ITO flexible memory device

Accepted Manuscript Effect of fatigue fracture on the resistive switching of TiO2-CuO film/ITO flexible memory device Jian-Chang Li, Bo Chen, Yu Qian PII:

S1567-1739(18)30133-0

DOI:

10.1016/j.cap.2018.05.009

Reference:

CAP 4750

To appear in:

Current Applied Physics

Received Date: 1 February 2018 Revised Date:

23 April 2018

Accepted Date: 8 May 2018

Please cite this article as: J.-C. Li, B. Chen, Y. Qian, Effect of fatigue fracture on the resistive switching of TiO2-CuO film/ITO flexible memory device, Current Applied Physics (2018), doi: 10.1016/ j.cap.2018.05.009. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Effect of fatigue fracture on the resistive switching of TiO2-CuO film/ITO flexible memory device

RI PT

Jian-Chang Li,* Bo Chen, and Yu Qian

SC

Vacuum and fluid engineering research center, Northeastern University, Shenyang 110819, P. R.

Corresponding author: Jian-Chang Li, Professor

AC C

EP

TE D

E-mail: [email protected]

M AN U

China

1

ACCEPTED MANUSCRIPT

Abstract We fabricated the GaIn/TiO2-CuO/ITO resistive memory and studied the effect of fatigue fracture on the switching performance. The device shows the stable bipolar

RI PT

resistive switching over 108 s under ambient condition. The ON/OFF ratio decreases seriously with increase of bending cycles. The main fatigue fracture caused by dynamic strain includes micro defect between nanoparticles, vertical crack along the

SC

film thickness and interfacial delamination between layers. Finite element analysis

M AN U

indicates that channel crack plays a key role to cause the interfacial delamination between function layer and ITO electrode. The channel crack and interfacial delamination can hinder the formation of tree−like conduction filaments. Moreover, oxygen via the cracks can be easily transformed to ions and reduce the density of

TE D

oxygen vacancies under the catalytic assistance of CuO. Our studies may provide some useful information for inorganic materials applied in flexible nonvolatile

EP

memory.

AC C

Keywords: TiO2−CuO, Bipolar resistive switching, Finite element analysis (FEA); Fatigue fracture; p−n heterojunction

2

ACCEPTED MANUSCRIPT

1. Introduction Resistive random access memory (RRAM) has received a great of attention for non-volatile memory application. Resistive switching were reported previously for

RI PT

metal oxide films of TiO2 [1], ZnO [2], Al2O3 [3] and CuxO [4, 5]. Among all the material system, the p-n heterostructure has been widely applied to the resistive memory and switching characteristic has been studied [6]. For example, Duan et al.

SC

demonstrated that a space charge region exists in the Pt/Cu2O/WO2/FTO memory

M AN U

heterojunction [7]. The leakage current of Ni/p-NiO/i-ZnO/n-ITO structure is found to be dominated by deep defects in the ZnO layer [8], while Li et al. attributed the switching mechanism of PET/ITO/p-ZnO/n-TiO2/Cu device to the formation and rupture of oxygen vaccines [9]. In addition to the extensive research of

TE D

heterostructured nanoparticles, the resistive switching of heterostructured nanowires and nanodots were also explored. Lee et al. reported bipolar switching of n-ZnO nanowires deposited on p-Si substrates, which can be ascribed to the gradual

EP

redistribution of oxygen vacancies and trapping of electrons in the ZnO nanowires

AC C

[10]. Current rectification and resistive switching were observed in the p-BiFeO3/n-Nb-SrTiO3 nanodots with polarization modulated p-n heterojunctions [11]. Interestingly, K.M.Kim found that the interface of n−TiO2/ p−NiO can control the formation and rupture of conduction filaments, which reduce the reset time and increase ON/OFF ratio [12]. It is shown that p-n heterostructure may play a key role in improving the stability and ON/OFF ratio of such memory devices [6].

3

ACCEPTED MANUSCRIPT With the development of wearable and transparent electronics, flexible RRAM attracts much attention for bendable systems [13-15]. The good flexibility and mechanical endurance of Ru/Lu2O3/ITO memory device was believed owing to the

RI PT

high ductility of Lu2O3 thin film and Ru electrode. Wang et al. studied the Ni/TiO2/Cu junctions at both tension and compression conditions, and found that the switching performance was slightly degraded [16]. Some groups investigated the flexibility of

SC

bilayer structures with resistive switching. The ON/OFF ratio of Al2O3/ZnO bilayer

M AN U

decreased after outer-bending 1000 times [17]. Electrical reliability and mechanical stability of CsPbBr3/PEDOT:PSS layers were studied, which shows good mechanical stability under 100 times bending [18]. However, the adhesion between the bilayers reduces largely with increase of bending cycles, which limits the device flexibility.

