Title: Atomic and electronic manipulation of ferroelectric polymorphs

Authors: Yonas Assefa Eshete1, Kyungrok Kang1, Seunghun Kang2, Yunseok Kim2, Suyeon
Cho3*, Heejun Yang4*

Affiliations:
1
Department of Energy Science, Sungkyunkwan University, Suwon 16419, Korea
2
School of Advanced Materials Science and Engineering, Sungkyunkwan University, Suwon
16419, Korea
3
Department of Chemical Engineering and Materials Science, Ewha Womans University, Seoul,
03760, Korea
4
Department of Physics, Korea Advanced Institute of Science and Technology (KAIST),
Daejeon 34141, Korea
*Correspondence to: [email protected] (S.C.), [email protected] (H.Y.)

Abstract
Polymorphism allows to design the symmetry of the lattice and spatial charge distribution of
atomically-thin materials. While distinct structural phases for superconducting, magnetic, and
topological characteristics have been extensively studied, polymorphic control challenges for
robust ferroelectricity in atomically-thin geometry. Here, we report on atomic and electric
manipulation of ferroelectric polymorphs in MoxW1-xTe. The atomic manipulation for the
polymorphic control via chemical pressure (i.e., substituting W for Mo atoms) and charge
carrier density realizes tunable polar lattice structures and robust ferroelectricity in the
atomically-thin geometry of the material. Effective breaking of the inversion symmetry and
hardening phonons, the ferroelectricity remains with a charge carrier density of up to 1.1×1013
cm-2, revealing the original phase diagram for ferroelectricity in atomically thin materials.

One sentence summary

The atomic and electronic manipulation of ferroelectric polymorphs in MoxW1-xTe2 exhibits
robust charge polarization and reveals ferroelectric phase diagram in atomically-thin materials.
Main Text
Polymorphism is a versatile aspect of layered transition metal dichalcogenides (TMDs), which
allows unprecedented electronic and energy devices with atomically-thin geometry (1-4).
Along with the most stable phase of each TMDs at a certain temperature and pressure, electric
doping, chemical pressure, and mechanical strain have demonstrated unique designs of
materials (i.e., polymorphs) for target applications (5). The polymorphic engineering can
modify the symmetry of lattices and electronic structures of TMDs with a bulk and atomically-
thin geometry (6), which is distinct from trivial defect formations or local electronic alterations
with non-TMD materials. Among various TMDs, MoTe2 is an attractive material that possesses
diverse structural and electronic phases covering a semiconducting, superconducting, and
topological semimetallic phases (polymorphs) with centrosymmetric (above T=250 K) and
non-centrosymmetric polar structures (below T=250 K) (3-5,7). The critical temperature
(T=250 K) for complete control of the atomic and electronic symmetries of the material
remains to be overcome for ferroelectric device applications at room temperature (8-10).

Achieving ferroelectricity in atomically-thin materials has been a long-standing issue in

physics and industry-level research. While highly integrated ferroelectric devices (e.g.,
nonvolatile memory, transistors, and sensors) require the electric polarity in extremely small
dimension, electrostatic depolarizing fields (11), charge screening (12), and surface-
reconstruction (13) fundamentally obstruct the appearance of the ferroelectricity in atomically-
thin materials (14,15). Several solutions, such as substrate engineering and charge doping, have
been applied to conventional ferroelectric materials in their thin film geometry (16,17),
however, dangling bonds and indistinct crystal structures of thin ferroelectric materials hinder
to study the ferroelectricity with atomically-thin geometry.

Layered polymorphic TMDs provide a breakthrough for atomically-thin ferroelectric materials.

The absence of dangling bonds, and the ability to exfoliate down to the atomic limit, stabilized
two-dimensional (2D) structures with polar lattice symmetry have demonstrated
unconventional ferroelectricity (Ref). Unprecedented device operations, such as the current
flow directly through ferroelectric materials and the gate control of the current, have been
reported with 2D materials, where inhomogeneous defects and small electric polarization
prevent systematic and practical ferroelectric devices. Thus, investigating new physical
parameters and phase diagram for the unconventional ferroelectricity are required to
fundamentally understand and utilize the emerging 2D ferroelectricity.
We report on the atomic and electronic manipulation of ferroelectric polymorphs in MoxW1-
xTe2, which achieves robust and gate-tunable ferroelectric switching in the atomically-thin
material above room temperature. Hexagonal, monoclinic, and orthorhombic phases are
involved in the polymorphic control without unstable and inhomogeneous defects. The
substitution of tungsten for molybdenum atoms in MoxW1-xTe2 plays a role for chemical
pressure and electric doping that allow robust ferroelectricity in the atomically-thin geometry
via breaking of inversion symmetry and phonon hardening above T=350 K. To reveal the
unique nature of the ferroelectricity, X-ray diffraction, X-ray absorption spectroscopy,
piezoresponse force microscopy (PFM), ferroelectric field-effect transistor measurements were
conducted in this study. Thus, we discovered a critical carrier density of 1.1×1013 cm-2 to
overcome the electric screening limit, manifesting the original phase diagram for
ferroelectricity in 2D materials.

