APPLIED PHYSICS LETTERS 96, 141908 共2010兲

Influence of Sn Migration on phase transition in GeTe and Ge2Se3 thin films

Archana Devasia,1,a兲 Santosh Kurinec,1 Kristy A. Campbell,2 and Simone Raoux3
1
Electrical and Microelectronic Engineering, Rochester Institute of Technology, Rochester,
New York 14623, USA
2
Electrical and Computer Engineering, Boise State University, Boise, Idaho 83725, USA
3
IBM T. J. Watson Research Center, Yorktown Heights, New York 10598, USA
共Received 8 February 2010; accepted 19 March 2010; published online 9 April 2010兲
Phase transitions in GeTe/SnSe and Ge2Se3 / SnTe are investigated using time resolved x-ray
diffraction. GeTe exhibits a structural transition from rhombohedral to the cubic phase at 300 ° C,
which is ⬃100 ° C lower than that of pure GeTe. This is facilitated by incorporation of Sn from
SnSe. Sn migration is observed explicitly in Ge2Se3 / SnTe by separation of SnSe phase. Amorphous
Ge2Se3 is also found to crystallize at a lower temperature of 300 ° C resulting in orthorhombic GeSe
and monoclinic GeSe2. Thus, inclusion of a Sn containing layer may offer a means to tailor phase
transition in Ge-chalcogenide thin films for phase change memory applications. © 2010 American
Institute of Physics. 关doi:10.1063/1.3385781兴

Phase change memory 共PCM兲, which relies on a change is observed. An exceedingly intense peak occurs at 2␪
in resistance with a concomitant change in phase,1 is touted = 39.18° which corresponds to 共002兲 Ta 共PDF No. 891545兲10
to become the nonvolatile memory mainstay of the future.2 indicating that the bottom electrode material is highly
Ge2Sb2Te5 共GST兲 has been the most widely studied PCM textured.
material.2–4 Several alternate chalcogenide compositions are Figure 1 shows the variation in intensity of the XRD
being explored in an attempt to discover materials with su- peaks during heating to 230 ° C and 270 ° C, respectively. At
perior performance.5–7 Investigation of PCM devices em- 170 ° C, the onset of a crystalline phase is discerned at 2␪
ploying bilayers of Ge-chalcogenide and Sn-chalcogenide = 35.14° and 2␪ = 51.69°. These peaks match the rhombohe-
have revealed the possibility of obtaining multistate behav- dral low temperature phase of GeTe 共PDF No. 471079兲.10
ior, enhanced ability to withstand thermal cycling and use of The crystallization temperature Tx = 170 ° C is in agreement
lower voltages for achieving desired phase change response.8 with that found in literature.5,11
In the present study, phase transition in response to applica- A sample heated to 350 ° C is illustrated in Fig. 2共a兲. A
tion of heat, which involves examining structural changes noticeable shift in the position of the GeTe peak at 2␪
with temperature variation in bilayers of GeTe/SnSe is ex- = 35.14° is observed at 300 ° C. A similar change is detected
plored. Previously, sublimation issues have been encountered at 290 ° C for the peak at 2␪ = 51.69° 关Fig. 2共b兲兴. This is
in studies on residual stress analysis of Ge2Se3 / SnTe bilay- indicative of a structural phase transition. On heating, GeTe
ers performed by this investigator.9 In order to preclude such undergoes a transformation from rhombohedral symmetry
a possibility, in the present study, phase transition in this 共R3m兲 to rocksalt cubic symmetry 共Fm3m兲. The transition
stack is analyzed using Al2O3 as a capping layer.
Tantalum, 100 nm thick, which serves as a bottom elec-
trode, was dc sputter deposited and patterned on oxidized
silicon substrate. 100 nm of plasma enhanced chemical vapor
deposited 共PECVD兲 SixNy was employed as the interlevel
dielectric. Ge-chalcogenide 共GeTe or Ge2Se3, 99.999% pu-
rity兲 of 30 nm thickness was thermally evaporated, followed
by 50 nm thick Sn-chalcogenide 共SnSe or SnTe, 99.999%
purity兲. Thickness was verified using cross-sectional SEM.
Samples of Ge2Se3 / SnTe bilayers were capped with 10 nm
Al2O3 by atomic layer deposition at a substrate temperature
of 150 ° C.
GeTe/SnSe and Ge2Se3 / SnTe stacks were examined us-
ing time-resolved x-ray diffraction 共XRD兲 at the National
Synchrotron Light Source. Incident beam wavelength was
1.797 Å. Measurements were carried out while samples were
heated at the rate of 1 ° C / s, under flowing He, to different
temperatures. The detector monitored the intensity of the
peaks over a ⫾7.5° 2␪ range.
In the as-deposited state, the bottom GeTe layer is
amorphous and only orthorhombic SnSe 共PDF No. 481224兲10
FIG. 1. 共Color online兲 Peak intensity variation with temperature when
a兲
Electronic mail: axd1830@rit.edu. sample is heated to, 共a兲 230 ° C and 2␪ = 37°, 共b兲 270 ° C and 2␪ = 50°.

