Switching mechanism of double forming process phenomenon in ZrOx/HfOy bilayer
resistive switching memory structure with large endurance
Chun-Yang Huang, Chung-Yu Huang, Tsung-Ling Tsai, Chun-An Lin, and Tseung-Yuen Tseng
Citation: Applied Physics Letters 104, 062901 (2014); doi: 10.1063/1.4864396 View online: http://dx.doi.org/10.1063/1.4864396
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/104/6?ver=pdfcov Published by the AIP Publishing
Articles you may be interested in
Low power W:AlOx/WOx bilayer resistive switching structure based on conductive filament formation and rupture mechanism
Appl. Phys. Lett. 102, 173503 (2013); 10.1063/1.4803462
Mechanism of resistive switching in Cu/AlOx/W nonvolatile memory structures J. Appl. Phys. 113, 164506 (2013); 10.1063/1.4803062
Resistive switching mechanisms relating to oxygen vacancies migration in both interfaces in Ti/HfOx/Pt memory devices
J. Appl. Phys. 113, 064510 (2013); 10.1063/1.4791695
Robust unipolar resistive switching of Co nano-dots embedded ZrO2 thin film memories and their switching mechanism
J. Appl. Phys. 111, 014505 (2012); 10.1063/1.3674322
Effect of ZrO x / HfO x bilayer structure on switching uniformity and reliability in nonvolatile memory applications Appl. Phys. Lett. 97, 172105 (2010); 10.1063/1.3491803
Switching mechanism of double forming process phenomenon in ZrO
x/HfO
ybilayer resistive switching memory structure with large endurance
Chun-Yang Huang, Chung-Yu Huang, Tsung-Ling Tsai, Chun-An Lin, and Tseung-Yuen Tsenga)
Department of Electronics Engineering and Institute of Electronics, National Chiao Tung University, Hsinchu 30010, Taiwan
(Received 6 November 2013; accepted 24 January 2014; published online 10 February 2014) In this Letter, the mechanism of double forming process phenomenon revealing in ZrO2/HfO2
bilayer resistive random access memory structure is investigated. This phenomenon caused by the formation of TiON interfacial layer can be well explained by using the energy band diagram. The TiON interfacial layer will be a tunneling barrier during the first forming process when a negative voltage applied on the device, while it will breakdown when applying a positive voltage. Besides, due to the double forming process, an asymmetric conductive filament with narrower size at ZrO2/HfO2interface is formed in the device. The point for formation and rupture of the conductive
filament can be confined at the ZrO2/HfO2 interface, and it will suppress the consumption of
oxygen ions during endurance test. Therefore, high speed (40 ns) and large endurance (107cycles) characteristics are achieved in this device structure. VC 2014 AIP Publishing LLC.
[http://dx.doi.org/10.1063/1.4864396]
Transition metal oxide (TMO) based resistive random access memory (RRAM) is one of the promising candidates to replace the current flash memory device as the next-generation nonvolatile memory application.1,2The resistive switching (RS) phenomenon in metal oxide layer is attrib-uted to the conductive filament consisting of migrated oxy-gen vacancies or ions under the applied electric field.3–6 Recently, the bilayer structure RRAM devices were pro-posed to improve endurance property, which revealed high performance RS than single layer devices.7–11However, the resistance ratio between high resistance state (HRS) and low resistance state (LRS) will tend to fluctuation and decreasing during switching cycles test.12,13This problem leads the en-durance degradation for its applications. Some literature reported that the endurance degradation is caused by the con-sumption of available oxygen ions for participating in RS, which may be caused by the oxygen ions migrating toward electrodes and release them from the electrodes under an applying electric field.14–18
In this study, a TaN/ZrO2/HfO2/TiN RRAM device was
proposed for the high performance RS properties. All of those materials for making metal-insulator-metal structure were compatible with complementary metal oxide semicon-ductor (CMOS) processes. In addition, we found that those devices cannot execute success resistance switching after forming process by applying a positive voltage. In addition, an interesting phenomenon of the double forming process was found in this bilayer structure. The conductive filament mechanism can be used to well explain the double forming process phenomenon. It was clearly indicated that the endur-ance fluctuation can be improved by controlling the point in the interface between HfO2 and ZrO2 layers for formation
and rupture of conductive filament. By the way, large endur-ance cycles (107cycles) with about 100 times of HRS/LRS
ratio under a fast speed (40 ns) were achieved in this TaN/ZrO2/HfO2/TiN device structure.