TE D

Junctions of p-n nanocomposite not only have lower set voltage and higher reset current [19], but can avoid shortcoming of poor adhesion existing in such bilayer structure under bending conditions.

EP

In our previous studies, Yuan et al. studied bending effect on the resistive

AC C

switching of NiO/ZnO nanocomposite film and revealed the crack in the function layer may influence the switching performance [20]. Nevertheless, the growth path of crack and adhesion between nanocomposite film and ITO were not taken into consideration. Herein, high-performance flexible RRAM was fabricated by coupling two semiconductor nanocomposites (n-TiO2 and p-CuO) in order to study the fatigue fracture of GaIn/TiO2-CuO/ITO flexible memory device under bending tests.

4

ACCEPTED MANUSCRIPT

2. Experiment All the chemical reagents are of analytical grade without further purification. Tetrabutyl titanate (C16H36O4Ti) (30ml), absolute ethyl alcohol (60ml) and

RI PT

concentrated nitric acid (3ml) were used as raw materials, solvent and sol stabilizer, respectively. First, to get the titanium sol, tetrabutyl titanate was added to the mixed

SC

solution of absolute ethyl alcohol and concentrated nitric acid. The mixture was stirred for 1h at room temperature to get faint yellow sols. Second, the sols were put

M AN U

into a poly-tetrafluoroethylene pressure vessel to obtain nanoparticles by hydrothermal treatment for 2h at 150 ºC. Finally, the product was purified by repeated centrifugation in pure ethyl alcohol to obtain TiO2 nanoparticles. Details of Cu(OH)2 sols fabrication were described elsewhere [21]. To fabricate CuO-TiO2 nanoparticles

TE D

(NPs), a solution of Cu(OH)2 sols and TiO2 sols were stirred thoroughly and then transferred into a poly-tetrafluoroethylene pressure vessel for 2h hydrothermal

EP

crystallization at 150 °C. The TiO2-CuO NPs were got after centrifugation in absolute ethyl alcohol. The mixed solution was spin-coated with 5 s for 800 rpm and 30 s for

AC C

2000 rpm onto the ITO. Finally, the nanocomposite film was heated on hotplate at 60 °C for 5 min.

The 200 nm-thick films on silicon were used for X-ray diffraction (XRD) and scanning electron microscopy (SEM) analysis, while samples on ITO and glass were subjected to electrical and optical characterizations, respectively. Current-voltage (I-V) curves were recorded using a LK-2005Z potentiostat with device structure of GaIn/TiO2-CuO film/ITO. The ITO bottom electrode was grounded, and the bias was 5

ACCEPTED MANUSCRIPT applied to the GaIn top electrode. The method of bending tests is described elsewhere [20]. All the processes were carried out under ambient conditions.

3. Result and discussion

RI PT

Fig. 1(a) shows the XRD pattern of the TiO2-CuO nanocomposite film, and the clear peak at θ=36.5º is indexed to the CuO (101). A weak diffraction peak at θ=48.6º

SC

is detected for TiO2 (-111). As calculated by Scherrer equation, the size of CuO and TiO2 nanoparticles are about 170.6 nm and 10 nm, respectively. Bipolar switching

M AN U

exits in the GaIn/TiO2-CuO/ITO device without any forming process (see Fig. 1(b)), which is due to the high-density oxygen deficient defects in the functional layer [22]. When the bias voltage swept from 0 to 4 V, a sudden current increase was observed at − 1.1 V (Vset), and the junction switched from a high-resistance state (HRS) to

TE D

low-resistance state (LRS). When the bias scanned toward 4 V, the device abruptly switched from LRS to HRS at 1.7 V (Vreset).

20

30

40

50

2θ (degree)

(b)

TiO2:CuO=1:1 OFF

1 Current (mA)

(-111)

AC C

EP

Intensity (a.u.)

(101)

(a)

0

ON

-1

60

70

-4

-2

0 Voltage (V)

2

4

6

ACCEPTED MANUSCRIPT (d) 2

ON/OFF ratio

3.0 3000

1.5

0.0

2500

-1.5 2000

3500

Von

Voff

ON/OFF ratio 3000

1 2500

0 2000

-1

ON/OFF ratio

Voff

ON/OFF ratio Threshold voltage (V)

Threshold voltage (V)

Von

1500

-2

RI PT

3500

(c)

1000

-3.0 2:1

1:1

0.5:1

0:1

TiO2: CuO

200

300

400

500

600

Film thickness (nm)

Fig. 1 (a) XRD pattern of the TiO2-CuO nanocomposite film. (b) Typical I-V curves of the

SC

GaIn/TiO2-CuO film/ITO device with schematic structure in the inset. (c) and (d) Effect of

M AN U

different TiO2/CuO ratio and nanocomposite film thickness on the threshold voltage and ON/OFF ratio, respectively.

