08001.0279ecst

ECS Transactions, 80 (1) 279-281 (2017)
10.1149/08001.0279ecst ©The Electrochemical Society
The Charge Trap Density Evolution in Wake-up and Fatigue Modes of FRAM
Damir R. Islamova,b, O. M. Orlovc, V. A. Gritsenkoa,b,d, and G. Ja. Krasnikovc
a
Rzhanov Institute of Semiconductor Physics SB RAS,
Novosibirsk 630090, Russian Federation
b
Novosibirsk State University, Novosibirsk 630090, Russian Federation
c
JSC Molecular Electronics Research Institute, Zelenograd 124460, Russian Federation
d
Novosibirsk State Technical University, Novosibirsk 630073, Russian Federation
We study the transport properties in different modes of FRAM
elements based on ferroelectric Hf0.5Zr0.5O2 thin films. The leakage
currents in Hf0.5Zr0.5O2 are described by phonon-assisted tunneling
between traps. Comparison experimental data with results of the
simulations allows us to extract the evaluation of the charge trap
density during endurance. A hypothesis about role of oxygen
vacancies in fatigue is discussed.
Introduction
Hafnia (HfO2) was considered as a paraelectric material until observation of ferroelectric
effect in thin films of doped HfO2 after high-temperature annealing (~1000 °C) (1, 2). Of
particular interest is the fact that ferroelectricity was also demonstrated in thin of the solid
solution Hf0.5Zr0.5O2 which requires annealing at significantly lower temperatures (3–6).
Ferroelectricity in these materials is associated with the ability to stabilize
noncentrosymmetric orthorhombic phase Pbc21 (4). It should be noted that hafnia-based
materials have many advantages over conventional ferroelectric regarding compatibility
with the technological processes used in microelectronics, and they have already
demonstrated their ability to provide a very high density of elements. Considering the
advantages of ferroelectric random-access memory (FRAM) as non-volatile, high-speed
performance, high number of switching cycles, the discovery of ferroelectric effect in
these materials gave an impetus for the development of the universal memory concept
which may lead to a significant breakthrough in the development of memory devices (7).
The unsolved problems in the way of development of FRAM-based universal
memory are the reason for Wake-up and Fatigue modes of FRAM elements. One of the
possible reasons for these effects is the presence of defects in Hf0.5Zr0.5O2 thin films. The
purpose of the present work is to study the evolution of charge trap density in
ferroelectric Hf0.5Zr0.5O2 after set/reset cycling (endurance). Purpose of the present work
is to identify the physical phenomena which cause FRAM degradation using the charge
transport studies of ferroelectric Hf0.5Zr0.5O2.
Experiment details
Transport
measurements
were
performed
for
TiN/Hf0.5Zr0.5O2/TiN
metal/insulator/metal (MIM) structures. Test structure were fabricated with atomic layer
deposition (ALD) technique. The 10-nm-thick TiN layer was deposited on oxidized Si
(100) substrate. Then the 10-nm-thick Hf0.5Zr0.5O2 films we deposited at 240 °C from
279
Downloaded on 2017-10-25 to IP 80.82.77.83 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
ECS Transactions, 80 (1) 279-281 (2017)
Tetrakis(EthylMethylAmino)Hafnium (TEMAHf), Tetrakis(EthylMethylAmino)Zirconium (TEMAZr), and H2O, as the Hf-precursor, Zr-precursor, and oxygen source,
respectively. The TEMAHf and TEMAZr precursors were mixed in single-cocktail
balloon. Laser ellipsometry and Rutherford backscattering spectroscopy confirmed the
thickness and stoichiometry of as deposited Hf0.5Zr0.5O2 films. A part of samples was
annealed at 400 °C in N2 environment during 30 sec (rapid thermal annealing, RTA). The
crystalline structures of as deposited and annealed films were examined by symmetrical
X-Ray diffraction (XRD) using ARL X'TRA tool (Thermo Scientific) utilizing CuKα
radiation. The TiN top electrodes (thickness of 10 nm) were deposited using the ALD
technique. For transport measurements, round (area is 7.854×103 µm2) electrodes were
formed by the photolithography process. The presence of ferroelectric properties of
Hf0.5Zr0.5O2 films is confirmed by observing the characteristic hysteresis on the
polarization-voltage (P-V) plate for TiN/Hf0.5Zr0.5O2/TiN structures.
The P-V and currents-voltage (I-V) dependences were measured by the standard
PUND technique at room temperature. The first “Positive” and the third “Negative”
impulses aligned dipoles in the ferroelectric films (set and reset, respectively) and they
were used to measure polarization, while, during the second “Up” and the forth “Down”
impulses, the leakage currents were measured. The leakage currents were extracted from
the current responses by removing displacement currents.
