Resistive switching characteristics of gallium oxide for nonvolatile
memory application
Jyun-Bao Yang
b, Ting-Chang Chang
a,b,⁎
, Jheng-Jie Huang
a, Shih-Ching Chen
a, Po-Chun Yang
b,
Yu-Ting Chen
b, Hsueh-Chih Tseng
a, Simon M. Sze
c, Ann-Kuo Chu
b, Ming-Jinn Tsai
da
Department of Physics, National Sun Yat-Sen University, Kaohsiung, Taiwan
b
Department of Photonics, National Sun Yat-Sen University, Kaohsiung, Taiwan
c
Institute of Electronics, National Chiao Tung University, Hsin-Chu, Taiwan
d
Electronics and Optoelectronics Research Laboratory, Industrial Technology Research Institute, Chutung, Hsinchu, Taiwan
a b s t r a c t
a r t i c l e i n f o
Available online 25 October 2012 Keywords:
RRAM Gallium oxide Oxygen vacancies
This study investigates bipolar resistance switching memory in a fabricated Pt/GaOx/TiN device. Gallium oxide sputtered in ambient Ar shows a change in resistance ratio of two orders of magnitude. The enhance-ment of resistance ratio is also observed in the gallium oxide layer when deposited in ambient Ar/O2. The X-ray photoelectron spectroscopy analysis shows that this gallium oxide in ambient Ar/O2can reduce the number of defects and enhance the stability of switching behavior. An analysis of current transport mecha-nism in the high resistance state indicates that the larger effective thickness can be attributed to the higher oxygen concentration, and can increase the resistance value of the high resistance state.
© 2012 Elsevier B.V. All rights reserved.
1. Introduction
Recently, the development of nonvolatile memory has been in-fluenced by device dimension scaling. When devices are continuously miniaturized, the tunneling oxide of thefloating-gate memory will become thinner[1]. Consequently, the stored charges can leak back to the substrate and lead to a loss of stored information[2–5]. In order to solve this problem, the next generation non-volatile memories must be developed, such as ferroelectric random access memory[6], magnet-ic random access memory[7], phase change random access memory[8], and resistance random access memory (RRAM)[9]. RRAM has attracted considerable interest due to its advantages of simple structure, low energy consumption, higher operating speed and higher endurance [10–14]. Therefore, many studies of promising materials have been reported, such as organic materials[15]and transition metal oxides [16–21].
2. Experiment
The substrate used in this experiment was TiN/SiO2/Si, with the
200-nm-thick SiO2(low temperature oxide, LTO) deposited on the
TiN. Moreover, we defined the size of via (4×4 μm) by etching the LTO layer. Secondly, a 23-nm-thick GaOxinsulator layer was
deposit-ed on the previous substrate by sputtering a Ga0.4O0.6target at room
temperature in ambient Ar (30 sccm), and labeled as sample A. In ad-dition, a 22.7-nm-thick GaOxwas also deposited on the substrate in
Ar/O2(30 sccm/10 sccm) for comparison as sample B. The sputtering
power, deposition rate and working pressure were set to 150 W, 23 Å/min and 0.53 Pa, respectively. Finally, the Pt/GaOx/TiN device
was completed by capping the top electrode (Pt) on the GaOx
shown in the inset ofFig. 1. The resistive switching characteristics of Pt/GaOx/TiN were measured by an Agilent B1500 semiconductor
⁎ Corresponding author at: Department of Physics, National Sun Yat-Sen University, Kaohsiung, Taiwan. Tel.: +886 7 5252000 3708; fax: +886 7 5253709.
E-mail address:[email protected](T.-C. Chang).
Fig. 1. The forming process of samples A and B. Inset is the Pt/GaOx/TiN device schematic
diagram. 0040-6090/$– see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.tsf.2012.10.026
Contents lists available atSciVerse ScienceDirect
Thin Solid Films
parameter analyzer, and the composition of the gallium oxidefilm was analyzed by X-ray photoelectron spectroscopy (XPS, JEOL JPS 9010 MX). The spectrometer source of Al Kα (1486.6 eV) radiation was used for XPS analysis. All of the peaks shape and parameters were fitting by using the software of XPSPEAK (Version 4.1) with Lorentzian–Gaussian functions. In addition, before gathering the XPS signal, the sample surface had been cleaning by argon ions beam. The beam voltage and beam current were 500 V and 8 mA, respectively.