The composite ratio plays a key role on switching characteristics of the TiO2-CuO thin film. We studied the ON/OFF ratio and threshold voltage with

TE D

different composite ratio. The ON/OFF ratio is defined as current in the low resistant divided by current in the high resistant. As shown in Fig. 1(c), with increase of CuO

EP

ratio from 33.3 % to 100 %, Von and Voff increase from 1.9 to 2.4 V and 2.1 to 2.8 V, respectively. However, the ON/OFF ratio decreases about 30%. The reason is that

AC C

there are more CuO nanoparticles as catalytic assistance to transform oxygen molecule to ions and reduce the density of oxygen vacancy [23]. Good device performance was obtained for the TiO2/CuO film with 2:1 ratio, which was thus selected for following studies. The effect of TiO2-CuO nanocomposite film thickness on the resistive switching is also taken into consideration. As given in Fig. 1(d), the ON/OFF ratio decreased nearly one time when the thickness of nanocomposite film increase from 200 nm to 7

ACCEPTED MANUSCRIPT 600 nm. These results can be ascribed to more oxygen vaccines in the thinner functional layer [24]. Therefore, 200 nm-thick TiO2-CuO nanocomposite film was chosen for the later bending tests. V

(b) 3000

reset

2 1 0

0

1500

5

10 Cycles

15

20

RI PT

2250

V set

3

ON/OFF ratio

Threshold voltage (V)

ON/OFF ratio

(a) 3000

2250

1500

ON/OFF ratio

0 0

5

10

15

20

0

10

102

103

104

105

bent unbent

106

107

108

Retention time (s)

M AN U

Cycles

SC

750 750

Fig. 2 (a) The ON/OFF ratio of GaIn/TiO2-CuO film/ITO device under 30 sweeping cycles. The inset shows the distribution of threshold voltage in the 30 cycles. (b) Retention properties of the device under bending and unbending conditions.

TE D

To exam the reliability of GaIn/TiO2-CuO film/ITO device, we performed endurance and retention measurements under ambient condition. As illustrated in Fig. 2(a), the switching performance is reversible under repeated bias sweeping, in which

EP

the ON/OFF ratio decreased sharply after the first two cycles. The Vset changed

AC C

erratically with increase of sweeping cycles (see insert of Fig. 2(a)). These can be attributed to that the Joule heating induced by voltage sweeping weakens the conduction filaments [25]. Fig. 2(b) shows the retention characteristics of device under flat and bending conditions. The ON/OFF ratio of flat sample retained above 108 s without any obvious degradation, while the bent device decreased about 40%. As discussed later, these can be explained that the fatigue fracture induced by bending hinders the formation of conduction filaments.

8

(c)

ON/OFF ratio

3000 1 0

2000

1000

-2

Von

15

17.5

∞

20

2 1 0 -1

-1.5

500

1000

2000

0

0

times

100 times

Log (I)

500 times

TE D

-4 -2.0

100

0

-1

-1.0

-0.5

EP

Log (I)

-3

1000

-2

(e)

unbent 20 mm 17.5 mm 15 mm 12.5 mm

-2

2000

Bending times

Bending radius (mm)

(d)

ON/OFF ratio

3000

-3

12.5

Voff

3

M AN U

-1

4

ON/OFF ratio

Voff

SC

Von

ON/OFF ratio Threshold voltage (V)

Threshold voltage (V)

(b) 2

RI PT

ACCEPTED MANUSCRIPT

Log (V)

0.0

1000 times

-2

2000 times

-3

-1.0

-0.5

0.0

Log (V)

AC C

Fig. 3 (a) Schematic illustration of bending test for the GaIn/TiO2-CuO film/ITO device. (b) and (c) show the effect of bending radius and times on the Von, Voff and ON/OFF ratio of the nanocomposite film, respectively. The double log of I−V curves with different (d) bending radius and (e) bending times, respectively.

