 19 h, some domains poled by 10 V tip bias are still in purple indicating the ZnO:Cu sample with a good PR maintainability. To further confirm these results, the correlations among the RS, PR and surface potential in ZnO:Cu (2 at.%) sample are further studied by conducting the “up-down” poling processes (Fig. S1(d), Supplementary Information). The current image shows that the area poled by 10 V sample bias is in LRS, whereas the area poled by 10 V sample bias is in HRS after removing the external electric field (Fig. 3(a)). The changes in the surface potential are also consistent with those observed from the “box-in-box” poling processes. Furthermore, Fig. 3(b) shows the surface potential can be easily removed by grounded tip scanning, indicating the charges in the LRS region are mainly screen charges. There are still some numbers of polarizations in the LRS region have re-rotated to the downward direction (Fig. 3(c)). However, it is also found that the switched polarizations cannot be maintained for long time in the LRS, the polarizations are switched back after about 14 h (Fig. 3(d)). Comparing with the PR retention effect investigated by “box-inbox” poling processes, the shorter PR remain time might due to varied poling processes, but also because of the different bias methods. 3.2. Mechanism of RS behavior It was assumed that the RS behavior in the ZnO:Cu films may be dominated by Schottky emissions [33,43]. This can be explained as following: Firstly, the Pt/ZnO:Cu interface is the Schottky contact

Please cite this article as: Xiao J et al., Correlation of resistance switching and polarization rotation in copper doped zinc oxide (ZnO:Cu) thin films studied by Scanning Probe Microscopy, Journal of Materiomics, https://doi.org/10.1016/j.jmat.2019.09.001

4

J. Xiao et al. / Journal of Materiomics xxx (xxxx) xxx

Fig. 1. Correlations between the RS, PR and surface potential in ZnO:Cu (2 at.% sample (PO2 ¼ 1  106 Torr) by 10 V/10 V (“box-in-box”) poling processes. A smaller DC sample bias of 1 V was applied to scan a 10  10 mm2 area after applying 10 V on the middle of 5  5 mm2 area (red square) and sequentially poling by 10 V on a central 2.5  2.5 mm2 area (green square). The scan rate is 1 Hz: (a) current image by c-AFM under sample bias; (b) the corresponding surface potential image by KPFM measurement under tip bias; (c) and (d) the corresponding PFM phase image and amplitude images under tip bias. All of the images are obtained on the same location.

because of the differences between the work functions of Pt (5.27 eV) and ZnO:Cu samples (4.48 and 5.01 eV for 0 at.% and 8 at.% respectively) [31]. Comparing with the one of ZnO:Cu (0 at.%) film, the enhanced work function of ZnO:Cu (8 at.%) film is due to the copper ions occupied on the Zn2þ sites [33], which can result in the formation of the trapping states and shifts the Fermi level toward the valence band [46]. The schematic energy band diagram without considering effect of spontaneous polarization in the ZnO:Cu films or external electric field is shown in Fig. 4(a). In addition, the most polarizations in un-biased region of ZnO:Cu films have the phase angle of approximately 135
(Figs. 1(c) and 2(c)), especially for the ZnO:Cu (8 at.%) film, this indicates that the majority of the polarizations are in the upward direction, i.e. in the direction of [0001] (yellow color). Hence, it is assumed that there is a positive Ebuilt-in in the ZnO:Cu films. Moreover, it also known that oxygen vacancies can act as donors that can release electrons to neutralize the positive charge bound by polarizations or annihilate with holes introduced by the copper ions to bend the conduction band downward [13,47,48]. This downward bended band structure can further augment work function difference, which may reduce the barrier height of the Schottky contact at the tip/film interface; whereas the negative polarization charges compensated by the positively charged oxygen vacancies may result in upward bended band structure and increase the Schottky barrier height at the bottom electrode/film interface. The schematic energy band diagram with considering effect of the Ebuilt-in and the partially screened spontaneous polarization in ZnO:Cu films is shown in

