Effect of ITO electrode with different oxygen contents on the electrical characteristics
of HfO
x
RRAM devices
Chia-Wen Zhong
b, Wen-Hsien Tzeng
b, Kou-Chen Liu
a,⁎
, Horng-Chih Lin
b, Kow-Ming Chang
b,c,
Yi-Chun Chan
a, Chun-Chih Kuo
a, Pang-Shiu Chen
d, Heng-Yuan Lee
e, Frederick Chen
e, Ming-Jinn Tsai
e aInstitute of Electronics Engineering, Chang Gung University, Tao-Yuan, 33302, Taiwan, ROC b
Institute of Electronics, Department of Electronics Engineering, National Chiao Tung University, HsinChu, 30010, Taiwan, ROC c
Department of Electronic Engineering, I-Shou University, No. 1, Sec. 1, Syuecheng Rd., Kaohsiung, 840, Taiwan, ROC d
Department of Material Science and Engineering, Minghsin University of Science and Technology, Hsinchu, 340, Taiwan, ROC eElectronics and Optoelectronics Research Laboratory, Industrial Technology Research Institute, Hsinchu, 310, Taiwan, ROC
a b s t r a c t
a r t i c l e i n f o
Available online 20 July 2012 Keywords:
Indium tin oxide (ITO) Transparent-RRAM (TRRAM) Oxygen content
In this study, the influence of indium tin oxide (ITO) top electrodes with different oxygen contents on the re-sistive switching characteristics of HfOx/TiN capacitor structure is investigated. Switching parameters,
includ-ing set and reset voltage values, and high and low resistance values are highly related to the properties of ITO thinfilms. Higher resistance values in both states can be obtained when ITO thin films with higher oxygen contents are used as top electrodes; such values are accompanied by larger set voltages andfluctuating tran-sient currents during the reset process. Based on the proposedfilament model, we suggest that the switching mechanism of HfOx/TiN structure is attributed to the formation and rupture of conductingfilamentary paths
near the anodic side, which is highly correlated with the properties of the top electrode. The top electrode must be well determined to obtain reliable switching properties.
Crown Copyright © 2012 Published by Elsevier B.V. All rights reserved.
1. Introduction
Resistive random access memory (RRAM) devices based on the different resistances of oxide materials have been extensively investi-gated in recent years. Because of their advantages, such as small cell size[1], fast switching speed[2], low power consumption[3], and compatibility with standard complementary metal oxide semiconductor (CMOS) technologies[4], thus RRAM devices have become one of several candidates that may potentially be utilized for the next generation of non-volatile memory devices. To this end, a transparent electronic device with a fully transparent ITO/ZnO/ITO capacitor structure, called a transparent-RRAM (T-transparent-RRAM), has been achieved[5]. The realization of T-RRAM is a milestone in memory technology and demonstrates the high potential of RRAM device compatibility. The use of indium-tin-oxide as a top elec-trode is highly applied to optical-electronic devices, such as solar cells and organic light-emitting diodes (OLEDs). Transparent conducting ox-ides (TCOs) are highly degenerate n-type semiconductors and high carri-er concentration matcarri-erials that exhibit low electrical resistivities of up to 2×10−4Ω-cm[6]. It has been reported that the degeneracy observed in TCO stems from oxygen vacancies and substitution of tin dopants within the ITOfilm that occur during deposition. However, the electrical proper-ties of ITOfilms depend on the film composition and deposition parame-ters[7]. Because switching properties are highly correlated with the
properties of the electrode materials, variations in ITO thinfilm properties may cause different switching characteristics in RRAMs.
In this paper, the electrical switching behavior of an ITO/HfOx/TiN
capacitor is investigated. HfOx-based memory cells have been
pro-posed to be candidates for nonvolatile memory applications due to their excellent switching characteristics, including low operation cur-rent, large resistance ratio, fast switching speed, reliable switching endurance, and data retention[8]. We previously investigated the effect of top electrodes on the electrical characteristics of HfOx/TiN RRAM
de-vices[9]. Here, the effect of different oxygen contents within the ITO electrode on the switching properties of the HfOx/TiN structure is
discussed to provide a better understanding of the behavior of the inter-mediate state. Based on the reportedfilament model, we suggest that the intermediate state during the RESET process is attributed to the par-tial rupture of the conductingfilamentary paths near the anodic side, which is highly dominated by the oxygen contents of the top electrode. 2. Experimental method
A TiN bottom electrode was deposited on a 100 nm SiO2/Si substrate
layer by dc sputtering at room temperature (RT). The thickness and sheet resistance of the electrode were 30 nm and 50–60 Ω/□, respec-tively. A 20-nm HfOxthinfilm was deposited by atomic layer deposition
on a TiN/SiO2/Si structure at 300 °C. Hafnium tetrachloride and water
were used as reactants for HfOxdeposition. A 1000 Å-thick TCO ITO
was used as a top electrode, and deposited by rf-magnetron sputtering
Surface & Coatings Technology 231 (2013) 563–566
⁎ Corresponding author. Tel.: +886 32118800x3152; fax: +886 32118507. E-mail address:[email protected](K.-C. Liu).