The mechanical strain is an important factor to affect flexible device application. We investigated the flexibility of GaIn/TiO2-CuO/ITO device by using a homemade bending system (see Fig. 3(a)). The substrate was bent to different radius and the

9

ACCEPTED MANUSCRIPT distribution of ON/OFF ratio was measured as a function (see Fig. 3(b)). The results show that the ON/OFF ratio reduced nearly two thirds, while the threshold voltage increased slightly with bending radius decreased from 22.5 to 12.5 mm. It is

RI PT

noteworthy that the ON/OFF ratio decreases sharply when the bending radius decrease from 20 to 17.5 mm. Cho et al. reported that bending radius plays a key role in affecting the peak strain applied on the surface of functional layer [26]. The

SC

following equation can be used to calculate the relationship of peak strain and

M AN U

bending radius for our device:

Strain= (df + dPET)/2R × 100

Where df and dPET are the thickness of TiO2-CuO/ITO bilayers and PET, respectively. The TiO2-CuO/ITO/PET multilayer film was bent to 20 and 17.5 mm, corresponding

TE D

to a peak strain of 0.43% and 0.5%, respectively. Beyond the critical bending radius, the ON/OFF ratio decreased sharply due to the generation of cracks, as detailed later. Dynamic bending test was performed at a fixed bending radius of 12.5 mm (see Fig.

EP

3(c)). The ON/OFF ratio decreases nearly 87% after bending 1000 cycles, while the

AC C

threshold voltages increases steadily with the increase of bending cycles. The reason can be ascribed to that more severe fatigue fracture caused by dynamic strain weakens the switching performance. To understand the effect of mechanical strain on the carrier transport, we plotted

the I−V curves to a double log scale. Fig. 3(d) and (e) show that I−V curves in HRS follow a superliner relationship. Further analysis indicates that the slope fluctuates in the scale of 1.4 to 2, corresponding to a space-charge-limited conduction (SCLC)

10

ACCEPTED MANUSCRIPT [27-29]. The carrier transport mechanism in our sample is thus unaffected by the

M AN U

SC

RI PT

mechanical strain.

Fig. 4 SEM images of the surface morphology of unbent sample (a) and 100 cyclic bent sample (b), respectively. The high magnification SEM were also given in the inset.

Compared with that of the sample affixed to a hemispherical mold (see inset of

TE D

Fig. 2(b)), the switching performance shows sharper degradation under repeated bending. This may indicate that dynamic strain causes more serious fatigue fracture.

EP

The surface morphology of the TiO2-CuO nanocomposite films were analyzed by SEM just before and after repeated bending. As shown in the Fig. 4(a), there are

AC C

hardly any cracks on the surface of unbent film. The high magnification SEM results show that some large voids can be detected on the film surface (see inset of Fig. 4(a)), which which provide the roots for the micro crack initiation [30,31]. After bending 100 cycles, as illustrated in Fig. 4(b), irregular micro cracks can be clearly observed on the surface of TiO2-CuO nanocomposite film. It is noted worthy that most cracks are along the grain boundaries of the TiO2 matrix (see inset of Fig. 4(b)) and the interfaces between TiO2 and CuO nanoparticles [32]. The 11

ACCEPTED MANUSCRIPT propagation of irregular micro cracks may suggest that the CuO nanoparticles play an obstacle role and change the growth path of the cracks [33]. As detailed later, these

SC

RI PT

micro cracks significantly affect the resistive switching performance.

M AN U

Fig. 5 The cross-sectional SEM image of the device without bending (a) and bent 100 cycles (b), respectively.

The grain boundary along the film thickness direction is an important factor to affect the formation of conduction filaments [34]. For unbent sample, as demonstrated

TE D

in the Fig. 5(a), cracks can be hardly observed either around the grain boundaries or at the interface between TiO2-CuO nanocomposite film and ITO. The Fig. 5(b) shows that a channel crack grows along the TiO2-TiO2 grain boundaries and causes

EP

interfacial crack between TiO2-CuO nanocomposite film and ITO after bending 100

AC C

times [35]. These fatigue fracture play a catastrophic factor on the formation of conduction paths between functional layer and electrodes. To clarify the device failure mechanism, a three-dimensional FEA model was

established to simulate the bending processes by using the commercial nonlinear finite element code ABAQUS. The paths were plotted to study stress distribution of the device. All materials were considered as homogeneous and linear elastic with parameters and thickness listed in Table 1. Since surface defects such as voids (see

12

ACCEPTED MANUSCRIPT Fig. 4(a) and Fig. 4(b)) are preferential sites for crack initiation, a pre-defined crack was thus assumed in the model (see Fig. 6(a)). We further found, upon bending, stress concentration was also found between CuO and TiO2 nanoparticles, and micro crack

RI PT

initiates along interface (see Fig. 6(b)). As shown in Fig. 6(c), the crack initiated and propagated along grain boundaries in the TiO2-CuO nanocomposite film subjected to seriously cyclic loading.