Fig. 4(b). Furthermore, the polarization rotation and the generated/ redistributed charges at the interfaces of the electrode/the ZnO:Cu thin film/tip may result in a varied Ebuilt-in which can alter the depletion region, and indirectly lead to the RS behavior in the film [16,17,49]. Fig. 4(c) and (d) is the schematic energy band diagram for the correlation between RS and PR with the considering of external electric field. When biased the bottom electrode with 10 V bias, the sample is in LRS and the polarization is rotated to downward (Fig. 4(c)). From the c-AFM, KPFM and PFM results, it is speculated that the abruptly occurred current is attributed to hole injections at the tip/film interface instead of the migration of positively charged ions. And the decreased width of the depletion region adds zero built-in field to the external electric field. Hence, the tip/film interface is in Ohmic-like contact. Whereas, when the bottom electrode is biased with 10 V (Fig. 4(d)), a built-in field antiparalleled to the external electric field is generated. On the one hand, the Ebuilt-in can reduce the external electric field. On the other hand, both of the re-rotated upward polarization and the migration/formation of positively charged oxygen vacancies can increase the width of the depletion region between the Pt tip/ZnO:Cu film. In other words, the Schottky barrier height at the tip/film interface is increased, so that switch the sample to high resistance state. In order to verify the built-in field in ZnO:Cu films, PFS experiments are conducted. The positive (Vp) and negative (Vn) coercive biases can be determined piezoresponse hysteresis loop measured by PFS experiments, hence, some critical parameters can be determined, for example, the imprint bias is defined as (jVpj- Vnj)/2

Please cite this article as: Xiao J et al., Correlation of resistance switching and polarization rotation in copper doped zinc oxide (ZnO:Cu) thin films studied by Scanning Probe Microscopy, Journal of Materiomics, https://doi.org/10.1016/j.jmat.2019.09.001

J. Xiao et al. / Journal of Materiomics xxx (xxxx) xxx

5

Fig. 2. Correlations between the RS, PR and surface charge distribution in ZnO:Cu (8 at.%) sample (PO2 ¼ 1  106 Torr) by 10 V/10 V (“box-in-box”) poling processes. A smaller DC sample bias of 1 V was applied to scan a 10  10 mm2 area after applying 10 V on the middle of 5  5 mm2 area (red square) and sequentially poling by 10 V on a central 2.5  2.5 mm2 area (green square). The scan rate is 1 Hz: (a) current image by c-AFM under sample bias; (b) the corresponding surface potential image by KPFM measurement under tip bias; (c) and (d) the corresponding PFM phase image and amplitude images under tip bias. All of the images are obtained on the same location; (e) the histograms of the 5  5 mm2 areas (subjected to 10 V) in Fig. 3(c) and Fig. 4(c), and schematic drawing of the polarization reversal by 10 V/-10 V sample voltage.

and the built-in field is Ebuilt-in ¼ Vbuilt-in/t ¼ (jVpj-jVnj)/t, t is the film thickness [13,49]. Fig. 5(a) and (b) are the PFS measured phase and amplitude loops for the ZnO:Cu (2 at.%) thin film sample deposited at PO2 ¼ 1  106 Torr. In this measurement, PFS loops are taken from eight random locations on a 2  2 mm2 area of each sample. The calculated piezoresponse hysteresis loop is shown in Fig. 5(c). It is clear that phase angles in the ZnO:Cu (2 at.%) sample change approximately 180
as the function of the applied voltage, and this PR phenomena may be attributed by two factors: (i) the partial substitution of host Zn2þ ions by smaller Cu2þ ions [33]; and (ii) the existence of oxygen vacancy [27,33,50]. Furthermore, it is worth to mention that there are two obvious asymmetries in the hysteresis loops (Fig. 5(a) and (b)). It is found that the positive coercive bias is 5.76 V and the negative coercive bias is 3.76 V, and this leads to a shift of 2.16 V along the bias axis. This shift is primarily corresponding to the built-in voltage. In addition, this PR behavior is also observed when the sample is placed in the closed cell filled with

flowing synthetic air (<5 ppm water) or flowing argon gas (<0.02 ppm water and <0.01 ppm oxygen) respectively, and these measurements in controlled gas environments can avoid the moisture and oxygen effects from the ambient air condition (Fig. 5(d)). It is noted that, when there is no moisture or oxygen in the environments, both positive and negative coercive biases are increased. In addition, the PR loops have become more symmetric and this indicates that the build-in voltage has been reduced when there is no moisture or oxygen in the environments. Furthermore, the built-in voltage, i.e., twice of imprint bias is also confirmed by conducting SS-PFM on the same sample. The imprint image can be determined from each hysteresis loop of 64  64 points on 2  2 mm2 area (Fig. S4 Supplementary Information). The imprint map shows that there are many points show larger imprint values (red points). It should be also noticed that the points in dark red or dark purple are the locations with un-saturated hysteresis loops [49]. Moreover, this PR behavior is observed in all of the ZnO:Cu

