ArticlePDF Available

Abstract and Figures

Van der Waals heterostructure superlattices of Sb2 Te1 and GeTe are strain engineered to promote switchable atomic disordering, which is confined to the GeTe layer. Careful control of the strain in the structures presents a new degree of freedom to design the properties of functional superlattice structures for data storage and photonics applications.
Content may be subject to copyright.
© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1
wileyonlinelibrary.com
CommuniCation
Phase-Change Memory Materials by Design: A Strain
Engineering Approach
Xilin Zhou,* Janne Kalikka, Xinglong Ji, Liangcai Wu, Zhitang Song,
and Robert E. Simpson*
Dr. X. Zhou, Dr. J. Kalikka, Prof. R. E. Simpson
ACTA Lab
Singapore University of Technology and Design
8 Somapah Road, 487372, Singapore
E-mail: xilin_zhou@sutd.edu.sg;
robert_simpson@sutd.edu.sg
X. Ji, Prof. L. Wu, Prof. Z. Song
State Key Laboratory of Functional Materials for Informatics
Shanghai Institute of Micro-system and Information Technology
Chinese Academy of Sciences
Shanghai 200050, China
DOI: 10.1002/adma.201505865
sub-layer atoms seems necessary to complete the transition,
the energy barrier for the vertical atomic motion of Ge domi-
nates the switching transition in iPCM,[18] and this has been
demonstrated experimentally and theoretically.[18–20] To increase
the efficiency of the iPCM switching process it is, therefore,
necessary to reduce the energy barrier for Ge atoms’ vertical
transition.
Until now the strategy for aiding the Ge atomic switching
in the superlattice took two paths: materials engineering and
strain engineering. The materials engineering approach was
based on creating vacancies in the Sb2Te3–GexTe 1x layer, which
was theorized to facilitate atomic movement of Ge.[21,22] For the
second path, the probability of atomic switching was increased
by biaxially straining the GeTe layer and consequently reducing
the activation energy of Ge atomic switching.[23]. Unexpectedly,
the atomic switching was only possible when the Sb2Te3–GeTe
iPCM structure was biaxially strained. Moreover, the switch was
confined to a 2D plane that was less than 5 Å thick.[23]
In this work, we substantially improve the transition speed,
switching energy, and cyclability of iPCM memory cells by
designing a tunable biaxially strained superlattice using Sb-rich
SbxTe1x (x > 0.4) layers. The strained iPCM cells, which are
composed of an Sb2Te1–GeTe superlattice heterostructure, dem-
onstrate fast and low power electrical switching capabilities.
Strain engineering prevailed as a key design knob in
sub-90 nm MOSFET devices and provided the scalability to
sustain Moore’s Law.[24,25] Straining a semiconductor crystal
via lattice mismatch or dielectric capping layer can modu-
late the carrier mobility and band structure in a favorable
way.[26] We hypothesized that straining the iPCM superlattice
structure is also a practical method to optimize its switching
energy because, first, the solid-state phase transition of the
iPCM prevents stress relaxation via melting, and second, the
layered heterostructure within the iPCM enables strain to be
applied biaxially. Moreover, the bond energy hierarchy in chal-
cogenides, such as GeTe,[27] provides a new paradigm to selec-
tively destabilize the GeTe layers of Sb2Te3–GeTe superlattice
using biaxial strain.[23]
It is especially interesting that there seems to be a general
trend for layered materials where the melting temperature of
the structure tends to decrease by increasing the in-plane lattice
parameter; see Figure 1a. This trend was discovered by sur-
veying the literature of layered chalcogenides covering the ele-
ments over groups IVB to VIA of the periodic table as listed
in Table S1 in the Supporting Information. These data points
reveal an underlying rule of the layered chalcogenides that
the larger the in-plane lattice parameter, the lower the melting
The potential discovery of a universal memory that exhibits fast
access speed, high-density storage, and nonvolatility has fueled
research into phase-change materials over the past decade.
Phase-change materials have been commercially applied to
optical and electrical data storage,[1–4] and now emerging appli-
cations that derive from the extreme optical contrast between
the amorphous and crystalline structural states include solid-
state displays,[5] all-photonic memories,[6] plasmonic-based
circuits,[7] optical modulators,[8] and neuromorphic com-
puting.[9–11] In prototypical “vertical” phase-change memory
(PCM) cell designs the active phase-change materials, which
tend to lie on the GeTe-Sb2Te3 tie-line, are sandwiched between
two conducting contacts that deliver current to thermally switch
the PCM cell via crystallization and melt-quenched amorphi-
zation on sub-nanosecond timescales.[12] The “mushroom”-
shaped amorphous region within the phase-change layer can be
visualized down to nanoscale dimensions.[11,13] The sub-optimal
atomic motions of the energy-sapping melt–quench amorphiza-
tion process dictate the high programming current of the reset
operation.[14] Thus fundamentally increasing the efficiency of
the reset process through materials engineering is the fore-
most challenge for wide spread commercialization of PCM
technology.
Recently highly efficient phase-change switching was dem-
onstrated in superlattice structures that exploit solid–solid
phase transformations rather than the typical melt–amorphiza-
tion–crystallization switching cycle. The superlattice consists
of rhombohedral Sb2Te3 and GeTe 2D crystals that are alter-
nately stacked along the out-of-plane growth axis, <111>.[15,16]
This “interfacial phase-change memory” (iPCM) superlattice
structure,[16] which is also known as the “topological-switching
random-access memory” (TRAM) structure,[17] is designed to
promote the vertical displacement of Ge atoms at the hetero-
structure interface. Even though a lateral motion of the GeTe
Adv. Mater. 2016,
DOI: 10.1002/adma.201505865
www.advmat.de
www.MaterialsViews.com
2wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
CommuniCation
temperature. This result thus supports our hypothesis that
biaxially straining the layered chalcogenides, especially the
tellurides, promotes interlayer atomic diffusion or structure
disordering, whereas compressing the a-axis increases the sta-
bility. Hence, for the SbxTe1x–GeTe superlattice heterostruc-
tures, an SbxTe1x layer with large in-plane a lattice parameter
Adv. Mater. 2016,
DOI: 10.1002/adma.201505865
www.advmat.de
www.MaterialsViews.com
Figure 1. a) Melting temperature of layered chalcogenides versus their in-plane lattice parameter, a. The dashed line shows an exponential fit to the
data points, which is intended to guide the readers’ eye. The melting temperatures and lattice parameters are listed in Supporting Information Table S1.
b) Schematic of the crystal structures. Left: The primitive rhombohedral GeTe cell with the (111) Miller plane indicated by shading. Middle: The hex-
agonal Sb2Te3 cell with the quintuple layers building block indicated. Right: The hexagonal Sb2Te1 cell with the indicated nonuple layers composed of
Sb2Te3 quintuple layers and Sb2 bilayers. The shaded planes in both Sb2Te3 and Sb2Te 1 structures suggest the (0 0 1) Miller planes of the hexagonal cell.
3
wileyonlinelibrary.com
© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
CommuniCation
will transfer in-plane tensile stress to the adjacent GeTe block
thus increasing its lattice parameter a; see Figure 1b.
The rhombohedral GeTe crystal has sub-sets of shorter and
longer bonds and exhibits a pronounced bonding energy hier-
archy. The layered structure has covalent-like bonded atoms
within GeTe layers and weaker layer-to-layer interactions. In a
hexagonal lattice configuration, with the c-axis along the <111>
direction of the primitive cell of GeTe, as shown in Figure 1b,
the in-plane lattice parameter aGe Te = 4.162 Å.[28]
The homologous series of crystals (Sb2)m–(Sb2Te3)n, with
m and n being integers, in the Sb–Te binary system, have the
trigonal long-period layered structures with cubic close-packed
stacking of Sb2 and Sb2Te3 blocks along the c-axis.[29,30] The in-
plane lattice parameter, a, of the structures is larger when the
Sb concentration in the Sb–Te compounds is higher.[30] There-
fore, we can increase the strain in the GeTe structure by incor-
porating Sb into the Sb2Te3 layer, which has a smaller lattice
constant, to create an SbxTe1x–GeTe superlattice.
