Bipolar resistive switching of chromium oxide for resistive random access memory
Shih-Cheng Chen
a, Ting-Chang Chang
b,c,⇑, Shih-Yang Chen
a, Chi-Wen Chen
b, Shih-Ching Chen
b, S.M. Sze
d,
Ming-Jinn Tsai
e, Ming-Jer Kao
e, Fon-Shan Yeh Huang
aa
Department of Electrical Engineering & Institute of Electronic Engineering, National Tsing Hua University, Taiwan, ROC
b
Department of Physics, National Sun Yat-Sen University, Kaohsiung 804, Taiwan, ROC
cCenter for Nanoscience & Nanotechnology, National Sun Yat-Sen University, Kaohsiung 804, Taiwan, ROC d
Institute of Electronics, National Chiao Tung University, Taiwan, Hsin-Chu 300, Taiwan, ROC
e
Nanoelectronic Technology Division, Electronics and Optoelectronics Research Lab/ITRI, 195 Sect. 4, Chunghsing Road, Chutung, Hsinchu 31040, Taiwan, ROC
a r t i c l e
i n f o
Article history:
Received 15 September 2010
Received in revised form 8 December 2010 Accepted 22 December 2010
Available online 4 February 2011 The review of this paper was arranged by Prof. A. Zaslavsky Keywords: Cr2O3thin film Resistance switching Nonvolatile memory
a b s t r a c t
This study investigates the resistance switching characteristics of Cr2O3-based resistance random access memory (RRAM) with Pt/Cr2O3/TiN and Pt/Cr2O3/Pt structures. Only devices with Pt/Cr2O3/TiN structure exhibit bipolar switching behavior after the forming process because TiN was able to work as an effective oxygen reservoir but Pt was not. Oxygen migration between Cr2O3and TiN was observed clearly before and after resistance switching from Auger electron spectroscopy (AES) analysis. Both low resistance state, ON state, and high resistance state, OFF state, of Pt/Cr2O3/TiN structures are stable and reproducible dur-ing a successive resistive switchdur-ing. The resistance ratio of ON and OFF state is over 102, on top of that, the retention properties of both states are very stable after 104s with a voltage of 0.2 V.
Ó 2010 Elsevier Ltd. All rights reserved.
1. Introduction
Conventional nonvolatile floating memories are expected to reach certain technical and physical limit in the future, therefore the next generation nonvolatile memory has been studied aggressively[1]. Furthermore, new concepts for high density and high-speed nonvolatile memory devices have been studied exten-sively, including a nanofloating gate memory (NFGM)[2], a polymer
random access memory (PoRAM) [3], a magneto-resistive RAM
(MRAM) [4] and a resistive RAM (RRAM) [5–10]. Among these
memories, RRAM was considered to be the most promising candi-date, owing to advantages of its simple structure, low operating power, fast switching speed and high density. Various materials have been demonstrated to possess resistive switching characteris-tics, such as Cu2S[5], Al2O3[6], NiO[7], HfO[8], PCMO [9], and
RbAg4I5[10].
In most studies, oxygen interaction between metal oxide and metal electrode has been reported as the dominant mechanism for switching behavior[6–8,11]. Due to the standard potential of different material’s effective influence on reaction and reduction, it is considered that the operation voltage will be dependent on the standard potential [12]. Thus, the standard potential plays
important roles in the resistance switching property. In many literatures, Al2O3and NiO have been proposed to show resistance
switching behavior [6,7,13,14]. The goal is to find a resistance switching layer with lower operation voltage and acquire the reliability that’ll be able to sustain the same on/off ratio after long period of operation. In this study, the Cr2O3dielectric was proposed
as a reversible resistance switching layer because the standard potential of Cr/Cr3+couple is 0.74 V vs. normal hydrogen electrode
(NHE) is between Al/Al3+couple is 1.66 V vs. NHE and Ni/Ni2+ cou-ple is 0.25 vs. NHE. Memory devices with Pt/Cr2O3/TiN and Pt/
Cr2O3/Pt structures were studied to discuss resistance switching
mechanism because TiN reacts with oxygen easily but Pt does not
[15]. In addition, Auger electron spectroscopy (AES) analysis was used to observe oxygen migration between Cr2O3and TiN before
and after resistance switching. It is helpful to understand the resis-tance switching mechanism. The result demonstrated stable and reproducible resistive switching phenomenon in Pt/Cr2O3/TiN
structure under atmospheric conditions with a resistance ratio above 102, illustrating that Cr
2O3 thin films have a promising
potential for NVM applications. 2. Experiment
The proposed resistive switching memory devices were fabricated on Pt/Ti/SiO2/Si and TiN/SiO2/Si substrates. The resistance switching layer, a 15-nm-thick Cr2O3 thin film, was
0038-1101/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.sse.2010.12.014
⇑Corresponding author at: Department of Physics, National Sun Yat-Sen Univer-sity, Kaohsiung 804, Taiwan, ROC. Tel.: +886 7 5252000x3708; fax: +886 7 5253709.
E-mail address:[email protected](T.-C. Chang).