Please cite this article as: Xiao J et al., Correlation of resistance switching and polarization rotation in copper doped zinc oxide (ZnO:Cu) thin films studied by Scanning Probe Microscopy, Journal of Materiomics, https://doi.org/10.1016/j.jmat.2019.09.001

6

J. Xiao et al. / Journal of Materiomics xxx (xxxx) xxx

Fig. 3. Correlations between the RS, PR and charge distribution in ZnO:Cu (2 at.%) sample (PO2 ¼ 1  106 Torr) by 10 V/0 V/10 V (“up-down”) poling processes. A smaller DC sample bias of 1 V was applied to scan a 6  6 mm2 area after applying 10V/0V/10V on the three adjacent areas of 3  1 mm2 (enclosed by red lines). The scan rate is 1 Hz: (a) current image by c-AFM under sample bias; (b) the corresponding surface potential image by KPFM measurement under tip bias; (c) PFM phase image; and (d) PFM phase image obtained 14 h later after the poling process. All of the images are obtained from the same location.

Fig. 4. Schematic energy band diagrams and Schottky contact for Pt tip/ZnO:Cu thin film/Pt bottom electrode system: (a) in an ideal thermal equilibrium condition without considering the spontaneous polarization in ZnO:Cu film and external electric field; (b) modified band diagram with considering effect of the Ebuilt-in and the partially screened spontaneous polarization in ZnO:Cu films; (c) When biased the bottom electrode with 10 V bias, the built-in field is zero and the polarization is rotated to downward direction towards the bottom electrode; and (d) when the bottom electrode is biased with 10 V, a built-in field antiparalleled to the external electric field is generated and the polarization is re-rotated to upward towards to the tip. In all of the figures, Wd indicates the width of the depletion region.

samples. Table 1 summarizes the average built-in voltages determined from the SS-PFM results for all of the samples. It is obvious that the Vbuilt-in increases in the ZnO:Cu sample with increasing PO2 from 1  106 Torr to 2  104 Torr. The Vbuilt-in is also increased in

the ZnO:Cu sample with increasing Cu concentration from 0 at.% to 8 at.%, whereas the values of Vbuilt-in is decreased when the Cu concentration of the sample increased to the maximum percentage of 12 at.% attributing to the formation of the secondary phase such

Please cite this article as: Xiao J et al., Correlation of resistance switching and polarization rotation in copper doped zinc oxide (ZnO:Cu) thin films studied by Scanning Probe Microscopy, Journal of Materiomics, https://doi.org/10.1016/j.jmat.2019.09.001

J. Xiao et al. / Journal of Materiomics xxx (xxxx) xxx

7

Fig. 5. (a) and (b): PFS measured phase and amplitude loops; (c) calculated piezoresponse hysteresis loops in eight random locations for ZnO:Cu (2 at.%) sample (PO2 ¼ 1  106 Torr), based on the equation of Pr ¼ A*cos(f), where A is the amplitude at bias off state, f is the phase angle at bias off state; (d) Hysteresis loops, Pr for the same sample by conducting PFS measurements in the ambient air, and in closed cell filled with flowing synthetic air (<5 ppm water) and flowing argon gas (<0.02 ppm water and <0.01 ppm oxygen) respectively.

Table 1 Summary of the average built-in voltages determined from the SS-PFM results for each of the samples. Po2(Torr)

1  106

1  106

2  104

2  104

2  104

2  104

Cu concentration Vbuilt-in(V)

2% 2.16

8% 3.18

0% 2.51

2% 5.26

8% 13.26

12% 11.92

as CuO in the sample [33].

appeared to influence the work reported in this paper.