0003-6951/2010/96共14兲/141908/3/$30.00 96, 141908-1 © 2010 American Institute of Physics

Downloaded 30 Jun 2010 to 220.227.207.12. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp
141908-2 Devasia et al. Appl. Phys. Lett. 96, 141908 共2010兲

FIG. 3. Variation of d-spacing of 共200兲 SnTe with temperature.

共200兲 peak of SnTe is monitored 共Fig. 3兲. The interplanar

spacing d200 is determined via peak profile fitting using the
Pearson Type VII function.15 The slump in d200 from 200 ° C
and its revival beyond 350 ° C is an indication of phase tran-
sition and/or solid solution formation. To validate this infer-
ence, an Al2O3 capped Ge2Se3 / SnTe sample is heated to
350 ° C. Several additional peaks are ascertained in the pro-
FIG. 2. 共Color online兲 Peak intensity variation with temperature when
file in Fig. 4. Peaks at 2␪ = 36.12°, 2␪ ⬇ 43.9° and 2␪ ⬇ 58°
sample is heated to, 共a兲 350 ° C and 2␪ = 37°, 共b兲 340 ° C and 2␪ = 50°. In can all be indexed to SnSe 共PDF No. 530527兲10 signifying
both cases GeTe peaks exist at room temperature because the samples were the migration of Sn from top to bottom layer. The peaks at
previously heated above 170 ° C 2␪ = 29.65° and 2␪ = 32.14° correspond to GeSe2 共PDF No.
421104兲10 while the peak at 2␪ = 30.14° is attributed to GeSe
temperature, Tc, has been determined by various investiga- 共PDF No. 751802兲.10 Thus, the Ge2Se3 specimen crystallizes
tors to be in the 350– 432 ° C range.5,12,13 When Ge atoms into orthorhombic GeSe and monoclinic GeSe2. These find-
are partially replaced by atoms of elemental analogs such as ings are in agreement with the fact that Ge2Se3 is equivalent
Sn or Pb, Tc decreases.12,14 It has been found that in GeTe– to 共GeSe兲0.33共GeSe2兲0.67.16 The crystallization onset 共Tx兲 for
SnTe solid solution, increasing SnTe content inhibits the the Ge2Se3 under investigation is found to be between
thermal stability of the rhombohedral phase with Tc reducing 300 ° C and 350 ° C and is lower than the bulk Tx of
to room temperature at 68 mol % SnTe.12 450 ° C.16 It has been reported that when Sn atoms are incor-
The driving mechanism of the phase transition in GeTe porated in a Ge–Se system, the tetravalent Sn combines only
is the softening of the zone-center transverse optic phonon with Se. Sn is bigger than Ge and the Sn–Se bond has a
mode, which causes a relative shift in the anion and cation larger ionicity than the Ge–Se bond. This causes lack of flex-
sublattices.13 To form a rock-salt structured lattice, the ratio ibility of the bond angle between Se and Sn atoms and thus,
of the effective cation to anion radii should be greater than addition of Sn introduces a strain in the network structure,
the equivalent of the Hume-Rothery 0.41, which as argued which gets reflected as a decrease in Tx.17
by Bierly et al.,12 is stimulated by an increase in temperature,
which increases the effective size of Ge more rapidly than Te