Results and discussion

Mo1-xWxTe2 system exhibit various crystal structures ranging from non-polar to polar phases
together with intriguing physical and chemical properties, that are tunable by carrier density
and strain [19]. In a series of Mo1-xWxTe2 alloy, there are two non-polar phases; hexagonal
(2H) and monoclinic (1T’) structure and one polar phase; orthorhombic (Td) structure and two
phase transitions of 2H-1T’ and 1T’- Td, occurred at room temperature. [20]. The nonpolar 2H
and 1T’ phases have centrosymmetric crystal structures with two-fold screw inversion centers,
in general, no spontaneous electric polarization would be observed. On the contrary to 2H and
1T’, orthorhombic (Td) phase has intrinsic ferroelectricity arisen from non-centrosymmetric
crystal structure. As shown in the Figure 1a, ferroelectric Td phase could be realized in Mo1-

xWxTe2 system with optimal W contents (0.4 ≤ x ≤ 1). Td-Mo1-xWxTe2 triggers polarization

switching which is manifested via small interlayer sliding of the middle layer[20,21] that
stabilizes pair of electric polarization states (red and blue arrows for up and down states),
therefore, we can obtain a ferroelectric hysteresis as applying an electric field across Td-Mo1-
xWxTe2. The transition between non-polar 1T’ and polar Td phase at x = 0.4 in Mo1-xWxTe2
alloy can be thought as a paraelectric-to-ferroelectric transition via symmetry breaking with
atomic manipulation.
A series of stoichiometric compound Mo1-xWxTe2 was prepared with a form of single crystal
via chemical vapor transport (CVT) method. The synthetic procedure was adopted from
reported articles (Ref) and the detailed conditions are described in the method section. The as-
grown single crystals with a composition of Mo1-xWxTe2 were characterized using X-ray
absorption spectroscopy (XAS), X-ray diffractometer (XRD) and Raman spectroscopy (see
Figure S3 and S4 in the supplementary). The crystal structures of Mo1-xWxTe2 were determined
by temperatures and W contents and this trend is consistent with reported articles [21]. It was
reported that pristine MoTe2 experiences the phase transition from Td to 1T’ around T=250 K.
We determined the critical temperature (Ttr) for the phase transition from Td to 1T’ using the
temperature-dependent resistivity measurements of Mo1-xWxTe2 single crystals and found the
critical W content (xC) to have Td phase at room temperature.

As shown in Figure 1b, the temperature-dependent resistivity of Mo1-xWxTe2 single crystals

with various W contents between 0.1 and 0.5 was recorded during cooling and heating
procedure. The Mo1-xWxTe2 (with 0.1 ≤ x ≤ 0.5) showed the resistivity anomaly with thermal
hysteresis, that are associated with first-order structural phase transition between 1T’ and Td
phase at the transition temperature (Ttr) [25]. The values of Ttr as a function of W content (x)
were extracted from the first derivative curves of temperature-dependent resistivity (see Figure
SXX in the supplementary). The Ttr exceeds room temperature above x = 0.4 (Figure SXX
inset) and the room-temperature ferroelectricity would be achieved above x = 0.4 by converting
nonpolar 1T’ phase to polar Td phase in Mo1-xWxTe2.

However, ferroelectricity rarely found in conventional metals because screening effect prevents
to manifest a spontaneous electric polarization. The temperature-dependent resistivity of Mo1-
xWxTe2 (with 0.1 ≤ x ≤ 0.5) in Figure 1b clearly shows a semimetallic behavior and the
resistivity decreases as increasing W contents and this semimetallic characters would obstruct
to achieve a 2D ferroelectricity. From Hall measurements, we confirmed that the majority
carriers are electrons and the electron density changed as a function of x. The carrier mobility
slightly decreases with W contents, probably due to increased number of scattering sources as
substituting W atoms in Mo1-xWxTe2 (see Figure SXX). As increasing W contents, the electron
density increases from 7 x 1019 cm-3 to 2 x 1021 cm-3, indicating robust electron doping effect
of W substitution in Mo1-xWxTe2 alloy.
Considering predicted electron density of 2.0 × 1020 cm−3 for the phase transition from 1T’ to
Td in MoTe2[33], it is reasonable that Mo1-xWxTe2 undergoes the phase transition from 1T’ to
Td around 5.0 × 1020 cm−3 (at x = 0.4) that could be a critical electron density (ne) to convert
1T’ to Td at room temperature in Mo1-xWxTe2 system. The electron density exceeds ~1021 cm-
3
with high W content (x ≥ 0.5), then, the ferroelectricity can be facilitated even above room