A 5 nm HfO2thin film as 1st RS layer was deposited on
TiN bottom electrode by sputtering at room temperature in 5 mTorr atmosphere with Ar/O2 mixture gas ratio of 10:3.
Subsequently, an oxygen annealing process at a furnace of 400C with O2gas flow of 30 SCCM for 30 min was
exe-cuted to improve the quality and reduce the amount of defects (oxygen vacancies) in HfO2layer. Then, a 3 nm ZrO2
thin film as 2nd RS layer was deposited by sputtering at room temperature in 10 mTorr atmosphere with Ar/O2
mix-ture gas ratio of 2:1. Finally, a 100 nm TaN top electrode with a diameter of 150 lm was deposited by sputtering at room temperature to form the TaN/ZrO2/HfO2/TiN
architec-ture. The direct current (DC) electrical characteristics meas-urements were performed by using an Agilent B1500A semiconductor parameter analyzer, and the alternating cur-rent (AC) pulse was generated by Agilent B1530A waveform generator/fast measurement unit (WGFMU). The voltage was applied on the TaN top electrode with the TiN bottom electrode grounded.
Figure1(a)shows the typical cross-section transmission electron microscopy (TEM) image of the TaN/ZrO2/HfO2/TiN bilayer RRAM device. Due to the
obvi-ously crystalline orientation formed in the TiN electrode, a thin interfacial layer TiON (2 nm) is observed. The DC sweep electrical I-V curve of the present device is shown in Fig. 1(b). Unlike regular bipolar RRAM devices, the TaN/ZrO2/HfO2/TiN device reveals the double forming
pro-cess phenomenon for switching its resistance from initial state (IS) to LRS. The double forming process can be well described in six steps: (1) First forming process, (2) Medium state (denoted as MS), (3) Second forming process, (4) LRS, (5) Reset process, and (6) HRS, as shown in Fig.1(b).
Step (1) is to construct a larger size conductive filament in ZrO2layer and a thinner one in HfO2layer. The typical
X-ray photoelectron spectra (XPS) analyses can confirm this
a)
phenomenon. The oxygen compositions of ZrO2and HfO2
layers are depicted in Figs.1(c)and1(d), respectively. The two major peaks in these figures can be fitted for ZrO2and
HfO2layers, respectively. The binding energies of the main
peaks at around 531.4 and 530.9 eV are attributed to the O2 ions in the stoichiometric Zr-O and Hf-O bonds, respec-tively, while the higher energies of 532.1 eV in these figures correspond to the oxygen vacancies in ZrO2and HfO2layers,
respectively.19The inset tables in Figs.1(c)and1(d)indicate the area proportion of each peak. Due to the post oxygen annealing process treated for the HfO2layer, the percentage
of oxygen vacancies existed in the HfO2layer is only 10%,
whereas that in the ZrO2layer is 25%. Therefore, compared
to ZrO2layer, less amount of the oxygen vacancies is stored
in HfO2layer, as shown in Fig.2(a). Besides, in our previous
study, the physical size of the conductive filament in an oxy-gen vacancies deficient layer is narrower than that in the ox-ygen vacancies rich layer.9Therefore, two different sizes of conductive filaments are formed and linked together to
connect top electrode and TiON layer (Fig.2(b)). However, no conductive filament is formed within the TiON layer in the step (1). It can be explained by the energy band diagram shown in Figs. 2(f)–2(h). Fig. 2(f) depicts the conduction band offset diagram. As a lot of defects (oxygen vacancies) exist in the ZrO2layer, the electrons can easily hop through
it from top electrode to HfO2layer when a negative voltage
is applied. Therefore, when the applied negative voltage increases to the first forming voltage ( 4 V), the defects (ox-ygen vacancies) generate and form conductive filament in HfO2layer. According to the band diagram in Fig. 2(g), the
electron passing through HfO2layer can easily slip along the
tilted conduction band in TiON layer to TiN bottom elec-trode. Therefore, a larger current reaches to compliance cur-rent as measured in Fig.1(b). However, the energy band will tend to be more flat when the applied voltage sweeps back to 0 V, thus the electrons located at the defects of HfO2layer
will face a barrier between HfO2and TiON layers. This
bar-rier causes a larger resistance when decreasing the applying
FIG. 1. (a) Typical cross-section TEM image of the TaN/ZrO2/HfO2/TiN
RRAM device. (b) DC sweep I-V curve of the double forming process phenomenon: (1) First forming pro-cess; (2) Medium state; (3) Second forming process; (4) Low resistance state; (5) Reset process; and (6) High resistance state. The O 1 s XPS spectra of (c) ZrO2and (d) HfO2layers.
