ArticlePDF Available

Abstract

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.
Low Energy Switching of Phase Change Materials Using a 2D
Thermal Boundary Layer
Jing Ning,*
,
Yunzheng Wang,
Ting Yu Teo, Chung-Che Huang, Ioannis Zeimpekis, Katrina Morgan,
Siew Lang Teo, Daniel W. Hewak, Michel Bosman, and Robert E. Simpson*
Cite This: ACS Appl. Mater. Interfaces 2022, 14, 41225−41234
Read Online
ACCESS Metrics & More Article Recommendations *
Supporting Information
ABSTRACT: 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 inecient. Here, we improve the energy eciency 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 MoS2and WS2
2D layers act as a thermal barrier, which eciently confines energy within the PCM
layer. Remarkably, the thermal insulation eect 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 diusion through the layer interface. Hence,
2D materials with stable vdW interfaces can be used to improve the thermal eciency of PCM-tuned Si photonic devices.
Furthermore, our waveguide simulations show that the 2D layer does not aect the propagating mode in the Si waveguide; thus, this
simple additional thin film produces a substantial energy eciency improvement without degrading the optical performance of the
waveguide. Our findings pave the way for energy-ecient laser-induced structural phase transitions in PCM-based reconfigurable
photonic devices.
KEYWORDS: phase change memory, van der Waals interfaces, 2D materials, thermal engineering, low energy switching
INTRODUCTION
Phase change materials (PCMs) have been commercialized for
both optical and electrical data storage because they exhibit a
large optical and electrical property contrast that can be
induced in nanoseconds, and once switched, the properties are
latched into a metastable state, i.e., they exhibit non-
volatility.
14
The data RESET process is achieved by
converting the crystalline phase to the amorphous phases
using heat by either a short current or a laser pulse to melt the
PCM. The molten state is then quenched at a high rate to
freeze in the disordered state.
5,6
The switching is reversible and
can be repeated billions of times.
7
The reversible transition
from the amorphous to the crystalline phase (SET) is induced
by heating the material to a temperature above the glass
transition temperature and below the melting temperature for
a relatively longer time.
PCMs have already been used in commercial products, from
DVD-RW optical discs
8
to advanced 3D X’Point electrical
memory,
9
and now they are widely studied for universal
memory and neuro-inspired computing.
7,10
Despite the
commercial successes and potential photonic applications,
the amorphization operation can be energy inecient because
the heat easily dissipates into the surroundings. Typically, only
1% of the supplied energy is used by the PCM, and this
energy ineciency limits the potential applications of PCMs.
11
Many eorts have been devoted to improving the energy
eciency in chalcogenide PCM switching. From an electrical
device perspective, one eective way is to reduce the contact
area between the bottom electrode and the PCM cell to
decrease the switching volume of PCMs. This is done by
replacing the typical mushroom structure with edge-contact-
type,
12
bridge-type,
13
or μTrench
14
structures or by applying a
nanoscale electrode such as carbon nanotubes.
5
Material
optimization can also improve the device switching energy
eciency. Doping is a good approach to modify the properties
of GeSbTe ternary PCMs, for example, by using Sc,
15
Ti,
16
C,
17
Cr,
18
etc. These dopant atoms diuse into the PCMs or
Received: July 19, 2022
Accepted: August 18, 2022
Published: August 31, 2022
Research Articlewww.acsami.org
© 2022 American Chemical Society 41225
https://doi.org/10.1021/acsami.2c12936
ACS Appl. Mater. Interfaces 2022, 14, 4122541234
Downloaded via Robert Simpson on September 20, 2022 at 10:42:22 (UTC).
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
partially substitute atoms to form local defects or distortions,
which prevent the nucleated crystals from growing into large
grains. Strong phonon scattering from the additional grain
boundaries results in low thermal conductivity, which enables
generated heat in the PCM to be trapped within it. This heat
trapping decreases the energy transfered to the surroundings,
and low-energy switching can be realized. However, incorpo-
rating dopants often produces phase separation, which is seen
as the material is cyclically switched between its amorphous
and crystalline states, thus shortening the lifespan of phase
change memory cells.
19,20
Stacking two dierent PCMs in a
superlattice-like structure can also lower the thermal
conductivity and resultant switching energy.
21
The lower
thermal conductivity is achieved at the interfaces, which are
produced by the alternating layers within the layered structure.
Interfacial phase change materials (iPCMs) have also been
used to lower entropic losses during the phase transition.
22
Strain engineering these iPCMs, by exploiting the lattice
mismatch of the Sb2Te3and GeTe layers, was applied to
further enhance the energy performance of the iPCM.
2325
However, growing the iPCM requires accurate control of the
physical vapor deposition system conditions. A more
straightforward approach to lowering the programming energy
of electrical phase change memory devices involves inserting
an interfacial layer between PCMs and heater electrodes.
Indeed, Ta2O5,
26
fullerene,
27
WO3,
28
and TiO2
29
have all
shown some promise at decreasing the programming voltage in
electrical devices. The low power switching originates from the
low thermal conductivity of these inserted layers, but the
downside of this approach is that the inserted layer increases
the electrical resistance of the electrical memory, which
negatively aects the overall device performance. Moreover,
Bi2Te3-inserted in Ge2Sb2Te5devices reduced the switching
current density and power due to the thermoelectric heating at
the interface.
30
Most of the eorts to increase the switching energy
performance of PCMs have focused on electrical memory
devices. However, with the increasing interest in PCM
programmable photonics, we need to start considering how
to make these devices switch eciently too. Some of the most
studied applications of PCM-programmable photonics include
Si waveguides and plasmonic metamaterials.
3135
In both
cases, the PCM is typically interfaced directly with high
thermal conductivity materials. For example, the thermal
conductivities of silicon and gold are, respectively, 140 and 318
W·m1·K1. Two dimensional (2D) van der Waals (vdW)
materials, such as graphene
36
and MoS2,
37
are known to have a
low out-of-plane thermal conductivity and are, therefore,
interesting to study as a way to increase the thermal boundary
resistance (TBR) between a material with high thermal
conductivity, such as a silicon waveguide, and a PCM. Indeed,
others have shown that the RESET energy of PCM electrical
Figure 1. (a) Schematic of the programmable optical structure consisting of a substrate, a 2D layer, and a phase change material layer. (b) Cross-
sectional TEM image of interfacial bilayer MoS2between a GeTe film and a silica-on-silicon substrate. (c) Crystallization temperature of only
GeTe, GeTe on MoS2, or GeTe on WS2, all of which are on the silica-on-silicon substrate. (d) Raman spectra of only MoS2, as-deposited
amorphous GeTe on MoS2, and annealed crystalline GeTe on MoS2, all on the silica-on-silicon substrate. The phonon modes for GeTe and MoS2
are highlighted and labeled using red and black, respectively.
ACS Applied Materials & Interfaces www.acsami.org Research Article
https://doi.org/10.1021/acsami.2c12936
ACS Appl. Mater. Interfaces 2022, 14, 4122541234
41226
memories can be lower when a 2D material is placed at the
interface of the electrical heater and the PCM. Until now, the
eect of 2D materials on the switching energy performance of
PCM-based photonic devices has not been studied but we
hypothesize that incorporating 2D materials into photonic
devices will eciently trap the heat in the PCM and could
radically lower the switching energy.
In PCM-integrated photonic devices, the PCM provides a
means to route and attenuate light in a photonic circuit.
3133
For PCM-integrated photonic devices, refractive index switch-
ing is realized by applying heat pulses to the PCM, which is in
contact with the waveguide. However, high power laser pulses
are needed to introduce the structural phase transitions and
concomitant refractive index changes in the PCM when it is
directly interfaced with a Si waveguide due to its high thermal
conductivity. When these PCM switches are incorporated in a
large-scale photonic network with an array of interconnected
waveguide meshes, the energy needed to precisely program the
network will be high and scale unfavorably with the number of
PCM-tuned elements, and this will ultimately limit the network
scalability. Indeed, in hardware neural networks, a single
programming pulse energy should be in the femtojoule
range,
38
but the current PCM programming pulse energy on
Si waveguides is in the picojoule range.
31
Thus, for these
devices to become practical, we must start considering how to
make them thermally energy ecient. Herein, we propose 2D
transition metal dichalcogenide (TMDC) crystal thermal
barriers as a possible means to limit heat diusion into the
substrate. TMDC sulfides have a higher melting temperature
and wider bandgap than their selenide and telluride counter-
parts. The higher melting temperature means that the layers
are stable even at the temperatures typically used during PCM
laser switching. The wide bandgap is necessary to minimize
absorption of the 1550 nm light in photonic integrated circuits.
Since silica-on-silicon and Si substrates are often, respectively,
used in plasmonics and photonic integrated circuits, we study
how 2D layers of MoS2and WS2on silica-on-silicon and Si
substrates influence the laser energy required to switch a PCM.
