TiN electrode-induced bipolar resistive switching of TiO2 thin films

Current Applied Physics 10 (2010) e71–e74

Contents lists available at ScienceDirect

Current Applied Physics journal homepage: www.elsevier.com/locate/cap

TiN electrode-induced bipolar resistive switching of TiO2 thin films Young Ho Do a,b, June Sik Kwak a, Yoon Cheol Bae c, Jong Hyun Lee a,b, Yongmin Kim d, Hyunsik Im d, Jin Pyo Hong a,b,c,* a

Novel Functional Materials and Devices Lab, Department of Physics, Hanyang University, Seoul 133-791, Republic of Korea The Research Institute for Natural Science, Hanyang University, Seoul 133-791, Republic of Korea c Division of Nano-semiconductor Engineering, Hanyang University, Seoul 133-791, Republic of Korea d Department of Semiconductor Science, Dongguk University, 3-26 Chung-Ku, Pil-Dong, Seoul 100-715, Republic of Korea b

a r t i c l e

i n f o

Article history: Received 19 August 2009 Received in revised form 18 September 2009 Accepted 29 September 2009 Available online 11 December 2009 Keywords: ReRAM Nonvolatile memory Resistive switching

a b s t r a c t The bipolar resistive switching characteristics in polycrystalline TiO2 thin films after regular forming process were studied using two different top or bottom TiN electrodes (Sample A: top TiN/TiO2/Pt, Sample B: Pt/TiO2/TiN bottom). The sample A and B clearly showed two different switching directions of counterclockwise (CCW) and clockwise (CW) bipolar switching behaviors, respectively, depending on the relative position of the TiN electrode. These switching characteristics in both samples could be understood by considering the forming and rupture of the conducting path due to the migration of oxygen ions between the TiO2 layer and the TiN electrode, which acts like the oxygen reservoir. In addition, both samples clearly display high reliable memory switching characteristics, such as stable switching speed (ls), endurance behaviors (>104), and long retention times (>104 s). Ó 2009 Elsevier B.V. All rights reserved.

1. Introduction Several novel types of nonvolatile memories (NVMs), such as phase change memory, magnetoresistive memory, nano floating gate memory, ferroelectric memory, and resistive random access memory (RAM) have been intensively studied [1–4]. These new types of memory are not only of paramount interest to companies that manufacture memory, but also to researchers in academia or small venture companies who seek to discover new ideas and processes outside of conventional memory. Many binary oxide materials capable of resistive switching behavior such as TiO2, NiO, MgO and CoO have attracted attention due to their possible applications for resistive random access memory (ReRAM) NVM devices. ReRAM features high density integration, long retention time, small size, and fast switching speed [5]. In addition, ReRAM is believed to be free from the scaling problems that are inherent to other NVMs. However, the optimal design for the resistive switching mechanism and the best materials for ReRAM applications are still unknown, even though they are the subject of a great deal of recent research [5–11]. Specifically, the switching mechanism must be elucidated before ReRAM can be used in next-generation NVMs. There is general agreement that the migration of oxygen ions or vacancies under an * Corresponding author. Address: Novel Functional Materials and Devices Lab, Department of Physics, Hanyang University, Seoul 133-791, Republic of Korea. Tel.: +82 2 2220 0911; fax: +82 2 2296 3738. E-mail address: [email protected] (J.P. Hong). 1567-1739/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.cap.2009.12.017

applied electric field plays a key role in resistive switching [12]. Therefore, there are a lot of papers focused on the redox reaction, due to some evidence obtained in the forming process [13–16]. However, other important issues for actualization of the ReRAM devices also exist and one of them is how to stabilize the resistive switching parameters, such as the operating voltages (set/reset voltage) and the resistance states in between high resistance state (HRS) and low resistance state (LRS). In this paper, we present two different bipolar resistive switching behaviors of TiO2 films in metal–insulator–metal (MIM) structures (TiN/TiO2/Pt/Ti/SiO2/Si and Pt/TiO2/TiN/SiO2/Si). Bipolar switching transition behavior is observed from the counter-clockwise (CCW) to the clockwise (CW) resistive switching in TiO2 films depending on TiN electrode position (top or bottom). In addition, both samples clearly display high reliable memory switching characteristics. Two different switching modes (CCW and CW) and high reliable memory switching characteristics are discussed in terms of the migration of oxygen ions between the TiO2 layer and the TiN electrode, which acts like the oxygen reservoir. In addition, typical device performance characteristics, such as switching speed, endurance, and retention testing, are discussed at room temperature (RT).