4. Summary and conclusion

Acknowledgments

In this work, the resistive switching behavior is observed in the ZnO:Cu thin films by“set” and “reset” process at a fixed area. With the increasing copper concentration, the film resistance is increased; and HRS and LRS are more distinguished. Increased charge storage and polarization rotation behavior are also observed by defects engineering with doping more copper. By comparing cAFM, PFM and KPFM results on the same location, clear couplings between RS behavior and PR behavior are demonstrated. Finally, the RS behavior affected by the existence of the built-in voltage is further confirmed by PFS measurements. The results show that both the copper and oxygen vacancy can tune the band structure and built-in voltage. These results give a better interpretation of the couplings among the RS behavior, the PR behavior and surface charges distribution in ZnO:Cu thin films; and the exploration of the underlying mechanisms of RS may imply the development of ZnO:Cu as next generation memory.

This work was supported by the Start-up Research Foundation of Hainan University [grant no. KYQD(ZR)1816]; The author also would like to thank for the support from Ministry of Education, Singapore, through National University of Singapore (NUS) under the Academic Research Fund (grant no. R265-000-496-112).

Conflicts of interest The authors declare that they have no known competing financial interests or personal relationships that could have

Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.jmat.2019.09.001. References [1] Sawa A. Resistive switching in transition metal oxides. Mater Today 2008;11: 28e36. [2] Pan F, Gao S, Chen C, Song C, Zeng F. Recent progress in resistive random access memories: materials, switching mechanisms, and performance. Mater Sci Eng R 2014;83:1e59. € an S, Avrutin V, Cho S-J, [3] Ozgür Ü, Alivov YI, Liu C, Teke A, Reshchikov M, Dog Morkoc H. A comprehensive review of ZnO materials and devices. J Appl Phys 2005;98(4):041301. [4] Chang W-Y, Lai Y-C, Wu T-B, Wang S-F, Chen F, Tsai M-J. Unipolar resistive switching characteristics of ZnO thin films for nonvolatile memory applications. Appl Phys Lett 2008;92(2):022110.

Please cite this article as: Xiao J et al., Correlation of resistance switching and polarization rotation in copper doped zinc oxide (ZnO:Cu) thin films studied by Scanning Probe Microscopy, Journal of Materiomics, https://doi.org/10.1016/j.jmat.2019.09.001