In this work, the 2D material was placed as an atomically
thin interfacial thermal barrier underneath the PCM, either on
a silicon substrate, or on a silica-on-silicon substrate, as shown
in Figure 1a,b. Since the inert vdW interfaces do not have any
dangling bonds, we expect they do not aect the structural
transition behavior of PCMs. We expect that a few atomic
layers (12 nm) of a dielectric material will not change the
optical performance of photonic devices. Thermally, on the
contrary, their eect is expected to be sizable, as the weak vdW
interfaces of the 2D TMDC layer should strongly limit heat
transport along the out-of-plane direction.
39
Hence, energy will
be confined within the small volume of PCM and greatly
reduce the power used to switch the PCM. We study by
experiment and simulation whether this inherent thermal
property is common to two dierent 2D TMDC materials,
MoS2and WS2, and show that these 2D TMDC layers are
indeed eective at increasing the optical switching power
eciency in PCM-tuned Si photonic devices. We believe that
this design is applicable to a wide range of PCM-based
photonic devices, including thin-film reflective displays,
40
programmable plasmonic devices,
34
and metasurfaces.
35
METHODS
Growth and Characterization. The stacked film structure
consists of either a 300 ±15 nm silica-on-silicon or a nonoxidized
Si substrate, a 2D TMDC thermal barrier, and a GeTe PCM layer, as
shown in Figure 1a. The MoS2and WS2were prepared using atomic
layer deposition (ALD) for the silica-on-silicon substrate and van der
Waals epitaxy (VdWE) for the Si substrate. For the MoS2growth, the
substrate was treated in a UV/O3reactor for 10 min. After that, MoO3
was grown using thermal ALD in a Cambridge Nanotech Savannah
S200 system using bis(tert-butylimido)-bis(dimethylamido) molybde-
num as a molybdenum precursor at 250 °C. The films were then
sulfurized in a tube furnace using an H2S/Ar gas mixture with a final
annealing temperature of 970 °C. Consequently, 23 layers of MoS2
(1.65 nm) were coated. The in-house developed VdWE apparatus
was used to grow monolayer WS2.
41,42
WCl6(99.9% pure from Sigma-
Aldrich) was used as the precursor, kept in a bubbler, and delivered by
Ar gas to the VdWE system to react with H2S gas to form a WS2
monolayer on the substrates at the set growth temperature of 900 °C.
A deposition time of 5 min was required to achieve uniform WS2
monolayer films (0.65 nm). Subsequently, a 30 nm GeTe layer was
deposited from a GeTe alloy target (2” diameter and 99.999% pure
from AJA International) using magnetron sputtering (AJA Orion5) in
an Ar atmosphere with a pressure of 3.7 mTorr (0.5 Pa) and a
working distance of 140 mm. We also placed the blank substrates
along with the MoS2/WS2deposited samples to act as the control
samples. The sample structure was measured using transmission
electron microscopy (TEM, FEI Titan) with an acceleration voltage
of 200 kV. Focused ion beam (FIB, FEI Helios Nanolab 450S) milling
was necessary to prepare the lamella for cross-sectional TEM imaging.
The crystalline GeTe films were first prepared by annealing the as-
deposited GeTe film using a temperature-controlled heating stage at
300 °C for 10 min. The crystallization temperature was found by
dierentiating the reflectivity curve, which was recorded while heating
the samples from room temperature to 300 °C with a 4 °C/min ramp
rate (Linkam Scientific Instruments Ltd.). To protect the film from
oxidation, which is known to influence its phase transitions,
43
the
anneal was performed in an Ar atmosphere flowing at 4 SCCM.
Raman spectra were collected at room temperature using a WITec
Alpha300R system equipped with a 633 nm wavelength excitation;
the incident laser intensity was kept low to minimize irradiation-
induced heating of the probed region. The thickness measurement
was carried out via atomic force microscopy (AFM, Asylum Research,
MFP-3D Origin). Our in-house developed static tester, which consists
of a low-power 635 nm probe laser and a relatively high-power 660
nm pump laser, was used to measure the switching power and time.
44
The system can simultaneously measure the reflection of the probe
laser from the sample while the pump laser pulses heat the sample.
The focused laser spot had a beam size of 0.8 μm (1/e2intensity) on
the sample. Here, we used the static tester to laser write an array of
crystallization and amorphization marks under dierent laser pulse
widths and incident powers. The reflected signal from the probe laser
was collected before and after the pump pulses. The crystallization or
amorphization was realized by a single pulse in the power-time-
reflectivity measurements.
Finite Dierence Simulations. The heat induced by laser pulses
can increase the temperature of PCMs and achieve crystallization or
amorphization. The transient temperature profile is obtained from the
unsteady heat conduction equation, as given in eq 1,
= · +cT x y z t
tT x y z t Q x y z t
( , , , ) ( , , , ) ( , , , )
(1)
where, T(x,y,z,t) is the temperature at a site of (x,y,z) and a certain
time t,ρis the mass density, cis the specific heat capacity, κis the
thermal conductivity, and Q(x,y,z,t) is the Joule heat brought by the
laser pulse, which can be expressed as eq 2, assuming a Gaussian beam
profile,
=
+
Q x y z t e P
wR e f t( , , , ) 2(1 ) ( )
zx y
w
in
2
2
2 2
2
(2)
where, Pin is the laser power, wis the 1/e2Gaussian beam radius, αis
the absorption coecient, Ris the reflectivity, and f(t) is the temporal
waveform.
ACS Applied Materials & Interfaces www.acsami.org Research Article
https://doi.org/10.1021/acsami.2c12936
ACS Appl. Mater. Interfaces 2022, 14, 4122541234
41227
Here, ρof GeTe (6.19 g·cm3) and cof GeTe (259.2 J·kg1·
K1) were used.
45
Meanwhile, ρof MoS2(5.06 g·cm3), cof MoS2
(379.6 J·kg1·K1),
46,47
ρof WS2(7.5 g·cm3), and cof WS2
(250 J·kg1·K1) were used,
46,48
respectively. The Rvalues of a-
GeTe (0.44), c-GeTe (0.68), a-GeTe/MoS2(0.47), c-GeTe/MoS2
(0.64), a-GeTe/WS2(0.54), c-GeTe/WS2(0.66) were measured with
a 660 nm laser. The αvalues of a-GeTe (1.8 ×107m1) and c-GeTe
(4.75 ×107m1) were calculated from their extinction coecient.
49
We used the κvalues of a-GeTe (0.204 W·m1·K1) and c-GeTe
(3.59 W·m1·K1).
50
Thermal boundary conductances of MoS2and
WS2layers were used as 16 and 5.5 MW·m2·K1.
51,52
The
thickness of the GeTe was 30 nm, and wwas measured as 0.8 μm.
Laser pulses (100 and 500 ns) with dierent powers were applied in
amorphization and crystallization simulations respectively.
Waveguide Simulation. The PCM-tuned Si waveguides used in
the optical simulation were optimized in the transverse electric (TE)
mode. The dimensions were chosen to ensure single mode
operation.
53,54
In the simulation model, refractive indices of the Si
waveguide, MoS2/WS2and GeTe layers were obtained from the
literature.
33,55,56
The refractive index values can be found in Figure
S6.To obtain the mode profile and overall eective index values of the
waveguide, we solved Maxwell’s equations on the waveguide cross-
section using the finite dierence eigenmode solver from Lumerical
Mode Solutions (LMS).
RESULTS AND DISCUSSION
The stacked substrate/TMDC/PCM structure was measured
by TEM, as shown in Figure 1b. This TEM cross-sectional
image confirms the sandwich structure design (see Figure 1a)
and indicates that the 2D crystal layer is not damaged by the
sputtering process. To ascertain that the 2D material layer is
chemically inert and does not influence the structural
transformation of GeTe, we measured the crystallization
temperature and phonon modes of the GeTe on top of the
2D TMDC layers. The crystallization temperature of
amorphous GeTe films, which were deposited directly on
top of the crystalline bilayer MoS2, monolayer WS2, and the
silicon substrate, was measured by recording the reflected
intensity of visible light from the films as a function of
temperature, as shown in Figure 1c. The sudden increase in
reflectivity corresponds to the crystallization of the material.
Both the as-deposited GeTe sample and the GeTe on MoS2or
WS2samples crystallized at 201 °C. The consistent
crystallization temperature indicates that the 2D materials do
not influence the GeTe phase transition. This is expected since
the vdW interfaces of the 2D layers are chemically inert and
stable. Hence, the GeTe layer is physically isolated from the
2D material layer as no dangling bonds are present to form
strong covalent bonds with the subsequent GeTe layer.
To further confirm that the MoS2layer does not aect the
local structural transformation in GeTe crystallization, we also
performed a Raman analysis. The Raman spectra of bilayer-
MoS2before and after GeTe deposition and the as-deposited
GeTe film with MoS2layers after annealing are presented in
Figure 1d. In Figure 1d, we highlight the GeTe and MoS2
phonon modes in red and black dot-and-dash lines,
respectively. The A1g(179 cm1), E2g2(230 cm1), E2g1(382
cm1), A1g(408 cm1), E1u2(417 cm1), and E2g2( 456 cm1)
modes seen in the MoS2sample were reported in the
literature.