temperature by stabilizing high temperature polar Td phase via substituting optimal amount of
W in Mo1-xWxTe2.

To verify the ferroelectricity at room temperature in Td-Mo1-xWxTe2, we conducted PFM

measurements on exfoliated Mo0.5W0.5Te2 with the thicknesses of mono-, bi- and tri-layers.
Figure 1c shows that the ferroelectric switching appears at bi- and tri-layered Mo1-xWxTe2
while no ferroelectric switching is found in mono-layered Mo1-xWxTe2, which is consistent to
the case of Td (or 1T’) monolayer that has no spontaneous polarization because of an inversion
symmetry [Ref]. Therefore, the thinnest ferroelectric materials can be achieved in bi-layered
Td-Mo0.5W0.5Te2 at room temperature.

To fabricate the ferroelectric device with the thinnest geometry, bi-layered Td-Mo0.5W0.5Te2
was used as a ferroelectric material, which was sandwiched by top and bottom dielectric h-BN
layers with the thicknesses of dt and db as shown in Figure 2a. The gate voltages of top (Vt)
and bottom (Vb) voltage were applied with effective vertical electric field (E⊥ ) across the
ferroelectric material. The vertical electric field triggers polarization switching manifested via
small interlayer sliding (20 pm) [20,21] that stabilizes pair of electric polarization states (up and
down) as shown in Figure 1a. Optical image of the bilayer device for conductance and hall
measurement is shown in Figure 2b.

The conductance (G) of bilayer Mo0.5W0.5Te2 with Td structure was collected as E⊥ is swept up
and down at different temperatures as shown in Figure 2c. In both sweep directions, the
conductance monotonically increases and suddenly drops at a certain critical electric field
Ec (up) and Ec (down) , demonstrating two stable and switchable polarization states: up (red
arrow) and down (blue arrow) electric polarization of Td-Mo0.5W0.5Te2. Asymmetric
ferroelectric switching appears with different critical electric fields for up and down electric
polarization probably due to the asymmetric dielectric properties of h-BN. Considering that
Td-WTe2 or defective 1T’-MoTe2 shows a weak ferroelectric switching, it is noteworthy that
reliable ferroelectric switching is observed in our Td-Mo0.5W0.5Te2 at high temperatures up to
T=400 K (see supporting Fig. s9).

To study the effect of external doping levels on the polarization switching, we plot the
conductance of the ferroelectric device at room temperature as functions of bottom gate voltage
and vertical electric field as shown in Figure 2d and 2e. Due to low carrier density of ultrathin
Td-Mo0.5W0.5Te2 device, the carrier density can be modulated via electrostatic gating. We found
that the polarization switching appeared as changing top gate voltage (Vt) from negative to
positive value, indicating that the ferroelectricity of Td-Mo0.5W0.5Te2 bilayer stabilized within
a certain range of external doping. As we discussed before, the electron density of Td-
Mo0.5W0.5Te2 exceeds ~1021 cm-3 that is 5-10 times higher than critical electron density for the
phase transition from 1T’ to Td. Therefore, effective electron and hole doping via electrostatic
gating on Td-Mo0.5W0.5Te2 bilayer might cause the reversible phase transition between 1T’ and
Td, leading to the electrostatic gating induced the transition between paraelectric (1T’)-
ferroelectric (Td) phase via small interlayer sliding as shown in Figure 1a.

We note that the critical electric fields (Ec) for up and down electric polarization are almost
invariant under vertical electric fields (E) as varying the carrier (electron and hole) doping with
various top gate bias. This indicates that the carrier doping in Td-Mo0.5W0.5Te2 bilayer does not
affect the critical field for the electric polarization switching even though the electric
polarization can be stabilized over considerable electron and hole doping range.