8

J. Xiao et al. / Journal of Materiomics xxx (xxxx) xxx

[5] Yang YC, Pan F, Liu Q, Liu M, Zeng F. Fully room-temperature-fabricated nonvolatile resistive memory for ultrafast and high-density memory application. Nano Lett 2009;9(4):1636e43. [6] Xu N, Liu LF, Sun X, Chen C, Wang Y, Han DD, Liu XY, Han RQ, Kang JF, Yu B. Bipolar switching behavior in TiN/ZnO/Pt resistive nonvolatile memory with fast switching and long retention. Semicond Sci Technol 2008;23(7):075019. [7] Simanjuntak FM, Panda D, Wei K-H, Tseng T-Y. Status and prospects of ZnObased resistive switching memory devices. Nanoscale Res Lett 2016;11(1): 368. [8] Zhu L, Zhou J, Guo Z, Sun Z. An overview of materials issues in resistive random access memory. J Materiomics 2015;1(4):285e95. [9] Herng TS, Kumar A, Ong CS, Feng YP, Lu YH, Zeng KY, Ding J. Investigation of the non-volatile resistance change in noncentrosymmetric compounds. Sci Rep 2012;2:587. [10] Qi J, Olmedo M, Zheng JG, Liu J. Multimode resistive switching in single ZnO nanoisland system. Sci Rep 2013;3:2405. [11] Xiao J, Ong WL, Guo Z, Ho GW, Zeng K. Resistive switching and polarization reversal of hydrothermal-method-grown undoped zinc oxide nanorods by using scanning probe microscopy techniques. ACS Appl Mater Interfaces 2015;7(21):11412e22. [12] Xiao J, Zeng K, Wong L-M, Wang S. Correlation of the resistive switching and polarization switching in zinc oxide thin films using scanning probe microscopy techniques. J Mater Res 2015;30(22):3431e42. [13] Yang CH, Seidel J, Kim SY, Rossen PB, Yu P, Gajek M, Chu YH, Martin LW, Holcomb MB, He Q, Maksymovych P, Balke N, Kalinin SV, Baddorf AP, Basu SR, Scullin ML, Ramesh R. Electric modulation of conduction in multiferroic Cadoped BiFeO3 films. Nat Mater 2009;8(6):485e93. [14] Jiang AQ, Wang C, Jin KJ, Liu XB, Scott JF, Hwang CS, Tang TA, Lu HB, Yang GZ. A resistive memory in semiconducting BiFeO(3) thin-film capacitors. Adv Mater 2011;23(10):1277e81. [15] Fan Z, Fan H, Lu Z, Li P, Huang Z, Tian G, Yang L, Yao J, Chen C, Chen D, Yan Z. Ferroelectric diodes with charge injection and trapping. Phys Rev Appl 2017;7(1):014020. [16] Yuan G-L, Wang J. Evidences for the depletion region induced by the polarization of ferroelectric semiconductors. Appl Phys Lett 2009;95(25):252904. [17] Gao P, Nelson CT, Jokisaari JR, Baek S-H, Bark CW, Zhang Y, Wang E, Schlom DG, Eom C-B, Pan X. Revealing the role of defects in ferroelectric switching with atomic resolution. Nat Commun 2011;2:591. [18] Kim Y-M, Morozovska A, Eliseev E, Oxley MP, Mishra R, Selbach SM, Grande T, Pantelides S, Kalinin SV, Borisevich AY. Direct observation of ferroelectric field effect and vacancy-controlled screening at the BiFeO3/LaxSr1 xMnO3 interface. Nat Mater 2014;13(11):1019e25. [19] Ievlev AV, Morozovska AN, Eliseev EA, Shur VY, Kalinin SV. Ionic field effect and memristive phenomena in single-point ferroelectric domain switching. Nat Commun 2014;5:4545. [20] Tan Z, Tian J, Fan Z, Lu Z, Zhang L, Zheng D, Wang Y, Chen D, Qin M, Zeng M, Lu X. Polarization imprint effects on the photovoltaic effect in Pb (Zr, Ti) O3 thin films. Appl Phys Lett 2018;112(15):152905. [21] Fan H, Chen C, Fan Z, Zhang L, Tan Z, Li P, Huang Z, Yao J, Tian G, Luo Q, Li Z. Resistive switching and photovoltaic effects in ferroelectric BaTiO3-based capacitors with Ti and Pt top electrodes. Appl Phys Lett 2017;111(25):252901. [22] Gruverman A, Wu D, Lu H, Wang Y, Jang H, Folkman C, Zhuravlev MY, Felker D, Rzchowski M, Eom C-B. Tunneling electroresistance effect in ferroelectric tunnel junctions at the nanoscale. Nano Lett 2009;9(10):3539e43. [23] Balke N, Jesse S, Li Q, Maksymovych P, Baris Okatan M, Strelcov E, Tselev A, Kalinin SV. Current and surface charge modified hysteresis loops in ferroelectric thin films. J Alloy Comp 2015;118(7):072013. [24] Gerra G, Tagantsev A, Setter N. Ferroelectricity in asymmetric metal/ferroelectric/metal heterostructures: a combined first principles phenomenological approach. Phys Rev Lett 2007;98(20):207601. [25] Afanasjev VP, Petrov AA, Pronin IP, Tarakanov EA, Ju Kaptelov E, Graul J. Polarization and self-polarization in thin PbZr1-xTixO3 (PZT) films. J Phys Condens Matter 2001;13(39):8755. [26] Xiao J, Herng TS, Ding J, Zeng K. Polarization rotation in copper doped zinc oxide (ZnO: Cu) thin films studied by Piezoresponse Force Microscopy (PFM) techniques. Acta Mater 2017;123:394e403. [27] Xiao J, Herng TS, Ding J, Zeng K. Resistive switching behavior in copper doped zinc oxide (ZnO: Cu) thin films studied by using scanning probe microscopy techniques. J Alloy Comp 2017;709:535e41. [28] Boppidi PKR, Raj PMP, Challagulla S, Gollu SR, Roy S, Banerjee S, Kundu S. Unveiling the dual role of chemically synthesized copper doped zinc oxide for resistive switching applications. J Appl Phys 2018;124(21):214901. [29] Qi J, Olmedo M, Ren J, Zhan N, Zhao J, Zheng J-G, Liu J. Resistive switching in single epitaxial ZnO nanoislands. ACS Nano 2012;6(2):1051e8. [30] Jin B, Bae S, Lee S, Im S. Effects of native defects on optical and electrical properties of ZnO prepared by pulsed laser deposition. Mater Sci Eng B 2000;71(1e3):301e5. [31] McCluskey MD, Jokela S. Defects in Zno. J Appl Phys 2009;106(7):10. [32] Jia C, Dong Q, Zhang W. Effect of incorporating copper on resistive switching