2. Experimental We have evaluated the resistive switching behaviors of two different test samples. One sample incorporates the TiN top electrode

e72

Y.H. Do et al. / Current Applied Physics 10 (2010) e71–e74

structure (TiN/TiO2/Pt/Ti/SiO2/Si: Sample A) and the other the TiN bottom electrode structure (Pt/TiO2/TiN/SiO2/Si: Sample B). The Pt bottom electrode (100 nm thick) of sample A was grown on Ti/SiO2/Si substrate by using a DC sputtering system. The TiN bottom electrode (100 nm thick) of sample B was deposited on a SiO2/ Si substrate using reactive RF sputtering under a flowing Ar and N2 gas mixture. In order to confirm the electrical resistivity of our TiN films, we performed the four probe measurement method, and both samples showed a resistivity of about 130 lX/cm. Subsequently, 50 nm-thick TiO2 films were deposited by a reactive RF sputtering technique onto a TiOx ceramic target in a flowing Ar and O2 gas mixture. The TiN and Pt top electrodes in the sample A and B were deposited using the same conditions mentioned above. The thickness and the area of each top electrode were 100 nm and 100 100 lm2, respectively. Structural properties and cross-sectional observation of the sample A and B are characterized by using X-ray diffraction (XRD) and scanning electron microscopy (SEM). The electrical characterization is done with a Keithley 4200 semiconductor system and an Agilent 81110A pulse generator at room temperature (RT). 3. Results and discussion

20

TiN (111)

Intensitnitsy (arb. uny)

TiO2 (110)

TiO2/TiN/SiO2/Si

30

40

Pt P (200)

TiO2/Pt/Ti/SiO2/Si

TiN (200)

Pt P (111)

Fig. 1 shows the XRD patterns of TiO2 thin films grown on TiN and Pt bottom electrodes. As shown in Fig. 1, the line of TiO2 (1 1 0) demonstrates a single texture of the TiO2 rutile phase, and strong lines of TiN (1 1 1) and (2 0 0) are also observed. Thus, the reactive RF sputtered TiN film has a good crystal quality. Moreover, the TiO2 thin films grown on TiN and Pt bottom electrodes showed nearly polycrystalline rutile characteristics. Although TiO2 thin films were fabricated on two different bottom electrodes, all the data exhibited that those TiO2 thin films have the same crystallinity. The cross-sectional SEM images of the sample A and B are shown in Fig. 2a and b, respectively. As shown, the TiN and Pt bottom electrode were clearly distinguished from the substrates, and the TiN film is polycrystalline with column-like structure. Moreover, clear interfaces between the TiO2 layers and top/bottom electrodes (TiN and Pt) were observed in both sample A and B. Fig. 3a shows the typical bipolar switching behavior of the sample A and B under DC sweeping mode, providing the bipolar switching phenomena after forming process. The forming process

50

2θ Fig. 1. XRD results of TiO2 thin films grown on Pt/Ti/SiO2/Si and TiN/SiO2/Si substrates.

Fig. 2. SEM cross-section images of the (a) sample A (TiN/TiO2/Pt/Ti/SiO2/Si) and (b) sample B (Pt/TiO2/TiN/SiO2/Si).

takes place at 4.5 V under the current limitation at 1 mA in voltage sweep trace (not shown). However, the sample A and B display resistive switching behavior in different directions after the forming process. The sample A shows only positive set and negative reset operations, but the sample B shows only negative set and positive reset operations. Thus, the behavior of sample A is defined as a counter-clockwise (CCW) direction switching event while sample B shows clockwise (CW) switching. There asymmetrical bipolar resistive switching behaviors could be attributed to the TiN electrode. In a previous works, we have reported a switching model where bipolar resistive switching in the Al/TiO2 structures was mostly governed by the characteristics of the redox reactions at the interfaces between the Al electrode and TiO2 layer due to the migration of oxygen ions [14]. Moreover, the Al bottom electrode has a CCW switching direction, and the Al top electrode has a CW switching direction. Compared with the Al electrode, in which active chemical reactions can occur at interfaces between the Al electrode and TiO2 layer, the TiN electrode can behave as an oxygen reservoir [17–19]. Therefore, the oxidation process in the TiN electrode did not take place at interfaces between the TiN electrode and TiO2 layer, in contrast to the experimental results of Al electrode. Work by Waser et al. demonstrated chemisorption and physisorption of oxygen in Pt grain boundaries [20,21]. However, our Pt/TiO2/Pt samples present only unipolar switching behavior that is common to binary oxide materials [14]. Therefore, the oxygen reservoir for Pt electrodes was not taken into consideration in

