Effects of the electroforming polarity on bipolar resistive switching
characteristics of SrTiO3−?films
X. B. Yan,1,2Y. D. Xia,1,2H. N. Xu,1,2X. Gao,2,3H. T. Li,1,2R. Li,2,3J. Yin,1,2and
Z. G. Liu1,2,a?
1Department of Materials Science and Engineering, Nanjing University, Nanjing 210093,
People’s Republic of China
2National Laboratory of Solid State Microstructures, Nanjing University, Nanjing 210093,
People’s Republic of China
3Department of Physics, Nanjing University, Nanjing 210093, People’s Republic of China
?Received 23 June 2010; accepted 22 August 2010; published online 13 September 2010?
The effects of the electroforming polarity on the bipolar resistive switching characteristics in
SrTiO3−?thin films have been investigated. The conduction mechanisms of high resistance state and
low resistance state are Poole–Frenkel emission and tunneling, respectively. The temperature
dependences of the resistance at high and low resistance state are both semiconductorlike. The
impact of the polarity of the electroforming voltage on the resistive switching mechanism and the
distribution of defects was discussed. A simple model describing the combination of bulk and the
interface effect was proposed to explain the resistive switching in this material. © 2010 American
Institute of Physics. ?doi:10.1063/1.3488810?
Resistive switching random access memory ?RRAM?
has recently attracted considerable investigations for their
possible application in next generation non-volatile memory,
which employs the reversal switching between the high re-
sistance state ?HRS? and the low resistance state ?LRS? as the
logic “0” or “1.”1,2Metal-insulator-metal ?MIM? heterostruc-
ture is the general configuration for RRAM. Normally, an
electroforming process is needed to activate resistive switch-
ing in the insulator. This process can bring two main
impacts,3–8?i? by applying a bigger electrical field stress be-
tween two electrodes, high density of defects can be created
though electromigration and ?ii? the oxygen vacancies segre-
gated at defects can form conductive channels to reduce the
resistance of the cells.
Up to now, the underlying mechanism for the resistive
switching effect is still a controvertible problem. Bulk effect,
interface effect or a combination of both constitutes the main
research focuses, but in any case the impact of defects are
not negligible for the resistive switching,9,10and it might be
based on different physical mechanisms in different insulator
materials and structures ?including different electrodes?.
SrTiO3films ?band gap around 3.5 eV? are an n-type semi-
conductor due to the self-doping by the formation of oxygen
vacancies. In this study, we investigated the bipolar resistive
switching performance of SrTiO3−?films with bottom Pt
??:5.65 eV? electrode and the Au ??:5.1 eV? top electrode.
The electroforming process can result in a variety of struc-
tural defects such as oxygen vacancies.11The polarity of the
electroforming voltage impacts the distribution of defects in
An STO single crystal was used as the target. The STO
films were prepared on Pt/Ti/Si substrates by using pulsed
laser deposition technique ?PLD? under the pressure of 3
?10−4Pa at 350 °C, and then the as-grown films were an-
nealed at 750 °C for 30 min. The thickness of the polycrys-
talline STO film is ?80 nm. The Au probe with about
20 ?m diameter was used as the top electrode. For electrical
measurements, a Keithley 2400 source-measure unit was
used to test the current-voltage ?I-V? and endurance charac-
teristic of the cells. The bottom Pt-electrodes were grounded
and the voltage was swept on the top electrodes for all the
At the initial state the resistance of the memory cells is
very larger, therefore a positive voltage sweep of 12 V as the
electroforming voltage with a ?5 mA compliance current
was employed to activate the memory cell, and after that, the
unipolar resistive switching can be obtained. Through this
treatment the resistance of the memory cell was reduced by
typically several orders of magnitude. After the electroform-
ing operation, the cell was switched to HRS as the series of
procedure by Lee et al.3then the programming voltage
sweep 0→3→0→−3→0 V was set to the memory cell for
obtaining the bipolar resistive switching, as shown by the
numbered arrows in Fig. 1?a?. We can observe obviously
from the current-voltage ?I-V? curve in Fig. 1?a? that in posi-
tive segment there is no resistive hysteresis where the
memory cell still stays at the HRS. And the memory cell can
only be switched from the HRS to LRS by the applied volt-
age with negative polarity. The sequent hysteresis of I-V
curve exhibits bipolar resistive switching behavior as shown
in Fig. 1?b?. The positive and the negative threshold voltage
are ?2.0 V and ?−1.2 V, respectively. Figure 1?c? demon-
strates the endurance property of Au/STO/Pt memory cell. It
can be repeated for more than 2?103times without failure.
A negative electroforming voltage was also employed to ac-
tivate the cells. The remarkable I-V hysteresis curve appears
as shown in Fig. 1?d?, similar to that in TiO2cells activated
by the negative electroforming voltage.5The differences of
resistive switching characteristics with Fig. 1?b? are that the
memory cell must be turned on under a positive voltage
stress. The resistance of LRS after the negative electroform-
ing is slightly different from that of LRS after the positive
electroforming, which might be resulted from the different
a?Electronic mail: firstname.lastname@example.org.