properties of ZnO films. J Alloy Comp 2012;520:250e4. [33] Herng TS, Wong MF, Qi D, Yi J, Kumar A, Huang A, Kartawidjaja FC, Smadici S, Abbamonte P, Sanchez-Hanke C, Shannigrahi S, Xue JM, Wang J, Feng YP, Rusydi A, Zeng K, Ding J. Mutual ferromagnetic-ferroelectric coupling in multiferroic copper-doped ZnO. Adv Mater 2011;23(14):1635e40. [34] Li T, Zeng K. Probing of local multifield coupling phenomena of advanced materials by scanning probe microscopy techniques. Adv Mater 2018;30(47): 1803064. [35] S.V. Kalinin, A. Gruverman, Scanning probe microscopy: electrical and electromechanical phenomena at the nanoscale, Springer2007. [36] Lu W, Wong L-M, Wang S, Zeng K. Local phenomena at grain boundaries: an alternative approach to grasp the role of oxygen vacancies in metallization of VO2. J Materiomics 2018;4(4):360e7. [37] Lu W, Wong L-M, Wang S, Zeng K. Effects of oxygen and moisture on the IV characteristics of TiO2 thin films. J Materiomics 2018;4(3):228e37. [38] Wang H, Zeng K. Domain structure, local surface potential distribution and relaxation of Pb (Zn1/3Nb2/3) O3e9% PbTiO3 (PZNe9% PT) single crystals. J Materiomics 2016;2(4):309e15. [39] Garcia V, Fusil S, Bouzehouane K, Enouz-Vedrenne S, Mathur ND, Barthelemy A, Bibes M. Giant tunnel electroresistance for non-destructive readout of ferroelectric states. Nature 2009;460(7251):81. [40] Seidel J, Maksymovych P, Batra Y, Katan A, Yang S-Y, He Q, Baddorf AP, Kalinin SV, Yang C-H, Yang J-C. Domain wall conductivity in La-doped BiFeO3. Phys Rev Lett 2010;105(19):197603. [41] Kalinin S, Bonnell D. Local potential and polarization screening on ferroelectric surfaces. Phys Rev B 2001;63(12). [42] Kim Y, Bae C, Ryu K, Ko H, Kim YK, Hong S, Shin H. Origin of surface potential change during ferroelectric switching in epitaxial PbTiO3 thin films studied by scanning force microscopy. Appl Phys Lett 2009;94(3):032907. [43] Kumar A, Herng TS, Zeng K, Ding J. Bipolar charge storage characteristics in copper and cobalt co-doped zinc oxide (ZnO) thin film. ACS Appl Mater Interfaces 2012;4(10):5276e80. [44] Herng TS, Qi DC, Berlijn T, Yi JB, Yang KS, Dai Y, Feng YP, Santoso I, S anchezHanke C, Gao XY, Wee ATS, Ku W, Ding J, Rusydi A. Room-temperature ferromagnetism of Cu-doped ZnO films probed by soft X-ray magnetic circular dichroism. Phys Rev Lett 2010;105(20). [45] Wong MF, Herng TS, Zhang Z, Zeng K, Ding J. Stable bipolar surface potential behavior of copper-doped zinc oxide films studied by Kelvin probe force microscopy. Appl Phys Lett 2010;97(23):232103. [46] Jesse S, Lee HN, Kalinin SV. Quantitative mapping of switching behavior in piezoresponse force microscopy. Rev Sci Instrum 2006;77(7):073702. [47] Jiang AQ, Wang C, Jin KJ, Liu XB, Scott JF, Hwang CS, Tang TA, Lu HB, Yang GZ. A resistive memory in semiconducting BiFeO3 thin-film capacitors. Adv Mater 2011;23(10):1277e81. [48] Osada M, Sakemi T, Yamamoto T. The effects of oxygen partial pressure on local structural properties for Ga-doped ZnO thin films. Thin Solid Films 2006;494(1e2):38e41. [49] Jesse S, Baddorf AP, Kalinin SV. Switching spectroscopy piezoresponse force microscopy of ferroelectric materials. Appl Phys Lett 2006;88(6):062908. [50] Scott J. Applications of modern ferroelectrics. Science 2007;315(5814):954e9.