The lattice parameter, a, of the Sb-rich Sb2Te1 crystal is larger
than that of GeTe crystal by 2.64%, and here we consider only
two representative compounds Sb2Te3 (m, n) = (0, 1) and Sb2Te1
(m, n) = (2, 1). The layered Sb2Te3 crystal consists of five-atom
layers stacked along the c-axis in the hexagonal lattice, called
quintuple layers. While the presence of nine-atom layers (non-
uple layers) composed of two Sb2 bilayers and Sb2Te3 quintuple
layers is identified in the hexagonal layered Sb2Te1 structure;
see Figure 1b. Interatomic covalent bonding within quintuple
layers and bilayers is strong, and the interblock coupling is a
much weaker van der Waals (vdW) type interaction,[31] which is
crucial for designing vdW heterostructures.[32]
The layered SbxTe1x 2D crystals play a key role as the scaf-
fold in building the vdW SbxTe1x–GeTe superlattice hetero-
structures. The c-axis oriented Sb2Te3 crystals were grown
by vdW heteroepitaxy.[33,34] The measured X-ray diffraction
(XRD) pattern of the Sb2Te3 crystal grown on Si (100) is
shown in Figure 2a. It exhibits strong preferential (0 0 l) ori-
entation. The (0 1 5) diffraction peak of Sb2Te3 at 28° (2
θ
), the
strongest diffraction peak of the randomly oriented Sb2Te3
powder, is not present, and only the peaks corresponding to
c-axis orientation, such as Sb2Te3 (0 0 3) and Sb2Te3 (0 0 9),
are observed. Rietveld refinement of the XRD pattern of the
layered Sb2Te3 crystal gives the lattice constants of aSb Te
23
=
4.26 Å and cSb Te
23
= 30.46 Å, indicating a fully relaxed
growth.[35] Crystal growth along the c-axis is generally driven
when the growth conditions are optimized to minimize the
structures’ total free energy.[36] Figure 2b presents the XRD
patterns of the Sb2Te3 films prepared at other sub-optimal
conditions, and the corresponding scanning electron micro-
scopy (SEM) images are available in the Supporting Infor-
mation (see Figure S1, Supporting Information). The inten-
sity of (0 0 l) peaks increased considerably with increasing
temperature, whereas the most prominent peaks of Sb2Te3
power diffraction, Sb2Te3 (0 1 5) and Sb2Te3 (0 1 10), became
progressively weaker. This suggests that the degree of c-axis
orientation of Sb2Te3 crystals is improved significantly under
optimized growth conditions. Figure 2c shows the SEM
image of the Sb2Te3 crystals prepared at 573 K, in which the
hexagonal crystals are observed with the (0 0 l) plane parallel
to the surface of the substrate.
Unlike Sb2Te3 crystals, highly (0 0 l) oriented Sb2Te1 crystals
could not be prepared directly on Si surfaces; see Figure S2 in
the Supporting Information. This is attributed to the lack of
weak Te–Te vdW interlayer bonds in the hexagonal Sb2Te1 unit
cell; see Figure 1b. According to the self-organized vdW epi-
taxial growth model of layered chalcogenides structures,[34] the
substrate must be prepared with an initial vdW gap between Te
layers for the crystal growth with strong out-of-plane orienta-
tion. Indeed, we found that it is possible to grow (0 0 l) oriented
Sb2Te1 crystal, if it is grown on a 10 nm thick Sb2Te3 buffer layer,
which presents a Te terminated surface, as shown in Figure 2a.
The c-axis oriented crystal was grown relaxed with the lattice
parameter of aSb2Te1 = 4.27 Å and cSb2Te1 = 17.64 Å.[37] Similarly,
the strongest powder diffraction peak of Sb2Te1 (1 0 3) is not
present, which confirms the strong preferred orientation of the
Sb2Te1 film. The atomic stacking period for both the layered
Sb2Te1 and Sb2Te3 crystals was found to be 5 and 9, respectively.
These values were calculated from the lattice spacing d(0 0 l)
of the highly oriented diffraction peaks, M and N,[38] as indi-
cated in Figure 2a.
The top-view SEM images of the Sb2Te1 crystals are given in
Figure 2c. Only triangular crystals were formed in the Sb2Te1
film with (0 0 l) facets parallel to the substrate surface, which
is in stark contrast to the hexagonal Sb2Te3 crystals. The mini-
mization of interface energy during the crystal growth accounts
for the triangular shape of the layered Sb2Te1 crystals since the
vdW interface between the quintuple-layer blocks of Sb2Te3 is
not present in the Sb2Te1 unit cell.[29,36]
Highly c-axis-oriented Sb2Te1–GeTe superlattice films were
grown by optimizing the growth conditions, as evident by the
XRD curves given in Figure 3a. The Sb2Te1 (1 0 3) and Sb2Te 1
(0 1 6) peaks, which are normally the most prominent dif-
fraction modes in the randomly oriented Sb2Te1 crystals,[39]
are suppressed in the out-of-plane direction of the superlat-
tices, indicating the layers parallel to the substrate plane. The
thickness of GeTe crystal layers in the superlattices was kept
constant at 1 nm, corresponding to the (GeTe)2 block and the
Ge-Te-Te-Ge layer sequence. The octahedral coordinated GeTe
layer is composed from the atomic layer sequence Ge-Te-Te-Ge
in the Sb2Te1–GeTe superlattice, which is required for electrical
switching.[16,20,40,41]
The thin layer of GeTe favors a growth of strained superlat-
tices since the misfit dislocation-induced film relaxation evolves
as film growth proceeds.[42] The epitaxially organized superlat-
tices are able to relax via the formation and displacement of
misfit dislocations in the crystals. A layer of the GeTe 2D crystal
that is thinner than the critical thickness for dislocation genera-
tion minimizes the nucleation and propagation of dislocations
in the vdW heterostructures,[43] thus effectively restraining the
relaxation process. Moreover, the XRD profiles of the Sb2Te1
GeTe superlattice films are similar to that of pure Sb2Te1 crys-
tals, and no diffraction peaks originating from GeTe crystal
are identified. The absence of diffraction peaks from the GeTe
layer was also found in the Sb2Te3–GeTe superlattice struc-
tures,[23,40,41] which suggested high-quality superlattice crys-
tals. Hence, in a similar manner to the atomic arrangement in
the Sb2Te3–GeTe superlattice model,[20,21] the Ge and Te atoms
are incorporated into the scaffold of the hexagonal Sb2Te1
structure. Therefore, the (GeTe)2 block in the Sb2Te1–GeTe
Adv. Mater. 2016,
DOI: 10.1002/adma.201505865
www.advmat.de
www.MaterialsViews.com
4wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
CommuniCation
Adv. Mater. 2016,
DOI: 10.1002/adma.201505865
www.advmat.de
www.MaterialsViews.com
Figure 2. a) XRD patterns of the highly (0 0 l) oriented Sb2Te3 and Sb2Te1 films deposited by sputtering at 573 K. The atomic stacking repetition of the
Sb2Te3 and Sb2Te1 crystals was calculated to be 5 and 9, respectively. The vdW Sb2Te3 buffer layer was required to prepare layered Sb2Te1 crystal along
the c-axis. b) XRD patterns of the Sb2Te3 films deposited at different substrate temperatures. The degree of c-axis orientation of the crystal was found
sensitive to the substrate temperature. c,d) SEM images of the Sb2Te3 and Sb2Te1 films with remarkable layered hexagonal and triangular crystals,
respectively. Scale bar: 100 nm.