The corresponding contour plot of conductance difference (ΔG) between the two-sweep
directions (i.e., forward and reverse sweep directions in Figure 2d) is shown in Figure 3a.
The nonzero conductance modulation marked by red and blue colors denotes the conductance
difference (ΔG) of the hysteretic region during up and downward polarization switching,
respectively. The inset in Figure 3a shows the line profile of the polarization switching taken
from the black dotted line.

The white dotted line extends across the stable polarization region (hysteretic region) and
unstable (non-hysteretic region) polarization range. To estimate the range of critical carrier
concentration of stable ferroelectric switching in the Td-Mo0.5W0.5Te2 device, we plotted the
conductance difference (G) along the white dotted line as a function of the calculated external
εhBNε0 (𝑉𝑏/𝑑𝑏+𝑉𝑡/𝑑𝑡)
doping concentration (ninduced = ) in Figure 3b. We checked the change of
𝑒
carrier density in the electrostatically gated Td-Mo0.5W0.5Te2 device using Hall measurements.
As shown in the increases of electron density denoted by red lined spheres in Figure 3b, Td-
Mo0.5W0.5Te2 bilayer has a gate induced charge density as varying Vt and Vb.

The ferroelectric switching is observed in an induced carrier doping range from an electron

density of 4.1 ⅹ 1012 cm-2 to a hole density of 8.2 ⅹ1012 cm-2 at room temperature as shown

in Figure 3b. Considering that Td phase becomes more stabilized with electron doping, this
dome-shaped diagram for the ferroelectricity overall the wide range of electron and hole doping
indicates that the gate-inducted ferroelectricity of Td-Mo0.5W0.5Te2 bilayer is hardly related
with the phase transition between 1T’ and Td. Therefore, we claim that the ferroelectric
switching in bi-layered Mo0.5W0.5Te2 retains polar Td phase under electrostatic gating. A
stronger polarization magnitude with a ΔG ≈ 9μS is observed at a (induced) hole concentration
of 5×1012 cm-2 while a weaker polarization (ΔG ≈ 4.5μS) is observed at the zero induced carrier
density (ninduced=0). The Hall measurement show that our Td-Mo0.5W0.5Te2 is a n-type semimetal
with electrons as majority carriers that can screen and suppress the electric polarization. The
electric gating modulates the charge carriers in the material whose Fermi level is modulated as
shown in the inset of Figure 3b. As the temperature decreases, the ferroelectric switching range
(in the domain of induced charge density) becomes more symmetric as shown in the Figure 3c,
which is ascribed to the temperature-dependent Fermi level modulation observed in Figure S13.

The carrier concentration to overcome electric screening and trigger ferroelectricity at various
temperatures is summarized in the phase diagram shown in Figure 3d. The ferroelectric phase
arises in a wider carrier concentration range at a lower temperature; the ferroelectricity appears
from a hole concentration of 1.1×1013 cm-2 to an electron concentration of 8.8×1012 cm-2 at
T=1.7 K. The carrier concentration range for ferroelectric Td-Mo0.5W0.5Te2 becomes narrower
as we increase the temperature, and the ferroelectricity disappears above T=450 K. This is
consistent with the structural phase transition temperature between 1T’ and Td shown in Figure
1b. The ferroelectric polymorph of Mo0.5W0.5Te2 could be clearly demonstrated in the phase
diagram in terms of temperature and carrier concentration inside the material. Polymorphs,
atomically-thin, hardening phonon, inversion symmetry breaking, atomic and electronic
manipulation of ferroelectric polymorphs