due to the former being lighter than the latter. On substitution
of Sn, the temperature necessary to produce the required ra-
tio of radii would be lowered since Sn cation
共rSn2+ = 1.02 Å14兲 is larger than Ge cation 共rGe2+ = 0.65 Å14兲.
As will be discussed in the next section, migration of Sn
to the Ge2Se3 layer during heat treatments has been observed
in Ge2Se3 / SnTe bilayers. In addition, switching behavior in
some PCM devices employing stacks of Ge-chalcogenide
and Sn-chalcogenide has been attributed to Sn migration to
the Ge-chalcogenide layer.8 Thus, in the sample under inves-
tigation, the Sn can be supplied by the top SnSe layer of the
stack resulting in the formation of Ge1−xSnxTe solid solution.
The addition of Sn facilitates the lowering of the temperature
of the rhombohedral to cubic transition to 290– 300 ° C, and
a substantial shift in lattice constant. Thus, beyond 300 ° C,
the 共202兲 rhombohedral peak at 2␪ = 35.14° becomes the
共200兲 cubic peak at 2␪ = 33.93° and the 共220兲 rhombohedral
peak at 2␪ = 51.69° becomes the 共220兲 cubic peak at 2␪
= 49.49°.
For the Ge2Se3 / SnTe bilayer, the as-deposited sample
exhibits peaks only due to the top SnTe layer 共PDF No. FIG. 4. 共Color online兲 共a兲 Peak intensity variation with temperature for
461210兲,10 with the exception of the most intense 共002兲 Ta sample heated to 350 ° C and 2␪ = 30°, 共b兲 Zoomed in XRD profile of
peak. To examine the phase transition in this bilayer, the sample after ramp to show detail.
Downloaded 30 Jun 2010 to 220.227.207.12. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp
141908-3 Devasia et al.
FIG. 5. 共Color online兲 共a兲 Peak intensity variation with temperature for
sample heated to 450 ° C and 2␪ = 35°, 共b兲 XRD profile of sample after
ramp.
When a Ge2Se3 / SnTe sample is heated to 450 ° C, sev-
eral existing peaks vanish. Figure 5 shows that the GeSe and
GeSe2 phases have dissolved while the 共200兲 SnTe peak has
become visibly weaker and the 共220兲 and 共222兲 peaks have
disappeared. In a GeSe2 – SnTe system, the existence of at
least two intermediate phases has been reported as a result of
complex physicochemical interactions,18 and this may be re-
sponsible for the peak disappearances. On the other hand, the
GeSe2 – SnSe system shows neither the existence of solid so-
lutions nor any intermediate compounds.19
Two previously unobserved peaks are determined in Fig.
5. The peak at 2␪ = 36.35° which appears at ⬃320 ° C and
grows strongly at ⬃400 ° C corresponds to SnSe. The peak
at 2␪ = 37.44° at ⬃440 ° C may be the 共310兲 peak of TaTe2
共PDF No. 04–004–3558兲10 indicating possible diffusion of
Ta or Te or both. Interdiffusion between phase change and
electrode material has also been observed at GST/W and
GST/TiN interfaces.20,21 Incorporating thin films of Ta2O5 or
TaSiNx or SiTiNx between the GST and electrode interface