0257-8972/$– see front matter. Crown Copyright © 2012 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.surfcoat.2012.07.039
Contents lists available atScienceDirect
Surface & Coatings Technology
at RT through a shadow mask with a diameter of 200μm. The sputtering target was a 3 inch ITO target which composed of In2O3:SnO2= 9:1 in
weight ratio (high-purity of 99.99%). To understand the effect of ITO oxygen contents on the switching characteristics of the device, three concentrations of oxygen with partial pressures of 0%, 15%, and 33% were sampled by controlling the O2sputtering gas from 0 to 12 sccm
during ITO deposition. For 15% and 33% samples, we only added the oxygen contents at an initial thickness of 100 Å, then the oxygenflow was controlled to 0% during the following 900 Å-thick ITOfilm deposi-tion. Theflow rate of the Ar sputtering gas was maintained at 24 sccm during deposition. A deposition power of 100 W and a process pressure of 5 mTorr were used for all samples under the same conditions. The electrical characteristics of the HfOx/TiN RRAM device were analyzed
with an Agilent 4156C semiconductor parameter analyzer. A bias volt-age was applied to the top electrode with a grounded TiN bottom elec-trode. All measurements were performed at RT.
3. Results and discussion
The resistive switching characteristics of RRAM devices require
electroforming with a current compliance (10μA) to activate the
switching behavior. After initial electroforming, RRAM devices switch to a low resistance state (LRS). As the applied voltage sweeps from 0 V to a certain positive voltage, the state switches back to a high resistance state (HRS), afterward the applied voltage sweeps from 0 V to a certain positive voltage, the HRS switches to low resistance state (LRS), the oper-ation and voltage are denoted as RESET process/voltage and SET process/ voltage respectively. This is called unipolar switching. We abbreviated our ITO/HfOx/TiN structure as IHT, and the different oxygen partial
pres-sures for ITO deposition are listed as IHT—0%, IHT—15%, and IHT—33%. Electroforming current–voltage (I–V) curves of the IHT samples are shown inFig. 1. The three samples exhibit almost similar leakage current values, a leakage current about 50 pA at 1 V and a breakdown voltage of about 5.5–6 V. No clear differences are observed between the electroforming processes of the three samples.
Fig. 2(a)–(c) shows the resistive switching characteristics of HfOx/TiN devices with ITO top electrodes fabricated under different
oxygen partial pressures of 0%, 15%, and 33%, respectively. A sequence of 5 cycles is shown to exhibit reproducible and reversible conducting behaviors. Thefigures show that the SET voltage increases slightly with increasing oxygen pressure. The above phenomenon maybe cause by formed a new HfOxthinfilm during formation of ITO top electrode
which the oxygen remained at initial thickness of 100 Å, the new HfOx
thinfilm is less oxygen-related defect, oxygen vacancy, causing extra
oxygen diffused into HfOxthinfilm which improves binding strength
so it need more energy to form conductivefilament. In addition, distinct conducting behaviors at the RESET operation are clearly observed. With increasing oxygen pressure of the ITO, an intermediate state during RESET becomes obvious. The intermediate state causes by random con-ductivefilaments are partial ruptured so the conduction current doesn't decrease abruptly in the intermediate state. When high enough energy Fig. 1. I–V curves of the electroforming process of the ITO/HfOx/TiN device with 0%,
15%, 33% oxygen content, respectively.