SC

Table 1 Material properties of layers utilized in the FEA. Young's modulus (GPa)

Poisson's ratio

PET

2.9

0.37

175 µm

ITO

118

0.30

30 nm

TiO2-CuO

210

0.27

200 nm

Thickness

AC C

EP

TE D

M AN U

Material

13

ACCEPTED MANUSCRIPT 100

1.4

(f)

80

1.2

Stress (GPa)

Stress (GPa)

1.3

1.1 1.0 0.9

60 40

0.8

20

0.7

0

0.6

2

4

6

8

10

12

14

0.00

0.02

RI PT

(e)

0.04

0.06

0.08

0.10

Crack length (µm)

Crack position

Fig. 6 (a) Finite element model of the nanoparticle interface and TiO2-CuO/ITO device. (b) Stress

SC

concentration was found at pre-defined crack and micro-crack along the grain boundary between

M AN U

TiO2 and CuO nanoparticles under bending. (c) Crack propagation in the nanocomposite film under bending. (d) Interface and interfacial delamination when the device is subjected to a tensile deformation. (e) Stress distribution along the grain boundaries between TiO2 and CuO, while stress distribution between TiO2-CuO film and ITO is shown in (f).

TE D

Interfacial delamination is also taken into consideration. For a very thin nanocomposite film，a surface-based cohesion model was introduced [36]. With

EP

increase of bending cycles, a resulting bending moment may induce a stress gradient in the adhesive region between the TiO2-CuO nanocomposite film and ITO electrode.

AC C

Once the stress satisfies damage initiation norm, interfacial delamination will be triggered and propagate alone the interface. Fig. 6(e) shows the stress distribution of the path 1, which indicates point 2 and point 14 formed stress concentration easily and micro defects existed along path 1. For path 2, interfacial delamination exists between the TiO2-CuO nanocomposite film and ITO, which has a serious effect on the carrier transport (see Fig. 6(f)). The FEA results are consistent with the above SEM observation. 14

SC

RI PT

ACCEPTED MANUSCRIPT

M AN U

Fig. 7 Schematic of the switching mechanism for unbent film under negative bias (a) and positive bias (b), respectively. (c) Schematic of the switching mechanism for bent sample under negative bias.

For our case, the switching behavior can be attributed to the tree-like filament

TE D

formation and rupture of oxygen vacancies [37]. As shown in Fig. 7(a), when a low negative bias is applied onto the GaIn electrode, oxygen ions move to the ITO electrode [38]. Oxygen vacancies are left and form a conduction path between the

EP

electrodes, resulting in transition from HRS to LRS. The growth path of conduction

AC C

filaments is related to grain boundaries [39]. Under positive bias, the oxygen ions moved back to the nanocomposite film and recombined with the oxygen vacancies, which made the conduction path destroyed (see Fig. 7(b)). Moreover, the Joule heating effect may accelerate ions migration. The rupture of the conductive path may be near the interface of GaIn/TiO2-CuO [40]. The effect of film fatigue on the resistive switching is further considered. Upon loading, the cracks initially originated from the voids at the nanocomposite surface, and then propagated in the

15

ACCEPTED MANUSCRIPT through-thickness direction along the grain boundaries. As shown in Fig. 7(c), under the same bias condition, the conduction path is more difficult to form due to the grain boundary cracks and micro-cracks in the film. The number of tree-like filaments

RI PT

reduced and linear conduction path may thus form between electrodes. Compared with that of the unbent film, the LRS current decreases and the ON/OFF ratio declines [38]. We suggest that, under the catalytic assistance of CuO [40,41], oxygen in air can

SC

be more easily absorbed onto the surface of TiO2 nanoparticles. This accelerates the

M AN U

transformation of oxygen molecules to ions and reduces the density of oxygen vacancies in the nanocomposite film.

The effect of delamination on the resistive switching is also taken into consideration. With increase of bending times, heat stress in the interfacial crack was

TE D

generated due to different thermal elastic modules of the ITO and TiO2-CuO nanocomposite [40,41]. As shown in Fig. 7(c), the delamination of TiO2-CuO nanocomposite film and ITO may happen along the interfacial direction due to heat

EP

stress and poor adhesion between them. The conduction paths of oxygen vacancies

AC C

are difficult to form at such delaminated interfaces.

16

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Fig. 8 Schematic of energy band diagram for the CuO/TiO2 interface at zero bias (a) and at forward bias before (b) and after bending (c).