e73


Current (A)

Reset

Reset

0.01

1E-3

Set Set

1E-4

(a) 1E-5

Sample A : CCW Sample B : CW 1E-6 -1

0

1

2

Sample A

(b)

Sample B

((c))

Voltage (V)

Current ((A)

1E-3

LRS HRS

1E-4

LRS HRS

1E-3

takes place, resulting in an increase the operating voltages and resistance ratio. Fig. 3b and c shows the different pulse widths of the on and off states in sample A and B. As described in this figure, the on and off states of our samples show a resistance ratio of about 10, and the LRS and HRS states are gradually stabilized with a decrease in pulse widths from 10 ms to 1 ls. These resistive switching properties were maintained up to 103 cycles (not shown in this letter), demonstrating excellent electrical stability. The distributions of operation voltage and resistance states in sample A and B are important factors for potential NVM applications. Fig. 4 shows the device-to-device distribution of the operating voltages (set/reset voltage) and resistance states (LRS and HRS) during 50-set/reset switching cycles. The sample A and B were measured using DC voltage sweep mode over the range from +3 to 3 V. As shown in Fig. 4a and b, the slope of the set voltage and reset voltage were nearly vertical, indicating a very stable set and reset process. Moreover, good resistance uniformity in sample A and B were observed with a resistance ratio >10 between HRS and LRS, and no overlap between LRS and HRS, as shown in Fig. 4c and d. Because there is a sharp difference between HRS and LRS in our samples, it is sufficient to use periphery circuits to probe the different resistance states. In addition, the operating voltages and resistance states of sample A shows a more dense distribution than those of sample B. This is due to the difference in stored oxygen ions between the TiN bottom electrode and the TiN top electrode. Initial stored oxygen ions in sample B (TiN bot-

1E-4 100 us

10 us

1 us

Pulse Width

( Cumulativee Probability (%)

our samples. In a previous study, we reported that the conducting paths, which consist of oxygen vacancies, were generated in the TiO2 thin film during the forming process [14]. In the case of sample A, the residual oxygen ions are reserved by the TiN top electrode, while conducting paths are generated in the TiO2 thin film. When an electrical field is applied to the TiN top electrode (reverse bias; V < 0), oxygen ions may be extracted from the TiN electrode, thereby restoring the oxygen vacancies in the conducting path near the interface region of the TiN electrode. As a result of these processes, the device switches into the HRS state (likely due to the rupture of the conducting path at the interface region). On the contrary, when a forward bias (V > 0) is applied to the TiN top electrode, the oxygen ions are ejected out from the interface region and can be reserved within the TiN top electrode, so the device switches into the LRS state (likely due to the formation of a conducting path at the interface region). Therefore, the bipolar resistive switching of the TiO2/TiN structure might be caused by the formation and rupture of the conducting paths at the interface region between the TiN electrode and the TiO2 layer. In addition, the operating voltages (set and reset voltages) and resistance ratio (LRS and HRS) of sample A shows larger values than those of sample B. This is due to the fact that the oxygen ions are fundamentally stored in the TiN bottom electrode (sample B). Those stored oxygen ions might be caused by the diffusion of oxygen during the TiO2 layer deposition or the inter-diffusion between the TiO2 and TiN bottom layers. Therefore, because of the oxygen ions initially stored in the TiN bottom layer, excessive absorption and extraction of oxygen ions between TiO2 layer and the TiN bottom electrode

-0 .6 100

((a))

100

(b) ( )

50

50

0

0 Sample A Sample B

0.6

0.8

1.0

1.1

1.2

1.3

Set Voltage (V) Reset Voltage (V)

Cumulativee Probability (%)

Fig. 3. (a) I–V characteristics of the sample A (TiN/TiO2/Pt/Ti/SiO2/Si) and sample B (Pt/TiO2/TiN/SiO2/Si). The LRS and HRS current value for different pulse widths showing significant changes in the current states as the pulse width varies. (b) sample A (set/reset voltage: +1/ 1.5 V) and (c) sample B (set/reset voltage: 1.5/ +2 V).