Fig. 2. The bipolar switching characteristics of samples A and B. Inset is the 100 cycle endurance characteristic, which is swept by DC voltage.
Fig. 3. The current transport models for sample B for (a) Ohmic conduction and (b) Schottky emission. The resistance values of (c) LRS and (d) HRS are observed at temperatures ranging from 300 K to 400 K.
Fig. 4. The resistance switching mechanism of Pt/GaOx/TiN during (a) the forming
3. Results and discussion
In this work, the voltage was applied on the TiN electrode and the Pt electrode was grounded. A + 10 V was applied with a current com-pliance of 1μA on both samples A and B for the purpose of device ac-tivation, namely the forming process. The voltage where the current increases abruptly is defined as the forming voltage. Fig. 1 shows that the forming voltage of sample A (6.2 V) is lower than that of sample B (7 V). Since the conduction paths are constructed during the forming process, the resistance state is switched to low resistance
state (LRS). By sweeping voltage from 0 V to−1.6 V, the resistance state is switched from LRS to high resistance state (HRS), named reset process. Subsequently, the set process is employed to switch the LRS by sweeping from 0 V to 2 V with a current compliance of 10 mA.
Fig. 2shows the resistive switching behavior of the gallium oxide film. The resistance ratio extracted at 0.2 V is approximately 2 orders of magnitude for sample A (in black), and is enhanced to about 2.5 orders for sample B (in red), as shown inFig. 2and its inset. Through this statistical comparison, the switching behavior of sample B can be clearly seen as being more stable.
In order to investigate the resistive switching behavior, the carrier transport mechanisms are analyzed. The analysis result indicates that the current of sample B is dominated by Ohmic conduction and by Schottky emission in LRS and HRS, respectively, as shown inFig. 3(a) and (b). When the device is switched to HRS, the gallium oxidefilm conduction path ruptures and the switching layer is formed. When volt-age is applied to the TiN electrode, the electrons in the conduction path acquire enough energy to overcome the barrier height between the Fermi level of the conduction path and the GaOxconduction band, as
shown in the inset ofFig. 3(b). Moreover, the carrier transport mecha-nisms of sample A are similar to that of sample B. In order to study the resistance switching mechanisms, the resistance variation trends for LRS and HRS are observed at temperatures ranging from 300 K to 400 K. The decreasing resistance value of LRS with the rising tempera-ture indicates that conduction paths have a semiconductor-like (oxygen vacancies) property[23], as shown inFig. 3(c). Furthermore, the con-duction mechanism can be confirmed to be the Schottky emission
Fig. 5. XPS analysis of (a) O 1s for sample A, (b) O 1s for sample B, (c) Ga 3d5/2for sample A and (d) Ga 3d5/2for sample B.
from the decreasing resistance in HRS while temperature rises, as shown inFig. 3(d).
According to the analysis above, we can conclude that the current transport mechanism of the Pt/GaOx/TiN structure is related to oxygen
vacancies[22,23], and a model responsible for the conduction mecha-nism is therefore proposed. Fig. 4 shows the resistance switching behaviors of the Pt/GaOx/TiN structure. During the forming process,
when applying a positive voltage on the TiN electrode, the oxygen va-cancy conduction path is formed due to the break of the Ga\O bonds by high energy electrons, as shown inFig. 4(a). Subsequently, the oxygen ions migrate toward the TiN because of the electric-field direc-tion and are reserved in the TiN electrode. When a negative bias is applied, the oxygen ions move away from the TiN/GaOxinterface and
re-combine with the oxygen vacancies along the conduction path. As a result, the resistance value is switched to HRS because the switching layer is formed between TiN and the conduction path. Conversely, when an enough positive bias is applied, the Ga\O bonds in the switching layer can be broken by energetic electrons and further induce the re-formation of oxygen vacancies, as shown inFig. 4(b).
In order to investigate the formation of the GaOxfilm, X-ray
pho-toelectron spectroscopy (XPS) analysis was performed.Fig. 5(a) and (b) shows the XPS analysis of O 1s (529.5 eV)[24].Fig. 5shows the composition ratio of O bonding to Ga 3d5/2bonding in samples A
and B is 1:1.07. This result indicates that the oxygen concentration of the GaOxlayer in sample B is indeed increased by sputtering in
ambient Ar/O2.Fig. 5(c) and (d) shows the XPS analysis of Ga 3d5/2.