57,58
The GeTe on the MoS2sample spectrum is a
combination of amorphous GeTe and MoS2peaks. The as-
deposited amorphous GeTe peaks occur at A (92 cm1), B
(123 cm1), C (162 cm1), and D (218 cm1) in the frequency
range 50250 cm1.
43,59
Upon crystallization, we observe a
weaker signal in bands C (162 cm1) and D (218 cm1). This
indicates a local structural change of Ge from a lower
tetrahedral coordination to an octahedral coordination, thus
confirming that GeTe crystallization has occurred. The MoS2
peaks are weakened by the 30 nm GeTe layer absorbing a
portion of the scattered intensity. This eect is more
substantial in the crystalline GeTe sample due to its higher
Figure 2. Switching behavior of GeTe film on the silica-on-silicon substrate with MoS2thermal barrier. (a) AFM topography of the write-mark
matrix written into GeTe on the MoS2bilayer on the silica-on-silicon substrate. (b) Raman spectra of the GeTe films on the MoS2bilayer in
dierent structural states. Red, blue, and purple curves correspond to as-deposited amorphous, optically crystallized, and ablated region. The signals
of GeTe and MoS2are labeled using red and black color in the range 50350 cm1. (c) Optical images of reamorphized laser-amorphized ”S”, ”U”,
”T”, and ”D” characters, which were sequentially written into the same area of a recrystallized GeTe film; scale bar, 10 μm.
ACS Applied Materials & Interfaces www.acsami.org Research Article
https://doi.org/10.1021/acsami.2c12936
ACS Appl. Mater. Interfaces 2022, 14, 4122541234
41228
absorption coecient, which is induced by a denser and more
compact crystalline structure upon annealing. From the
spectral measurements, we see that the MoS2Raman modes
are unaected by the GeTe layer and are able to withstand the
deposition and heat-induced crystallization process. A similar
eect was observed in the Raman patterns of WS2and GeTe
on WS2, as shown in Figure S1.
60
Hence, we conclude that the
MoS2layers can be used in conjunction with telluride PCMs
without any further altering the layer bonds or stoichiometry.
Moreover, the thermal stability of MoS2with Te-based PCMs
also indicates that other switching energy optimization
strategies, such as superlattice or strain engineering, may be
used to further lower the PCM switching energy.
The aim of this work is to study whether the 2D TMDC
layers can lower the heat power required for GeTe layers on
thermally conductive substrates to crystallize or amorphize. To
test this, we used a laser to write amorphous marks into the
crystalline GeTe film on MoS2and WS2layers. An amorphous-
mark pulse power-time-reflectivity matrix was made by
controlling the laser power and pulse duration. A microscope
image of the resultant amorphization matrix for crystalline
GeTe on MoS2is shown in Figure S2b. The laser pulse power
was set in the range 033 mW, and the pulse duration was
from 10 to 100 ns. A visible reflectivity change was observed in
the optical microscope image when the laser power reached
13.45 mW. The white and blue areas are where the GeTe
amorphized or ablated, respectively. We used an AFM to
confirm whether the material truly amorphized or whether it
had ablated (see Figure 2a). We observe a 2 nm increase in
thickness upon amorphization, which corresponds to a 7%
thickness change between crystalline and amorphous states, a
result that is consistent with previous reports.
61
GeTe ablation
occurred for pulse powers above 17.6 mW and pulse durations
longer than 70 ns. Ablation is visible as a slackening and
squeezing from the center of the irradiated mark, which results
in a microbasin forming, as shown in the inset line profile on
the AFM height map. We then compare the Raman spectra of
GeTe surfaces that consist of as-deposited regions, laser
crystallized regions, and ablated regions, which were damaged
by high power laser pulses. The Raman spectra are shown in
Figure 2b. The signals from the laser crystallized GeTe regions
match the annealed crystalline GeTe sample, as shown in
Figure 1d. We observe that the ablated regions have a stronger
MoS2signal because the GeTe surface has been removed and
is, therefore, unable to eciently absorb the MoS2Raman
scattered photons. These measurements confirm that the MoS2
structure is still intact after laser switching.
To demonstrate that GeTe on MoS2can be reversibly
switched and that consistently low power laser pulses can be
used for amorphization, we amorphized dot-matrix-images of
the characters ”S”, ”U”, ”T”, and ”D” from a crystalline region
of the sample using low power laser pulses. Figure 2c shows
optical micrographs of our rewritable pattern where the
annealed crystalline GeTe on MoS2sample was selected as the
rewritable canvas. The blue background shows crystalline
GeTe while the lighter shade of blue indicates GeTe
amorphization. To make this pattern, we used 50 ns, 15.5
mW laser pulses to amorphize the film, and 800 ns, 7.46 mW
laser pulses to crystallize it. Although not studied here, the
write-erase cycling ability of data storage PCMs is known to
increase as the energy used to switch the PCM goes down.
62
Hence, we expect higher cycleability for the TMDC-interfaced
PCMs compared to PCMs without the TMDC layer. However,
to increase the reflectivity contrast we used laser pulse powers
that were 15 and 35% higher than the minimum power
required to amorphize and crystallize the GeTe sample, which
may slightly compromise the GeTe cycleability.
Thus far we have shown that interfacing MoS2with GeTe
does not influence structural transformations, and we have
confirmed that the GeTe can be amorphized and recrystallized
by laser switching. We now quantify the enhancement caused
by the MoS22D layer on the switching energy of GeTe films
using our laser static tester.
44
The laser static tester was used to
amorphize the crystalline GeTe film with dierent pulse
powers and widths. The optical contrast gradually appeared
with increasing pulse power and width (Figure S2), indicated
by a decrease in reflectivity, as shown in Figure 3a,b. For a
GeTe layer deposited on a silica-on-silicon substrate, the
amorphization power threshold was 21.95 mW for 70 ns
pulses. In contrast, amorphizing the GeTe interfaced with the
MoS22D layer on a silica-on-silicon substrate only required
13.45 mW and 50 ns, as shown in Figure 3b. The laser
switching power used to amorphize the samples was reduced
by 40% by adding a MoS22D layer. The greatly reduced
amorphization power is attributed to the ultrahigh thermal
boundary resistance of the MoS2interfaces, which confines the
laser heat inside the small volume of the GeTe such that it
rapidly reaches its melting temperature. Moreover, if a 33.42
mW pulse is used to amrophize the GeTe on MoS2, then the
amorphization time is reduced by 67% from 30 to 10 ns.
Similarly, GeTe on WS2on top of a silica-on-silicon substrate
also produced a decrease in switching energy and time. We
direct the interested reader to Figure S3, where the
corresponding power-time-reflectivity plots for laser amorph-
ization are included.
It is interesting that the switching energy performance
enhancement due to the TMDC layer is only seen for
amorphization and not for crystallization. We found that the
GeTe with and without MoS2is crystallized by a pulse with the
same power. This eect is due to the lower laser pulse power
and longer time required for crystallization. This means heat
can diuse further through the stacked layers into the
substrate, which results in a smaller temperature gradient
through the sample. Since the thermal conductivity of SiO2is
relatively low at 1.4 W·m1·K1, this limits the heat loss to
some extent and makes the switching energy reduction
unapparent.
We have found that adding 2D TMDC layers between GeTe
and a substate is eective at lowering the RESET
(amorphization) power of GeTe without influencing its local
atomic structure nor the crystallization temperature. We
hypothesize, therefore, that the 2D TMDC layers must
Figure 3. Laser amorphization power-time-reflectivity measurement
of (a) only GeTe and (b) GeTe with MoS2on the silica-on-silicon
substrate.
ACS Applied Materials & Interfaces www.acsami.org Research Article
https://doi.org/10.1021/acsami.2c12936
ACS Appl. Mater. Interfaces 2022, 14, 4122541234
41229
introduce an enormous TBR, and it is this TBR that improves
the switching energy eciency by trapping heat in the PCM.
To study how MoS2and WS2layers can act as an ecient
thermal boundary, we performed finite-dierence simulations
to model the heat transport between the interfaces of the
substrate/TMDC/PCM stack. We simplified the model to a
sandwich structure consisting of substrate, an infinitely thin 2D
TBR interfacial layer, and the GeTe phase change material.
The heat transport simulation for the amorphization and
crystallizaiton process are presented in Figure 4 and Figure S4.