5
wileyonlinelibrary.com
© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
CommuniCation
superlattice is bi-axially strained. In addition, an Sb2Te1 layer
thickness dependent shift of the (0 0 l) diffraction peaks is
observed, and a larger separation between the M and N dif-
fraction peaks is apparent for the thicker Sb2Te1 films. This
indicates that the atomic stacking period along c-axis of the
superlattice[38] is suppressed and the in-plane lattice param-
eters of the Sb2Te1–GeTe structures are increased for thick
Sb2Te1 layers. Therefore, the GeTe 2D crystal is stretched in the
plane of the layers, thus increasing the spacing of the Ge and
Te atoms along the a-axis. Density functional theory models of
bulk cubic GeTe show that the rate of change in Ge–Te bond
length with respect to biaxial strain is greater for the shorter
Ge–Te bonds (see Figure S3, Supporting Information). How-
ever, we expect that the longer Ge–Te bonds to be more sus-
ceptible to strain induced distortion triggered loss of order due
to the bonds’ “resonant” character and sensi-
tivity to p-orbital overlap.[27]
The dependence of Sb2Te1–GeTe hetero-
structure lattice parameters on the thickness
of the Sb2Te1 blocks was also confirmed
by Rietveld analysis (see Figure S4, Sup-
porting Information). The biaxial strain of
the GeTe layer is readily tuned by adjusting
the thickness of the Sb2Te1 layer within the
superlattice. The in-plane biaxial strain in
the vdW heterostructure was defined as
ε
= (aSL aGeTe )/aGeTe , where aSL denotes the
in-plane lattice constant of the superlattices.
The in-plane biaxial strain of GeTe is plotted
as a function of Sb2Te1 layer thickness in
Figure 3b. The measurements of the Sb2Te3
GeTe superlattices are also included for com-
parison. It is evident that the level of biaxial
strain in the Sb2Te1–GeTe superlattices is
significantly higher than that of Sb2Te3-based
superlattice structures, which is ascribed par-
tially to the larger in-plane lattice parameter
of the Sb2Te1 crystal, and partially to the weak
vdW interactions between adjacent quintuple
layers in the Sb2Te3 crystal that enable strain
relaxation. The in-plane biaxial strain of the
GeTe layers in the Sb2Te1–GeTe superlat-
tice is larger than 1.5%, a level that is well
above the critical strain value for Ge atomic
switching in the superlattice according to the
atomic switching strain map.[23] Hence, the
strained Sb2Te1–GeTe superlattice hetero-
structures with thick Sb2Te1 blocks promise
fast and low energy switching in the iPCM
cells.
The strain caused by the lattice mismatch
of Sb2Te1 and GeTe layers compresses Sb2Te1
and stretches GeTe in the in-plane direction.
This tends to destabilize GeTe as its bonds
are elongated making the atoms less tightly
bound to each other. On the other hand, the
Sb2Te1 layer is stabilized due to atoms having
less free space around them because of the
compression. This is analogous to the influ-
ence of pressure on the melting temperature of most bulk
materials; in general, the melting temperature tends to increase
with increasing hydrostatic pressure. It can also be viewed as
interface premelting at the interface between the Sb2Te1 and
GeTe blocks.[23,44]
By considering the bulk biaxial strain applied to GeTe from
the Sb2Te1 layer, the energy for disordering the structure is
reduced by 10.4 meV per atom; see Figure S4 in the Supporting
Information. Comparing with the 120 meV per atom energy dif-
ference between the bulk cubic GeTe and distortion amorphous
GeTe,[27] this strain is expected to lower the thermal energy
required for disordering. The disordering of the layers was
studied with a density functional molecular dynamics (DF/MD)
superlattice model with 4 nm thick Sb2Te1 layer and 1 nm thick
GeTe layer. The structure was heated at 2 K ps1 from 600 K,
Adv. Mater. 2016,
DOI: 10.1002/adma.201505865
www.advmat.de
www.MaterialsViews.com
Figure 3. a) XRD patterns of the highly (0 0 l) oriented [Sb2Te1–GeTe]n superlattices, n is the
number of the periodic structure. The XRD data of the layered Sb2Te1 crystal are included for
comparison. The lattice constants, a, listed above the curves were calculated by the Rietveld
refinement procedure. b) The in-plane biaxial strain of GeTe layer versus the SbxTe1–x layer thick-
ness. The strained levels in all strained superlattices meet the demand for atomic switching.
6wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
CommuniCation
and the crystallinity of the atoms was monitored throughout
the simulation. The average crystallinity of the layers is plotted
in Figure 4a, and it shows that the GeTe layer is disordered at
a significantly lower temperature than the Sb2Te1 layer. A pre-
melting temperature was defined as the temperature where the
exponential fit to the layer crystallinity crosses 80% crystallinity.
The premelting temperature of GeTe layer was 942 K while the
premelting temperature for Sb2Te1 layer was 1085 K. It is clear
that there is a wide temperature gap between the disordering
of GeTe and Sb2Te1 layers, allowing the GeTe layer to switch
without affecting the stability of Sb2Te1 layer. This difference in
disordering temperatures is also visible in the visualizations of
the simulation structure during melting shown in Figure 4b.
At 900 K both layers are crystalline, and the
appreciable distortions of GeTe layer start at
938 K. At 1020 K the GeTe layer is slightly
disordered, and at 1080 K it is considerably
disordered while the Sb2Te1 layer is still crys-
talline. At 1110 K both layers are disordered
and the crystallinity is close to 0. Hence,
the GeTe layer in the Sb2Te1–GeTe superlat-
tice disorders prior to the Sb2Te1 layer. This
indicates that with careful heat pulses we can
disorder the GeTe layers in the strained
superlattice while keeping the Sb2Te1 scaffold
intact. Although the softening of GeTe layer
in the strained iPCM makes Ge switching
more complicated than a simple 1D transi-
tion. We speculate that precise Ge atomic
motions, which have been suggested by
others,[18] are possible when the GeTe layer is
in a softened state at 1020 K; see Figure 4b.
The GeTe layer of the structure contains
relatively few atoms and this contributes
to the sudden fluctuations in the simulated
crystallinity. However, the GeTe layer crystal-
linity stays above 90% until 938 K. Whereas
the Sb2Te1 is thick enough to allow for partial
and more gradual disordering, which pro-
duces a smoother decrease in crystallinity.
To demonstrate the potential of strain
engineering in the iPCM cells, the resist-
ance–voltage (RV) switching characteristic
of the strained iPCM devices was measured,
and the results are given in Figure 5. A typ-
ical T-shaped device structure is employed
as illustrated in Supporting Information
Figure S5, which is compatible with the
deposition of the layered structure. Figure 5a
displays the electrical switching behavior of
the strained Sb2Te1–GeTe iPCM cells with
voltage pulse width of 200 ns, which is typi-
cally used to test non-superlattice Ge2Sb2Te5
based memory cells. For such long pulses,
the iPCM is heated above its melting tem-
perature and the superlattice layers inter-
diffuse, as observed in Figure 4b (1110 K).
Starting from the reset (high resistance) state
of the first reversible programming, the cell
resistance is abruptly switched to a low resistance state when
the pulse amplitude exceeded 1.0 V, marking a set transition.
As the voltage is increased and approaches the transition point,
the cell resistance increases abruptly, returning to the initial
resistance level, and a reset transition achieved. Repeating
the long 200 ns duration pulses causes the set programming
curves to iteratively shift to higher voltages. This is shown by
the arrow in Figure 5a, which is indicative of a programming
history dependent switching process of the device. The dra-
matic increase of the set programming energy with switching
events strongly indicates the evolution of interlayer disordering
and mixing of superlattice layers within the iPCM cells. This
implies that the layered structure of the iPCM cell is destroyed
Adv. Mater. 2016,
DOI: 10.1002/adma.201505865
www.advmat.de
www.MaterialsViews.com
Figure 4. a) Premelting of the layers of Sb2Te1–GeTe superlattice as a function of temperature.
The order parameter is calculated using both the first and second coordination shells for each
atom of interest. The red and blue dashed lines are 1 exp (x) fits to the Sb2Te1 and GeTe layer
crystallinities, respectively. The GeTe layers premelt at a temperature 40 K lower than the
Sb2Te1 blocks. b) Visualizations of the simulation structures at 900, 1020, 1080, and 1110 K.
The Sb2 bilayers and Te–Te vdW gap are clear at 900 K. At 1020 K the GeTe layer softens and at
1080 K the GeTe layer disorders but the Sb2Te1 layer remains intact. At 1110 K both layers are
disordered and interdiffusive.