Set Voltage (V) Reset Voltage (V) -0. 8 -1. 0 -1 .0 -1 .1 -1.2


1 ms

Sample A p B Sample

100

100

50

50

(c)

0 1.9

2.0

2.1 2.7

(d) 3.0

3.3

0


10 ms

3.6

L HRS ((ohm) h ) L LRS (ohm) ( h ) Log. Log. Fig. 4. Distribution of (a) Set voltages and (b) Reset voltages in the sample A (TiN/ TiO2/Pt/Ti/SiO2/Si) and sample B (Pt/TiO2/TiN/SiO2/Si). Distribution of the resistance state of (c) LRS and (d) HRS switching in the sample A and B. (Read voltage: 0.2 V).

e74


Cu urrent (A)

1E-3

without using an external bias voltage. This indicates the potential for the use of our samples for nonvolatile applications, extrapolating to a data retention time of more than 10 years.

Set/reset voltage : +1/-1 +1/-1.55 V Read voltage : 0.2 V

LRS HRS

1E-3

LRS HRS

1E-4

(b)

Sample B

10

Two different resistive switching directions (CCW and CW characteristics) in polycrystalline TiO2 thin films were investigated with the top and bottom TiN electrodes. The resistive switching directions of each sample clearly depend on the position of the TiN electrode in the MIM structure. The bipolar switching behaviors seem to be attributable to the formation and rupture of the conducting path at the interfaces region between the TiO2 layer and TiN electrode due to the migration of oxygen ions between the TiO2 layer and the TiN electrode, which acts like the oxygen reservoir. In addition, high reliable resistive switching parameters (operating voltages and resistance states), stable switching speed (ls), endurance behaviors (>104), and long retention times (>104 s) were confirmed for possible nonvolatile memory applications.

Set/reset voltage g : -1.5/+2 V Read voltage : 0.2 V

1E-5 0

4. Conclusion

(a)

Sample A

1E-4

10

1

10

2

10

3

10

4

Current (m C mA)

Number of Pulse 2.0 1.5 1.0 0.5 0.0

LRS HRS

Set/reset voltage : +1/-1.5 V Read voltage : 0.2 V 10 years

(c)

Sample A

1.0

Acknowledgment

LRS HRS

0.5

Set/reset voltage : -1.5/+2 V Read voltage : 0.2 V

This project was supported by the research fund of Hanyang University (HY-2006-G).

10 years

0.0 10

0

10

1

10

2

10

3

10

4

10

5

10

Sample B

(d)

6

8

10

7

10

10

9

Times (s) Fig. 5. Endurance properties of the (a) sample A (TiN/TiO2/Pt/Ti/SiO2/Si, set/reset/ read voltage: +1/ 1.5/+0.2 V) and (b) sample B (Pt/TiO2/TiN/SiO2/Si, set/reset/read voltage: 1.5/+2/+0.2 V). Retention test of (c) sample A (set/reset/read voltage: +1/ 1.5/+0.2 V) and (d) sample B (set/reset/read voltage: 1.5/+2/+0.2 V).

tom electrode) are expected to be formed during TiO2 layer deposition on the TiN bottom electrode. But in sample A (TiN top electrode), oxygen ions for bipolar resistive switching are only induced from the TiO2 thin film used to apply the bias voltage. Therefore, the formation and rupture of the conducting paths occurred more randomly due to the initial storage of oxygen ions from the TiN bottom electrode (sample B). To further evaluate the performances of the sample A and B, the cycling endurance characteristics were measured, as shown in Fig. 5a and b. The operating voltages of sample A and B were +1 (set)/ 1.5 (reset) V and 1.5 (set)/+2 (reset) V, respectively. The current states are read at 0.2 V in each DC sweep. The device could be switched between on (LRS) and off (HRS) states for more than 104 cycles without switching failure. Moreover, after more than 104 repeated cycles, the operating voltages (set/reset voltages) and resistance ratio (LRS and HRS) were similar to those of the first cycle, indicating good reproducible behavior of our samples. The retention characteristics of the fabricated samples were measured at RT. Fig. 5c and d shows the current alteration over time for both the HRS and the LRS of sample A and B. Moreover, sample A and B have specific operating voltages (sample A: Vset = +1 V, Vreset = 1.5 V and sample B: Vset = 1.5 V, Vreset = +2 V), with a 0.2 readout voltage. As can be seen, the HRS and LRS were retained for 104 s