We modeled GeTe amorphization using the measured
threshold power for amorphization with 100 ns pulses. We
see from the amorphization matrices in Figure S2 that the
threshold laser powers are 13.45 and 21.95 mW with and
without the MoS2layer, respectively. Both the GeTe and the
GeTe with MoS2samples reached a similar temperature after
100 ns as shown in Figure 4, and this temperature is above the
GeTe melting temperature, which is necessary for amorphiza-
tion. Importantly, the modeled laser power necessary to melt
GeTe on MoS2is 40% less than that required to melt GeTe on
a silica-on-silicon substrate. The temperature distribution plots
in Figure 4b show that the MoS2layer causes the heat to be
eciently confined within the PCM layer. Moreover, the GeTe
with MoS2experiences a higher heating rate during the 100 ns
laser pulse, and higher quench rate after the pulse ended, as
shown in Figure 4c. At first glance, this high quench rate may
seem counterintuitive because the MoS2interfacial layer has a
large TBR. However, the substrate temperature is much lower
when the MoS2layer is included (see Figure 4b), and there is
less thermal energy provided to the whole structure. This
means that only the GeTe layer needs to cool substantially, and
the SiO2can act as a heat sink and absorb the small amount of
thermal energy that is trapped in the GeTe layer. We conclude
that just 1 nm of MoS2can eectively prevent heat transfer to
Figure 4. Heat transport simulation of GeTe on MoS2on a silica-on-silicon substrate during laser amorphization. (a) Axial temperature distribution
in GeTe films on a silica-on-silicon substrate without and with MoS2interfacial layers using dierent laser powers. (b) Temperature distribution in
cross-section of GeTe and GeTe on MoS2samples with dierent power pulses. (c) Maximum temperature on the surface of GeTe and GeTe on
MoS2samples after a 100 ns laser pulse with dierent powers.
Figure 5. WS2eect on laser switching and amorphization. Power-time-reflectivity measurements for (a) crystallization of GeTe on Si, (b)
crystallization of GeTe on WS2on silicon, and (c) reamorphization of GeTe on WS2on silicon. Simulated temperature of GeTe on WS2on silicon
during amorphization. (d) Axial temperature distribution in GeTe films on Si without and with a WS22D layer using same laser pulse power. (e)
Cross-sectional temperature distribution of GeTe on Si and GeTe on WS2on silicon with the same laser pulse power. (f) Maximum temperature on
the surface of GeTe on silicon and GeTe on WS2on silicon after a 100 ns laser pulse with the same power.
ACS Applied Materials & Interfaces www.acsami.org Research Article
https://doi.org/10.1021/acsami.2c12936
ACS Appl. Mater. Interfaces 2022, 14, 4122541234
41230
the substrate during 100 ns laser pulses. Indeed, in terms of
thermal isolation, the 1 nm thick MoS2is equivalent to 100
nm of SiO2.
In the previous analysis, we simulated the heating using the
dierent laser pulse powers for amorphization. However, if the
same laser power is used for samples with and without the 2D
TMDC layers, and if the pulse times are longer, we can see that
the samples reach thermal equilibrium at dierent temper-
atures.
Based on the observed threshold power for crystallization,
we simulated a 5.54 mW 500 ns laser pulse in Figure S4. This
pulse condition is similar to that required for GeTe
crystallization. MoS2causes the GeTe to equilibrate at 700
K. In constrast, the GeTe on a silica-on-silicon substrate
saturates at 600 K. Both temperatures are above the 573 K
required for GeTe crystallization, but since GeTe crystal-
lization is limited by the nucleation time, we do not see a
significant dierence in the overall crystallization time in the
experiments. The sample with a 300 nm thick SiO2layer and
a MoS2layer equilibrates at 700 K rather than 600 K, which
occurs without the MoS2layer. However, we would expect the
dierence in equilibrating temperatures to be much more
pronounced for TMDCs interfaced directly with a highly
thermally conductive Si waveguide.
In order to establish the generality of this thermal barrier
property among 2D materials and distinguish the improvement
in crystallization, we grew GeTe on top of another 2D material,
WS2. Moreover, the samples are grown on silicon, which is
more relevant to silicon photonics, rather than the silica-on-
silicon substrate. WS2has a similar structure and properties as
MoS2with weak vdW interfaces. Si is 100×more thermally
conductive (140 W·m1·K1) than SiO2, which facilitates
faster heat dissipation. In our previous measurement, the
crystallization temperature of GeTe on WS2(201 °C) and
Raman spectra were measured, as seen in Figure 1c and Figure
S1, which means that the WS2monolayer also had no chemical
reaction with GeTe and good thermal stability. Moreover, WS2
substantially decreased the switching energy of GeTe on the
silica-on-silicon substrate (Figure S3). We should expect the
influence of WS2on the switching power to be even more
dramatic for Si substrates because its thermal conductivity is an
order of magnitude greater than that of SiO2. We measured the
laser switching power and time for GeTe thin films with and
without the WS2interfacial layer on silicon. Again, both
samples came from the same GeTe sputtering batch. In our
laser write-mark matrix switching experiment, the crystalliza-
tion power is 7.45 mW for GeTe grown on WS2while it is
13.45 mW for GeTe deposited directly on a Si substrate, as
shown in Figure 5a,b. This is in agreement with simulations
where we see more than 40% reduction in the switching power
for the modeled crystallization process, as shown in Figure S5.
This also conforms with the switching eciency improvement
seen in the GeTe/MoS2sample during amorphization. We also
studied the amorphization processes where the annealed
crystalline GeTe layer with a 0.65 nm-thick interfacial WS2
layer became less reflective after being exposing to an
amorphizing 21.95 mW, 50 ns laser pulse. Figure 5c shows
the corresponding laser pulse power-time-reflectivity plots with
the WS2layer. In contrast, we found that amorphization of
GeTe directly on silicon was not possible, even at our system’s
power limit of 33.42 mW. Thus, a suciently thick oxide layer
or a 0.65 nm TMDC layer are required to limit heat transfer to
the substrate. Indeed, the WS2layer was necessary to
amorphize the GeTe layer on Si, although the laser pulse
power is higher than that required for GeTe on WS2on the
silica-on-silicon substrate (12.42 mW, 80 ns). Hence, the WS2
layer is extremely eective at reducing the RESET energy on
silicon substrates. As shown by our measurements and
simulations, this improvement results from the TBR at the
WS2interface layer. The weak vdW interaction restricts the
heat generated by laser pulses from dissipating in the out-of-
plane direction. Surprisingly, the eect of WS2on the heat
transport between the Si and the GeTe layer is so pronounced
that it was even possible to ablate the GeTe layer on WS2on
Si. This is remarkable considering that GeTe deposited directly
on Si cannot even be heated to induce amorphization. The
thermal simulations of the GeTe-WS2-Si stacks provide some
insight into this substantial dierence. Parts df of Figure 5
show that 21.95 mW laser heating pulses cause the GeTe on
silicon to marginally heat because the heat rapidly dissipates
into the Si substrate. Indeed, after approximately 10 ns, the
temperature of the GeTe saturates at 349 K, which is a
negligible temperature rise. However, adding the subnanom-
eter-thick WS2monolayer confines the heat within the GeTe
layer and the temperature saturates at 1070 K in 30 ns. These
results explain the reason GeTe on Si could not be laser
amorphized in our laser static tester system but could be
readily amorphized when the subnanometer thick WS2layer
was inserted between the GeTe and the silicon. This result is
especially relevant to Si photonics, where the PCM is usually
placed in direct contact with the Si waveguide.
These results indicate that incorporating subnanometer
thick TMDC layers into PCM-based reconfigurable photonic
integrated circuit (PIC) devices will allow for ecient PCM
switching. However, ideally, the TMDC material should have
negligible interaction with the optical mode propagating in the
waveguide. To demonstrate the compatibility of subnanometer
thick TMDCs with photonic devices, we compare the changes
in optical mode confinement of photonic waveguides with and
without the TMDC layer using finite dierence eigenmode
(FDE) calculations. Figure 6a shows the waveguide simulation
model with the WS2layer. The resulting mode pattern and
eective refractive index, neff, of the waveguides at the 1550 nm
wavelength, which is in the telecommunication c-band, are
shown in Figure 6be. We observe that the thin WS2layer
negligibly changes the real part of the eective refractive index
by less than 0.3% for both amorphous and crystalline GeTe-
tuned waveguides. The absolute eective refractive index
values are also shown in Figure 6be. Moreover, there are no
Figure 6. (a) Schematic of the GeTe-tuned Si waveguide model with
a WS22D TBR layer. The corresponding mode patterns and eective
refractive index, neff, values for amorphous and crystalline GeTe
without and with WS2are shown in (be).
ACS Applied Materials & Interfaces www.acsami.org Research Article
https://doi.org/10.1021/acsami.2c12936
ACS Appl. Mater. Interfaces 2022, 14, 4122541234
41231
discernible changes in the mode patterns. The nondiscernible
change in mode pattern is partly due to WS2having a close
refractive index value to Si and being nonabsorbing in the
infrared due to its large bandgap (Figure S6).
55,56
Therefore,
the WS2layer can be incorporated into PCM-based PIC
devices with minimal optical eect yet produce a dramatic
reduction in the PCM switching energy, which is a highly
desirable trait. For reliable and repeatable cycling performance,
the PCM should be encapsulated with a barrier against
oxidation, such as Al2O3.
