Embed Size (px)
Citation preview
www.elsevier.com/locate/apsusc
Applied Surface Science 254 (2008) 2281–2284
Au doped Sb3Te phase-change material for C-RAM device
Feng Wang a,b,*, Ting Zhang b, Chun-liang Liu a, Zhi-tang Song b,Liang-cai Wu b, Bo Liu b, Song-lin Feng b, Bomy Chen c
a Key Laboratory of Physical Electronics and Devices of Ministry of Education, Xi’an Jiaotong University, Xi’an, 710049 ShaanXi, Chinab Laboratory of Nanotechnology, Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, 200050 Shanghai, China
c Silicon Storage Technology, Inc., 1171 Sonora Court, Sunnyvale, CA 94086, USA
Received 5 September 2007; accepted 5 September 2007
Available online 11 September 2007
Abstract
Au doped Sb3Te phase-change films have been investigated by means of in situ temperature-dependent resistance measurement. Crystallization
temperature of 2 at.% Au doped Sb3Te has been enhanced to 161 8C, which leads to a better data retention. The physical stability of the film has
been improved evidently after adding Au as well. Resistance contrast has been improved to 1.1 � 104, one order of magnitude higher than that of
pure Sb3Te. X-ray diffraction patterns indicate the polycrystalline Au–SbTe series have hexagonal structure, similar with pure Sb3Te alloy, when
Au doping dose is less than 9 at.%.
# 2007 Elsevier B.V. All rights reserved.
PACS : 64.70.Kb; 61.43.Dq; 73.61.Jc
Keywords: Phase-change; Crystallization temperature; Data retention; Au doped Sb3Te
1. Introduction
In recent years, chalcogenide random access memory (C-
RAM) has been developed to be one of the most promising
candidates for the next generation nonvolatile memories due to
its many advantages. The advantages of C-RAM include
nonvolatility, high density, ability for scaling down and
compatibility with complementary metal-oxide semiconductor
(CMOS) process compared with other new memory technol-
ogies [1–4].
At present, high speed and good stability are desired
qualities in C-RAM development. For stability, many
researches focused on improving phase-change material [5–
7]. The essential is to optimize the material’s crystallization
properties. There are two important aspects of the optimization,
one is thermal stability and the other is physical stability.
Preferable thermal stability and physical stability would make
archival life longer and the phase-change film more reliable.
* Corresponding author. Tel.: +86 021 62511070x8408;
fax: +86 021 62134404.
E-mail address: [email protected] (F. Wang).
0169-4332/$ – see front matter # 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.apsusc.2007.09.013
However, improving stability may deteriorate other material
properties such as resistance contrast or crystallization speed
[5]. Suitable phase-change material should have good stability
and other advantages at the same time.
Eutectic Sb–Te material is widely used for C-RAM device
and optical storage disk, and it has been proved to be suitable in
high-speed applications. The most superior property of Sb–Te
alloys compared with other materials is the fast crystallization
speed due to its growth dominated crystallization mechanism
which would lead to crystallization speed scales inversely with
the size of contact area [8–10]. It has been found that the speed
of crystallization increases with the Sb/Te ratio, but the thermal
stability scales inversely with it [8,11]. In our former study,
Sb3Te has preferable performance compared with other Sb–Te
alloys. Hence, Sb3Te is discussed in this work.
Although eutectic Sb–Te alloys have the above-mentioned
superior properties, it has some obvious deficiencies, such as a
relatively poor thermal stability. Thus, our investigation is
mainly focused on the improvement of Sb–Te material
performance combining high thermal stability with good
physical stability and large resistance contrast by adding
foreign element Au. Furthermore, the influence of Au content
will be discussed to give the best composition.
Fig. 2. Measured d(Log R)/dT as a function of temperature at a heating rate of
15 8C/min, where (A) represents pure Sb3Te, (B) 2 at.% Au doped Sb3Te, (C)
4 at.% Au doped Sb3Te and (D) 9 at.% Au doped Sb3Te.