Dr. Juanxiu Xiao is an associate professor at State Key Laboratory of Marine Resource Utilization in South China Sea, Hainan University. She received her Ph.D degree from Department of Mechanical Engineering, National University of Singapore in January 2016. Then she continued her research career at Materials Science and Engineering Department, National University of Singapore. Her current interests are mainly focused on the functionalities of transitional metal oxide materials and devices, and their applications in information storage, environment and energy. Dr. Xiao has published 28 journal papers and with Hindex of 14.

Dr. Tun Seng Herng is a research fellow at Department of Materials Science and Engineering, National University of Singapore. He received B.Eng. (2003) and P.hD (2008) from in Electrical and Electronics Engineering from Nanyang Technological University (NTU). Then he carried out research career in National University of Singapore (Materials Science and Engineering Department). His research interests are 3D printing, magnetic materials, multiferroic and electronics sensor. Dr. Tun has published more than 100 journal papers and with H-index of 24.

Please cite this article as: Xiao J et al., Correlation of resistance switching and polarization rotation in copper doped zinc oxide (ZnO:Cu) thin films studied by Scanning Probe Microscopy, Journal of Materiomics, https://doi.org/10.1016/j.jmat.2019.09.001

J. Xiao et al. / Journal of Materiomics xxx (xxxx) xxx Ms. Yang Guo is a graduate student at State Key Laboratory of Marine Resource Utilization in South China Sea, Hainan University. She received her bachelor degree from Department of Applied Chemistry, North China Electric Power University in 2018. Her research area is preparing transitional metal oxide related materials and devices, and exploring their applications in information storage, environment and energy.

Dr. Ding Jun is a professor at Department of Materials Science and Engineering, National University of Singapore. Prof Ding Jun has been working in the area of nanomagnetics and spintronics for many years. He has paid a particular attention on these materials and devices in different applications, including spintronic structures in information storage, nanoparticles in biomedical and environmental applications, magnetic sensors, magnetic energy harvesters and metamaterials. Recently, His research has also been concentrated on additive manufacturing (3D Printing). His research has been focused on the development of starting mateirals for fabrication of multi-functioanal devices/structures of metal, ceramics, polymer and composite. Dr. Ding has published more than 390 journal papers and with H-index of 73.

9 Dr. Ning Wang is a professor at State Key Laboratory of Marine Resource Utilization in South China Sea, Hainan University. He got his Ph.D in materials science from Tsinghua University in January 2007. Then he joined University of Electronic Science and Technology of China as an associate professor. He worked as a post-doctoral fellow at Nagoya University, Japan during November 20 08 to November 2011 and a visiting scholar in University of California, Berkeley, USA during November 2013 to November 2014. Prof. Wang's research group aims to using multidisciplinary to address the issues in environment and energy. Currently, his main research interests are solar cell, extraction uranium/lithium from seawater and anti-bifouling.

Dr. Kaiyang Zeng is an associate professor at Department of Mechanical Engineering, National University of Singapore. He got his Ph.D in materials science from Royal Institute Technology (KTH), Sweden. His main research areas include using Scanning Probe Microscopy based techniques to study the multifield coupling phenomena in advanced materials, such as high performance piezo/ferroelectric materials, biomaterials, supramolecular materials, oxide materials, and energy storage materials. Dr. Zeng has published more than 220 journal papers and with H-index of 45.

Please cite this article as: Xiao J et al., Correlation of resistance switching and polarization rotation in copper doped zinc oxide (ZnO:Cu) thin films studied by Scanning Probe Microscopy, Journal of Materiomics, https://doi.org/10.1016/j.jmat.2019.09.001