Fig. 2. Switching characteristics of the I–V curve of an HfOx/TiN capacitor with a top electrode of (a) ITO—0%, (b) ITO—15%, and (c) ITO—33% in 5 consecutive cycles. 564 C.-W. Zhong et al. / Surface & Coatings Technology 231 (2013) 563–566
ruptures remaining conductive filament the intermediate state is changed to high resistance state (HRS). The random conductive fila-ment is a major issue for device reliability. In the IHT—33% sample, the electrical characteristicsfirst switch to the intermediate state, and then maintain almost the same current with increasing voltage. When the voltage reaches a large enough value, about 3 V, the sec-ond switch to the HRS occurs. The intermediate state number is usu-ally 1 or 2. This phenomenon is observed less in the IHT—0% sample
and even less in the IHT—15% sample. We define that the voltage
value corresponding to Imaxis called Vm, and Vresetis defined as the
voltage at which the state completely switches back to the HRS. The set voltage is defined as Vset. This definition is different from
that in other reports because of the specific electrical characteristics of our IHT device.Table 1lists the Vset, Vm, and Vresetof the IHT
sam-ples. The respective average values for Vset, Vm, and Vresetare about
4.59, 2.35, and 4 V for IHT—0%, 5.02, 1.87, and 4 V for IHT—15%, and 5.90, 1.81, and 4.03 V for IHT—33%. It is obvious that IHT—33% has the highest Vresetand the lowest Vm. The Vresetis almost the
same for all samples. Larger Vsetvalues indicate that the SET operation
experiences difficulties in forming a conducting filament within the TMOfilm, and lower Vmvalues indicate that the conductingfilament
is easier to rupture during the RESET operation. We suggest that the shift in the operating voltage is related to the influence of the oxygen contents within the ITOfilm.
Fig. 3compares the endurance of the IHT—0% and IHT—33% de-vices. The poor high-to-low resistance ratio of the IHT—0% device limits the application of this ITOfilm as the top electrode for RRAM devices and confines the development of T-RRAM. Fortunately, device
performance is greatly improved in the IHT—33% sample. Compared
with IHT—0%, the IHT—33% sample shows higher resistance values
in both the HRS and LRS, and larger resistance ratio (1 order of mag-nitude), which may be attributed to the higher resistivity of the ITO film. The incorporation of oxygen will lead to a decrease in oxygen va-cancies in the ITOfilms and hence an increase in the resistivity. The larger the amount of oxygen ions that could be supplied from the ITO electrode of the IHT—33% device, the easier it is for filament rup-ture during the RESET operation. Therefore, switching stability could be improved.
It has been reported that oxygen contents play an important role in the resistive switching of RRAM devices [2,10,11]. Studies on RRAM devices based on HfOx/TiN substrates under unipolar switching
indicate that the switching mechanism may be attributed to the fila-ment model[12]. In thefilament model, the formation and rupture of
the conductingfilaments are the main switching mechanisms. The
switching process occurs near the anode electrode was proposed by Kim et al.[13]and Lee et al.[14]. We suggest that the intermediate state is highly dependent on the oxygen content of the electrode, imply-ing that thefilaments are influenced by oxygen ions within the ITO film. Schematic diagrams of the electrode effect are shown inFig. 4(a)–(e). During the SET operation, oxygen ions are extracted to the top electrode, contributing to oxygen vacancies within the HfOxfilm, and the injected
electrons conduct under positive bias. This causes a switch to the LRS, as shown inFig. 4(a). Larger Vsetvalues are observed in the IHT—33%
sam-ple; this may be attributed to oxygen ion migration difficulties due to the higher resistivity of the ITO—33% film at the anodic interface, as shown inFig. 4(c). Given a positive bias, electrons transported through oxygen vacancies contribute to the current in the LRS. As the dc sweep voltage
increases, larger amounts of electrons are transported through a tiny filament, which induced the Joule heating effect and the rearrangement of atoms. Then, the state was switched back to the HRS. In our previous report, we concluded that rupture of the conductingfilaments occurs at the anodic side of HfOxfilms because switching characteristics are
high-ly correlated with electrode properties. During the RESET process as shown inFig. 4(e), large amounts of non-lattice oxygen within the ITO—
33% film could influence the rearrangement of atoms during Joule
heating. The intermediate state may be attributed to the inconsistent rupture of conductingfilaments, as shown inFig. 4(d). Thesefindings
match the smaller Vm of IHT—33% well because larger amounts of
non-lattice oxygen ions could be easily triggered under thermal Joule heating. This may lead tofluctuations in the diameter of the conducting filaments and the intermediate states in a dispersed phenomenon. Since the switching properties of the IHTfilm are highly correlated with the number of oxygen ions or vacancies within the electrode and insulator thinfilm, a well-packaged fabricated RRAM device free from such is necessary to promote the reliable switching behavior of ITO/HfOx/TiN
RRAM devices. 4. Conclusion
We conclude that ITO top electrodes with different oxygen contents play significant roles in the switching characteristics of an HfOx/TiN
structure. The RESET behavior changes from the abrupt type to the Table 1
Operation voltages of the HfOxfilm obtained from ITO top electrodes with different oxygen contents.