References
[1] P. Zurcher, R. E. Jones, P. Y. Chu, D. J. Taylor, B. E. White, S. Zafar, B. Jiang, Y.-J. T.
Lii, S. J. Gillespie, IEEE Transactions on Components, Packaging, and Manufacturing
Technology: Part A 1997, 20, 175.
[2] A. I. Khan, A. Keshavarzi, S. Datta, Nature Electronics 2020, 3, 588.
[3] K. Mistewicz, Journal of Nanomaterials 2018, 2018, 2651056.
[4] H. Zhang, T. Wei, Q. Zhang, W. Ma, P. Fan, D. Salamon, S.-T. Zhang, B. Nan, H. Tan,
Z.-G. Ye, Journal of Materials Chemistry C 2020, 8, 16648.
[5] G. Fiori, F. Bonaccorso, G. Iannaccone, T. Palacios, D. Neumaier, A. Seabaugh, S. K.
Banerjee, L. Colombo, Nature Nanotechnology 2014, 9, 768.
[6] J. Junquera, P. Ghosez, Nature 2003, 422, 506.
[7] Y. Zhang, G.-P. Li, T. Shimada, J. Wang, T. Kitamura, Physical Review B 2014, 90,
184107.
[8] D. Zhao, T. Lenz, G. H. Gelinck, P. Groen, D. Damjanovic, D. M. de Leeuw, I. Katsouras,
Nature Communications 2019, 10, 2547.
[9] C. Gattinoni, N. Strkalj, R. Härdi, M. Fiebig, M. Trassin, N. A. Spaldin, Proceedings of
the National Academy of Sciences 2020, 117, 28589.
[10] P. Gao, H.-J. Liu, Y.-L. Huang, Y.-H. Chu, R. Ishikawa, B. Feng, Y. Jiang, N. Shibata,
E.-G. Wang, Y. Ikuhara, Nature Communications 2016, 7, 11318.
[11] D. Pesquera, E. Parsonnet, A. Qualls, R. Xu, A. J. Gubser, J. Kim, Y. Jiang, G. Velarde,
Y.-L. Huang, H. Y. Hwang, R. Ramesh, L. W. Martin, Advanced Materials 2020, 32,
2003780.
[12] R. Liu, J. G. Ulbrandt, H.-C. Hsing, A. Gura, B. Bein, A. Sun, C. Pan, G. Bertino, A.
Lai, K. Cheng, E. Doyle, K. Evans-Lutterodt, R. L. Headrick, M. Dawber, Nature
Communications 2020, 11, 2630.
[13] N. Setter, D. Damjanovic, L. Eng, G. Fox, S. Gevorgian, S. Hong, A. Kingon, H.
Kohlstedt, N. Y. Park, G. B. Stephenson, I. Stolitchnov, A. K. Taganstev, D. v Taylor, T.
Yamada, S. Streiffer, Journal of Applied Physics 2006, 100, 051606.
[14] W. Choi, N. Choudhary, G. H. Han, J. Park, D. Akinwande, Y. H. Lee, Materials Today
2017, 20, 116.
[15] L. Li, M. Wu, ACS Nano 2017, 11, 6382.
[16] W. Wan, C. Liu, W. Xiao, Y. Yao, Applied Physics Letters 2017, 111, 132904.
[17] K. Chang, J. Liu, H. Lin, N. Wang, K. Zhao, A. Zhang, F. Jin, Y. Zhong, X. Hu, W. Duan,
Q. Zhang, L. Fu, Q.-K. Xue, X. Chen, S.-H. Ji, Science 2016, 353, 274.
[18] S. N. Shirodkar, U. v Waghmare, Physical Review Letters 2014, 112, 157601.
[19] J. Park, I. W. Yeu, G. Han, C. S. Hwang, J.-H. Choi, Scientific Reports 2019, 9, 14919.
[20] H. Wang, X. Qian, npj Computational Materials 2019, 5, 119.
[21] Q. Yang, M. Wu, J. Li, The Journal of Physical Chemistry Letters 2018, 9, 7160.
[22] S. Yuan, X. Luo, H. L. Chan, C. Xiao, Y. Dai, M. Xie, J. Hao, Nature Communications
2019, 10, 1775.
[23] P. Sharma, F.-X. Xiang, D.-F. Shao, Z. Dawei, E. Tsymbal, A. Hamilton, J. Seidel,
Science Advances 2019, 5, eaax5080.
[24] Z. Fei, W. Zhao, T. A. Palomaki, B. Sun, M. K. Miller, Z. Zhao, J. Yan, X. Xu, D.
H. Cobden, Nature 2018, 560, 336.
[25] W. G. Dawson, D. W. Bullett, Journal of Physics C: Solid State Physics 1987, 20, 6159.
[26] K. Zhang, C. Bao, Q. Gu, X. Ren, H. Zhang, K. Deng, Y. Wu, Y. Li, J. Feng, S. Zhou,
Nature Communications 2016, 7, 13552.
[27] P. Paufler, Acta Crystallographica Section A - ACTA CRYSTALLOGR A 2004, 60, 641.
[28] H.-J. Kim, S.-H. Kang, I. Hamada, Y.-W. Son, Physical Review B 2017, 95, 180101.
[29] J. M. Wesselinowa, St. Kovachev, Physical Review B 2007, 75, 45411.
[30] Y. Qi, P. G. Naumov, M. N. Ali, C. R. Rajamathi, W. Schnelle, O. Barkalov, M. Hanfland,
S.-C. Wu, C. Shekhar, Y. Sun, V. Süß, M. Schmidt, U. Schwarz, E. Pippel, P. Werner, R.
Hillebrand, T. Förster, E. Kampert, S. Parkin, R. J. Cava, C. Felser, B. Yan, S. A.
Medvedev, Nature Communications 2016, 7, 11038.
[31] H. Sakai, K. Ikeura, M. S. Bahramy, N. Ogawa, D. Hashizume, J. Fujioka, Y. Tokura, S.
Ishiwata, Science Advances 2016, 2, e1601378.
[32] H. Takahashi, T. Akiba, K. Imura, T. Shiino, K. Deguchi, N. K. Sato, H. Sakai, M. S.
Bahramy, S. Ishiwata, Physical Review B 2017, 95, 100501.
[33] C. Yue, J. Hu, X. Liu, A. M. Sanchez, Z. Mao, J. Wei, ACS Nano 2016, 10, 429.
[34] G. F. Nataf, M. Guennou, J. M. Gregg, D. Meier, J. Hlinka, E. K. H. Salje, J. Kreisel,
Nature Reviews Physics 2020, 2, 634.
[35] P. S. Bednyakov, B. I. Sturman, T. Sluka, A. K. Tagantsev, P. v Yudin, npj
Computational Materials 2018, 4, 65.
[36] Z. Zheng, Q. Ma, Z. Bi, S. de la Barrera, M.-H. Liu, N. Mao, Y. Zhang, N. Kiper, K.
Watanabe, T. Taniguchi, J. Kong, W. A. Tisdale, R. Ashoori, N. Gedik, L. Fu, S.-Y. Xu,
P. Jarillo-Herrero, Nature 2020, 588, 71.
[37] K. Yasuda, X. Wang, K. Watanabe, T. Taniguchi, P. Jarillo-Herrero, Science 2021,
372, 1458.
[38] Y. A. Eshete, N. Ling, S. Kim, D. Kim, G. Hwang, S. Cho, H. Yang, Advanced
Functional Materials 2019, 29, 1904504.
Methods