Using the Lorentzian–Gaussian functions analysis method, two peaks are present in both samples for the Ga 3d5/2analysis. InFig. 5(c) and
(d), those peaks are verified to be Ga\O bonding (19.2 eV) and Ga\Ga bonding (18.2 eV)[24]. In order to investigate the influence of the additional oxygen atoms for the gallium oxidefilm, the composition of Ga 3d5/2bonding was analysis. The composition ratio of the resistive
switching layers can be identified by calculating the peak areas in the XPS analysis, with results indicating that, in comparison to sample A, the composition ratio of Ga\O bonding to Ga\Ga bonding increases in sample B. The additional oxygen atoms in sample B enhance the oxidation of Ga atoms, leading to the decreasing number of defects in the gallium oxidefilm. This analysis indicates the switching behavior of sample B is more stable because the additional oxygen can reduce the defects in the gallium oxide and decrease the distribution of conduction paths, as shown in the inset ofFig. 2.
In addition, as noted before,Fig. 3(b) shows that the conduction mechanism of HRS is dominated by the Schottky emission. The thick-ness of the switching layer is defined as effective thickness (deff), as
shown inFig. 4(b), and can be estimated from the Schottky emission equation: J¼ AT2
exp −q φðB−pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiqV=4πεideffÞ
kT
. The effective thickness can be estimated from the slope valuekTq ffiffiffiffiffiffiffiffiffiffiffiffi4πεq
ideff
q
in the plot of ln TJ2
ver-suspffiffiffiffiV. The k,εiand deffare the Boltzmann constant, permittivity and
effective thickness of the resistive switching layer, respectively. The slopes of samples A and B are extracted to be 8.01 and 7.53, as shown inFig. 6and inFig. 3(b). The estimated deffof sample B is
about 2.53 nm, larger than that of sample A (2.23 nm). This result indicates that the oxygen-rich composition of sample B can provide sufficient oxygen ions to re-combine with the oxygen vacancies among the switching layer. Hence, the effective thickness of sample B is larger than that of sample A.
Fig. 7(a) shows the endurance over 104cycles. When applying a + 1.2 V with a 400 ns pulse width, the resistance state is switched from HRS to LRS. Conversely, the resistance state is switched from LRS to HRS by the −1.6 V pulse with a 600 ns pulse width. The obtained results reveal that both samples A and B can withstand at least 104cycles and still retain approximately 1.5 orders in their
respective resistance ratios. For the retention characteristics, after the 104cycle endurance test, the LRS and HRS for both samples
retains stable without degradation longer than 104s at 85 °C, as
shown inFig. 7(b). 4. Conclusion
In summary, we deposited a GaOx insulator by sputtering a
Ga0.4O0.6target in ambient Ar (sample A), and in Ar/O2(sample B)
in order to enhance the concentration of oxygen atoms. The resis-tance value of sample B is increased in HRS, because the additional O2gas can provide sufficient oxygen ions to re-combine the oxygen
vacancies along the switching layer, and a larger effective thickness can be formed. Furthermore, the switching behavior of sample B be-comes more stable by reducing the defects in the gallium oxide.
Acknowledgment
This work was performed at National Science Council Core Facilities Laboratory for Nano-Science and Nano-Technology in Kaohsiung– Pingtung area and was supported by the National Science Council of the Republic of China under contract nos. NSC 100-2120-M-110-003 and NSC-97-2112-M-110-009-MY3.
Fig. 7. (a) Endurance for 104
cycles as tested by AC sweeping, and (b) the retention for 104
References
[1] T.C. Chang, F.Y. Jian, S.C. Chen, Y.T. Tsai, Mater. Today 14 (2011) 608.
[2] F.M. Yang, T.C. Chang, P.T. Liu, U.S. Chen, P.H. Yeh, Y.C. Tu, J.Y. Lin, S.M. Sze, J.C. Lou, Appl. Phys. Lett. 90 (2007) 222104.