had been found to inhibit interdiffusion. In order to gain
further insight into the formation of TaTe2 in our films, TEM
and PEELS analysis will be performed in the future.
Downloaded 30 Jun 2010 to 220.227.207.12. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp
Appl. Phys. Lett. 96, 141908 共2010兲
Incorporation of a Sn layer can be used to tune the phase
transition characteristics of Ge-chalcogenide thin films. Sn
ions play a major role in reducing the crystallization tem-
perature 共Tc or Tx兲 of the Ge-chalcogenide layer, and this
property can be exploited for low power PCM applications.
However, the possibility of intermediate phase formation in
Ge2Se3 / SnTe and the stability of these phases when sub-
jected to repeated heat treatments or electrical switching re-
quires further investigation.
This research has been supported by NASA EPSCoR
under Grant No. NNX07AT60A. The work was carried
out in part at the National Synchrotron Light Source,
Brookhaven National Laboratory, which is supported by the
U.S. Department of Energy, under Contract No. DE-AC02-
98CH10886.
1
A. Pirovano, A. Lacaita, A. Benvenuti, F. Pellizzer, and R. Bez, IEEE
Trans. Electron Devices 51, 452 共2004兲.
2
H. Lv, P. Zhou, Y. Lin, T. Tang, B. Qiao, Y. Lai, J. Feng, B. Cai, and B.
Chen, Microelectron. J. 37, 982 共2006兲.
3
A. Lacaita, J. Solid State Electrochem. 50, 24 共2006兲.
4
S. Hudgens and B. Johnson, MRS Bull. 29, 829 共2004兲.
5
S. Raoux, B. Munoz, H.-Y. Cheng, and J. L. Jordan-Sweet, Appl. Phys.
Lett. 95, 143118 共2009兲.
6
S. Raoux and M. Wuttig, Phase Change Materials Science and Applica-
tions 共Springer, New York, 2009兲.
7
K. Chung, D. Wamwangi, M. Woda, M. Wuttig, and W. Bensch, J. Appl.
Phys. 103, 083523 共2008兲.
8
K. A. Campbell and C. M. Anderson, Microelectron. J. 38, 52 共2007兲.
9
A. Devasia, F. Bai, M. Davis, K. A. Campbell, S. Gupta, and S. Kurinec,
Thin Solid Films 517, 6516 共2009兲.
10
International Centre for Diffraction Data.
11
M. Chen, K. A. Rubin, and R. W. Barton, Appl. Phys. Lett. 49, 502
共1986兲.
12
J. N. Bierly, L. Muldawer, and O. Beckman, Acta Metall. 11, 447 共1963兲.
13
T. Chattopadhyay, J. X. Boucherle, and H. G. von Schnering, J. Phys. C
20, 1431 共1987兲.
14
N. Kh. Abrikosov, O. G. Karpinskii, L. E. Shelimova, and E. S. Avilov,
Izv. Akad. Nauk SSSR, Neorg. Mater. 16, 237 共1980兲 Inorg. Mater-
.共USSR兲 共English Transl.兲 16, 150 共1980兲.
15
S. K. Gupta, J. Appl. Crystallogr. 31, 474 共1998兲.
16
A. Feltz, Amorphous Inorganic Materials and Glasses 共VCH, New York,
1993兲.
17
I. Haruvi-Busnach, J. Dror, and N. Croitoru, J. Mater. Res. 5, 1215
共1990兲.
18
V. Vassilev, L. Aljihmani, and V. Paravanova, J. Therm Anal. Calorim. 76,
727 共2004兲.
19
V. Vassilev, K. Tomova, and V. Paravanova, J. Therm Anal. Calorim. 86,
199 共2006兲.
20
H. Cheng, Y. Chen, C. Lee, S. Wang, and T. Chin, Thin Solid Films 515,
990 共2006兲.
21
H. Cheng, Y. Chen, R. Chung, and T. Chin, Semicond. Sci. Technol. 21,
1196 共2006兲.