7
wileyonlinelibrary.com
© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
CommuniCation
by significant Joule heating caused by the long duration voltage
pulses, leading to a typical crystalline–amorphous Ge2Sb2Te5
alloy-like switching mechanism.
In contrast, substantially shorter voltage pulses of 20 ns dura-
tion produced a highly repeatable, and reversible RV switching
characteristic in the strained Sb2Te1–GeTe iPCM cells, as shown
in Figure 5b. There is no iterative increase in programming
voltage, and electrical pulses of just 14 ns can switch the iPCM
devices reversibly between resistance states that are separated
by more than two orders of magnitude. Atomic switching at
the interfaces could be achieved in the strained iPCM cells
with sub-20 ns pulses because diffusive atomic transitions are
confined to the GeTe blocks within the Sb2Te1 scaffold of the
strained superlattices.[23] The iPCM structure is switched by
premelting the GeTe layer at temperatures substantially lower
than the disordering temperature of Sb2Te1. Indeed, Figure 5c
shows that the strained Sb2Te1–GeTe iPCM cells have a substan-
tially lower set voltage and a greater resistance contrast than
Adv. Mater. 2016,
DOI: 10.1002/adma.201505865
www.advmat.de
www.MaterialsViews.com
Figure 5. Electrical switching curves of the strained iPCM devices. a) RV characteristics of strained Sb2Te1–GeTe iPCM cells with a constant 200 ns
pulse. b) RV characteristics of strained Sb2Te1–GeTe iPCM cells under sub-20 ns pulses. c) RV characteristics of different memory cells with 20 ns
pulse. d) The set time as a function of set voltage of different memory cells. The dashed lines give the trend of the data points in an exponential fit.
A 48% lower set switching voltage and a five times faster switching speed are obtained in the strained iPCM cells with respect to the Ge2Sb2Te5-based
devices. e,f) Endurance-cycling measurements for the strained Sb2Te 1–GeTe iPCM cells with reset (3 V, 50 ns) and set (1.5 V, 50 ns) pulses (e), and
conventional mushroom-type Ge2Sb2Te5 PCM devices with reset (5.6 V, 300 ns) and set (3.6 V, 500 ns) pulses (f). Each curve in (a)–(c) is representa-
tive of more than five iPCM cells.
8wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
CommuniCation
Adv. Mater. 2016,
DOI: 10.1002/adma.201505865
www.advmat.de
www.MaterialsViews.com
similar devices using the Ge2Sb2Te5 alloy and the Sb2Te 3–GeTe
superlattice. The 20 ns duration pulse is found insufficient to
switch a fully conductive state in the devices based on crystal-
line–amorphous mechanism. The dashed line exponential fits
in Figure 5d reveal the excellent switching performance of the
strain engineered iPCM cells over the melting-quench domi-
nated Ge2Sb2Te5 -based devices.[45] The strained iPCM cells,
which enable interlayer atomic switching, only need 48% of the
set voltage yet show five times faster switching than Ge2Sb2Te5
alloy-based PCM devices; see Figure 5d.
The cyclability of the strained Sb2Te1–GeTe iPCM cells is sub-
stantially greater than that of Ge2Sb2Te5-based cells. The cyclic
switching endurance of the strained Sb2Te1–GeTe iPCM cells
was performed by alternating reset–set pulses of 50 ns dura-
tion. The set and reset resistance levels, which are shown in
Figure 5e, are well separated during the reversible switching for
more than 2 × 104 cycles. The resistance ratio of reset to set state
is more than one order of magnitude. The cyclability of identical
memory cell designs that employ the Ge2Sb2Te5 alloy is signifi-
cantly lower than the Sb2Te1–GeTe iPCM cells; see Figure 5f.
The Sb2Te1–GeTe superlattice devices required substantially
shorter voltage pulse times and lower pulse amplitudes than
the Ge2Sb2Te5 alloy-based cells. This is attributed to the in-plane
biaxial strain-induced interface premelting, which facilitates
the GeTe switching in the superlattice via localized disordering
while preserving the stability of the Sb2Te1 layer. The entropic
losses associated with the switching are therefore reduced,
allowing the observed lower power switching, and the increased
crystallographic texture within the high resistance state that
enables the faster recrystallization of the GeTe layers.[46] The
improved number of switching cycles of the Sb2Te1–GeTe
devices is also due to the confined premelting switching mecha-
nism, which minimizes the atomic migration and phase segre-
gation that can occur in conventional PCM devices.[16,47,48]
We expect that the cyclability of the strained iPCM cells can
be further improved by reducing the thermal energy that is
deposited into the memory cell.[3] Clearly, this could be achieved
by reducing the bottom electrode area and using shorter pro-
gramming pulse durations of 14 ns, which is the shortest pulse
that reversibly switched the strained Sb2Te1–GeTe iPCM cells.
To conclude, strain engineering is an effective method to
optimize the switching performance of iPCM memory devices.
By growing highly (0 0 l) textured stoichiometric Sb2Te1–GeTe
superlattices on Sb2Te3 surfaces, we have significantly reduced
the required pulse voltage and pulse durations for switching
Ge-Sb-Te-based memory devices.
The Sb2Te1–GeTe superlattices exhibit strong preferred (0 0 l)
orientation and the planar lattice parameters are strongly
dependent on the Sb2Te1 layer thicknesses. The GeTe blocks
within the hexagonal superlattice structures are subjected to
varying degrees of biaxial strain and this enabled us to design
strain-tuned superlattices.
By incorporating the strained Sb2Te1–GeTe superlattices into
prototype memory devices, the switching time was reduced
to just 14 ns. Moreover, the switching voltage was lowered by
48% relative to the sub-optimal Ge2Sb2Te5-based devices. The
improved device performance is due to the confined GeTe
premelting within the Sb2Te1 scaffold. We found that strained
2D GeTe crystal layers undergo an interfacial premelting-like
phase transition at a temperature lower than the disordering
temperature of the Sb2Te1 layer. Thus carefully controlling the
Joule heat delivered to the superlattice structure can disorder
the GeTe layer without disrupting the Sb2Te1 superlattice scaf-
fold, and since the Sb2Te1 does not undergo a phase transi-
tion, entropic losses are reduced, allowing the observed lower
switching voltage. The increased crystallographic order within
the amorphous state then allows for the observed high-speed
recrystallization of the GeTe layers. We want to clarify that
although our strained iPCM model does not need an external
electric field to disorder the GeTe layer, we suspect that the
electric field can induce favorable atomic motions[18] when the
GeTe is in the softened state at 1020 K.
We have demonstrated that strain engineering provides a
new route to create high-performance phase-change materials
based on layered 2D crystals of nontraditional phase-change
materials, such as Sb2Te1. This demonstration strongly sug-
gests that previously studied PCMs, which were once rejected
due to their low switching speed or high-energy consumption,
may still perform well in a strained superlattice structure and
should be revisited. The iPCM structure opens up new strain,
compositional, and structural design degrees of freedom to
tailor phase-change materials for specific applications in data
storage, active photonics, and logic computing.
This conclusion implies Sb-Te can be replaced by other 2D
scaffold materials that can provide the necessary strain to desta-
bilize the GeTe layer. The suitable replacements must have an
in-plane lattice parameter that is larger than GeTe, yet be stable
at the GeTe disordering temperature. By analyzing Figure 1a
and Table S1 in the Supporting Information, we see that Bi2Te3
and Bi2Se3 are therefore promising candidate scaffolds. Note
that both Bi2Te3 and Bi2Se3 are topological insulators and may
allow new memory concepts such as TRAM.[17]
Experimental Section
Deposition of Highly Oriented Films: Highly (0 0 l) oriented films of
100 nm thick were deposited on Si (100) substrate by sputtering using
alloy targets of Sb2Te3, Sb2Te 1, and GeTe. Sputter deposition is a low-
cost, industrially well-established, and scalable technology with respect
to molecular beam epitaxy[49,50] and other chemical strategies.[51–53] The
background pressure of the vacuum system was better than 2.5 × 107 Pa,
and the optimized substrate temperature was 573 K. After deposition,
the films were cooled down to room temperature naturally in the vacuum
system. In order to prepare thin films with strong preferred orientation,
the native silicon oxide on Si wafer was cleaned by argon plasma in
a negative self-bias mode, allowing the self-organized vdW epitaxial
growth.[33,34] To ensure the formation of highly (0 0 l) textured superlattice
films, a 10 nm Sb2Te3 vdW layer was grown on the substrate before the
superlattices deposition. The Sb2Te1–GeTe superlattice films were formed
by alternately stacking the ultra-thin Sb2Te1 and GeTe layers, in which the
thickness of Sb2Te1 layer was 1, 2, and 4 nm in different samples while
the GeTe block thickness kept constant (1 nm). The film compositions of
each layer were almost same as the sputter-target composition. The XRD
measurements were performed using Cu K
α
radiation (
λ
= 1.54 Å) in
θ
/2
θ
geometry. The diffraction data were collected over the 2
θ
range of 5–60°.
