Phase Change Memories challenges: a material and process perspective
S. Maitrejean
1
, G. Ghezzi
1,2
, E. Gourvest
1,3,4
,
G. Betti Beneventi
1
,
A. Fantini
1
, N. Pashkov
1
, G. Navarro
1
, A. Roule
1
, F. Fillot
1
, P. Noé
1
, S. Lhostis
4
, O. Cueto
1
, C. Jahan
1
,
JF Nodin
1
, A. Persico
1
, M. Armand
1
, L. Dussault
3
, C. Vallé
3
, Ph. Michallon
1
, R. Morel
5
, A. Brenac
5
,
M. Audier
2
, JY Raty
6
, F. Hippert
2
, L. Perniola
1
,
V. Sousa
1
, B. de Salvo
1
1
CEA, LETI, MINATEC Campus, 17 rue des Martyrs, 38054 Grenoble, France
2
Laboratoire des Matériaux et du Génie Physique (Grenoble-INP, CNRS), Minatec, 3 Parvis L. Neel,
38016 Grenoble, France
3
LTM-Leti, Minatec Campus, F38054 Grenoble, France
4
STMicroelectronics, 38926 Crolles, France
5
INAC/SP2M and Université Joseph Fourier, CEA Grenoble, 17 rue des Martyrs, 38054 Grenoble, France
6
Physics Department B5, University of Liège, B-4000 Sart-Tilman, Belgium
ABSTRACT
Among all the new memories concepts, Phase Change
Memories (PCM) is one of the most promising. However,
various challenges remain. This paper reviews the materials
and processes required to face these challenges.
As an example, attention will be made on the effect of Phase
change material composition on stability of the amorphous
phase i.e. on the retention of the information. Additionally,
it is showed how specific processes such as CVD or ALD
can be developed in order to minimize the current required
to amorphize the phase change material i.e. to reset the
device. Finally, with the perspectives of the advanced
integration nodes, experimental results on the effect of
scaling on phase transformation are presented and discussed.
INTRODUCTION
As stated in various technological review papers [1-3], new
concepts are required to face the scaling challenges of
advanced memories. Phase Change Memories (PCM) rely
on the large resistivity variation that occurs during the
reversible amorphous-to-crystalline phase transition of
chalcogenide material such as GeTe or GeSbTe (GST). The
memory cell is a simple resistor as illustrated in the insert of
Fig. 1. Current pulses with various intensities are used to
switch the material from one phase to the other. Information
is read thanks to the resistance variation. During the past 10
years, PCM attracts large attention thanks to (i) their
maturity, i.e. industrial prototypes have been released, and
(ii) their amazing performances. They combine scalability,
high speed, high cyclability and non-volatility [4].
Nevertheless, numerous challenges remain. Indeed, a
switching PCRAM module requires:
- A high retention time at elevated temperature. This implies
a high thermal stability of the amorphous phase.
- A high transformation speed with a low pulse current in
order to minimize driving transistor foot print i.e. memory
cell footprint.
- A high cyclability of the cell. Thereby, reversible phase
transformation between amorphous and crystalline phase are
mandatory.
- A high scalability. The switching properties have to be
maintained at dimensions well below tenth of nm.
In the following parts, we will review the materials and
processes required facing these challenges.
Fig. 1: Main, Simulation of the effect of confinement on PCM cell
resistance switching (from crystal state to amorphous). Insert, schematic
cross section of plug and confined cell. [14]
PHASE CHANGE MATERIAL ENGINEERING
Two main axes have been explored in order to increase the
stability of the amorphous phase and consequently the
ability of the cell to retain information.
One can change or adjust the film composition: compared to
classical Ge
2
Sb
2
Te
5
, the binary compound GeTe
demonstrates higher retention time [5]. If you explore non-
stoichiometric composition, you can further increase
crystallization temperature and make functional devices [6].
However, during crystallization, you may have precipitation
that may compromise the device reliability during cycling
[7, 8]. It is worth noting that usually, a tradeoff between
speed and data retention needs to be made. Recently, the
demonstration of a “golden composition” of GST film was
announced. [9]
300 250 200 150 100
20 22 24 26 28 30 32 34
10
0
10
1
10
2
10
3
10
4
10
5
10
6
10
7
10
8
101 °C
E
a
= 3.0 eV
87 °C
E
a
= 2.9 eV
124 °C
E
a
= 2.26 eV
154 °C
E
a
= 3.1 eV
Fail time [s]
1/kT [eV
-1
]
GST225
GeTe
GeTeN 2%
GeTeN 4%
(b)
Temperature [°C]
Fig. 2: Effect of doping element on retention time. Arrehenius extrapolation
at 10 years for GST, GeTe and GeTeN devices: GeTeN2% extrapolated fail
temperature after 10 years is 154 °C. [10]
978-1-4673-1137-3/12/$31.00 ©2012 IEEE