ARTICLES

PUBLISHED ONLINE: 23 JUNE 2013 | DOI: 10.1038/NMAT3687

Mobility engineering and a metal–insulator

transition in monolayer MoS2
Branimir Radisavljevic and Andras Kis*

Two-dimensional (2D) materials are a new class of materials with interesting physical properties and applications ranging from
nanoelectronics to sensing and photonics. In addition to graphene, the most studied 2D material, monolayers of other layered
materials such as semiconducting dichalcogenides MoS2 or WSe2 are gaining in importance as promising channel materials for
field-effect transistors (FETs). The presence of a direct bandgap in monolayer MoS2 due to quantum-mechanical confinement
allows room-temperature FETs with an on/off ratio exceeding 108 . The presence of high-κ dielectrics in these devices enhanced
their mobility, but the mechanisms are not well understood. Here, we report on electrical transport measurements on MoS2
FETs in different dielectric configurations. The dependence of mobility on temperature shows clear evidence of the strong
suppression of charged-impurity scattering in dual-gate devices with a top-gate dielectric. At the same time, phonon scattering
shows a weaker than expected temperature dependence. High levels of doping achieved in dual-gate devices also allow the
observation of a metal–insulator transition in monolayer MoS2 due to strong electron–electron interactions. Our work opens up
the way to further improvements in 2D semiconductor performance and introduces MoS2 as an interesting system for studying
correlation effects in mesoscopic systems.

range2,10 . Charge traps25 present at the interface between the

M
olybdenum disulphide (MoS2 ) is a layered transition-
metal dichalcogenide semiconductor1 with potential substrate and the MoS2 layer have recently been proposed as the
applications that could complement those of graphene. dominant cause for such low room-temperature mobility in MoS2
As neighbouring layers in transition-metal dichalcogenide crystals devices. Understanding the origin of this mobility degradation
are weakly bound through van der Waals interactions, single and finding a way to restore the mobility to bulk values or even
atomic crystals composed of one or several layers can be extracted further enhance it would allow us to unlock the full technological
using the micromechanical cleavage technique2 or liquid-phase potential of this material.
exfoliation3,4 . Large-area MoS2 can also be grown using techniques The encapsulation of monolayer MoS2 in a high-κ dielectric
such as chemical vapour deposition5,6 . The strong covalent bonding environment26 was shown to result in an increase of the room-
between metal and chalcogenide atoms results in a high mechanical temperature mobility10 . This was tentatively assigned to reduced
strength7 of MoS2 membranes8 and electrical breakdown current Coulomb scattering due to the high-κ dielectric environment26 and
densities at least 50 times higher than in copper9 . In contrast possible modification of phonon dispersion in MoS2 monolayers.
to graphene, the presence of a bandgap in monolayer MoS2 and An increase of mobility with the dielectric deposition, similar to
other semiconducting dichalcogenides allows the fabrication of that in monolayers, was also observed in multilayer samples27,28
transistors that can be turned off and used as switches10 . Logic and monolayer samples with polymer gating29 . Previous mobility
circuits11 and amplifiers12 with high gain based on monolayer estimates for monolayer MoS2 are however based on two-contact
MoS2 have also been demonstrated, and superconductivity in measurements and lack the information on their temperature
20-nm-thick MoS2 was achieved at high electron concentrations dependence. More accurate measurements are needed to gain a
using ionic-liquid gating13 . better understanding of the various mechanisms that could limit
Monolayer MoS2 has electronic and optical properties that are the mobility in monolayer MoS2 . In the phonon-limited high-
fundamentally different from those of thicker layers owing to temperature part, the mobility is expected to follow a µ ∼ T −γ
quantum-mechanical confinement14,15 . Bulk MoS2 is an indirect temperature dependence with γ = 1.69 and mobility reaching
gap semiconductor whereas single-layer MoS2 has a direct gap14–17 . a room-temperature value ∼410 cm2 V s−1 according to first-
The lack of inversion symmetry results in strong coupling of principles calculations30 . The deposition of a top-gate dielectric is
spin and valley degrees of freedom18–20 and could be used in expected to mechanically quench the homopolar phonon mode and
devices based on the valley Hall effect21 . The atomic-scale thickness reduce the coefficient γ to 1.52.
(6.5 Å) of monolayer MoS2 , smaller than the screening length, also Here, we report on mobility measurements in monolayer MoS2
allows a large degree of electrostatic control over the electrical based on the Hall effect. This allows us to remove the effect of
conductivity. Together with the absence of dangling bonds, this contact resistance and also directly measure the gate-modulated
would allow transistors based on monolayer MoS2 to outperform charge density and gate capacitance necessary for the accurate
silicon transistors at the scaling limit22,23 . measurements of the field-effect mobility. Our devices are FETs in
Previous measurements have shown that the room-temperature single- and dual-gate configurations shown in Fig. 1. Degenerately
mobility of bulk MoS2 is in the 200–500 cm2 V s−1 range and is doped Si wafers covered with 270 nm thermally grown SiO2 serve as
limited by phonon scattering24 . Exfoliation of single layers onto the substrate and back-gate. MoS2 flakes are shaped into Hall bars
SiO2 results in a decrease of mobility down to the 0.1–10 cm2 V s−1 using oxygen plasma etching. A 30-nm-thick HfO2 layer deposited