[3] F.M. Yang, T.C. Chang, P.T. Liu, P.H. Yeh, Y.C. Yu, J.Y. Lin, S.M. Sze, J.C. Lou, Appl. Phys. Lett. 90 (2007) 132102.
[4] W.R. Chen, T.C. Chang, P.T. Liu, J.L. Yeh, C.H. Tu, J.C. Lou, C.F. Yeh, C.Y. Chang, Appl. Phys. Lett. 91 (2007) 082103.
[5] T.C. Chen, T.C. Chang, F.Y. Jian, S.C. Chen, C.S. Lin, M.H. Lee, J.S. Chen, C.C. Shih, IEEE Electron. Device Lett. 30 (2009) 834.
[6] G. Asano, H. Morioka, H. Funakubo, Appl. Phys. Lett. 83 (2003) 5506. [7] A. Ney, J.S. Harris Jr., Appl. Phys. Lett. 86 (2005) 013502.
[8] T.C. Chong, L.P. Shi, R. Zhao, P.K. Tan, J.M. Li, H.K. Lee, X.S. Miao, A.Y. Du, C.H. Tung, Appl. Phys. Lett. 88 (2006) 122114.
[9] Y.T. Tsai, T.C. Chang, W.L. Huang, C.W. Huang, Y.E. Syu, S.C. Chen, S.M. Sze, M.J. Tsai, T.Y. Tseng, Appl. Phys. Lett. 99 (2011) 092106.
[10] L.W. Feng, C.Y. Chang, Y.F. Chang, W.R. Chen, S.Y. Wang, P.W. Chiang, T.C. Chang, Appl. Phys. Lett. 96 (2010) 052111.
[11] S.C. Chen, T.C. Chang, S.Y. Chen, C.W. Chen, S.C. Chen, S.M. Sze, M.J. Tsai, M.J. Kao, F.S. Yeh, Solid-State Electron. 62 (2011) 40.
[12] H.Y. Lee, P.S. Chen, T.Y. Wu, C.C. Wang, P.J. Tzeng, C.H. Lin, F. Chen, M.J. Tsai, C. Lien, Appl. Phys. Lett. 92 (2008) 142911.
[13] P.C. Yang, T.C. Chang, S.C. Chen, Y.S. Lin, H.C. Huang, D.S. Gan, Electrochem. Solid-State Lett. 14 (2011) H93.
[14] K.C. Chang, T.M. Tsai, T.C. Chang, Y.E. Syu, C.C. Wang, S.L. Chuang, C.H. Li, D.S. Gan, S.M. Sze, Appl. Phys. Lett. 99 (2011) 263501.
[15] K.M. Kim, B.J. Choi, B.W. Koo, S. Choi, D.S. Jeong, C.S. Hwang, Electrochem. Solid-State Lett. 9 (2006) G343.
[16] L.P. Ma, J. Liu, Y. Yang, Appl. Phys. Lett. 80 (2002) 2997.
[17] Y.E. Syu, T.C. Chang, T.M. Tsai, Y.C. Hung, K.C. Chang, M.J. Tsai, M.J. Kao, S.M. Sze, IEEE Electron Device Lett. 32 (2011) 545.
[18] L.W. Feng, Y.F. Chang, C.Y. Chang, T.C. Chang, S.Y. Wang, P.W. Chiang, C.C. Lin, S.C. Chen, S.C. Chen, Thin Solid Films 519 (2010) 1536.
[19] H. Shima, F. Takano, H. Muramatsu, H. Akinaga, I.H. Inoue, H. Takagi, Appl. Phys. Lett. 92 (2008) 043510.
[20] K.M. Kim, B.J. Choi, C.S. Hwang, Appl. Phys. Lett. 90 (2007) 242906.
[21] K. Jung, J. Choi, Y. Kim, H. Lm, S. Seo, R. Jung, D.C. Kim, J.S. Kim, B.H. Park, J.P. Hong, J. Appl. Phys. 103 (2008) 034504.
[22] M. Liu, Z. Abid, W. Wang, X. He, Q. Liu, W. Guan, Appl. Phys. Lett. 94 (2009) 233106.
[23] X. Gao, Y. Xia, J. Ji, H. Xu, Y. Sum, H. Li, C. Yang, H. Guo, J. Yin, Z. Liu, Appl. Phys. Lett. 97 (2010) 193501.