63
Since the WS2on silica-on-silicon
substrates showed a similar amorphization performance to that
of MoS2on silica-on-silicon, we also expect MoS2to show a
similar improvement as WS2if it is placed directly on the Si
waveguide. Similarly, a MoS2interfacial 2D layer causes a
negligibe change in the eective refractive index and
concomitant modes of the GeTe-tuned PCM waveguide
(Figure S7).
CONCLUSIONS
In conclusion, we have demonstrated a radical reduction in the
switching energy of GeTe on dierent substrates by employing
interfacial subnanometer thick 2D TMDC crystal layers. We
expect these performance enhancements to be broadly
applicable to programmable photonics, especially program-
mable plasmonic metamaterials and Si photonics, where the
PCM is often placed in direct contact with materials of high
thermal conductivity. The enhancement in switching energy
eciency is due to the 2D material vdW bonds confining heat
in the PCM layer. We demonstrated that the PCMs interfaced
with a 2D TMDC layer consumed less energy in both pulsed
laser crystallization and amorphization operations. There is an
over 40% reduction in power when the MoS2layer is interfaced
between the PCM and a silica-on-silicon substrate. The
improvement when WS2is placed between the Si and the
PCM is even more pronounced. However, we were unable to
quantify the relative enhancement factor because the PCM
could not be switched without the 2D TMDC layer due to the
power requirement being too high. However, simulations show
that the equilibrium temperature for 21.95 mW laser pulses is
increased by more than 700 K when a WS2layer is included
between the GeTe layer and the Si. We found that, in PCM-
programmed Si waveguide simulations, these 2D TMDC layers
have a negligible eect on the mode pattern and the waveguide
eective refractive index. These results show that 2D TMDC
layers should be included when designing ecient PCM-
programmable devices, such as photonic memories, all-optical
neural networks, and plasmonic metasurfaces.
ASSOCIATED CONTENT
*
Supporting Information
The Supporting Information is available free of charge at
https://pubs.acs.org/doi/10.1021/acsami.2c12936.
Figures of Raman spectra of WS2, amorphous GeTe on
WS2, and thermally crystallized GeTe on WS2, optical
micrographs of amorphization matrices on crystalline
GeTe and GeTe with MoS2, amorphization power-time-
reflectivity measurement of GeTe with WS2on the silica-
on-silicon substrate, crystallization heat transport simu-
lation of GeTe on MoS2and WS2layers, refractive
indexes of the materials used in the waveguide
simulation model, and GeTe-tuned Si waveguide
simulation model with MoS2layer (PDF)
AUTHOR INFORMATION
Corresponding Authors
Jing Ning Singapore University of Technology and Design
(SUTD), 487372, Singapore; Department of Materials
Science &Engineering, National University of Singapore,
117575, Singapore; orcid.org/0000-0001-7887-5083;
Email: jing_ning@mymail.sutd.edu.sg
Robert E. Simpson Singapore University of Technology and
Design (SUTD), 487372, Singapore;
Email: robert_simpson@sutd.edu.sg
Authors
Yunzheng Wang Singapore University of Technology and
Design (SUTD), 487372, Singapore
Ting Yu Teo Singapore University of Technology and Design
(SUTD), 487372, Singapore
Chung-Che Huang Optoelectronics Research Centre,
University of Southampton, Southampton SO17 1BJ, United
Kingdom
Ioannis Zeimpekis Optoelectronics Research Centre,
University of Southampton, Southampton SO17 1BJ, United
Kingdom
Katrina Morgan Optoelectronics Research Centre, University
of Southampton, Southampton SO17 1BJ, United Kingdom
Siew Lang Teo Institute of Materials Research and
Engineering (IMRE), Agency for Science Technology and
Research (A*STAR), Innovis 138634, Singapore
Daniel W. Hewak Optoelectronics Research Centre,
University of Southampton, Southampton SO17 1BJ, United
Kingdom
Michel Bosman Department of Materials Science &
Engineering, National University of Singapore, 117575,
Singapore; Institute of Materials Research and Engineering
(IMRE), Agency for Science Technology and Research
(A*STAR), Innovis 138634, Singapore; orcid.org/0000-
0002-8717-7655
Complete contact information is available at:
https://pubs.acs.org/10.1021/acsami.2c12936
Author Contributions
J.N. and Y.W. contributed equally to this work.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
The SUTD research was funded by a Singapore MoE Project
“Electric-field induced transitions in chalcogenide monolayers
and superlattices”, grant MoE 2017-T2-1-161 and an A*STAR
AME project: ”Nanospatial Light Modulators (NSLM)”,
A18A7b0058. The 2D materials work was supported by the
UK’s Engineering and Physical Sciences Research Council
through the Future Photonics Manufacturing Hub (EPSRC
EP/N00762X/1), the Chalcogenide Photonic Technologies
(EPSRC EP/M008487/1), and ChAMPChalcogenide Ad-
vanced Manufacturing Partnership (EPSRC EP/G060363/1).
J.N. is grateful for her MoE PhD scholarship.
REFERENCES
(1) Wuttig, M.; Steimer, C. Phase Change Materials: From Material
Science to Novel Storage Devices. Appl. Phys. A: Mater. Sci. Process.
2007,87, 411417.
ACS Applied Materials & Interfaces www.acsami.org Research Article
https://doi.org/10.1021/acsami.2c12936
ACS Appl. Mater. Interfaces 2022, 14, 4122541234
41232
(2) Lencer, D.; Salinga, M.; Grabowski, B.; Hickel, T.; Neugebauer,
J.; Wuttig, M. A Map for Phase-change Materials. Nat. Mater. 2008,7,
972.
(3) Siegrist, T.; Jost, P.; Volker, H.; Woda, M.; Merkelbach, P.;
Schlockermann, C.; Wuttig, M. Disorder-induced Localization in
Crystalline Phase-change Materials. Nat. Mater. 2011,10, 202.
(4) Kolobov, A. V.; Fons, P.; Frenkel, A. I.; Ankudinov, A. L.;
Tominaga, J.; Uruga, T. Understanding the Phase-change Mechanism
of Rewritable Optical Media. Nat. Mater. 2004,3, 703708.
(5) Xiong, F.; Liao, A. D.; Estrada, D.; Pop, E. Low-power Switching
of Phase-change Materials with Carbon Nanotube Electrodes. Science
2011,332, 568570.
(6) Zallo, E.; Wang, R.; Bragaglia, V.; Calarco, R. Laser Induced
Structural Transformation in Chalcogenide Based Superlattices. Appl.
Phys. Lett. 2016,108, 221904.
(7) Zhang, W.; Mazzarello, R.; Wuttig, M.; Ma, E. Designing
Crystallization in Phase-change Materials for Universal Memory and
Neuro-inspired Computing. Nat. Rev. Mater. 2019,4, 150168.
(8) Ohta, T. Phase-change Optical Memory Promotes the DVD
Optical Disk. J. Optoelectron. Adv. Mater. 2001,3, 609626.
(9) Hruska, J. Intel, Micron Reveal Xpoint, a New Memory
Architecture That Could Outclass DDR4 and NAND. https://www.
extremetech.com/extreme/211087-intel-micron-reveal-xpoint-a-new-
memory-architecture-that-claims-to-outclass-both-ddr4-and-nand.
(10) Tuma, T.; Pantazi, A.; Le Gallo, M.; Sebastian, A.; Eleftheriou,
E. Stochastic Phase-change Neurons. Nat. Nanotechnol. 2016,11,
693699.
(11) Sadeghipour, S. M.; Pileggi, L.; Asheghi, M. Phase Change
Random Access Memory, Thermal Analysis. Thermal and Thermo-
mechanical Proceedings 10th Intersociety Conference on Phenomena in
Electronics Systems, San Diego, CA, May 30June 2, 2006; pp 660
665.
(12) Ha, Y.; Yi, J.; Horii, H.; Park, J.; Joo, S.; Park, S.; Chung, U.-I.;
Moon, J. An Edge Contact Type Cell for Phase Change RAM
Featuring Very Low Power Consumption. Symposium on VLSI
Technology. Digest of Technical Papers (IEEE Cat. No. 03CH37407),
Kyoto, Japan, June 1012, 2003; pp 175176.
(13) Chen, Y.; Rettner, C.; Raoux, S.; Burr, G.; Chen, S.; Shelby, R.;
Salinga, M.; Risk, W.; Happ, T.; McClelland, G., et al. Ultra-thin
Phase-change Bridge Memory Device Using GeSb. 2006 International
Electron Devices Meeting, San Francisco, December 1113, 2006; pp
14.
(14) Pellizzer, F.; Benvenuti, A.; Gleixner, B.; Kim, Y.; Johnson, B.;
Magistretti, M.; Marangon, T.; Pirovano, A.; Bez, R.; Atwood, G. A
90nm Phase Change Memory Technology for Stand-alone Non-
volatile Memory Applications. 2006 Symposium on VLSI Technology,
Honolulu, HI, June 1315, 2006; pp 122123.
(15) Wang, Y.; Zheng, Y.; Liu, G.; Li, T.; Guo, T.; Cheng, Y.; Lv, S.;
Song, S.; Ren, K.; Song, Z. Scandium Doped Ge2Sb2Te5for High-
speed and Low-power-consumption Phase Change Memory. Appl.