References [1] S. Lai, T. Lowrey, Int. Electron Devices Meet. Tech. Dig. (2001) 803. [2] G.A. Prinz, Science 282 (1998) 1660. [3] T. Hayashi, Y. Igarashi, D. Inomata, T. Ichimori, T. Mitsuhashi, K. Ashikaga, T. Ito, M. Yoshimaru, M. Nagata, S. Mitarai, H. Godaiin, T. Nagahama, C. Isobe, H. Moriya, M. Shoji, Y. Ito, H. Kuroda, M. Sasaki, Int. Electron Devices Meet. Tech. Dig. (2002) 543. [4] A. Chen, S. Haddad, Y.C. Wu, T.N. Fang, Z. Lan, S. Avanzino, S. Pangrle, M. Buynoski, M. Rathor, W. Cai, N. Tripsas, C. Bill, M. VanBuskirk, M. Taguchi, Int. Electron Devices Meet. Tech. Dig. (2005) 746. [5] R. Waser, M. Aono, Nat. Mater. 6 (2007) 833. [6] T. Fujii, M. Kawasaki, A. Sawa, H. Akoh, Y. Kawazoe, Y. Tokura, Appl. Phys. Lett. 86 (2005) 012107. [7] A. Beck, J.G. Bednorz, C. Gerber, C. Rossel, D. Widmer, Appl. Phys. Lett. 77 (2000) 139. [8] M.J. Rozenberg, I.H. Inoue, M.J. Sánchez, Phys. Rev. Lett. 92 (2004) 178302-1. [9] Y.H. Do, J.S. Kwak, J.P. Hong, H.S. Im, B.H. Park, J. Korean Phys. Soc. 55 (2009) 1009. [10] Y.K. Seo, D.J. Lee, W.S. Choi, D.W. Kim, Y.S. Lee, J. Korean Phys. Soc. 55 (2009) 129. [11] S.H. Joo, J.H. Sok, J. Korean Phys. Soc. 53 (2008) 3427. [12] Y.B. Nian, J. Strozier, N.J. Wu, X. Chen, A. Ignatiev, Phys. Rev. Lett. 98 (2007) 146403. [13] A. Odagawa, Y. Katoh, Y. Kanzawa, Z. Wei, T. Mikawa, S. Muraoka, T. Takagi, Appl. Phys. Lett. 91 (2007) 133503. [14] Y.H. Do, J.S. Kwak, J.P. Hong, K.H. Jung, H.S. Kim, J. Appl. Phys. 104 (2008) 114512. [15] Y.H. Do, J.S. Kwak, Y.C. Bae, K.H. Jung, H.S. Im, J.P. Hong, Appl. Phys. Lett. 95 (2009) 093507. [16] M.K. Yang, J.W. Park, T.K. Ko, J.K. Lee, Appl. Phys. Lett. 95 (2009) 042105. [17] M. Mandl, H. Hoffman, P. Kucher, J. Appl. Phys. 68 (1990) 2127. [18] M. Fujimoto, H. Koyama, M. Konagai, Y. Hosoi, K. Ishihara, S. Ohnishi, N. Awaya, Appl. Phys. Lett. 89 (2006) 223509. [19] B. Sun, Y.X. Liu, L.F. Liu, N. Xu, Y. Wang, X.Y. Liu, R.Q. Han, J.F. Kang, J. Appl. Phys. 105 (2009) 061630. [20] D.S. Jeong, H. Schroeder, R. Waser, J. Appl. Phys. 104 (2008) 123716. [21] D.S. Jeong, H. Schroeder, R. Waser, Phys. Rev. B 79 (2009) 195317.