Solid-State Electronics 62 (2011) 40–43
Contents lists available atScienceDirect
Solid-State Electronics
deposited by sputtering a chromium target (99.9% pure) with a RF power of 100 W in argon and oxygen mixed gas ambient (Ar/O2= 30 sccm/30 sccm) at room temperature. After the Cr2O3
deposition, a 140-nm-thick Pt top electrode was deposited through a metal shadow mask by sputtering. The current–voltage (I–V) characteristics of memory devices were measured by Keithley 4200 semiconductor characterization system at room temperature. During the electrical measurement, a bias voltage was applied to the top electrode, while the bottom electrode was grounded.
3. Discuss and result
The composition of the as-deposited chromium oxide thin film on TiN/SiO2/Si subtract without Pt top electrode was analyzed by
X-ray photoelectron spectroscopy (XPS). It was performed by using a monochromatic Al K
a
(1486.6 eV) X-ray and calibrated by C 1 s peak at 284.5 eV. From the XPS analysis, the binding energies for Cr 2p3/2and 2p1/2were obtained at 576.4 eV and 586.8 eV,respec-tively. It was estimated to be Cr2O3[16,17].Fig. 1a and b show the
current–voltage (I–V) curves of a device with Pt/Cr2O3/TiN and
Pt/Cr2O3/Pt structure under the cycling DC voltage sweeping
oper-ations. Before cycling operations, the irreversible forming pro-cesses, as shown in the inserts, were performed by applying a negative DC voltage to the top Pt electrode with a current compli-ance of 1 mA. Due to the formation of conduction path(s) the apparent increase of current occurred at the voltage of 10 V in Pt/Cr2O3/TiN and 8 V in Pt/Cr2O3/Pt device can be seen. During
the forming process, the initial resistance state is transformed from an initial high resistance state, IHRS, to a low resistance state, LRS. After the forming process, the repeated hysteretic resistance switching behavior was observed only in the Pt/Cr2O3/TiN device,
shown in four steps in the figure. When the applied voltage raises from 0 to 1.3 V during the reset process, most resistance states start to transform from LRS to high resistance state, HRS, at a reset voltage of 0.6 V. Conversely, as the applied voltage sweeps from 0 to 1.5 V in the set process, an abrupt increase in current is ob-served at about 0.7 V. Most resistance states start to transform from HRS to LRS and achieved the nonvolatile resistance switching. In the set process, compliance current of 1 mA is applied to prevent a device breakdown. The resistance ratio of two resistance states, HRS/LRS, is about 100 times at a reading voltage of 0.2 V. How-ever, Pt/Cr2O3/Pt devices do not exhibit bipolar resistive switching
behaviors as shown inFig. 1b. After the forming process, the resis-tance state transforms from IHRS to LRS. But the resisresis-tance state does not transform from LRS to HRS even with another reset pro-cess. It should be noted that Pt/Cr2O3/TiN devices exhibit bipolar
resistive switching behaviors but Pt/Cr2O3/Pt devices do not. TiN
can be an effective oxygen reservoir since it can react with oxygen readily. On the contrary, Pt is an inert metal which does not easily react with oxygen[15].
Fig. 2a and b show the Auger electron spectroscopy analysis of Pt/Cr2O3/TiN devices with LRS and HRS. Comparing to HRS, the
oxygen intensity increases in TiN and decreases in Cr2O3 in LRS.
It is considered that a negative voltage is applied to the top elec-trode in the set process. The electrical field ruptures Cr–O bonds and induces the oxygen ions moved from Cr2O3to reserve in TiN.
It is due to the fact that the Cr2O3film is nonstoichiometric and
there are a large amount of oxygen vacancies in the Cr2O3film.
On the contrary, in the reset process, oxygen ions are extracted from TiN and moved from TiN to Cr2O3. The oxygen intensity
decreases in TiN and increases in Cr2O3in HRS for that reason.
Based on the above observations, a physical model is proposed to illustrate the resistive switching behaviors with a different oper-ating voltage. During the forming process, a negative voltage is ap-plied onto the top electrode, and soft-breakdown occurs while Cr– O bonds rupture at a critical voltage. Thereby, an oxygen vacancy filament is formed in the insulating oxide film to act as a conduc-tion channel, as shown in Fig. 3a. Carriers can travel through vacancies by hopping and the devices would then switch to LRS. Contrarily, when applying a positive bias, oxygen ions are ex-tracted from TiN bottom electrode and recovered with the oxygen vacancies (i.e., Cr–O bonds form again) near the TiN interface, as shown inFig. 3b. The conducting oxygen vacancy filament ruptures at the interface between TiN and Cr2O3and devices switched to
HRS in the reset process. As a result, by applying a different polar-ity bias on devices, the bipolar resistive switching behavior can easily be obtained due to the generation and recovery of oxygen vacancies at the interface of TiN electrode and Cr2O3dielectric. In
previous literature, similar using oxygen vacancies generation and recombination to format/rupture a conduction filament to switch the resistance states is reported in some oxide-based RRAM
[18,19].