Results and discussions
The P-V measurements show that after 10 set/reset cycles, residual polarization 2Pr
rises from 19 µC/cm2 to 20.6 µC/cm2 (wake-up mode). After 103 set/reset cycles, 2Pr
began to decrease (Fatigue mode) and, after 106 cycles, it reached the value of 13 µC/cm2.
Note that, at a ~104 cycle, the residual polarization exhibits a local maximum.
The leakage current depends on the voltage exponentially and can be described by the
phonon-assisted tunneling between traps (8)–(10):
eg
,
s2
æ 2s 2m *W
æ Wopt - Wt ö
2p !Wt
t
÷÷ expç expçç ç
!
2kT ø
kT (Wopt - Wt )
è
è
J=
g=
m* s 2
ö
÷ sinhæç eFs ö÷
÷
è 2kT ø
ø
[1]
with the thermal (Wt = 1.25 eV) and optical (Wopt = 2.5 eV) trap energies related to the
oxygen vacancy in Hf0.5Zr0.5O2 (11)–(13). Here J is the current density, e is the
elementary charge, s is the mean distance between traps, γ is the probability rate of
charge carrier tunneling between traps, ħ is the Planck constant, m* is the effective mass,
k is the Boltzmann constant, T is the local temperatures, F is the local electric field in the
dielectric medium.
The slope of the initial I-V curves on the semi-log plot corresponds to the trap density
of N = s-3 = 1.3×1020 cm-3. After 104 set/reset cycles, the current increases spasmodically
and the slope of the log(I)-V curves corresponds to N = 1.5×1020 cm-3. During further
cycles the current rises at a particular voltage on the test structure. After 106 cycles, the
leakage current is described by the phonon-assisted tunneling between the traps with trap
density N = 1.8×1020 cm-3.
Despite that ab initio calculations predict amphoteric nature of the charge traps (13),
the capacitance-voltage measurements on trapped charge accumulation demonstrate that
280
Downloaded on 2017-10-25 to IP 80.82.77.83 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
ECS Transactions, 80 (1) 279-281 (2017)
mostly holes are involving in transport processes (11). One can assume that holes,
involved on the transport process, contribute to the formation of loosening Zr-O anf Hf-O
bonds, which leads to forming more and more defects, namely oxygen vacancies. On the
one hand, the trapped charge can pin dipoles and “switch them off” from the ferroelectric
response (14). In the other hand, an oxygen vacancy might be formed by removing and
oxygen atom, which is form a dipole by its movement in the lattice. This phenomenon
leads to decreasing ferroelectricity of the films, i.e. degradation of the Hf0.5Zr0.5O2 film in
the FRAM element (fatigue mode).
Summary
In conclusion, this study demonstrates that the fatigue of a Hf0.5Zr0.5O2-based FRAM
element is caused by spasmodically formation of new defects, namely oxygen vacancies,
in the active medium. The trap density can be extracted from the curve slope in the
log(I)-V plate depicted for leakage currents, extracted from the standard PUND
measurements. Models of the defects formation and fatigue were proposed.
Acknowledgments
The work was supported by the Russian Science Foundation, grant #14-19-00192.
References
1.
2.
3.
4.
5.
6.
T.S. Böscke, J. Müller, D. Bräuhaus et al., Appl. Phys. Lett. 99, 102903 (2011).
S. Mueller, J. Mueller, A. Singh et al., Adv. Function. Mater. 22, 2412 (2012).
J. Müller, T.S. Böscke, D. Bräuhaus et al., Appl. Phys. Lett. 99, 112901 (2011).
J. Müller, T.S. Böscke, U. Schröder et al., Nano Lett. 12, 4318 (2012).
M.H. Park, H.J. Kim, Y.J. Kim et al., Appl. Phys. Lett. 102, 112914 (2013).
A. Chernikova, M. Kozodaev, A. Markeev et al., Microelectron. Eng. 147, 15
(2015).
7. M.H. Park, Y.H. Lee, H.J. Kim et al., Adv. Mater. 27, 1811 (2015).
8. K.A. Nasyrov. V.A. Gritsenko, JETP 112, 1026 (2011).
9. K.A. Nasyrov, V.A. Gritsenko, J. Appl. Phys. 109, 093705 (2011).
10. D.R. Islamov, V.A. Gritsenko, A. Chin, Optoelectron., Instrum. Data Process. 53,
184 (2017).
11. D.R. Islamov, T.V. Perevalov, V.A. Gritsenko et al., Appl. Phys. Lett. 106,
102906 (2015).
12. D.R. Islamov, A.G. Chernikova, M.G. Kozodaev et al., JETP Lett. 102, 544 (2015).
13. D.R. Islamov, A.G. Chernikova, M.G. Kozodaev et al., ECS Trans. 75, 123 (2017).
14. M. Pešić, F.P.G. Fengler, L. Larcher et al., Adv. Funct. Mater. 26, 4601 (2016).
281
Downloaded on 2017-10-25 to IP 80.82.77.83 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).