Vset Vm Vreset
IHT—0% 4.59 2.35 4
IHT—15% 6.02 1.76 4
IHT—33% 6.75 1.54 4.03
Fig. 3. Endurance tests of the (a) ITO—0%/HfOx/TiN and (b) ITO—33%/HfOx/TiN capaci-tor structures.
565 C.-W. Zhong et al. / Surface & Coatings Technology 231 (2013) 563–566
gradual or two-step types, and the high and low resistance increases with increasing oxygen content in the ITO thinfilm. In addition, the high-to-low resistance ratio and set voltage values all increase when
ITO films with higher oxygen contents are used as top electrodes.
From the reportedfilament model, we propose a possible switching
mechanism based on the formation and rupture of conductingfilaments inside the HfOxfilms to elucidate distinctions between the switching
parameters of different samples. Top electrodes with larger oxygen contents may increase the resistivity of ITO thinfilms, resulting in the set voltage increased during the SET process. Larger amounts of non-lattice oxygen ions are easily reoxidized by oxygen vacancies under Joule heating, which is related to transitions between resistance states during the RESET process. An adequate top electrode may improve the switching characteristics of a device during operation.
Acknowledgment
This work was supported by J. P. Lin in Nanya Technology Corporation and by M. J. Tsai, P. S. Chen and H. Y. Lee in the Electronics and Optoelec-tronics Research Laboratory of Industrial Technology Research Institute and by the National Sciences Council (NSC 97_2221_E_182005).
References
[1] M. Lundstrom, Science (2003) 210.
[2] M. Fujimoto, H. Koyama, M. Konagai, Y. Hosoi, K. Ishihara, S. Ohnishi, N. Awaya, Appl. Phys. Lett. 89 (2006) 223509.
[3] K. Tsunoda, K. Kinoshita, H. Noshiro, Y. Yamazaki, T. Iizuka, Y. Ito, A. Takahashi, A. Okano, Y. Sato, T. Fukano, M. Aoki, Y. Sugiyama, IEDM Tech. Dig. (2007) 767. [4] A. Chen, S. Haddad, Y.C. Wu, Z. Lan, T.N. Fang, S. Kaza, Appl. Phys. Lett. 91 (2007)
123517.
[5] J.W. Seo, J.W. Park, K.S. Lim, J.H. Yang, S.J. Kang, Appl. Phys. Lett. 93 (2008) 223505. [6] C. Park, Y. Seo, J. Jung, D.W. Kim, J. Appl. Phys. 103 (2008) 054106.
[7] C.Y. Lin, C.Y. Wu, C.Y. Wu, T.C. Lee, F.L. Yang, C. Hu, T.Y. Tseng, IEEE Electron Device Lett. 28 (2007) 366.
[8] H.Y. Lee, P.S. Chen, T.Y. Wu, Y.S. Chen, C.C. Wang, P.J. Tzeng, C.H. Lin, F. Chen, C.H. Lien, M.J. Tsai, IEDM Tech. Dig. (2008) 297.
[9] K.M. Chang, W.H. Tzeng, K.C. Liu, W.R. Lai, ECS Trans. 28 (2010) 119.
[10] S. Seo, M.J. Lee, D.H. Seo, E.J. Jeoung, D.S. Suh, Y.S. Joung, I.K. Yoo, I.R. Hwang, S.H. Kim, I.S. Byun, J.-S. Kim, J.S. Choi, B.H. Park, Appl. Phys. Lett. 85 (2004) 5655. [11] D.C. Kim, M.J. Lee, S.E. Ahn, S. Seo, J.C. Park, I.K. Yoo, I.G. Baek, H.J. Kim, E.K. Yim, J.E. Lee,
S.O. Park, H.S. Kim, U-In Chung, J.T. Moon, B.I. Ryu, Appl. Phys. Lett. 88 (2006) 232106. [12] H.Y. Lee, P.S. Chen, C.C. Wang, S. Maikap, P.J. Tzeng, C.H. Lin, L.S. Lee, M.J. Tsai, Jpn.
J. Appl. Phys. 46 (2007) 2175.
[13] K.M. Kim, B.J. Choi, Y.C. Shin, S. Choi, C.S. Hwang, Appl. Phys. Lett. 91 (2007) 012907. [14] 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.
Fig. 4. Schematic pictures of the HfOx/TiN structure at (a) LRS and (b) HRS with ITO—0% as the top electrode, and (c) LRS, (d) IRS, and (e) HRS with ITO—33% as the top electrode. Solid and empty circles represent the oxygen ions and oxygen vacancies, respectively.