The morphologies of the films were analyzed by field-emission SEM.
Fabrication of iPCM Devices: The strained Sb2Te1–GeTe iPCM devices
were fabricated with a typical “mushroom” type (T-shape) structure. The
diameter of the W electrode of the devices is 190 nm. To ensure strong
preferred crystallographic orientation, a 10 nm thick Sb2Te3 vdW buffer
9
wileyonlinelibrary.com
© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
CommuniCation
[1] M. Wuttig, Nat. Mater. 2005, 4, 265.
[2] M. Wuttig, N. Yamada, Nat. Mater. 2007, 6, 824.
[3] G. W. Burr, M. J. Breitwisch, M. Franceschini, D. Garetto,
K. Gopalakrishnan, B. Jackson, B. Kurdi, C. Lam, L. A. Lastras,
A. Padilla, B. Rajendran, S. Raoux, R. S. Shenoy, J. Vac. Sci. Technol.
B 2010, 28, 223.
[4] H.-S. Wong, S. Raoux, S. Kim, J. Liang, J. P. Reifenberg,
B. Rajendran, M. Asheghi, K. E. Goodson, Proc. IEEE 2010, 98,
2201.
[5] P. Hosseini, C. D. Wright, H. Bhaskaran, Nature 2014, 511, 206.
[6] C. Ríos, M. Stegmaier, P. Hosseini, D. Wang, T. Scherer,
C. D. Wright, H. Bhaskaran, W. H. Pernice, Nat. Photonics 2015, 9,
725.
[7] M. Rude, R. E. Simpson, R. Quidant, V. Pruneri, J. Renger, ACS
Photonics 2015, 2, 669.
[8] L. Waldecker, T. A. Miller, M. Rude, R. Bertoni, J. Osmond,
V. Pruneri, R. E. Simpson, R. Ernstorfer, S. Wall, Nat. Mater. 2015,
14, 991.
[9] C. D. Wright, Y. Liu, K. I. Kohary, M. M. Aziz, R. J. Hicken, Adv.
Mater. 2011, 23, 3408.
[10] T. H. Lee, D. Loke, K.-J. Huang, W.-J. Wang, S. R. Elliott, Adv. Mater.
2014, 26, 7493.
[11] D. Kuzum, R. G. Jeyasingh, B. Lee, H.-S. P. Wong, Nano Lett. 2011,
12, 2179.
[12] D. Loke, T. Lee, W. Wang, L. Shi, R. Zhao, Y. Yeo, T. Chong, S. Elliott,
Science 2012, 336, 1566.
[13] A. Sebastian, M. Le Gallo, D. Krebs, Nat. Commun. 2014, 5, 4314.
[14] J. Tominaga, R. Simpson, P. Fons, A. Kolobov, in 2010 European
Symp. on Phase Change and Ovonic Science (E\PCOS 2010), E\PCOS
2010, p. 6, http://www.epcos.org/library/papers/pdf_2010/Oral/
C04-Tominaga.pdf.
[15] J. Tominaga, P. Fons, A. Kolobov, T. Shima, T. C. Chong, R. Zhao,
H. K. Lee, L. Shi, Jpn. J. Appl. Phys. 2008, 47, 5763.
[16] R. Simpson, P. Fons, A. Kolobov, T. Fukaya, M. Krbal, T. Yagi,
J. Tominaga, Nat. Nanotechnol. 2011, 6, 501.
[17] M. Tai, T. Ohyanagi, M. Kinoshita, T. Morikawa, K. Akita, S. Kato,
H. Shirakawa, M. Araidai, K. Shiraishi, N. Takaura, in 2014 Symp. on
VLSI Technology (VLSI-Technology): Dig. Tech. Pap., IEEE, Piscataway,
NJ, USA, 2014, pp. 1–2, DOI:10.1109/VLSIT.2014.6894436.
[18] X. Yu, J. Robertson, Sci. Rep. 2015, 5, 12612.
[19] J. Tominaga, A. V. Kolobov, P. J. Fons, X. Wang, Y. Saito, T. Nakano,
M. Hase, S. Murakami, J. Herfort, Y. Takagaki, Sci. Technol. Adv.
Mater. 2015, 16, 014402.
[20] T. Ohyanagi, N. Takaura, M. Tai, M. Kitamura, M. Kinoshita,
K. Akita, T. Morikawa, S. Kato, M. Araidai, K. Kamiya, T. Yamamoto,
K. Shiraishi, Ohyanagi, 2013 IEEE International Electron Devices
Meeting (IEDM), IEEE, Piscataway, NJ, USA, 2013, 30.5.1–30.5.4,
DOI: 10.1109/IEDM.2013.6724725.
[21] N. Takaura, T. Ohyanagi, M. Tai, M. Kinoshita, K. Akita,
T. Morikawa, H. Shirakawa, M. Araidai, K. Shiraishi, Y. Saito,
J. Tominaga, 2014 IEEE International Electron Devices Meeting
(IEDM), IEEE, Piscataway, NJ, USA 2014, pp.29.2.1–29.2.4,
DOI: 10.1109/IEDM.2014.7047132.
[22] M. Tai, T. Ohyanagi, M. Kinoshita, T. Morikawa, K. Akita, M. Takato,
H. Shirakawa, M. Araidai, K. Shiraishi, N. Takaura, 2015 Symp. VLSI
Technology (VLSI Technology), IEEE, Piscataway, NJ, USA, 2015,
pp. T96–T97, DOI: 10.1109/VLSIT.2015.7223707.
[23] J. Kalikka, X. Zhou, E. Dilcher, S. Wall, J. Li, R. E. Simpson, 2015,
unpublished.
[24] M. Yang, M. Ieong, L. Shi, K. Chan, V. Chan, A. Chou, E. Gusev,
K. Jenkins, D. Boyd, Y. Ninomiya, D. Pendleton, Y. Surpris,
D. Heenan, J. Ott, K. Guarini, C. Emic, M. Cobb, P. Mooney,
B. To, N. Rovedo, J. Benedict, R. Mo, H. Ng, in 2003 IEEE Int.
Electron Devices Meeting (IEDM), IEEE, Piscataway, NJ, USA,
2003, p. 453.
Adv. Mater. 2016,
DOI: 10.1002/adma.201505865
www.advmat.de
www.MaterialsViews.com
layer was deposited on the device base, which is immediately followed
by the growth of 39 nm thick [Sb2Te1–GeTe]13 superlattice films at 573 K.
The thickness of the repetitive blocks of Sb2Te1 and GeTe was 2 and
1 nm, respectively. The TiN (20 nm) and Al (300 nm) films were deposited
sequently as top electrodes. The TiN film also protected the GeTe layer
from oxidation in subsequent fabrication process since the structural
transition capability of GeTe material was found sensitive to oxygen
incorporation.[54] The PCM devices using Sb2Te3–GeTe superlattice
(39 nm) and Ge2Sb2Te5 alloy (39 nm) films were also fabricated for
comparison. All electrical measurements on the devices were done on a
custom-made characterization system with integrated probe stage. Device
programming was performed with pulsed voltages using an arbitrary
waveform generator (Tektronix AWG5002B) and a Keithley-2400 meter.