To further understand the effect of micro crack on the carrier transport among

TE D

the p-n junction, we plotted a schematic energy band of CuO/TiO2 interface. When the CuO and TiO2 nanoparticle formed a p-n junction, as given in the Fig. 8(a), the

EP

difference of work functions promotes electrons transferring from TiO2 to CuO until Fermi level align. When a negative bias applied onto the top electrode, electron

AC C

aggregates in the conduction band of CuO and moves to TiO2, while the holes accelerate to transfer at valence band (see Fig. 8(b)) [42]. For bent device, the micro defects acted as blocking role and hindered injected electrons transferring from TiO2 to CuO (see Fig. 8(c)). As detailed above, these fatigue fracture would weaken the resistive switching of the device.

17

ACCEPTED MANUSCRIPT

4. Conclusions In conclusion, we fabricated TiO2-CuO nanocomposite film by hydrothermal method and investigated how fatigue fracture influences the resistive switching of

RI PT

GaIn/TiO2-CuO/ITO device. The device shows the stable bipolar switching performance over 30 cycles and retention properties ~108 s under flat condition. The ON/OFF ratio decreases sharply with the increase of bending times. We show that

SC

micro defects between grain boundaries, channel crack and interfacial delamination

M AN U

between layers are the main fatigue fracture caused by repeated bending. The FEA results indicate that length of interfacial delamination and bending radius are correlated. The channel crack and interfacial delamination can hinder the formation of conduction paths. Moreover, under the catalytic assistance of CuO, oxygen were

TE D

easily absorbed into the film via the cracks and transformed to ions, leading to the reduction of oxygen vaccines density. The micro defects among the nanoparticles act

AC C

EP

as interfacial barrier and restrain electrons transferring from CuO to TiO2.

Acknowledgement This research is supported financially by the Project of National Natural Science Foun dation of China (Grant no. 51773030).

18

ACCEPTED MANUSCRIPT

References [1] H. Yao, W. Fu, L. Liu, X. Li, D. Ding, P. Su, S. Feng, H. Yang, Hierarchical photoanode of rutile TiO2 nanorods coupled with anatase TiO2 nanosheets array for

206-211.

RI PT

photoelectrochemical application, Journal of Alloys and Compounds. 680 (2016)

[2] F. Raffone, F. Risplendi, G. Cicero, A New Theoretical Insight Into ZnO NWs

SC

Memristive Behavior, Nano Lett. 16 (2016) 2543-2547.

M AN U

[3] K.M. Kim, B.J. Choi, B.W. Koo, S. Choi, D.S. Jeong, C.S. Hwang, Resistive Switching in Pt/Al2O3/TiO2/Ru Stacked Structures, Electrochemical and Solid-State Letters. 9 (2006) G343.

[4] K. Park, J.S. Lee, Flexible resistive switching memory with a Ni/CuOX/Ni

TE D

structure using an electrochemical deposition process, Nanotechnology. 27 (2016) 125203.

[5] J.J. Diaz Leon, D.M. Fryauf, R.D. Cormia, M.X. Zhang, K. Samuels, R.S.

EP

Williams, N.P. Kobayashi, Reflectometry-Ellipsometry Reveals Thickness, Growth

AC C

Rate, and Phase Composition in Oxidation of Copper, ACS Appl Mater Interfaces. 8 (2016) 22337-22344.

[6] J. Wu, J. Wang, Diodelike and resistive hysteresis behavior of heterolayered BiFeO3/ZnO ferroelectric thin films, Journal of Applied Physics. 108 (2010) 094107. [7] W.J. Duan, J.B. Wang, X.L. Zhong, H.J. Song, B. Li, Switchable Cu2O/WOX p–n junction for high density crossbar arrays, RSC Adv. 6 (2016) 102603-102607.

19

ACCEPTED MANUSCRIPT [8] Y. Vygranenko, K. Wang, A. Nathan, Low leakage p-NiO/i-ZnO/n-ITO heterostructure ultraviolet sensor, Applied Physics Letters. 89 (2006) 172105. [9] H. Li, W. Dong, X. Wu, J. Xi, Z. Ji, Resistive switching characteristics of

RI PT

ZnO/a-TiO2 bilayer film fabricated on PET/ITO transparent and flexible substrates, Materials Research Bulletin. 84 (2016) 449-454.

[10] S. Chang Lee, Q. Hu, Y.-J. Baek, Y. Jin Choi, C. Jung Kang, H. Ho Lee, T.-S.

SC

Yoon, Analog and bipolar resistive switching in pn junction of n-type ZnO nanowires

M AN U

on p-type Si substrate, Journal of Applied Physics. 114 (2013) 064502. [11] L. Zhao, Z. Lu, F. Zhang, G. Tian, X. Song, Z. Li, K. Huang, Z. Zhang, M. Qin, SujuanWu, X. Lu, M. Zeng, X. Gao, J. Dai, J.M. Liu, Current rectifying and resistive switching in high density BiFeO3 nanocapacitor arrays on Nb-SrTiO3 substrates, Sci

TE D

Rep. 5 (2015) 9680.