Phys. Lett. 2018,112, 133104.
(16) Zhu, M.; Wu, L.; Rao, F.; Song, Z.; Ren, K.; Ji, X.; Song, S.;
Yao, D.; Feng, S. Uniform Ti-doped Sb2Te3Materials for High-speed
Phase Change Memory Applications. Appl. Phys. Lett. 2014,104,
053119.
(17) Zhou, X.; Xia, M.; Rao, F.; Wu, L.; Li, X.; Song, Z.; Feng, S.;
Sun, H. Understanding Phase-change Behaviors of Carbon-doped
Ge2Sb2Te5for Phase-change Memory Application. ACS Appl. Mater.
Interfaces 2014,6, 1420714214.
(18) Wang, Q.; Liu, B.; Xia, Y.; Zheng, Y.; Huo, R.; Zhang, Q.; Song,
S.; Cheng, Y.; Song, Z.; Feng, S. Cr-doped Ge2Sb2Te5for Ultra-long
Data Retention Phase Change Memory. Appl. Phys. Lett. 2015,107,
222101.
(19) Xia, M.; Zhu, M.; Wang, Y.; Song, Z.; Rao, F.; Wu, L.; Cheng,
Y.; Song, S. TiSbTe Alloy: A Candidate for Fast and Long-life
Phase-change Memory. ACS Appl. Mater. Interfaces 2015,7, 7627
7634.
(20) Borisenko, K. B.; Chen, Y.; Cockayne, D. J. H.; Song, S. A.;
Jeong, H. S. Understanding Atomic Structures of Amorphous C-
doped Ge2Sb2Te5Phase-change Memory Materials. Acta Mater. 2011,
59, 43354342.
(21) Chong, T.; Shi, L.; Zhao, R.; Tan, P.; Li, J.; Lee, H.; Miao, X.;
Du, A.; Tung, C. Phase Change Random Access Memory Cell with
Superlattice-like Structure. Appl. Phys. Lett. 2006,88, 122114.
(22) Simpson, R. E.; Fons, P.; Kolobov, A. V.; Fukaya, T.; Krbal, M.;
Yagi, T.; Tominaga, J. Interfacial Phase-change Memory. Nat.
Nanotechnol. 2011,6, 501.
(23) Kalikka, J.; Zhou, X.; Dilcher, E.; Wall, S.; Li, J.; Simpson, R. E.
Strain-engineered Diffusive Atomic Switching in Two-dimensional
Crystals. Nat. Commun. 2016,7, 11983.
(24) Zhou, X.; Kalikka, J.; Ji, X.; Wu, L.; Song, Z.; Simpson, R. E.
Phase-change Memory Materials by Design: A Strain Engineering
Approach. Adv. Mater. 2016,28, 30073016.
(25) Zhou, X.; Behera, J. K.; Lv, S.; Wu, L.; Song, Z.; Simpson, R. E.
Avalanche Atomic Switching in Strain Engineered Sb2Te3GeTe
Interfacial Phase-change Memory Cells. Nano Futures 2017,1,
025003.
(26) Matsui, Y.; Kurotsuchi, K.; Tonomura, O.; Morikawa, T.;
Kinoshita, M.; Fujisaki, Y.; Matsuzaki, N.; Hanzawa, S.; Terao, M.;
Takaura, N., et al. Ta2O5Interfacial Layer between GST and W Plug
Enabling Low Power Operation of Phase Change Memories. 2006
International Electron Devices Meeting, San Francisco, CA, December
1113, 2006; pp 14.
(27) Kim, C.; Suh, D.-S.; Kim, K. H.; Kang, Y.-S.; Lee, T.-Y.; Khang,
Y.; Cahill, D. G. Fullerene Thermal Insulation for Phase Change
Memory. Appl. Phys. Lett. 2008,92, 013109.
(28) Rao, F.; Song, Z.; Gong, Y.; Wu, L.; Feng, S.; Chen, B.
Programming Voltage Reduction in Phase Change Memory Cells with
Tungsten Trioxide Bottom Heating Layer/ Electrode. Nanotechnology
2008,19, 445706.
(29) Xu, C.; Song, Z.; Liu, B.; Feng, S.; Chen, B. Lower Current
Operation of Phase Change Memory Cell with a Thin TiO2Layer.
Appl. Phys. Lett. 2008,92, 062103.
(30) Khan, A. I.; Kwon, H.; Islam, R.; Perez, C.; Chen, M. E.;
Asheghi, M.; Goodson, K. E.; Wong, H.-S. P.; Pop, E. Two-fold
Reduction of Switching Current Density in Phase Change Memory
Using Bi2Te3Thermoelectric Interfacial Layer. IEEE Electron Device
Lett. 2020,41, 16571660.
(31) Rudé, M.; Pello, J.; Simpson, R. E.; Osmond, J.; Roelkens, G.;
van der Tol, J. J.; Pruneri, V. Optical Switching at 1.55 μm in Silicon
Racetrack Resonators Using Phase Change Materials. Appl. Phys. Lett.
2013,103, 141119.
(32) Ríos, C.; Stegmaier, M.; Hosseini, P.; Wang, D.; Scherer, T.;
Wright, C. D.; Bhaskaran, H.; Pernice, W. H. Integrated All-photonic
Non-volatile Multi-level Memory. Nat. Photonics 2015,9, 725732.
(33) Wuttig, M.; Bhaskaran, H.; Taubner, T. Phase-change Materials
for Non-volatile Photonic Applications. Nat. Photonics 2017,11, 465
476.
(34) Cao, T.; Wei, C.; Simpson, R. E.; Zhang, L.; Cryan, M. J. Rapid
Phase Transition of a Phase-change Metamaterial Perfect Absorber.
Optical Materials Express 2013,3, 11011110.
(35) Lu, L.; Dong, Z.; Tijiptoharsono, F.; Ng, R. J. H.; Wang, H.;
Rezaei, S. D.; Wang, Y.; Leong, H. S.; Lim, P. C.; Yang, J. K.
Reversible Tuning of Mie Resonances in the Visible Spectrum. ACS
Nano 2021,15, 19722.
(36) Ahn, C.; Fong, S. W.; Kim, Y.; Lee, S.; Sood, A.; Neumann, C.
M.; Asheghi, M.; Goodson, K. E.; Pop, E.; Wong, H.-S. P. Energy-
efficient Phase-change Memory with Graphene as a Thermal Barrier.
Nano Lett. 2015,15, 68096814.
(37) Neumann, C. M.; Okabe, K. L.; Yalon, E.; Grady, R. W.; Wong,
H.-S. P.; Pop, E. Engineering Thermal and Electrical Interface
Properties of Phase Change Memory with Monolayer MoS2.Appl.
Phys. Lett. 2019,114, 082103.
(38) Wang, Z.; Wu, H.; Burr, G. W.; Hwang, C. S.; Wang, K. L.; Xia,
Q.; Yang, J. J. Resistive Switching Materials for Information
Processing. Nat. Rev. Mater. 2020,5, 173195.
ACS Applied Materials & Interfaces www.acsami.org Research Article
https://doi.org/10.1021/acsami.2c12936
ACS Appl. Mater. Interfaces 2022, 14, 4122541234
41233
(39) Kim, K.; He, J.; Ganeshan, B.; Liu, J. Disorder Enhanced
Thermal Conductivity Anisotropy in Two-dimensional Materials and
van der Waals Heterostructures. J. Appl. Phys. 2018,124, 055104.
(40) Dong, W.; Liu, H.; Behera, J. K.; Lu, L.; Ng, R. J.; Sreekanth, K.
V.; Zhou, X.; Yang, J. K.; Simpson, R. E. Wide Bandgap Phase Change
Material Tuned Visible Photonics. Adv. Funct. Mater. 2019,29,
1806181.
(41) Felix, J. F.; da Silva, A. F.; da Silva, S. W.; Qu, F.; Qiu, B.; Ren,
J.; de Azevedo, W. M.; Henini, M.; Huang, C.-C. A Comprehensive
Study on the Effects of Gamma Radiation on the Physical Properties
of a Two-dimensional WS2Monolayer Semiconductor. Nanoscale
Horizons 2020,5, 259267.
(42) Gordo, V. O.; Balanta, M.; Gobato, Y. G.; Covre, F.; Galeti, H.;
Iikawa, F.; Couto, O.; Qu, F.; Henini, M.; Hewak, D.; et al. Revealing
the Nature of Low-temperature Photoluminescence Peaks by Laser
Treatment in van der Waals Epitaxially Grown WS2Monolayers.
Nanoscale 2018,10, 48074815.
(43) Zhou, X.; Du, Y.; Behera, J. K.; Wu, L.; Song, Z.; Simpson, R. E.
Oxygen Tuned Local Structure and Phase-change Performance of
Germanium Telluride. ACS Appl. Mater. Interfaces 2016,8, 20185
20191.