The cell resistance after applying voltage pulses was recorded at a
constant read voltage of 0.1 V.
Computational Methods: The premelting was studied using DF/MD
simulations with Vienna Ab initio Simulation Package.[55] The PAW
pseudopotentials from the associated library,[56] periodic boundary
conditions, single point in the Brillouin zone (k = 0), and plane-wave
basis with 220 eV cutoff were used for the simulations. Additionally, for
MD simulations, canonical ensemble (NVT), 3 fs time step, and velocity
rescaling for temperature control were used. The simulation system
consisted of 4 nm of Sb2Te1 and 1 nm of GeTe, totaling 207 atoms.
Crystallinity was measured with bond orientational (BO) order
parameter of Steinhardt et al.[57] which is generalizable to many
symmetries, and has been utilized before in crystallinity simulations.[58]
The BO order parameter was calculated by projecting the bond vectors
onto a basis of spherical harmonics Ylm, l = 4 was used for cubic lattice.
The order parameter Q1 is defined as:
Qi lNi Ni Yr
blm ij
j
Ni
k
Ni
ml
lb
() 4
21
1
()
1
() ()
l
1
()
0
() 2
π
=+
===−
(1)
where rij is the vector between atoms i and j, N (i) is the number of
neighbors for atom i, and Nb includes the atom (i) and its neighbors.
The
Qi()
4 value for ideal, vacancy-free rock-salt lattice is 0.764, and to
compensate for thermal fluctuations a “crystalline atom” is defined as
having
Qi()
4 0.6. This method results in a good agreement with the
atoms marked as crystalline by the order parameter and the atoms that
look crystalline in the visualization. In the Supporting Information we
have also included BO calculations for the first coordination cell which
increases the spatial resolution (see Figure S6, Supporting Information).
After relaxation system was heated from 600 K until melting at
2 K ps1. The crystallinity of atoms was calculated for each time step.
The simulation cell was divided into two parts based on the initial
coordinates of the GeTe and Sb2Te1 layers. The crystallinity of a layer was
then defined as the average crystalline fraction of atoms in the volume
initially occupied by that layer.
Supporting Information
Supporting Information is available from the Wiley Online Library or
from the author.
Acknowledgements
This work was supported by the SUTD-MIT International Design Centre
(IDC), Designer Chalcogenides Project (IDCSF1200108OH). X.J., L.W.,
and Z.S. acknowledge the support from the Strategic Priority Research
Program of Chinese Academy of Sciences (XDA09020402).
Received: November 26, 2015
Revised: December 27, 2015
Published online:
10 wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
CommuniCation
[25] R. Arghavani, Z. Yuan, N. Ingle, K. Jung, M. Seamons,
S. Venkataraman, V. Banthia, K. Lilja, P. Leon, G. Karunasiri, S. Yoon,
A. Mascarenhas, IEEE Trans. Electron Devices 2004, 51, 1740.
[26] M. L. Lee, E. A. Fitzgerald, M. T. Bulsara, M. T. Currie, A. Lochtefeld,
J. Appl. Phys. 2005, 97, 011101.
[27] A. Kolobov, M. Krbal, P. Fons, J. Tominaga, T. Uruga, Nat. Chem.
2011, 3, 311.
[28] L. Shelimova, O. Karpinskii, E. Avilov, M. Kretova, Inorg. Mater.
1993, 29, 1291.
[29] L. Shelimova, O. Karpinskii, M. Kretova, V. Kosyakov, V. Shestakov,
V. Zemskov, F. Kuznetsov, Inorg. Mater. 2000, 36, 768.
[30] K. Kifune, T. Fujita, T. Tachizawa, Y. Kubota, N. Yamada,
T. Matsunaga, Cryst. Res. Technol. 2013, 48, 1011.
[31] H. Zhang, C.-X. Liu, X.-L. Qi, X. Dai, Z. Fang, S.-C. Zhang, Nat.
Phys. 2009, 5, 438.
[32] A. Geim, I. Grigorieva, Nature 2013, 499, 419.
[33] A. Koma, Thin Solid Films 1992, 216, 72.
[34] Y. Saito, P. Fons, A. V. Kolobov, J. Tominaga, Phys. Status Solidi B
2015, 252, 2151.
[35] W.-S. Kim, J. Alloys Compd. 1997, 252, 166.
[36] C. Thompson, Annu. Rev. Mater. Sci. 2000, 30, 159.
[37] V. Agafonov, N. Rodier, R. Ceolin, R. Bellissent, C. Bergman,
J. Gaspard, Acta Crystallogr. C 1991, 47, 1141.
[38] K. Kifune, Y. Kubota, T. Matsunaga, N. Yamada, Acta Crystallogr. B
2005, 61, 492.
[39] X. Zhou, L. Wu, Z. Song, F. Rao, Y. Cheng, C. Peng, D. Yao, S. Song,
B. Liu, S. Feng, B. Chen, Appl. Phys. Lett. 2011, 99, 032105.
[40] T. Ohyanagi, M. Kitamura, M. Araidai, S. Kato, N. Takaura,
K. Shiraishi, Appl. Phys. Lett. 2014, 104, 252106.
[41] N. Takaura, T. Ohyanagi, M. Kitamura, M. Tai, M. Kinoshita,
K. Akita, T. Morikawa, S. Kato, M. Araidai, K. Kamiya, T. Yamamoto,
K. Shiraishi, in 2013 VLSI Technology Symp. (VLSI Technology), IEEE,
Piscataway, NJ, USA 2013, p. T130.
[42] C. V. Falub, H. von Känel, F. Isa, R. Bergamaschini, A. Marzegalli,
D. Chrastina, G. Isella, E. Müller, P. Niedermann, L. Miglio, Science
2012, 335, 1330.
[43] J. Narayan, B. Larson, J. Appl. Phys. 2003, 93, 278.
[44] Y. Mishin, M. Asta, J. Li, Acta Mater. 2010, 58, 1117.
[45] X. Zhou, M. Xia, F. Rao, L. Wu, X. Li, Z. Song, S. Feng, H. Sun, ACS
Appl. Mater. Interfaces 2014, 6, 14207.
[46] R. Simpson, P. Fons, A. Kolobov, M. Krbal, J. Tominaga, Appl. Phys.
Lett. 2012, 100, 021911.
[47] J.-B. Park, G.-S. Park, H.-S. Baik, J.-H. Lee, H. Jeong, K. Kim, J. Elec-
trochem. Soc. 2007, 154, H139.
[48] X. Zhou, L. Wu, Z. Song, F. Rao, B. Liu, D. Yao, W. Yin, J. Li, S. Feng,
B. Chen, Appl. Phys. Express 2009, 2, 091401.
[49] J. E. Boschker, J. Momand, V. Bragaglia, R. Wang, K. Perumal,
A. Giussani, B. J. Kooi, H. Riechert, R. Calarco, Nano Lett. 2014, 14,
3534.
[50] Y. Jiang, Y. Sun, M. Chen, Y. Wang, Z. Li, C. Song, K. He, L. Wang,
X. Chen, Q.-K. Xue, X. Ma, S. B. Zhang, Phys. Rev. Lett. 2012, 108,
066809.
[51] W. Shi, L. Zhou, S. Song, J. Yang, H. Zhang, Adv. Mater. 2008, 20,
1892.
[52] R. J. Mehta, Y. Zhang, C. Karthik, B. Singh, R. W. Siegel,
T. Borca-Tasciuc, G. Ramanath, Nat. Mater. 2012, 11, 233.
[53] S. Singh, S. Hong, W. Jeon, D. Lee, J.-Y. Hwang, S. Lim, G. D. Kwon,
D. Pribat, H. Shin, S. W. Kim, S. Baik, Chem. Mater. 2015, 27, 2315.
[54] X. Zhou, W. Dong, H. Zhang, R. E. Simpson, Sci. Rep. 2015, 5,
11150.
[55] G. Kresse, J. Hafner, Phys. Rev. B 1993, 47, 558.
[56] G. Kresse, D. Joubert, Phys. Rev. B 1999, 59, 1758.