Electrical Engineering Institute, Ecole Polytechnique Federale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland. *e-mail: andras.kis@epfl.ch

NATURE MATERIALS | VOL 12 | SEPTEMBER 2013 | www.nature.com/naturematerials 815

© 2013 Macmillan Publishers Limited. All rights reserved.
ARTICLES NATURE MATERIALS DOI: 10.1038/NMAT3687

a b

Vds
Ids

Vbg

MoS2

Vds Vtg
HfO2
Ids

Vbg

Figure 1 | Fabrication of single-gated and dual-gated MoS2 devices. a, Optical image of the MoS2 dual-gated device used in our measurements. The inset
shows the single-gate version of the same device before ALD deposition of HfO2 and top-gate electrode fabrication. Scale bars, 5 µm. b, Cross-sectional
views of devices based on single-layer MoS2 in a single-gate (top) and dual-gate (bottom) configuration. Gold leads are used for the source, drain and
voltage probes (V1 ,V2 ,V3 and V4 ). Voltage probes have been omitted from the drawing. The silicon substrate, covered with a 270-nm-thick SiO2 layer was
used as the back gate. The top-gate dielectric is a 30-nm-thick HfO2 layer.

by atomic layer deposition (ALD) forms the top-gate dielectric. The at Vbg = 30 V with negligible hysteresis for all temperatures and
optical image of one of our top-gated devices is shown in Fig. 1a. nonlinear behaviour vanishing completely for temperatures above
We have performed measurements on multiple devices in single- 40 K, excluding the possibility of the contact resistance or Schottky
and dual-gate configurations (Fig. 1b; for more details on devices barrier influencing the mobility extraction. Figure 2c shows the
see Supplementary Table S1). By using the top gate we can induce temperature dependence of mobility in this device. Mobility is
stronger electrostatic doping of our monolayer MoS2 owing to the extracted from the conductance curves in the 30–40 V range of
higher dielectric constant and smaller thickness of the HfO2 layer back-gate voltage Vbg , using the expression for field-effect mobility
(εr2 ∼ 19, dox2 (HfO2 ) = 30 nm) compared with the bottom-gate µ = [dG/dVbg ] × [L12 /(WCox1 )]. The temperature dependence is
SiO2 (εr1 ∼ 3.9, dox1 (SiO2 ) = 270 nm). For both types of device, we characterized by a peak at ∼200 K. Below 200 K, we observe a
measure the four-probe conductance defined as G = Ids /(V1 –V2 ), decrease of the mobility as the temperature is lowered down to
where Ids is the drain current and V1 –V2 is the measured voltage 4 K. This behaviour is consistent with mobility limited by scattering
difference between the voltage probes. from charged impurities31 . Increasing the temperature above 200 K
A typical conductance, G, dependence on the gate voltage also results in a strong decrease of the mobility from the peak value
for a single-gate device is shown in Fig. 2a, measured up to of 18 cm2 V s−1 , related to electron–phonon scattering that becomes
the back-gate voltage Vbg = 40 V that corresponds to a charge the dominant scattering mechanism at higher temperatures30 . We
concentration of n2D ∼ 3.6 × 1012 cm−2 calculated using the fit this part of the curve with the generic temperature dependence
parallel-plate capacitor model, with n2D = Cox1 1Vbg /e, where of the mobility µ ∼ T −γ , where the exponent depends on the
Cox1 = ε0 εr1 /dox1 , ε0 = 8.85 × 10−12 F m−1 , e = 1.602 × 10−19 C is dominant phonon scattering mechanism. From the fit we find the
the elementary charge and 1Vbg = Vbg –Vbg,th . The value of value of γ ≈ 1.4, in good agreement with theoretical predictions for
threshold voltage Vbg,th varies for each device and is close to its monolayer MoS2 (γ ≈ 1.69; ref. 30).
pinch-off voltage estimated from the conductance curves. We find We now examine dual-gated devices. Figure 3a shows a typical