Materials Synthesis and Characterization

A series of stochiometric Mo1−xWxTe2 single crystals were prepared by the chemical vapor
transport (CVT)[38] method using iodine as the transport medium. According to stoichiometry
composition of each specific sample, 99.95 % pure Mo (sigma Aldrich), 99.98 % pure W
(sigma Aldrich) powder and 99.95 % pure Te (sigma Aldrich) pellets were weighted, grinded
and sealed in a vacuum quartz tube together with iodine powder (3 mg; 99.99% purity) (sigma
Aldrich). The quartz tube was kept in the middle of 2 zone CVT furnace with a temperature
gradient of 100 0C (1150 ◦C at one end and the other end was kept at a temperature of 1050 ◦C)
for 11-14 days. The temperature was slowly cooled to room temperature to which could give a
chance to crystalize the 2H phase and obtained gray shiny crystals of the 2H, 1T’ and Td phase
with typical dimensions of several millimeter length (2 to 5 mm). By means of controlling and
optimized the growth condition, large size and high-quality crystals can be obtained.

The bulk and exfoliated thin film of Mo1−xWxTe2 alloy was subjected to several structural and
compositional characterization, The elemental distribution (fig S1) and stoichiometric analysis
(fig S2) of the alloy were performed by energy dispersive X-ray spectroscopy (EDS) analysis
conducted on scanning electron microscope (SEM) (regulus8240). X-ray diffraction (XRD)
measurements were conducted on the single crystals sample using an X-ray diffractometer
(Rigaku, Smart Lab). The X-ray Generator, 9kW and 2θ scanned from 10° to 70° (angle) were
used to collect XRD data. The phonon properties (Raman spectroscopy) of the sample were
studied by confocal Raman spectroscopy (NT-MDT, NTEGRA) using the 531 and 633 nm
laser as an optical source. All magnetotransport, Hall-effect (Hall) and magnetoresistance (MR)
and temperature dependent resistivity (R-T) measurements were performed using quantum
design - physical property measurement system (QD PPMS-14 T) with base temperature of
1.5K and a 14 T magnet. Further composition dependent measurement X-ray absorption
spectroscopy (XAS) was performed to investigate electronic and lattice structure of each
constituent element. Pezoresponse force microscopy (PFM) used to measure ferroelectric
phase and amplitude of thin film Mo1−xWxTe2 alloy at room temperature.