FIG. 2. Schematic description of the conducting mechanism of the double forming process of the TaN/ZrO2/HfO2/TiN RRAM device on
(a) initial state, (b) first forming pro-cess, (c) second forming propro-cess, (d) reset process, and (e) set process. The energy band diagram at (f) conduction band offset, (g) applying a negative voltage, and (h) applying a positive voltage on the top electrode in the present device. The work functions of TaN (3.4 eV) and TiN (4.5 eV); the electron affinities of ZrO2 (2.8 eV),
HfO2 (2.65 eV), and TiON (3.2 eV)
are found in Refs.21–24.
voltage. Therefore, the device switches from IS to MS except regular LRS in step (2), and the conductive mechanism is dominated by the Fowler-Nordheim (FN) tunneling.20 In addition, the ln(I/V2) versus 1/V curve shows a straight fit-ting line with a negative slope at large voltage region (more negative) in HRS, which proves the conductive mechanism dominated by FN tunneling in step (2), as shown in Fig.3(a). The property in step (2) is also shown in the HfO2single
layer device; however, no further RS behavior reveals in it (not shown here).
Besides, compared to the TaN/ZrO2/HfO2/TiN device
under the double forming process in Fig.1(b), the another same fresh device can also execute a single forming process by applying a positive voltage. However, if we attempt to switch such a device from LRS to HRS by applying a nega-tive bias, the device switches to much lower resisnega-tive LRS rather than switches to HRS, as shown in curves (iii) and (iv) of Fig.3(b). Fig.2(h)illustrates the energy band diagram of positive forming process induced hard breakdown. Due to the asymmetric barriers at TiN/TiON (1.3 eV) and TaN/ZrO2
(0.6 eV) interfaces, the higher barrier at TiN/TiON interface causes that the FN tunneling phenomenon in TiON is not obvious during the positive forming process. On the other hand, due to the work function difference between TaN and TiN electrodes, there is a build-in electric field created from TaN toward TiN in the resistive switching layers. When a positive voltage is applied on the TaN top electrode, the external electric field associated with build-in one will enhance the effective field in the layers. Therefore, the TiON layer will tend to breakdown and generate more oxygen vacancies in it during the positive forming process. On the other hand, oxygen vacancies also generate in HfO2layer in
this situation. Those oxygen vacancies will tend to migrate to form the conductive filament. The shape of the conductive filament is illustrated in inset of the Fig.3(b). Therefore, the device cannot switch to HRS because of a large sized
con-TaN/ZrO2/HfO2/TiN device cannot operate suitably under a
single positive forming process.
After completing the first forming process, we also try a negative voltage to switch the device from MS to a much higher resistance state; unfortunately, a permanent hard breakdown occurs in it. On the other hand, a successful re-sistance switching for memory property is executed when a positive voltage applies on the device in MS as shown in Fig.1(b). The successful bipolar RS characteristic with 100 cycles is shown in Fig. 3(c). Compared to the smaller set voltage (1 V) in Fig.3(c), a larger threshold voltage about 2.8 V is measured. Therefore, we call this process as the sec-ond forming process with a secsec-ond forming voltage of 2.8 V in step (3) of Fig.1(b). After the second forming process, the device can switch from MS to LRS, as shown in step (4). Finally, the regular reset process is effectively executed by applying a negative voltage after the double forming process steps in the present device, as shown in steps (5) and (6). Due to a positive voltage applies on the top electrode, the ox-ygen vacancies with positive charge in HfO2layer will be
repelled toward bottom electrode. By the way, as described before in Figs. 2(h)and 3(b), a positive voltage applied on the device will cause a breakdown in the TiON layer. Therefore, the conductive filament will re-construct itself to a large size at near TiN and a small size at near the ZrO2/HfO2 interface. Besides, the size of conductive
fila-ment in ZrO2layer is still large because of the existing large
amount of oxygen vacancies in it during step (1). Hence, an asymmetric shape of conductive filament with a weakest point located at the ZrO2/HfO2 interface is designed and
formed in the present device, as shown in Fig.2(c). Thus, the device switches from MS to LRS due to the formation of complete conductive filament. Finally, the repeatable resist-ance switching properties can be achieved in this device.