(44) Behera, J. K.; Zhou, X.; Tominaga, J.; Simpson, R. E. Laser
Switching and Characterisation of Chalcogenides: Systems, Measure-
ments, and Applicability to Photonics. Optical Materials Express 2017,
7, 37413759.
(45) Madelung, O.; Rössler, U.; Schulz, M. Germanium Telluride
(GeTe) Heat Capacity, Density, Melting Point. Non-Tetrahedrally
Bonded Elements and Binary Compounds I; Springer-Verlag Berlin
Heidelberg, 1998.
(46) Haynes, W. 100 Key Points. CRC Handbook of Chemistry and
Physics, 96th ed.; CRC Press, 2015.
(47) Suryavanshi, S. V.; Gabourie, A. J.; Barati Farimani, A.; Pop, E.
Thermal Boundary Conductance of Two-dimensional MoS2Inter-
faces. J. Appl. Phys. 2019,126, 055107.
(48) Ruppert, C.; Chernikov, A.; Hill, H. M.; Rigosi, A. F.; Heinz, T.
F. The Role of Electronic and Phononic Excitation in the Optical
Response of Monolayer WS2after Ultrafast Excitation. Nano Lett.
2017,17, 644651.
(49) Jafari, M.; Rais-Zadeh, M. Zero-static-power Phase-change
Optical Modulator. Opt. Lett. 2016,41, 11771180.
(50) Ghosh, K.; Kusiak, A.; Noé, P.; Cyrille, M.-C.; Battaglia, J.-L.
Thermal Conductivity of Amorphous and Crystalline GeTe Thin Film
at High Temperature: Experimental and Theoretical Study. Phys. Rev.
B2020,101, 214305.
(51) Yu, Y.; Minhaj, T.; Huang, L.; Yu, Y.; Cao, L. In-plane and
Interfacial Thermal Conduction of Two-dimensional Transition-metal
Dichalcogenides. Physical Review Applied 2020,13, 034059.
(52) Gertych, A. P.; Czerniak-Łosiewicz, K.; Łapinska, A.; Swiniarski,
M.; Ojrzynska, M.; Judek, J.; Zdrojek, M. Phonon and Thermal
Properties of Thin Films Made from WS2Mono-and Few-layer
Flakes. J. Phys. Chem. C 2021,125, 1444614452.
(53) Teo, T. Y.; Krbal, M.; Mistrik, J.; Prikryl, J.; Lu, L.; Simpson, R.
E. Comparison and Analysis of Phase Change Materials-based
Reconfigurable Silicon Photonic Directional Couplers. Optical
Materials Express 2022,12, 606621.
(54) Xu, P.; Zheng, J.; Doylend, J. K.; Majumdar, A. Low-loss and
Broadband Nonvolatile Phase-change Directional Coupler Switches.
ACS Photonics 2019,6, 553557.
(55) Edwards, D. F. Silicon (Si). In Handbook of Optical Constants of
Solids; Palik, E. D., Ed.; Academic Press, 1997; pp 547569.
(56) Liu, H.-L.; Yang, T.; Chen, J.-H.; Chen, H.-W.; Guo, H.; Saito,
R.; Li, M.-Y.; Li, L.-J.; et al. Temperature-dependent Optical
Constants of Monolayer MoS2, MoSe2, WS2, and WSe2: Spectro-
scopic Ellipsometry and First-principles Calculations. Sci. Rep. 2020,
10, 15282.
(57) Tuschel, D. Resonance Raman and Photoluminescence
Spectroscopy and Imaging of Few-layer MoS2.Spectroscopy 2015,
30 (3), 1429.
(58) Windom, B. C.; Sawyer, W.; Hahn, D. W. A Raman
Spectroscopic Study of MoS2and MoO3: Applications to Tribological
Systems. Tribol. Lett. 2011,42, 301310.
(59) Andrikopoulos, K.; Yannopoulos, S.; Voyiatzis, G.; Kolobov, A.;
Ribes, M.; Tominaga, J. Raman Scattering Study of the a-GeTe
Structure and Possible Mechanism for the Amorphous to Crystal
Transition. J. Phys.: Condens. Matter 2006,18 (3), 965.
(60) Berkdemir, A.; Gutiérrez, H. R.; Botello-Méndez, A. R.; Perea-
López, N.; Elías, A. L.; Chia, C.-I.; Wang, B.; Crespi, V. H.; López-
Urías, F.; Charlier, J.-C.; et al. Identification of individual and few
layers of WS2 using Raman Spectroscopy. Sci. Rep. 2013,3(1), 18.
(61) Zhou, X.; Dong, W.; Zhang, H.; Simpson, R. E. A Zero Density
Change Phase Change Memory Material: Gete-o Structural
Characteristics upon Crystallisation. Sci. Rep. 2015,5, 11150.
(62) Burr, G. W.; Breitwisch, M. J.; Franceschini, M.; Garetto, D.;
Gopalakrishnan, K.; Jackson, B.; Kurdi, B.; Lam, C.; Lastras, L. A.;
Padilla, A.; et al. Phase Change Memory Technology. J. Vac. Sci.
Technol., B: Nanotechnol. Microelectron.: Mater., Process., Meas.,
Phenom. 2010,28, 223262.
(63) Zhang, Y.; Fowler, C.; Liang, J.; Azhar, B.; Shalaginov, M. Y.;
Deckoff-Jones, S.; An, S.; Chou, J. B.; Roberts, C. M.; Liberman, V.;
et al. Electrically Reconfigurable Non-volatile Metasurface Using Low-
loss Optical Phase-change Material. Nat. Nanotechnol. 2021,16, 661
666.
ACS Applied Materials & Interfaces www.acsami.org Research Article
https://doi.org/10.1021/acsami.2c12936
ACS Appl. Mater. Interfaces 2022, 14, 4122541234
41234
Recommended by ACS
Paraffin-Enabled Compressive Folding of Two-Dimensional
Materials with Controllable Broadening of the Electronic
Band Gap
Weifeng Zhang, Nan Liu,
et al.
AUGUST 19, 2021
ACS APPLIED MATERIALS & INTERFACES READ
Realization of 4-Bit Multilevel Optical Switching in
Ge2Sb2Te5 and Ag5In5Sb60Te30 Phase-Change Materials
Enabled in the Visible Region
Mozhikunnam Sreekrishnan Arjunan, Anbarasu Manivannan,
et al.
DECEMBER 04, 2020
ACS APPLIED ELECTRONIC MATERIALS READ
Tri-state Metasurface-Based Electromagnetic Screen with
Switchable Reflection, Transmission, and Absorption
Functionalities
André de Lustrac, Shah Nawaz Burokur,
et al.
FEBRUARY 26, 2021
ACS APPLIED ELECTRONIC MATERIALS READ
A Non-volatile Quasi-Continuous All-Optical Fiber
Programmable Platform Based on GST-Coated
Microspheres
Rui Duan, Zhihai Liu,
et al.
FEBRUARY 23, 2022
ACS PHOTONICS READ
Get More Suggestions >
... Similarly, a "superlattice-like" structure, which consists of non-epitaxial Sb 2 Te 3 and GeTe stacked layers, was proposed to lower the switching energy by decreasing the thermal conductivity via interfacial phonon scattering [16]. More recently, several reports have demonstrated how two-dimensional crystals, such as graphene, MoS 2 , and WS 2 can be placed at the interface between the PCM and the electrode or substrate to confine heat through increased thermal boundary resistance [17][18][19]. (2) Decrease entropic losses; Interfacial phase change materials (iPCMs) improve the switching performance by limiting the movement of the atoms during the phase transition. Less movement freedom reduces entropic losses [20] and concomitantly decreases the energy required to switch the material. ...
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.
Article
Reconfigurable waveguide components are promising building blocks for photonic neural networks and as an optical analogue to Field-Programmable Gate Arrays. By changing the effective index of the waveguide, reconfigurable waveguide components can achieve on-chip light routing and modulation. In this paper, we design and demonstrate an Sb2S3-reconfigurable InP membrane Mach-Zehnder Interferometer (MZI) on a silicon substrate. Sb2S3, which has tunable refractive index and low absorption in the near-infrared spectrum, was patterned the InP waveguide MZIs to make an optical switch in the telecomms conventional-band. By laser induced crystallisation of the Sb2S3, it was possible to control interference in the MZI and achieve 18 dB on/off switching at 1540~nm. Laser reamorphisation and reversible switching of the Sb2S3 layer resulted in damage to the waveguide structure. However, simulations show that a transition metal di-chalcogenide (TMDC) 2-dimensional crystal layers can act as efficient thermal barriers that prevent thermal damage to the waveguide during laser amorphisation. Therefore, combining Sb2S3 with InP waveguides seems to be a feasible approach to achieve low-loss reprogrammable waveguide components for on-chip photonics routing and neural networks.