F. Wang et al. / Applied Surface Science 254 (2008) 2281–22842282
2. Experiments
The Au doped Sb3Te films have been deposited by co-
sputtering single element Sb, Te and Au targets on SiO2/Si
(1 0 0) substrates. The size of sample is about 3 cm � 3 cm.
Different power was applied on Au target in order to achieve
different composition. In this experiment, the fixed Sb/Te ratio
was obtained by applying the specific DC power applied on Sb
and Te targets. The measured ratio is changeless with the value
of 3:1 according to energy dispersive spectroscopy (EDS)
measurement. The background pressure in the sputtering
process was below 2 � 10�4 Pa, and the sputtering Ar pressure
was 0.27 Pa. The thickness of the film is about 220 nm
according to scanning electron microscope (SEM) cross-
section observation. The films were annealed in N2 atmosphere
at 250 8C for 1 min and 300 8C for 2 min in order to carry out
X-ray diffraction (XRD) measurement and SEM surface
observation, respectively. XRD was employed to characterize
the structure of the film. The XRD patterns were taken in the 2u
range of 20–708 using a Cu target with a scanning step of 0.028,and the data acquisition time in each step is 0.3 s. In situ
temperature-dependent resistance measurement has been
carried out in a vacuum chamber, inside which the temperature
is regulated by a refrigerator [6]. In the refrigerator, high purity
nitrogen is employed for refrigeration by Joule-Thomson
effect, and a resistance wire is for Joule heating. The
dependence of electrical resistance on temperature has been
measured in discrete model with step of 2 K.
3. Results and discussion
Fig. 1 shows the XRD patterns of pure and various Au doped
Sb3Te after annealed at 250 8C for 1 min. When Au content is
less than 4 at.%, it could be found that the crystalline structure
of Au–SbTe series material is similar to pure Sb3Te. It indicates
that Au doping does not change the type of lattice structure. All
the materials are with hexagonal structure, and it accords with
the reported research [12]. The peaks of Au–Sb compound
would appear in the material with high Au content.
Fig. 1. X-ray diffraction patterns of Au–SbTe films on Si (1 0 0) substrate,
annealed at 250 8C for 1 min, where (A) represents pure Sb3Te, (B) 2 at.% Au
doped Sb3Te, (C) 4 at.% Au doped Sb3Te and (D) 9 at.% Au doped Sb3Te.
An important effect of Au content is to increase thermal
stability. Thermal stability is a basic property of phase-change
material, and it will result in better data retention (long-term
archival stability) and less thermal crosstalk in the C-RAM
devices. In order to increase data retention of the phase-change
material, a relatively high crystallization temperature (Tc) is
desired [13]. In our research, Tc is obtained from in situ
resistance measurement. The crystallization temperature is
determined by the minimum in the derivative (d(Log R)/dT)
[14]. Fig. 2 shows analyzed curves of d(Log R)/dT as a function
of T, where R represents measured resistance and T
temperature. Table 1 shows the crystallization temperature of
Sb3Te and Au–SbTe series materials. For 2 and 4 at.% Au
doped Sb3Te films, the crystallization temperature is 161 and
148 8C, respectively, when heating rate is fixed at 15 8C/min.
Nevertheless Tc of pure Sb3Te is only 137 8C which is too low
for practical applications. No obvious crystallization has been
observed when Au concentration exceeds 9 at.%.