APPLIED PHYSICS LETTERS 97, 112101 ?2010?
0003-6951/2010/97?11?/112101/3/$30.00© 2010 American Institute of Physics
tunneling barrier height due to different work function of top
and bottom electrode.
Figures 2?a? and 2?b? demonstrate the fitting results for
the I-V curves in Fig. 1?b? at the HRS and LRS to clarify the
conduction mechanism of the memory cell. As for the I-V
curve of the HRS of Fig. 1?b?, the relation ln?I/V? versus
V1/2can be fitted linearly in high voltage region as shown in
Fig. 2?b?, indicating that the conduction mechanism is
Poole–Frankel ?PF? emission. The PF mechanism is bulk ef-
fect dominated, which comes from a lowering of the Cou-
lomb potential barrier of a trap site due to electric field.
Therefore the conduction of the HRS in the high field region
is governed by electron hopping between the trap states
probably existing at grain boundaries or structural defects
induced by oxygen vacancies.11This is consistent with the
prediction of the model based on the formation of the con-
ducting channels in the HRS of the cells.12Moreover, elec-
tron tunneling cannot occur at HRS, because the thickness of
STO films is as large as 80 nm. The I-V curve of the LRS in
Fig. 1?b? can be fitted as I=A sinh?BU?, as shown in Fig.
2?b?. Here, A=0.016 and B=0.853 are the fitting constants.
This function relation is in agreement with the characteristic
of the electron tunneling model, similar to the behavior of
TiO2−xin memristor device.13
The temperature dependence of the resistance in HRS
and LRS was measured as shown in Fig. 3. It can be
observed that the resistances of HRS and LRS decreases as
the temperature increased, implying a semiconductor behav-
ior rather than a metallic one. So we can rule out the forma-
tion and rupture of conductive metallic nanofilaments ac-
counting for the resistive switching. This temperature
dependence behavior is different from that in single crystals
SrTiO3samples. This might come from different density of
oxygen vacancies resulted from different electroforming
A theoretical model has been proposed to interpret the
resistive switching effect in MIM structured RRAMs,15in
which the insulating medium was divided into three regions.
The middle region of the insulator medium is called “cen-
tral” domain. In our memory cells, the layered structure was
also divided into three domains. The Schottky barriers are
thought to be formed at both top and bottom interfaces due to
the higher work function of the electrodes. However, because
the oxygen vacancies migrate to the bottom electrode after
the action of the positive electroforming voltage and pile up
on the bottom electrode, and enough oxygen vacancies accu-
mulated at “bottom” domain will make the Schottky barrier
at bottom interface eliminated, as reported in the study of
TiO2,4the bottom interface between the bottom electrode
and SrTiO3−?film became Ohmic contact after the action of
the positive electroforming voltage, The changes in the re-
sistance contributed by “bottom” domain under different po-
larity of bias is therefore negligible, as described in Ref. 7.
Moreover, the symmetric I-V curves in HRS and LRS indi-
FIG. 1. ?Color online? The I-V characteristic of the cell
with SrTiO3−?films activated by the positive electro-
forming voltage ?a? the first voltage sweep operation,
?b? the sequent voltage sweep operation, ?c? the endur-
ance of the memory cell. ?d? The I-V characteristic of
the memory with SrTiO3−?films activated by the nega-
tive electroforming voltage. The inset is the equivalent
circuit of memory cell with SrTiO3−?films activated by
the negative electroforming voltage.
FIG. 2. ?Color online? ?a? ln?I/V? as a function of V1/2
at the HRS in high field region of Fig. 1?b?. ?b? The I-V
curve of the LRS of Fig. 1?b? can be fitted as I
=A sinh?BU? function. Here, A=0.016 and B=0.853.
112101-2Yan et al.Appl. Phys. Lett. 97, 112101 ?2010?
cate that the Schottky barrier at top interface reduced due to
the creation of high density defects near the interface be-
tween the top electrode and SrTiO3−?film, as described for
Sr2TiO4films.10As a result, in HRS the main resistance
comes from the bulk resistance of the “central” domain, and
the conduction mechanism is bulk effect PF emission. The
current jump in negative segment of Fig. 1?a? can be inter-
preted that a negative voltage sweep applied to the top elec-
trode drive positive charged vacancies in the SrTiO3−?films
to move from the bottom electrode toward the top electrode.
This movement will increase the local concentration of oxy-
gen vacancies segregate near dislocations or grain bound-
aries in the central domain, and finally conductive channel?s?
will be formed. The memory cell was turned to the LRS.