[57] P. J. Steinhardt, D. R. Nelson, M. Ronchetti, Phys. Rev. B 1983, 28,
784.
[58] J. Kalikka, J. Akola, J. Larrucea, R. O. Jones, Phys. Rev. B 2012, 86,
144113.
Adv. Mater. 2016,
DOI: 10.1002/adma.201505865
www.advmat.de
www.MaterialsViews.com
... Subsequent studies, however, revealed a crystalline-to-amorphous phase-transition behavior in a similar SL structure, indicating the possible controversy in the switching mechanism. Different mechanisms such as enhanced thermal efficiency, [10][11][12][13][14] stacking fault [15] or strain-assisted transition, [16,17] and partial amorphization [18] have also been suggested in more recent works. Still, studies so far lack a detailed analysis of the crystalline state in the set state. ...
... The reset and set processes occur in the (111)-oriented FCC GST and amorphous states, respectively. Therefore, the low-current amorphization due to the high thermal efficiency of SL suggested by other groups [4,13,17] cannot occur in this case. ...
Article
Full-text available
This work demonstrates the atomic layer deposition (ALD) of Sb2 Te3 /GeTe superlattice (SL) film on planar and vertical sidewall areas containing TiN metal and SiO2 insulator. The peculiar chemical affinity of the ALD precursor to the substrate surface and the two-dimensional nature of the Sb2 Te3 enabled the growth of an in-situ crystallized SL film with a preferred orientation. The SL film showed a reduced reset current of ∼ 1/7 of the randomly oriented Ge2 Sb2 Te5 alloy. The reset switching was induced by the transition from the SL to the (111)-oriented face-centered-cubic (FCC) Ge2 Sb2 Te5 alloy and subsequent melt-quenching-free amorphization. The in-plane compressive stress, induced by the SL-to-FCC structural transition, enhanced the electromigration of Ge along the [111] direction of FCC structure, which enabled such a significant improvement. Set operation switched the amorphous to the (111)-oriented FCC structure. This article is protected by copyright. All rights reserved.
... Indeed, the GeTe layers embedded within the Sb 2 Te 3 or Sb 2 Te 1 scaffold exhibit premelting, which is similar to amorphization but limited to the GeTe layers. The Sb 2 Te 3 films remain crystalline and the layered structure is uncompromised [22,30]. It is, therefore, important to confirm whether the Ti 3.6 -(Sb 2 Te 3 ) 96.4 -GeTe superlattice also exhibits premelting. ...
Preprint
Full-text available
Phase change memory devices are typically reset by melt-quenching a material to radically lower its electrical conductance. The high power and concomitantly high current density required to reset phase change materials is the major issue that limits the access times of 3D phase change memory architectures. Phase change superlattices were developed to lower the reset energy by confining the phase transition to the interface between two different phase change materials. However, the high thermal conductivity of the superlattices means that heat is poorly confined within the phase change material, and most of the thermal energy is wasted to the surrounding materials. Here, we identified Ti as a useful dopant for substantially lowering the thermal conductivity of Sb2Te3-GeTe superlattices whilst also stabilising the layered structure from unwanted disordering. We demonstrate via laser heating that lowering the thermal conductivity by doping the Sb2Te3 layers with Ti halves the switching energy compared to superlattices that only use interfacial phase change transitions and strain engineering. The thermally optimized superlattice has (0 0 l) crystallographic orientation yet a thermal conductivity of just 0.25 W/m.K in the "on" (set) state. Prototype phase change memory devices that incorporate this Ti-doped superlattice switch faster and and at a substantially lower voltage than the undoped superlattice. During switching the Ti-doped Sb2Te3 layers remain stable within the superlattice and only the Ge atoms are active and undergo interfacial phase transitions. In conclusion, we show the potential of thermally optimised Sb2Te3-GeTe superlattices for a new generation of energy-efficient electrical and optical phase change memory.
... Another exotic characteristic of iPCM structure is that it can exhibit a topologically nontrivial band structure, and its properties can be engineered through strain controlling approach. 244,245 Then in 2019, Okabe et al. investigated the switching mechanism of GeTe−Sb 2 Te 3 iPCM and found that the thermal properties of iPCM account only for ∼13% reduction of RESET current change when compared with traditional GST alloys. 246 Another significant cause of the reduced RESET energy is the void migration process, in which the random voids distributed in iPCM move and concentrate around BEC, leading to smaller BEC area and RESET current. ...
Article
Phase transitions can occur in certain materials such as transition metal oxides (TMOs) and chalcogenides when there is a change in external conditions such as temperature and pressure. Along with phase transitions in these phase change materials (PCMs) come dramatic contrasts in various physical properties, which can be engineered to manipulate electrons, photons, polaritons, and phonons at the nanoscale, offering new opportunities for reconfigurable, active nanodevices. In this review, we particularly discuss phase-transition-enabled active nanotechnologies in nonvolatile electrical memory, tunable metamaterials, and metasurfaces for manipulation of both free-space photons and in-plane polaritons, and multifunctional emissivity control in the infrared (IR) spectrum. The fundamentals of PCMs are first introduced to explain the origins and principles of phase transitions. Thereafter, we discuss multiphysical nanodevices for electronic, photonic, and thermal management, attesting to the broad applications and exciting promises of PCMs. Emerging trends and valuable applications in all-optical neuromorphic devices, thermal data storage, and encryption are outlined in the end.
... Previous reports showed that when the GeTe strain is controlled in the superlattice structure, the bonding strength of the Ge-Te changes, and the crystallization energy changes accordingly. [47][48][49] However, in the trigonal GST structure, no structure exists in which the GeTe layer is strained between Sb 2 Te 3 layers. The SLs in the structure consisting solely of t-GST after heat treatment no longer had fast and low energy switching characteristics because of the t-GST stability. ...
Article
Interfacial phase-change memory (iPCM), comprising alternating layers of two chalcogenide-based phase-change materials—Sb2Te3 (ST) and GeTe (GT)—has demonstrated outstanding performance in resistive memories. However, its comprehensive understanding is controversial. Herein, the phase-change characteristic of iPCM is identified using atomic scale imaging, X-ray diffraction, and chemical analysis with first-principles density functional theory (DFT) calculations. By inducing laser pulsing, the ST/GT superlattice structure in the low-resistance state tends to reversibly convert into the modified metastable face-centered cubic (fcc) GeSbTe structure in the high-resistance state. This transition is driven by Ge atom rearrangement to pre-existing vacancy layers and ordered vacancy-layer formation. DFT atomistic modeling shows that the resistance difference of 10² orders between low- and high-resistance states is a direct consequence of the intercalation of Ge atoms into the vacancy layer. These results provide insights into iPCM phase-change mechanisms and phase-change random access memory design with low energy and high speed.
... As the number of grain boundaries increases, the crystal diffusion and slippage can be reduced. Hence, the residual stress in the bulk of films can be degraded [24,25]. Moreover, the increased grain boundaries provide a phonon and electron scattering center, and the decreased thermal and electrical conductivity will improve the energy efficiency of the Joule heating [26]. ...
Article
Full-text available
Phase change memory (PCM), due to the advantages in capacity and endurance, has the opportunity to become the next generation of general-purpose memory. However, operation speed and data retention are still bottlenecks for PCM development. The most direct way to solve this problem is to find a material with high speed and good thermal stability. In this paper, platinum doping is proposed to improve performance. The 10-year data retention temperature of the doped material is up to 104 °C; the device achieves an operation speed of 6 ns and more than 3 × 105 operation cycles. An excellent performance was derived from the reduced grain size (10 nm) and the smaller density change rate (4.76%), which are less than those of Ge2Sb2Te5 (GST) and Sb2Te3. Hence, platinum doping is an effective approach to improve the performance of PCM and provide both good thermal stability and high operation speed.