The data retention of C-RAM is generally determined from
the thermal stability of phase-change material. Therefore, Au–
SbTe materials promise better data retention due to the higher
Tc which offers better stability against spontaneous recrys-
tallization at room temperature. Obtaining characteristic of data
retention in this work is to evaluate the failure time (tf) at a
certain annealing temperature [6,15,16]. Here tf is defined as
the duration when the resistance of the material drops to 10%
the value of amorphous state in annealing process. Resistance
of films as a function of time at various temperatures is shown
in Fig. 3. It is obvious that the Au–SbTe films have better data
retention compared with Sb3Te since longer time needed to
complete phase transition at the same temperature. Fig. 4 shows
failure time of amorphous 2 at.% Au doped Sb3Te and pure
Table 1
Crystallization temperature of various Au–SbTe films with different Au content
Composition Crystallization temperature (8C)
Sb3Te 137
2 at.% Au doped Sb3Te 161
4 at.% Au doped Sb3Te 148
Fig. 3. Time-dependent electrical resistance at specific temperature: (a) Sb3Te
at specific temperature from 135 to 150 8C. (b) 2 at.% Au doped Sb3Te at
specific temperature from 135 to 155 8C.
Fig. 4. Failure time as a function of temperature.
Fig. 5. SEM image of film surface, annealed at 300 8C for 2 min. (a) Image of
Sb3Te film. (b) Image of Au–SbTe film.
F. Wang et al. / Applied Surface Science 254 (2008) 2281–2284 2283
Sb3Te. The failure time is extrapolated to 80 8C by fitting the
data to an Arrhenius equation [8]. Although there would be
some errors since it is far away from the measured range, the
fitted curve indicates that doping Au can increase the failure
time by several orders of magnitude at 80 8C. tf of Au–SbTe at
80 8C is above 2.7 � 1011 s, four orders of magnitude longer
than that of pure Sb3Te of which is approximately 4.3 � 107 s.
The reason of choosing 10% criteria is mainly based on the
consideration that it will be very difficult to distinguish between
low state and high state if the resistance drops to less than 10%
of high state value. The result seems smaller than reported one
because the different criteria of failure time. For example, tf was
defined as the duration for complete crystallization in reported
research [15].
Physical stability of film is a parameter which the device
reliability is related to. Many factors such as volatilization or
stress would result in crack of film. It would deteriorate the
device reliability if the phase-change material film cracks.
Fig. 5 shows the surfaces of annealed samples observed by
SEM. Both of the films have been annealed at 300 8C for 2 min
in N2 atmosphere. For Au–SbTe film, the film is with perfect
surface, whereas many flaws turn up on the surface of Sb3Te
film. This may be explained by release of stress inside the film
during the annealing process, for there would be evident change
in density of phase-change material during phase-change
Fig. 6. Temperature-dependent electrical resistance of Sb3Te and Au–SbTe
films: (a) pure Sb3Te and 2 at.% Au doped Sb3Te films. (b) 4 at.% Au doped
Sb3Te and 9 at.% Au doped Sb3Te films.
F. Wang et al. / Applied Surface Science 254 (2008) 2281–22842284
process [17]. Since doping Au could increase the physical
stability of Sb3Te film during phase-change process, Au doped
Sb3Te would offer the potential for better reliability.
The ratio of amorphous state resistance (Rhigh) and
crystalline state resistance (Rlow) is referred to as resistance
contrast. In practical applications, the resistance contrast must
be large enough to distinguish the resistance of the material
within memory chips correctly and easily. For 2 at.% Au doped
Sb3Te, the ratio reaches to 1.1 � 104 with Rhigh 4.6 � 105 V
and Rlow 41 V, an order of magnitude higher than pure Sb3Te
with ratio of 6.2 � 103. Fig. 6 shows the electrical resistance of
Au–SbTe film versus temperature measured in vacuum with
heating rate of 15 8C/min. Before crystallization, the resistance
decreases steadily when the temperature rises. It is obvious that
the resistance decreases sharply near crystallization tempera-
ture. The resistance becomes not sensitive to temperature after
crystallization.