that the temperature variation of G in a single-gate monolayer top-gating dependence of the four-contact G and sheet conductivity
device (Fig. 2b), in the high-temperature regime (80 K ≤ T ≤ 280 K), σ , defined as σ = GL12 /W with L12 = 1.55 µm and W = 1.9 µm
can be modelled with thermally activated transport: being the distance between the voltage probes and the device
width, respectively. The use of the top gate allows a higher degree
G = G0 (T )e−Ea /kB T of doping, up to n2D ∼ 3.6 × 1013 cm−2 (Supplementary Fig. S1),
much higher than typical values for single-gated devices (n2D ∼ 4 ×
where Ea is the activation energy, kB is the Boltzmann constant and 1012 cm−2 ). We observe here an insulating behaviour that persists
G0 (T ) is the temperature-dependent parameter extracted from the until Vtg = 2.2 V. At this point, corresponding to n2D ∼ 1×1013 cm−2
fitting curves. The good agreement of the data with the activation (as measured from the Hall effect), monolayer MoS2 enters a
transport model at higher temperatures is suggestive of charge metallic state and the associated metal–insulator transition32 (MIT)
transport that is thermally activated. At temperatures T ≤ 80 K is observed, the first of its kind in a 2D semiconductor (Fig. 3a).
we observe that the variation of G weakens for almost all Vbg This transition manifests itself as a crossing-over between conduc-
values. This can be explained with hopping through localized tance versus gate voltage curves acquired at different temperatures,
states becoming dominant at lower temperatures25 , driving the indicating two different regimes. For gate voltages corresponding to
system into a strongly localized regime. The inset of Fig. 2a shows charge densities smaller than n2D ∼ 1×1013 cm−2 , monolayer MoS2
double sweeps of Ids –Vds characteristics for several temperatures behaves as a classical semiconductor with conductance decreasing

816 NATURE MATERIALS | VOL 12 | SEPTEMBER 2013 | www.nature.com/naturematerials

© 2013 Macmillan Publishers Limited. All rights reserved.
NATURE MATERIALS DOI: 10.1038/NMAT3687 ARTICLES
a 35 T = 4.1 K Vbg = 30 V
to MIT. Fig. 3b shows the temperature dependence of the device
Vds = 100 mV 20 K
400 40 K conductance for different values of the charge density n2D . We
T = 4.1 K 60 K
30 20 K 200
100 K
120 K can see here more clearly that above the critical charge density
200 K

Ids (nA)
60 K
0
260 K
of 1 × 1013 cm−2 , the conductance of monolayer MoS2 increases
25 100 K
140 K ¬200
with decreasing temperature, which is the manifestation of metallic
20
180 K
¬400
behaviour. For charge densities smaller than 1×1013 cm−2 , the con-
G (µS)

220 K
260 K ductance decreases with the temperature, corresponding to semi-
¬0.2 ¬0.1 0.0 0.1 0.2
15 300 K conducting behaviour. This striking feature occurs when the con-
Vds (V)
ductivity is of the order of the quantum conductance e 2 /h, the min-
10 imum of metallic conductivity, which was considered not to exist in
2D electronic systems according to the scaling theory of localization
5
based on non-interacting electronic gases proposed in 197933 .
0 The first step in our analysis is to define the critical point of
¬20 ¬10 0 10 20 30 40 the MIT. Inspecting Fig. 3a, we can see that each two consecutive
Vbg (V) isotherms of G(Vtg ) cross each other at some value of Vtg . These
b 4 intersections are temperature dependent, so an unambiguous
G (Vbg) = G0exp (¬Ea/kT) Vbg
determination of the transition is not possible. Fortunately,
2V
2
5V at temperatures under 80 K, the crossing point seems to be
9V independent of the temperature and emerges at a well-defined
10¬5 13 V
8 20 V point Vtg = 2.2 V, clearly separating the metallic and insulating
6 25 V phases. This transition point is the direct consequence of quantum
30 V
interference effects of weak and strong localization. At lower
G (S)