Article
Low power and high switching ratio are the development direction of the next generation of resistive random access memory (RRAM). Previous techniques could not increase the switching ratio while reducing the SET power. Here, we report a method to fabricate low-power and high-switching-ratio RRAM by adjusting the interstice radius (rg) between the van der Waals (vdW) layers of transitional-metal dichalcogenides (TMDs), which simultaneously increases the switching ratio and reduces the SET power. The SET voltage, SET power, switching ratio and endurance of the device are strongly correlated with rg. When the ratio of rg to the radius of the metal ions that form the conductive filaments (rg/rAg+) is near 1, the SET voltage and SET power vertically decrease while the switching ratio vertically rises with increasing rg/rAg+. For the fabricated Ag/[SnS2/poly(methyl methacrylate)]/Cu RRAM with an rg/rAg+ of 1.04, the SET voltage, SET power and switching ratio are 0.14 V, 10-10 W and 106, respectively. After 104 switching cycles and a 104 s retention time, the switching ratio of the device can still be stable above 106. Bending has no influence on the performance of the device when the bending radius is not <2 mm.
Article
Chemical mechanical planarization (CMP) is receiving a growing interest in the fabrication of phase change memory in order to achieve a highly scaled confine cell structure and global planarization within...
Article
Full-text available
The switchable optical and electrical properties of phase change materials (PCMs) are finding new applications beyond data storage in reconfigurable photonic devices. However, high power heat pulses are needed to melt-quench the material from crystalline to amorphous. This is especially true in silicon photonics, where the high thermal conductivity of the waveguide material makes heating the PCM energy inefficient. Here, we improve the energy efficiency of the laser-induced phase transitions by inserting a layer of two-dimensional (2D) material, either MoS2 or WS2, between the silica or silicon substrate and the PCM. The 2D material reduces the required laser power by at least 40% during the amorphization (RESET) process, depending on the substrate. Thermal simulations confirm that both MoS2 and WS2 2D layers act as a thermal barrier, which efficiently confines energy within the PCM layer. Remarkably, the thermal insulation effect of the 2D layer is equivalent to a ∼100 nm layer of SiO2. The high thermal boundary resistance induced by the van der Waals (vdW)-bonded layers limits the thermal diffusion through the layer interface. Hence, 2D materials with stable vdW interfaces can be used to improve the thermal efficiency of PCM-tuned Si photonic devices. Furthermore, our waveguide simulations show that the 2D layer does not affect the propagating mode in the Si waveguide; thus, this simple additional thin film produces a substantial energy efficiency improvement without degrading the optical performance of the waveguide. Our findings pave the way for energy-efficient laser-induced structural phase transitions in PCM-based reconfigurable photonic devices.
Article
For improving the three-dimensional structure of phase-change memory devices, Ovonic threshold switch devices have received renewed attention as selectors owing to a simple production process, good scalability, and excellent performance. These can replace transistors and diodes in the available technology. Here, we have studied the GeSe-based chemical mechanical polishing process. The different concentrations of hydrogen peroxide and lysine interacting with GeSe in chemical mechanical polishing were investigated. Material characterization was performed by scanning electron microscopy and atomic force microscopy. In addition, the reaction mechanism in the chemical mechanical polishing process was analyzed by electrochemical experiments and X-ray photoelectron spectroscopy.
Article
Full-text available
We study the switching process in chalcogenide superlattice (CSL) phase-change memory materials by describing the motion of an atomic layer between the low and high resistance states. Two models have been proposed by different groups based on high-resolution electron microscope images. Model 1 proposes a transition from Ferro to Inverted Petrov state. Model 2 proposes a switch between Petrov and Inverted Petrov states. For each case, we note that the main transition is actually a vertical displacement of a Ge layer through a Te layer, followed by a lateral motion of GeTe sublayer to the final, low energy structure. Through calculating energy barriers, the rate-determining step is the displacive transition.
Article
Full-text available
The extreme electro-optical contrast between crystalline and amorphous states in phase-change materials is routinely exploited in optical data storage and future applications include universal memories, flexible displays, reconfigurable optical circuits, and logic devices. Optical contrast is believed to arise owing to a change in crystallinity. Here we show that the connection between optical properties and structure can be broken. Using a combination of single-shot femtosecond electron diffraction and optical spectroscopy, we simultaneously follow the lattice dynamics and dielectric function in the phase-change material Ge2Sb2Te5 during an irreversible state transformation. The dielectric function changes by 30% within 100 fs owing to a rapid depletion of electrons from resonantly bonded states. This occurs without perturbing the crystallinity of the lattice, which heats with a 2-ps time constant. The optical changes are an order of magnitude larger than those achievable with silicon and present new routes to manipulate light on an ultrafast timescale without structural changes.
Article
Full-text available
Oxygen-doped germanium telluride phase change materials are proposed for high temperature applications. Up to 8 at.% oxygen is readily incorporated into GeTe, causing an increased crystallisation temperature and activation energy. The rhombohedral structure of the GeTe crystal is preserved in the oxygen doped films. For higher oxygen concentrations the material is found to phase separate into GeO2 and TeO2, which inhibits the technologically useful abrupt change in properties. Increasing the oxygen content in GeTe-O reduces the difference in film thickness and mass density between the amorphous and crystalline states. For oxygen concentrations between 5 and 6 at.%, the amorphous material and the crystalline material have the same density. Above 6 at.% O doping, crystallisation exhibits an anomalous density change, where the volume of the crystalline state is larger than that of the amorphous. The high thermal stability and zero-density change characteristic of Oxygen-incorporated GeTe, is recommended for efficient and low stress phase change memory devices that may operate at elevated temperatures.
Article
Full-text available
The ability to manipulate light propagation at the nanoscale is of vital importance for future integrated photonic circuits. In this work we exploit the high contrast in the optical properties of the phase change material Ge2Sb2Te5 to control the propagation of surface plasmon polaritons at a Au/SiO2 interface. Using grating couplers, normally incident light at λ = 1.55 μm is converted into propagating surface plasmons on a Au waveguide. Single laser pulses (λ = 975 nm) are applied to a thin film of Ge2Sb2Te5 placed on top of the device, which, upon transition from its amorphous to crystalline structural phase, dramatically increases both its refractive index and absorption coefficient, thus inhibiting propagation of the plasmonic mode. This effect is investigated for different interaction lengths between the phase change material and the Au waveguide, and contrast values in the transmitted intensity up to several tens of percents are demonstrated.Keywords: phase change materials; surface plasmon; nanophotonics; nonvolatile; chalcogenides
Conference Paper
A 50nm topological-switching random-access memory (TRAM) was fabricated for the first time. A high-quality Ge x Te 1−x /Sb 2 Te 3 superlattice film enabled set and reset voltages of TRAM to be less than 40% of those of PRAM. Statistical analysis of 16kb data showed the reset voltage to be less than 1.2 V, the lowest as a TRAM test chip.
Article
Implementing on-chip non-volatile photonic memories has been a long-term, yet elusive goal. Photonic data storage would dramatically improve performance in existing computing architectures by reducing the latencies associated with electrical memories and potentially eliminating optoelectronic conversions. Furthermore, multi-level photonic memories with random access would allow for leveraging even greater computational capability. However, photonic memories have thus far been volatile. Here, we demonstrate a robust, non-volatile, all-photonic memory based on phase-change materials. By using optical near-field effects, we realize bit storage of up to eight levels in a single device that readily switches between intermediate states. Our on-chip memory cells feature single-shot readout and switching energies as low as 13.4 pJ at speeds approaching 1 GHz. We show that individual memory elements can be addressed using a wavelength multiplexing scheme. Our multi-level, multi-bit devices provide a pathway towards eliminating the von Neumann bottleneck and portend a new paradigm in all-photonic memory and non-conventional computing.
Article
Highly oriented SbTe films were successfully deposited by RF-magnetron sputtering on both crystalline and amorphous substrates. A novel deposition mechanism and method are proposed based on van der Waals epitaxy. Due to the selective reactivity of the top surface atoms of the substrate with sputtered atoms, a Te monolayer is the first layer formed on the substrate, resulting in the subsequent layer-by-layer growth of the SbTe film independent of the crystallinity of the substrates. We believe that this method can be applied to the mass production of a wide range of various van der Waals solids, such as transition metal dichalcogenides and topological insulators for future electronics devices.