[12] K.M. Kim, S.J. Song, G.H. Kim, J.Y. Seok, M.H. Lee, J.H. Yoon, J. Park, C.S. Hwang, Collective Motion of Conducting Filaments in Pt/n-Type TiO2/p-Type NiO/Pt

AC C

1587-1592.

EP

Stacked Resistance Switching Memory, Advanced Functional Materials, 21 (2011)

[13] S. Lee, H. Kim, D.-J. Yun, S.-W. Rhee, K. Yong, Resistive switching characteristics of ZnO thin film grown on stainless steel for flexible nonvolatile memory devices, Applied Physics Letters. 95 (2009) 262113. [14] J. Won Seo, J.-W. Park, K.S. Lim, S.J. Kang, Y.H. Hong, J.H. Yang, L. Fang, G.Y. Sung, H.-K. Kim, Transparent flexible resistive random access memory fabricated at room temperature, Applied Physics Letters. 95 (2009) 133508.

20

ACCEPTED MANUSCRIPT [15] L.G. Wang, Z.Y. Cao, X. Qian, L. Zhu, D.P. Cui, A.D. Li, D. Wu, Atomic Layer Deposited

Oxide-Based

Nanocomposite

Structures

with

Embedded

CoPtx

Nanocrystals for Resistive Random Access Memory Applications, ACS Appl Mater

RI PT

Interfaces. 9 (2017) 6634-6643. [16] H. Wang, Y. Zhu, D. Fu, Coexistence of bipolar and unipolar resistive switching characteristics of thin TiO2 film grown on Cu foil substrate for flexible nonvolatile

SC

memory device, Journal of Alloys and Compounds. 695 (2017) 2669-2671.

M AN U

[17] S. Park, K. Cho, S. Kim, Memory characteristics of flexible resistive switching devices with triangular-shaped silicon nanowire bottom electrodes, Semiconductor Science and Technology. 30 (2015) 055019.

[18] D. Liu, Q. Lin, Z. Zang, M. Wang, P. Wangyang, X. Tang, M. Zhou, W. Hu,

TE D

Flexible All-Inorganic Perovskite CsPbBr3 Nonvolatile Memory Device, ACS Appl Mater Interfaces. 9 (2017) 6171-6176.

[19] F. Yang, M. Wei, H. Deng, Bipolar resistive switching characteristics in

EP

CuO/ZnO bilayer structure, Journal of Applied Physics. 114 (2013) 134502.

AC C

[20] H.-L. Yuan, J.-C. Li, Effect of bending on resistive switching of NiO/ZnO nanocomposite thin films, Journal of Alloys and Compounds. 709 (2017) 752-759. [21] G. Kaur, K. Saini, A.K. Tripathi, V. Jain, D. Deva, I. Lahiri, Room temperature growth and field emission characteristics of CuO nanostructures, Vacuum. 139 (2017) 136-142.

21

ACCEPTED MANUSCRIPT [22] R. Lontio Fomekong, D. Lahem, M. Debliquy, S. Yunus, J. Lambi Ngolui, A. Delcorte, Ni0.9Zn0.1O/ZnO nanocomposite prepared by malonate coprecipitation route for gas sensing, Sensors and Actuators B: Chemical, 231 (2016) 520-528.

RI PT

[23] Gul. F, E. H, ZnO and ZnO1−x, based thin film memristors: The effects of oxygen deficiency and thickness in resistive switching behavior, Ceramics International. 43 (2017)10770-10775.

SC

[24] J.-C. Li, Q. Cao, X.-Y. Hou, Ru-Al codoping to mediate resistive switching of

M AN U

NiO:SnO2 nanocomposite films, Applied Physics Letters. 104 (2014) 113511. [25] D.-Y. Cho, K.-H. Kim, T.-W. Kim, Y.-J. Noh, S.-I. Na, K.-B. Chung, H.-K. Kim, Transparent and flexible amorphous InZnAlO films grown by roll-to-roll sputtering for acidic buffer-free flexible organic solar cells, Organic Electronics. 24 (2015)

TE D

227-233.

[26] J.-C. Li, X.-Y. Hou, Q. Cao, Effect of Zn/Sn ratio on structure and properties of

219-224.