High density oxygen vacancies drift closer to the top elec-
trode under negative voltage. The segregated oxygen vacan-
cies make the Schottky barrier narrower and lower, as shown
in Fig. 1?b?. As a result, the electrons can easily tunnel
through it, as schematically depicted in Fig. 4?a?.16There-
fore, the resistive behavior of the cell was dominated by the
tunneling of electrons through Schottky barrier between the
top Au electrode and the SrTiO3−?film. Conversely When
the polarity of the applied voltage sweep is reversed, the
oxygen vacancies can be pushed away gradually from the
interface, as schematically depicted in Fig. 4?b?. The oxygen
vacancies segregating near dislocations or grain boundaries
in the center bulk part ?central domain? will move back to
bottom electrode, leading to the breakdown of the conductive
channel?s?. As described above, after the action of a negative
voltage electroforming the top interface can be thought to be
Ohmic contact and the Schottky barrier between bottom
electrode and the SrTiO3−?films also reduced as depicted in
above situation under the positive electroforming voltage.
Therefore, the switching processes after different polarity of
electroforming voltage correspond with each other inversely
as shown in Figs. 1?b? and 1?d?. This resistive switching
characteristic provides the evidence that the distribution of
the defects is impacted by different polarity of the electro-
Resistance–capacitor ?RC? parallel equivalent circuit is
contributed by trap assisted carrier transport known as
Frankel–Poole process.17According to the analysis of the
resistive switching mechanism for our memory cells, the
equivalent circuit of a RRAM cell activated by the positive
electroforming voltage can be further simplified as the circuit
shown in Fig. 4?c?. Three circuit elements correspond to
“top,” “central,” and “bottom” domain in cells, respectively.
The middle RC parallel circuit represents the current trans-
port process with PF trapping effect. And under this situation
the Schottky diode in the top interface dominates electronic
transport in LRS. The equivalent circuit for the negative volt-
age electroforming also can demonstrated by the inset of the
Fig. 1?d? correspondingly.
This work was financially supported by the State Key
Program for Basic Research of China under Grant No.
2007CB935401, 863 project of China with Grant No.
2009ZX02023-5-4?, National Natural Science Foundation of
China under Grant No. 10804048.
1R. Waser and M. Aono, Nature Mater. 6, 833 ?2007?.
2A. Sawa, Mater. Today 11, 28 ?2008?.
3M. H. Lee, K. M. Kim, G. H. Kim, J. Y. Seok, S. J. Song, J. H. Yoon, and
C. S. Hwang, Appl. Phys. Lett. 96, 152909 ?2010?.
4J. R. Jameson, Y. Fukuzumi, Z. Wang, P. Griffin, K. Tsunoda, G. I. Meijer,
and Y. Nishi, Appl. Phys. Lett. 91, 112101 ?2007?.
5C. Nauenheim, C. Kuegeler, A. Ruediger, and R. Waser, Appl. Phys. Lett.
96, 122902 ?2010?.
6J. Joshua Yang, F. Miao, M. D. Pickett, D. A. A. Ohlberg, D. R. Stewart,
C. N. Lau, and R. S. Williams, Nanotechnology 20, 215201 ?2009?.
7A. Shkabko, M. H. Aguirre, I. Marozau, T. Lippert, and A. Weidenkaff,
Appl. Phys. Lett. 95, 152109 ?2009?.
8S. F. Karg, G. I. Meijer, J. G. Bednorz, C. T. Rettner, A. G. Schrott, E. A.
Joseph, C. H. Lam, M. Janousch, U. Staub, F. La Mattina, S. F. Alvarado,
D. Widmer, R. Stutz, U. Drechsler, and D. Caimi, IBM J. Res. Dev. 52,
9M. Hamaguchi, K. Aoyama, S. Asanuma, Y. Uesu, and T. Katsufuji, Appl.
Phys. Lett. 88, 142508 ?2006?.
10K. Shibuya, R. Dittmann, S. Mi, and R. Waser, Adv. Mater. 22, 411
11K. Szot, W. Speier, G. Bihlmayer, and R. Waser, Nature Mater. 5, 312
12C. C. Lin, B. C. Tu, C. C. Lin, and C. H. Lin, IEEE Electron Device Lett.
27, 9 ?2006?.
13J. J. Yang, M. D. Pickett, X. Li, D. A. A. Ohlberg, D. R. Stewart, and R.
S. Williams, Nat. Nanotechnol. 3, 429 ?2008?.
14P. Calvani, M. Capizzi, F. Donato, S. Lupi, P. Maselli, and D. Peschiaroli,
Phys. Rev. B 47, 8917 ?1993?.
15M. J. Rozenberg, I. H. Inoue, and M. J. Sánchez, Phys. Rev. Lett. 92,
16V. V. Zhirnov and R. K. Cavin, Nat. Nanotechnol. 3, 377 ?2008?.
17S. T. Hsua, T. Li, and N. Awaya, J. Appl. Phys. 101, 024517 ?2007?.
FIG. 3. ?Color online? The temperature dependence of the resistance in ?a?
HRS and ?b? LRS.
FIG. 4. ?Color online? The positively charged oxygen vacancies can pile up
at the interface to reduce the width of the barrier resulting in the tunneling of
electrons ?a? or push away from the interface to increase the width of the
barrier ?b?. ?c? The equivalent circuit of memory cell with SrTiO3−?films
activated by the positive electroforming voltage.
112101-3 Yan et al.Appl. Phys. Lett. 97, 112101 ?2010?