Device Fabrication

We have prepared two different layout of van der Waals heterostructures device of few-layer
Mo0.5W0.5Te2: (1) Mo0.5W0.5Te2 heterostructures device with top and bottom graphite gate (as
shown in fig 2a); (2) WTe2 heterostructures device that top gate replaced with graphene with
four Au contact (as shown in fig:4a inset). Van der Waals heterostructures of few-layer
Mo0.5W0.5Te2, hexagonal boron nitride (bulk Taniguchi’s h-BN) and graphite were fabricated
by the layer-by-layer dry transfer method. In this process, few-layer Mo0.5W0.5Te2, hBN and
graphite were mechanically exfoliated from bulk crystals onto silicon substrates (280nm
oxide). Graphite flakes 3–5 layers thick were chosen for the top and bottom gates and 20-30nm
thick h-BN flakes were chosen for the top and bottom dielectric sheet. Then the device
preparation divides in to two ways (Supporting fig S6): Part-1(bottom part), includes
heterostructure of graphite (bottom gate)/hBN(bottom gate)/Pt contact electrode, The graphite,
hBN and Mo0.5W0.5Te2 flakes were picked up by a stamp consisting of a thin film of
polycarbonate (PC) on polydimethylsiloxane (PDMS) respectively. The stack was then
released onto a Si substrate. The graphite/hBN/Pt stacks were patterned hall bar structure
Exposing A positive e-beam resist, poly(methylmethacrylate) (PMMA) using e-beam
lithography machine (Nanometer Pattern Generation System for JSM-6510, JEOL). The
samples were developed with MIBK (Methyl Isobutyl Ketone) solution. Pt (8 nm) contact
electrodes were deposited on the developed hall bar structure. the heterostructure were
subsequently left in acetone for 30 min to remove the PMMA residues followed by vacuum
annealing at (10-5 torr) for 4 hours at 200 0C. Part-2 (top part), incudes heterostructure of few
layers Mo0.5W0.5Te2/hBN/Graphite. The top graphite and top h-BN was picked up first and
brought into oxygen and water molecule free glove box (Ar field). After exfoliation and
selecting appropriate Mo0.5W0.5Te2 flake inside the glove box, the second part Gr/Hbn
heterostructure was used to pick up the flake and released into pre-patterned Pt contact (on the
first part). Final step of e-beam lithography and metal deposition was used to have electrical
contact of pads Cr/Au (5/50 nm) with the Pt contacts (7 nm) and the top and bottom graphite
gates.

Device measurements and Analysis

All transport measurements of the device were performed using a Heliox 3He cryostat system
(Oxford instruments) and Lakeshore cryogenic prob station. The cryogenic system provides
stable temperatures ranging from 1.5 to 300 K and magnetic fields up to 12 T, or Lakeshore
low temperature hall measurement system with cryogenic probe station (chamber pressure = 5
× 10−5 to10−7 Torr) cooled with closed-circuit refrigerator (CCR) loop of liquid helium and
connected to a Keithley 4200-SCS parameter analyzer. In most device measurement was done
in two and four terminal device structures, both essentially yielded the same results, except
relatively lower conductance was observed in two terminal data, implying the contact
resistance effect. To illustrate this, in Supporting fig S11 penal a and penal b indicate switching
behavior measured with two- and four-terminal configurations, respectively. Temperature
dependent four terminal resistance measurement revealed insulating behavior if the bilayer
device (fig S12).
 a 2H (0 ≤ x < 0.1) 1T’ (0.1 ≤ x < 0.4) Td (0.4 ≤ x ≤ 1)

Ferroelectric
W switching
W

P
c c
E
a a
Te Mo

Centrosymmetric (nonpolar) phase Non-centrosymmetric (polar) phase

b c
Td 1T’ x = 0.1 Mo1-xWxTe2 with x = 0.5 4

Amplitude (pm/V)
1.0 200
Resistivity (mW·cm)

0.2 Phase (o) 3L 3L 3

0.3
100 1L
0.5 0.4
2L 2
0.5

0 2L 1
Mo1-xWxTe2
0
0 100 200 300 400 -2 0 2 -2 0 2
Temperature (K) Voltage (V) Voltage (V)

Figure 1. Tunable polar structural phase transitions in MoTe2. a, Evolution of crystal

structure in the Mo1-xWxTe2 alloy as a function of W content (x). Shaded blue rectangles denotes
the unit cell of each crystal structure and the two-fold screw inversion centers were marked in
2H and 1T’ structure. The polarization switching in Td phase manifested via interlayer sliding
of the middle layer relative to its non-ferroelectric position (highlighted in shaded structure).
The red (or blue) arrows represent the net polarization with up (or down) state.
b, The temperature-dependent resistivity of Mo1-xWxTe2 (0.1 ≤ x ≤ 0.5) was recorded during
cooling and heating procedure. The structural phase transition between 1T’ and Td undergoes
at a critical temperature marked as colored inverted triangles. c, Thickness dependent PFM
phase (left penal) and amplitude (right penal) hysteresis loop during the switching process. In
the ferroelectric hysteresis loop, strong polarization observed in the trilayer sample (black
curve) demonstrating complete 1800 phase reversal whereas ferroelectric switching vanished
in the monolayer sample which is consistent with recovery of inversion symmetry. The pink
arrows indicate the direction of the hysteresis loop.
 a c d
1100 300 K Vt = 3V
Ef