The electric field can be enhanced in the conductive fila-ment due to its larger conductivity than other regions of the
FIG. 3. (a) F-N tunneling curve fitting of the TaN/ZrO2/HfO2/TiN RRAM
de-vice in medium state. (b) I–V curve of positive voltage forming process. The inset shows the schematic diagram. (c) Typical bipolar I-V curves with 100 times switching cycles. (d) AC endur-ance characteristic at room tempera-ture and the inset shows the read disturbance behavior at 85C with a
ions existed in ZrO2layer will migrate to HfO2layer when a
negative voltage is applied on top electrode. In addition, the current flowing through the narrowest size of asymmetric conductive filament (the ZrO2/HfO2interface) will cause a
serious local Joule heating in it. Consequently, the migrated oxygen ions associated with local thermal effect will easily recombine (oxidation) with oxygen vacancies at the ZrO2/HfO2 interface or top of the conductive filament in
HfO2layer. Thus, dissolution of the asymmetric conductive
filament causes it switching from LRS to HRS, as shown in Fig. 2(d). On the other hand, when a positive voltage is applied on the device in Fig.2(c), the oxygen ions existed in HfO2layer will also tend to migrate to ZrO2layer. However,
compared to HfO2 layer, the conductive filament in ZrO2
layer is too large to rupture it. Therefore, the device cannot switch to HRS by applying a positive voltage on it. When a positive voltage is applied on it during the set process, a soft breakdown and reduction reaction (oxygen vacancies are generated) occur in the region above top of the conduction filament in HfO2layer. The escaped oxygen ions due to
gen-erated oxygen vacancies migrate back to ZrO2layer to
oxi-dize part of the oxygen vacancies to re-form the conductive filament. The device switches back to LRS, as shown in Fig.
2(e). Therefore, the formation and rupture point of conduc-tive filament for RS property can be limited at a very narrow region at the ZrO2/HfO2interface.
Due to the oxygen vacancies generation and recombina-tion can be confined at the ZrO2/HfO2interface except near
the electrode during RS cycles test, the problem of escaping oxygen ions from electrodes18 can be suppressed in the de-vice. Therefore, it can be expected that the highly stable en-durance characteristic can be achieved in the present device structure. Fig. 3(d) shows the high speed AC endurance property by using a pulse height of 3 V for set process and 3.3 V for reset process under a pulse width of 40 ns. Large endurance of more than 107 switching cycles is achieved. About 100 times resistance ratio with less fluctuation between HRS and LRS is revealed in the ZrO2/HfO2bilayer
device. In addition, the read disturbance property under a constant voltage stress of 0.3 V at 85C is shown in the inset of Fig.3(d). No degradation property in both HRS and LRS is maintained for more than 104s.
In summary, the TaN/ZrO2/HfO2/TiN bilayer RRAM
structure is fabricated in this study. The structure reveals the double forming process phenomenon before resistive switch-ing cycles. The energy band diagram and conductive fila-ment model can well explain this interesting phenomenon. Due to the formation of TiON thin layer between HfO2and
TiN interface, it causes a dielectric breakdown only by using positive voltage applied on the device. An asymmetric con-ductive filament is formed after the double forming process, which can confine the point for formation and rupture of con-ductive filament at the ZrO2/HfO2interface during resistive
switching. It can suppress the resistance state degradation caused by escaping the oxygen ions from the electrodes.
Therefore, the large endurance (107cycles) with high opera-tion speed (40 ns) is observed in this device structure, which is a good candidate for next generation nonvolatile memory application.
This work was supported by National Science Council, Taiwan, under Project No. NSC 102-2221-E-009-134-MY3.
1T. Y. Tseng and S. M. Sze, “An introduction to nonvolatile memories,” in
Nonvolatile Memories: Materials, Devices, and Applications, edited by T. Y. Tseng and S. M. Sze (Am. Scientific Pub., CA, USA, 2012), Vol. 1, pp. 1–9.
2Y. Y. Chen, L. Coux, S. Clima, B. Govoreanu, R. Degraeve, G. S. Kar, A.
Fantini, G. Groesenken, D. J. Wouters, and M. Jurczak, IEEE Trans. Electron Devices60, 1114 (2013).
3B. Gao, B. Sun, H. Zhang, L. Liu, X. Liu, R. Han, J. Kang, and B. Yu, IEEE Electron Device Lett.30, 1326 (2009).