The retention and endurance were tested to further investigate the reliability characteristics of the memory device. The endurance properties of the memory device is shown inFig. 4a in which short pulse applied are 1.6 V amplitude 4
l
s wide and 1.7 V amplitude 5l
s to switch the devices to on state and off state, respectively. The resistances were extracted at a reading voltage of 0.2 V at room temperature. The result indicates that there is no apparent degradation on resistance ratio after 6 104operation cycles. Inaddition, after 100 DC sweep cycling the retention properties of LRS and HRS at 85 °C were measured as shown inFig. 4b. The resis-tance values of HRS and LRS were very stable even after 104s. The
device with Pt/Cr2O3/TiN structure owns a good reliability for
memory application.
Fig. 1. Typical bipolar I–V characteristics at room temperature of (a) Pt/Cr2O3/TiN memory device and (b) Pt/Cr2O3/Pt memory device.
4. Conclusions
The Pt/Cr2O3/TiN and Pt/Cr2O3/Pt structures were fabricated for
the nonvolatile resistance switching memory application by
sputtering a chromium target in an Ar/O2 environment at room
temperature. Before the resistance switching, a forming process is necessary. Only the device with TiN electrode exhibits bipolar resistive switching behaviors because TiN is an effective oxygen reservoir. Observing oxygen migration from Auger electron spectroscopy analysis, a physical model about oxygen vacancy filaments formation and rupture is proposed to illustrate the
resistive switching behaviors. The ratio of resistance of ON and OFF state is over 102. The endurance and retention results indicate
that the proposed memory device has excellent device reliability. Therefore, Cr2O3has a high potential for application in resistance
random access memory in the future. Acknowledgements
This work was performed at National Science Council Core Facilities Laboratory for Nano-Science and Nano-Technology in Kaohsiung-Pingtung area and supported by the National Science Fig. 2. Auger electron spectroscopy (AES) analysis of Pt/Cr2O3/TiN devices with (a) low resistance state (LRS) and (b) high resistance state (HRS).
Fig. 3. Distribution of oxygen vacancies in the device: (a) low resistance state (LRS) and (b) high resistance state (HRS).
Fig. 4. (a) Endurance performance of the Cr2O3memory device at room temperature and (b) retention characteristics of the device at 85 °C. Reading voltage was 0.2 V.
Council of the Republic of China under Contract Nos. NSC-98-3114-M-110-001 and NSC 97-2112-M-110-009-MY3.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in the online version, atdoi:10.1016/j.sse.2010.12.014.
References
[1] Meindl JD, Chen Q, Davis JA. Science 2001;293():2044.
[2] Park S, Im H, Kim I, Hiramoto T. Jpn J Appl Phys Part 1 2006;45:638. [3] Jung JH, Kim J-H, Kim TW, Song MS, Kim Y-H, Jin S. Appl Phys Lett
2006;89:122110.
[4] Worledge DC. Appl Phys Lett 2004;84:4559.
[5] Chen L, Xia Y, Liang X, Yin K, Yin J, Chen Y, et al. Appl Phys Lett 2007;91:073511.
[6] Lin C-Y, Wu C-Y, Wu C-Y, Hu C, Tseng T-Y. J Electrochem Soc 2007;154:G189. [7] Seo S, Lee MJ, Seo DH, Jeoung EJ, Suh D-S, Joung YS, et al. Appl Phys Lett
2004;85:5655.
[8] Lee H-Y, Chen P-S, Wang C-C, Maikap S, Tzeng P-J, Lin C-H, et al. Jpn J Appl Phys Part 1 2007;46:2175.
[9] (a) Sakai J, Imai S. J Appl Phys 2005;97:10H709;
(b) Odagawa A, Kanno T, Adachi H. J Appl Phys 2006;99:016101. [10] Liang XF, Chen Y, Chen L, Yin J, Liu ZG. Appl Phys Lett 2007;90:022508. [11] Xu N, Gao B, Liu LF, Sun B, Liu XY, Han RQ, et al. Symp VLSI Technol 2008:100. [12] Douglas A. Skoog, Donald M. West. Fundamentals of analytical chemistry. 8th
ed. 2004.
[13] Lin Chih-Yang, Lee Dai-Ying, Wang Sheng-Yi, Lin Chun-Chieh, Tseng Tseung-Yuen. Surface Coat Technol 2008;203:628–31.
[14] Seo S, Lee MJ, Seo DH, Jeoung EJ, Suh D-S, Joung YS, et al. Appl Phys Lett 2006;86:093509.
[15] Lin CY, Wu CY, Wu CY, Lee TC, Yang FL, Hu C, et al. IEEE Electron Device Lett 2007;28:366.
[16] Hassel M, Hemmerich I, Kuhlenbeck H, Freund H-J. Surface Sci Spectra 1998;4(3).
[17] Cheng Ruihua, Xu B, Borca CN, Sokolov A, Yang C -S, Yuan L, et al. Appl Phys Lett 2001;79(19).
[18] Waser Rainer, Aono Masakazu. Nat Mater 2007;6(833–840):2.
[19] Szot Krzysztof, Speier Wolfgang, Bihlmayer Gustav, Waser Rainer. Nat Mater 2006;5(312–320):21.