Condactance (S)
1090 2V
350 K
t-Graphite 1V
1080 330 K 0V
dt t-hBN Vt -1V
1070
-2V
300 K 1050
E Td-Mo0.5W0.5Te2 -3V
1060

Conductance (mS)
-4V
-5V
1050
db -8 -4 0 4 8
b-hBN
Vb
77 K
e Vb(V)
1100
b-Graphite Vt = 4V

Condactance (S)
3V
b 30
2V
1V
5 μm
0V
-1V
1.7 K -2V
-3V
1050 -4V
20 -5V

-6V
-0.1 0 0.1
E V/nm) -0.2 0.0 0.2
E (v/nm)

Fig. 2. Ferroelectric switching in Td phase Mo0.5W0.5Te2. a, Schematic cross-sectional view

of device geometry. bottom and top dielectric h-BN and graphite gate is used to apply vertical
𝑉𝑏/𝑑𝑏−𝑉𝑡/𝑑𝑡
electric field ( E⊥ = ) through encapsulated bilayer Td-Mo0.5W0.5Te2 flake. b,
2

Optical image of representative device, red shaded area marks the flake. c, Conductance of

undoped Mo0.5W0.5Te2 a device as a function of vertical electric field, E⊥ swept up (positive)

and dawn(negative). black arrows show the conductance loop, showing bistability (opposite
slop). The red (or blue) arrows represent the net polarization with up (or down) state.
d, Conductance 𝐺 of Mo0.5W0.5Te2 in dual gate function of Vb and Vt (Vb kept as sweep bias
and Vt as step bias) at room temperature. Forward and reverse sweep direction shown pink
arrow whereas the polarization a switching shown in the red and blue arrow. Inset shows
schematic electronic band diagram of n-type Td-Mo0.5W0.5Te2. e, Conductance of
Mo0.5W0.5Te2 in dual gate function (the top penal, Fig. 2d) plotted as function of electric field

E⊥.
 a b
ΔG(μS) 12
8
10
5 Ef 6
G(S)

n (X1012cm-2)
0
8

G (S)
-4
-6 -4 -2 0 2 4 6
Vt (V)

Vb (V)

0 0 Ef 3
Ef
4
0

-5
-8 0 0
-10 -5 0 5
-5 0 5 External doping, (1012cm-2)
Vb(V)

c 300 K d
77 K
20 1.7 K 400
Temperature ()

15
G (S)

10 200 Non- FE
Ferroelectric FE
5

0
-10 -5 0 5 10 -12 -6 0 6 12
12 -2
External doping (1012cm-2) External doping (10 cm )

Fig. 3. Gate tunable ferroelectric switching in Td-MoTe2 at Room temperature. a, Contour

plot of conductance difference(ΔG) between the two-sweep direction (forward and reverse,
pink arrow in Fig. 2d ) in a dual gated device. Dashed contour lines drawn across gate range of
stable polarization (white line) and polarization switching (black line) respectively. The inset
shows the conductance deference profile (ΔG) for upward and downward polarization (taken
from the black contour line). b, The Conductance difference profile (black curve, taken from
white contour line in Fig. 3b) and total carrier concentration (red curve) as a function of
externally doped carrier concentration via electrostatic gating. The purple double arrow
indicates stable polarization range as function of external doping. The inset shows schematic
band diagram of fermi level modulation as function of external carrier doping. c, The
Conductance difference profile (ΔG) at different representative temperatures, 300K (black
curve), 77K (red curve) and 1.7K (blue curve). The green dashed arrow indicates the peak
polarization states of each respective temperature. The light red and green shaded area
represents hole and electron doping regions, respectively. d, Ferroelectricity phase diagram as
a function of carrier concentration(doping). The green shaded area indicates the doping range
of stable ferroelectricity. In the inset, Electron (right) and hole (left) doping band diagram.
Fig. 4. Theoretical insight of gate tunable ferroelectric switching in Td-Mo0.5W0.5Te2.