Although doping Au into Sb3Te would improve the
performance, too high dose of dopant would make the
performance just go to the opposite. Considering Tc shown
in Fig. 2 and resistance contrast shown in Fig. 6, it is obvious
that 2 at.% Au doped Sb3Te has best performance among Au–
SbTe series, whereas the performance of 9 at.% Au doped
Sb3Te becomes very poor. In addition, too much Au element
would lead to very low resistance. It would go against
identifying the states from high resistance state and low
resistance state if the resistance of phase-change film is too low,
considering the driver circuit of C-RAM chip is�1 kV. So, the
limited content of Au dopant should be 4 at.% in practical
applications.
4. Conclusions
The properties of Au doped Sb–Te material have been
investigated in detail. Good performance can be achieved from
Au–SbTe series which combined good data retention with high
physical stability and large resistance contrast. For 2 at.% Au
doped Sb3Te, better performance has been achieved compared
with pure Sb3Te, for Tc increases to 161 8C and the resistance
contrast increases to 1.1 � 104. Failure time of 2 at.% Au doped
Sb3Te is above 2.7 � 1011 s at 80 8C, several orders longer than
that of pure one. Physical stability of film has been improved
significantly as well. All these indicate that the Au–SbTe would
be a suitable phase-change material for practical C-RAM
applications.
Acknowledgements
This work is supported by 973 Program (2007CB935400,
2006CB302700), National High Technology Development
Program of China (2006AA03Z360), Science and Technology
Council of Shanghai (06QA14060, 06XD14025, 0652nm003,
06DZ22017).
References
[1] M. Wuttig, C. Steimer, Appl. Phys. A 87 (2007) 411.
[2] G. Muller, T. Happ, M. Kund, G.Y. Lee, N. Nagel, R. Sezi, IEEE P. Int.
Electron. Devices Meet. (2004) 567.
[3] A. Pirovano, A. Lacaita, A. Benvenuti, F. Pellizzer, S. Hudgens, R. Bez,
IEEE P. Int. Electron. Devices Meet. (2003) 699.
[4] S.G. Alberici, R. Zonca, B. Pashmakov, Appl. Surf. Sci. 231–232 (2004)
821.
[5] K. Kao, H. Cheng, C. Jong, C. Lan, T. Chin, IEEE Trans. Mag. 43 (2007)
930.
[6] T. Zhang, Z.T. Song, F. Wang, B. Liu, S.L. Feng, B. Chen, Jpn. J. Appl.
Phys. 46 (2007) L602.
[7] T. Zhang, Z.T. Song, F. Rao, G.M. Feng, B. Liu, S.L. Feng, B. Chen, Jpn. J.
Appl. Phys. 46 (2007) L247.
[8] M. Lankhorst, L. Pieterson, M. Schijndel, B. Jacobs, J. Rijpers, Jpn. J.
Appl. Phys. 42 (2003) 863.
[9] E.R. Meinders, M. Lankhorst, Jpn. J. Appl. Phys. 42 (2003) 809.
[10] O. Noritake, A. Sumio, N. Naomasa, et al. Jpn. J. Appl. Phys. 41 (2002)
1695.
[11] M. Lankhorst, B. Ketelaars, R. Wolters, Nat. Mater. 4 (2005) 347.
[12] C. Sun, J. Lee, M. Youm, Y. Kim, Jpn. J. Appl. Phys. 45 (2006) 9157.
[13] A. Pirovano, A. Redaelli, F. Pellizzer, F. Ottogalli, M. Tosi, D. Ielmini,
A.L. Lacaita, R. Bez, IEEE Trans. Dev. Mater. Reliab. 4 (3) (2004)
422.
[14] I. Friedrich, V. Weidenhof, W. Njoroge, P. Franz, M. Wuttig, J. Appl. Phys.
87 (2000) 4130.
[15] L. Pieterson, M. Schijndel, J. Rijpers, Appl. Phys. Lett. 83 (2003) 1373.
[16] T. Zhang, Z.T. Song, B. Liu, S.L. Feng, B. Chen, Solid-State Electron. 51
(2007) 950.
[17] H. Lyeo, D. Cahill, B. Lee, J. Abelson, Appl. Phys. Lett. 89 (2006)
151904.