4
B. Gao, J. Kang, L. Liu, X. Liu, and B. Yu,Appl. Phys. Lett.98, 232108 (2011).
5H.-S. P. Wong, H. Y. Lee, S. Yu, Y. S. Chen, Y. Wu, P. S. Chen, B. Lee,
F. T. Chen, and M. J. Tsai,Proc. IEEE100, 1951 (2012).
6
C. Y. Lin, S. Y. Wang, D. Y. Lee, and T. Y. Tseng,J. Electrochem. Soc.
155, H615 (2008).
7Y. Bai, H. Wu, Y. Zhang, M. Wu, J. Zhang, N. Deng, H. Qian, and Z. Yu, Appl. Phys. Lett.102, 173503 (2013).
8
T. Ninomiya, S. Muraoka, Z. Wei, R. Yasuhara, K. Katayama, and T. Takagi,IEEE Electron Device Lett.34, 762 (2013).
9M. H. Lin, M. C. Wu, C. Y. Huang, C. H. Lin, and T. Y. Tseng,J. Phys. D: Appl. Phys.43, 295404 (2010).
10
D. Y. Lee, T. L. Tsai, and T. Y. Tseng,Appl. Phys. Lett.103, 032905 (2013).
11J. Lee, E. M. Bourim, W. Lee, J. Park, M. Jo, S. Jung, J. Shin, and H.
Hwang,Appl. Phys. Lett.97, 172105 (2010).
12
S. Yu, Y. Wu, Y. Chai, J. Provine, and H.-S. P. Wong, in 2011 International Symposium on VLSI Technology, Systems and Applications (VLSI-TSA), 2011, p. 1.
13G. H. Kim, J. H. Lee, J. Y. Seok, S. J. Song, J. H. Yoon, K. J. Yoon, M. H.
Lee, K. M. Kim, H. D. Lee, S. W. Ryu, T. J. Park, and C. S. Hwang,Appl. Phys. Lett.98, 262901 (2011).
14W. Shen, R. Dittmann, U. Breuer, and R. Waser,Appl. Phys. Lett.93,
222102 (2008).
15
Y. Lu, B. Chen, B. Gao, Z. Fang, Y. H. Fu, J. Q. Yang, L. F. Liu, X. Y. Liu, H. Y. Yu, and J. F. Kang, in 2012 IEEE International Reliability Physics Symposium (IRPS)(IEEE, 2012), pp. MY.4.1–MY.4.4.
16J. J. Yang, F. Miao, M. D. Pichett, D. A. A. Ohlberg, D. R. Stewart, C. N.
Lau, and R. S. Williams,Nanotechnology20, 215201 (2009).
17
B. Chen, J. F. Kang, B. Gao, Y. X. Deng, L. F. Liu, X. Y. Liu, Z. Fang, H. Y. Yu, X. P. Wang, G. Q. Lo, and D. L. Kwong,IEEE Electron Device Lett.34, 1292 (2013).
18
S. Yu, X. Guan, and H.-S. P. Wong, Tech. Dig. -Int. Electron Devices Meet.2012, 26.1.1.
19J. C. Dupin, D. Gonbeau, P. Vinatier, and A. Levasseur, Phys. Chem. Chem. Phys.2, 1319 (2000).
20
H. Y. Lee, P. S. Chen, T. Y. Wu, C. C. Wang, P. J. Tzeng, C. H. Lin, F. Chen, M. J. Tsai, and C. Lien,Appl. Phys. Lett.92, 142911 (2008).
21J. K. Lee, S. Jung, J. Park, S. W. Chung, J. S. Roh, S. J. Hong, I. H. Cho,
H. I. Kwon, C. H. Park, B. G. Park, and J. H. Lee,Appl. Phys. Lett.101, 103506 (2012).
22
C. S. Kang, H. J. Cho, Y. H. Kim, R. Choi, K. Onishi, A. Shahriar, and J. C. Lee,J. Vac. Sci. Technol. B21, 2026 (2003).
23S. Y. Wang, D. Y. Lee, T. Y. Tseng, and C. Y. Lin,Appl. Phys. Lett.
95, 112904 (2009).
24
H. Choi, J. Yi, S. Hwang, S. Lee, S. Song, S. Lee, J. Lee, D. Son, J. Park, S. J. Kim, J. Y. Kim, S. Lee, J. Moon, C. Kim, J. Park, M. Joo, J. S. Roh, S. Park, S. W. Chung, J. Rhee, and S. J. Hong, in 2011 3rd IEEE International Memory Workshop (IMW)(IEEE, 2011), pp. 1–4.