EP

ZnO–SnO2 nanocomposite films, Journal of Alloys and Compounds. 611 (2014)

AC C

[27] K.J. Han, K.S. Kang, Y. Chen, K.H. Yoo, J. Kim, Effect of annealing temperature on the conduction mechanism for a sol–gel driven ZnO Schottky diode, Journal of Physics D: Applied Physics. 42 (2009) 125110. [28] Y. Li, Z. Hu, S. Liu, X. Duan, B. Wang, Comparison of the preparation methods for a highly efficient CuO/TiO2 photocatalyst for hydrogen generation from water, Reaction Kinetics, Mechanisms and Catalysis. 112 (2014) 559-572.

22

ACCEPTED MANUSCRIPT [29] Y. Leterrier, L. Médico, F. Demarco, J.A.E. Månson, U. Betz, M.F. Escolà, M. Kharrazi Olsson, F. Atamny, Mechanical integrity of transparent conductive oxide films for flexible polymer-based displays, Thin Solid Films. 460 (2004) 156-166.

RI PT

[30] H. Li, X. Xi, J. Ma, K. Hua, A. Shui, Low-temperature sintering of coarse alumina powder compact with sufficient mechanical strength, Ceramics International. 43 (2017) 5108-5114.

SC

[31] M. Mazur, A. Poniedziałek, D. Kaczmarek, D. Wojcieszak, J. Domaradzki, D.

M AN U

Gibson, Investigation of various properties of HfO2-TiO2 thin film composites deposited by multi-magnetron sputtering system, Applied Surface Science. 421 (2017) 170-178.

[32] L.-c. Feng, N. Xie, W.-z. Shao, L. Lv, J. Zhong, L. Zhen, Mechanical properties

TE D

of cermet composites with various geometrical tortuosity of metal phase: Fractal characterization, Materials Science and Engineering: A. 607 (2014) 236-244. [33] S. Chao, V. Petrovsky, F. Dogan, Complex Impedance Study of Fine and Coarse

AC C

3031-3034.

EP

Grain TiO2 Ceramics, Journal of the American Ceramic Society. 93 (2010)

[34] L.H. Liang, X.N. Li, H.Y. Liu, Y.B. Wang, Y.G. Wei, Power-law characteristics of damage and failure of ceramic coating systems under three-point bending, Surface and Coatings Technology. 285 (2016) 113-119. [35] D.-P. Tran, H.-I. Lu, C.-K. Lin, Effects of cyclic deformation on conductive characteristics of indium tin oxide thin film on polyethylene terephthalate substrate, Surface and Coatings Technology. 283 (2015) 298-310.

23

ACCEPTED MANUSCRIPT [36] S.-W. Yeom, H.J. Ha, J. Park, J.W. Shim, B.-K. Ju, Transparent bipolar resistive switching memory on a flexible substrate with indium-zinc-oxide electrodes, Journal of the Korean Physical Society. 69 (2016) 1613-1618.

RI PT

[37] S.-W. Yeom, S.W. Park, I.-s. Jung, M. Kim, H.J. Ha, J.H. Shim, B.-k. Ju, Highly flexible titanium dioxide-based resistive switching memory with simple fabrication, Applied Physics Express. 7 (2014) 101801.

SC

[38] S. Yu, X. Guan, H.S.P. Wong, Conduction mechanism of TiN/HfOx/Pt resistive

M AN U

switching memory: A trap-assisted-tunneling model, Applied Physics Letters. 99 (2011) 063507.

[39] S. Park, J.H. Lee, H.-D. Kim, S.M. Hong, H.-M. An, T.G. Kim, Resistive switching characteristics of sol-gel based ZnO nanorods fabricated on flexible

TE D

substrates, physica status solidi (RRL) - Rapid Research Letters. 7 (2013) 493-496. [40] J. Deng, L. Wang, Z. Lou, T. Zhang, Design of CuO–TiO2 heterostructure nanofibers and their sensing performance, J. Mater. Chem. A. 2 (2014) 9030-9034.

EP

[41] L. Li, J. Lei, T. Ji, Facile fabrication of p–n heterojunctions for Cu2O

AC C

submicroparticles deposited on anatase TiO2 nanobelts, Materials Research Bulletin. 46 (2011) 2084-2089.

[42] N. Helaili, Y. Bessekhouad, A. Bouguelia, M. Trari, Visible light degradation of Orange II using xCuyOz/TiO2 heterojunctions, J Hazard Mater. 168 (2009) 484-492.

24

ACCEPTED MANUSCRIPT Highlights:
The GaIn/TiO2-CuO/ITO resistive memory was fabricated.
The fatigue fracture of the device is clearly induced by bending.

AC C

EP

TE D

M AN U

SC

RI PT


The fatigue fracture affects largely on the device switching performance.