Article
Full-text available
We first report the temperature phonon properties of a large-area thin film (lateral size: ∼mm, thickness ∼50 nm) made from WS2 mono and few layers deposited on the SiO2/Si substrate. We show that temperature phonon properties probed by Raman spectroscopy significantly differ for bulk and thin-film systems in a temperature range of 90–450 K; however, both exhibit strong temperature-dependent phonon energy nonlinear behavior. Employing the optothermal Raman method, we also estimate thin-film thermal conductivity (κ = 4.3 W/(m K)) and thermal interface conductance (g = 5.5 MW/(m² K)) at room temperature. We find that the thermal properties of thin films (κ and g) are 1 order of magnitude lower than that for bulk or monolayer WS2, which is caused by a reduction in flake size and an increase in boundary scattering. This finding is important for heat management in future applications of thin films made from two-dimensional (2D) nanoflakes.
Article
Full-text available
Active metasurfaces promise reconfigurable optics with drastically improved compactness, ruggedness, manufacturability and functionality compared to their traditional bulk counterparts. Optical phase-change materials (PCMs) offer an appealing material solution for active metasurface devices with their large index contrast and non-volatile switching characteristics. Here we report a large-scale, electrically reconfigurable non-volatile metasurface platform based on optical PCMs. The optical PCM alloy used in the devices, Ge2Sb2Se4Te (GSST), uniquely combines giant non-volatile index modulation capability, broadband low optical loss and a large reversible switching volume, enabling notably enhanced light–matter interactions within the active optical PCM medium. Capitalizing on these favourable attributes, we demonstrated quasi-continuously tuneable active metasurfaces with record half-octave spectral tuning range and large optical contrast of over 400%. We further prototyped a polarization-insensitive phase-gradient metasurface to realize dynamic optical beam steering. An electrically reconfigurable optical metasurface using a Ge2Sb2Se4Te phase change material shows half an octave spectral tuning and promising performances for optical beam steering applications.
Article
Full-text available
High switching current density has been a key bottleneck for phase change memory (PCM) technology. Here, we demonstrate interfacial thermoelectric heating (TEH) as a promising way of tackling this challenge. We use TEH induced by a thin Bi <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> Te <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> interfacial layer to demonstrate ~2× reduction of reset current density (J <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">reset</sub> ) and power (Preset) compared to control PCM devices based on Ge <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> Sb <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> Te <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">5</sub> (GST). Measurements of polarity-dependent reset current and power in well-cycled devices reveal the strong TEH caused by the Bi <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> Te <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> interfacial layer. The TEH origin of J <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">reset</sub> reduction is further confirmed by electrothermal simulations. Such TEH-engineered PCM devices are scalable with the bottom electrode diameter and thus could be promising for high density data storage applications.
Article
Full-text available
The temperature-dependent ($T$ = 4.5 - 500 K) optical constants of monolayer MoS$_2$, MoSe$_2$, WS$_2$, and WSe$_2$ were investigated through spectroscopic ellipsometry over the spectral range of 0.73 - 6.42 eV. At room temperature, the spectra of refractive index exhibited several anomalous dispersion features below 800 nm and approached a constant value of 3.5-4.0 in the near-infrared frequency range. With a decrease in temperature, the refractive indices decreased monotonically in the near-infrared region due to the temperature-dependent optical band gap. The thermo-optic coefficients at room temperature had values from 6.1 $\times$ 10$^{-5}$ to 2.6 $\times$ 10$^{-4}$ K$^{-1}$ for monolayer transition metal dichalcogenides at a wavelength of 1200 nm below the optical band gap. The optical band gap increased with a decrease in temperature due to the suppression of electron-phonon interactions. On the basis of first-principles calculations, the observed optical excitations at 4.5 K were appropriately assigned. These results provide basic information for the technological development of monolayer transition metal dichalcogenides-based photonic devices at various temperatures.
Article
Full-text available
Thermal transport properties bear a pivotal role in influencing the performance of phase change memory (PCM) devices, in which the PCM operation involves fast and reversible phase change between amorphous and crystalline phases. In this paper, we present a systematic experimental and theoretical study on the thermal conductivity of GeTe at high temperatures involving fast change from amorphous to crystalline phase upon heating. Modulated photothermal radiometry (MPTR) is used to experimentally determine thermal conductivity of GeTe at high temperatures in both amorphous and crystalline phases. Thermal boundary resistances are accurately taken into account for experimental consideration. To develop a concrete understanding of the underlying physical mechanism, rigorous and in-depth theoretical exercises are carried out. For this, first-principles density functional methods and linearized Boltzmann transport equations (LBTE) are employed using both direct and relaxation time based approach (RTA) and compared with that of the phenomenological Slack model. The amorphous phase experimental data has been described using the minimal thermal conductivity model with sufficient precision. The theoretical estimation involving direct solution and RTA method are found to retrieve well the trend of the experimental thermal conductivity for crystalline GeTe at high temperatures despite being slightly overestimated and underestimated, respectively, compared to the experimental data. A rough estimate of vacancy contribution has been found to modify the direct solution in such a way that it agrees excellently with the experiment. Umklapp scattering has been determined as the significant phonon-phonon scattering process. Umklapp scattering parameter has been identified for GeTe for the whole temperature range which can uniquely determine and compare Umklapp scattering processes for different materials.
Article
Full-text available
This article reports the effects of gamma radiation on the structural, optical and magnetic properties of monolayer tungsten disulfide (WS2) grown by a scalable Van der Waals epitaxial (VdWE) process on SiO2 coated Si substrate. We found that ionizing radiation (gamma ray) interacts strongly with two-dimensional WS2, which induces effectively p-doping in the samples. As the radiation dose increases, the p-doping concentration increases substantially. In addition, in the small radiation dose regime, the WS2 monolayers exhibit usual diamagnetic behavior. However, a remarkable ferromagnetic hysteresis emerges when the WS2 monolayer is irradiated with 400 Gys. It is attributed to the presence of irradiation-induced complex vacancies composed of one tungsten and a pair of its nearby sulfurs. Moreover, these results have shown that the detector based on the large scale monolayer VdWE-grown two-dimensional WS2 is an appealing candidate for sensing high-energy photon at small radiation doses.
Article
The unique optical properties of phase change materials (PCMs) can be exploited to develop efficient reconfigurable photonic devices. Here, we design, model, and compare the performance of programmable 1 × 2 optical couplers based on: Ge2Sb2Te5, Ge2Sb2Se4Te1, Sb2Se3, and Sb2S3 PCMs. Once programmed, these devices are passive, which can reduce the overall energy consumed compared to thermo-optic or electro-optic reconfigurable devices. Of all the PCMs studied, our ellipsometry refractive index measurements show that Sb2S3 has the lowest absorption in the telecommunications wavelength band. Moreover, Sb2S3 -based couplers show the best overall performance, with the lowest insertion losses in both the amorphous and crystalline states. We show that by growth crystallization tuning at least four different coupling ratios can be reliably programmed into the Sb2S3 directional couplers. We used this effect to design a 2-bit tuneable Sb2S3 directional coupler with a dynamic range close to 32 dB. The bit-depth of the coupler appears to be limited by the crystallization stochasticity. © 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
Article
We elucidate the dependence of the in-plane and interfacial thermal conduction of two-dimensional (2D) transition-metal dichalcogenide (TMDC) materials (including MoS2, WS2, and WSe2) on the materials’ physical features, such as size, layer number, composition, and substrates. The in-plane thermal conductivity k is measured at suspended 2D TMDC materials and the interfacial thermal conductance g is measured at materials supported on substrates, both through Raman thermometry techniques. The thermal conductivity k increases with the radius R of the suspended area following a logarithmic scaling as k∼log(R). k also shows a substantial decrease from monolayer to bilayer, but only changes slightly with a further increase in the layer number. In contrast, the interfacial thermal conductance g has a negligible dependence on the layer number, but g increases with the strength of the interaction between 2D TMDC materials and the substrate, substantially varying among different substrates. The result is consistent with theoretical predictions and clarifies much inconsistence in the literature. This work provides useful guidance for thermal management in 2D TMDC materials and devices.
Article
The rapid increase in information in the big-data era calls for changes to information-processing paradigms, which, in turn, demand new circuit-building blocks to overcome the decreasing cost-effectiveness of transistor scaling and the intrinsic inefficiency of using transistors in non-von Neumann computing architectures. Accordingly, resistive switching materials (RSMs) based on different physical principles have emerged for memories that could enable energy-efficient and area-efficient in-memory computing. In this Review, we survey the four physical mechanisms that lead to such resistive switching: redox reactions, phase transitions, spin-polarized tunnelling and ferroelectric polarization. We discuss how these mechanisms equip RSMs with desirable properties for representation capability, switching speed and energy, reliability and device density. These properties are the key enablers of processing-in-memory platforms, with applications ranging from neuromorphic computing and general-purpose memcomputing to cybersecurity. Finally, we examine the device requirements for such systems based on RSMs and provide suggestions to address challenges in materials engineering, device optimization, system integration and algorithm design. Resistive switching materials enable novel, in-memory information processing, which may resolve the von Neumann bottleneck. This Review focuses on how the switching mechanisms and the resultant electrical properties lead to various computing applications.