4
35 V
40 V carrier concentrations (< n2D ∼ 1 × 1013 cm−2 ) the system is in
2
the insulating state and strong localization34 prevails. This charge
density is comparable to that recorded for 20-nm-thick MoS2
10¬6
8 (ref. 13). As the top-gate bias is increased above Vtg = 2.2 V
6
(concentration above n2D ∼ 1×1013 cm−2 ), the system is driven into
4
a metallic phase and weak localization seems to be the dominant
0.004 0.008 0.012 0.016 0.020 effect. The observed quantum critical point of MIT in our devices
1/T (K¬1) is the consequence of a strongly correlated 2D electron gas35 . As the
c 2 system is confined in two dimensions, strong Coulomb interactions
between electrons could cause a large ratio36 rs between potential
(Coulomb, EC ) and kinetic (Fermi, EF ) energy:
EC nv nv m∗ e 2
rs = = ∗√ = √
µ (cm2 V¬1 s¬1)

EF aB πn2D 4πε h̄2 πn2D

10 µ ~ T¬1.4 where nv is the number of degenerate valleys in the spectrum,

9 a∗B = (4πε h̄2 )/(m∗ e 2 ) is the effective Bohr radius, with ε being
8
the dielectric constant and m∗ is the effective electron mass. A
system with rs  1 cannot be considered as non-interacting and
7
the conclusion of scaling theory of localization33 is not valid in
6 this regime. In the case of monolayer MoS2 we obtain rs ≈ 4.2,
3 4 5 6 78 2 3 4 5 6 78 2 3 4 5
10 100 considering m∗ = 0.45mo (ref. 22), ε = 7.3εo (ref. 37), a double-
T (K) degenerate conduction band around the K point (nv = 2) and an
electron concentration at the transition of n2D ∼ 1 × 1013 cm−2 .
Figure 2 | Electron transport in single-gate monolayer MoS2 supported on This value is similar to those previously measured in the case of
SiO2 . a, G as a function of Vbg for a single-gate monolayer MoS2 device GaAs/AlGaAs heterostructures38 (rs ∼ 4–5) and Si metal–oxide–
acquired at different temperatures. The inset shows double sweeps of semiconductor FETs39 (rs ∼ 8). This shows that monolayer MoS2
Ids –Vds characteristics for several temperatures at Vbg = 30 V with is a very attractive 2D system with strong Coulomb interactions,
negligible hysteresis for all temperatures. b, Arrhenius plot of G for different making the high-rs regime easier to reach than in cleaner Si
values of Vbg . Solid lines are linear fits to the data showing activated metal–oxide–semiconductor FETs or n-GaAs-based devices40 . The
behaviour for limited regions of T and Vbg (charge density). c, The reason for this is the relatively high effective mass (for Si: 0.19 mo ;
dependence of µ on T shows a pronounced low-temperature regime n-GaAs: 0.07 mo ) and lower dielectric constant (for Si: 11.7 εo ;
consistent with transport dominated by scattering from charged impurities. n-GaAs: 12 εo ) of monolayer MoS2 .
Above ∼200 K, µ is limited by phonon scattering and follows a µ ∼ T −1.4 We can now also investigate the Ioffe–Regel criterion41–43 for
dependence. Error bars are estimated on the basis of uncertainties in 2D semiconductors, which predicts the existence of a MIT when
determining the voltage drop across the channel. the parameter kF · le satisfies
√ the criterion kF · le ∼ 1, with the
Fermi wave vector kF = 2πn2D , and mean free path of electrons
as the temperature is decreased, whereas for higher gate voltages, le = h̄kF σ /n2D e 2 . According to this criterion, for kF ·le  1 the phase
corresponding to charge densities above n2D ∼ 1 × 1013 cm−2 , the is metallic whereas for kF · le  1, the phase is insulating. For our
conductance increases as the temperature is decreased, which is the device, at the crossing point of Vtg = 2.2 V, we have kF · le ∼ 2.5, in
hallmark of metallic behaviour. In the inset of Fig. 3a are shown good agreement with the theory. Our other devices also exhibit kF ·le
double sweeps of Ids –Vds characteristics for several temperatures at close to 2 (Supplementary Table S1).
Vtg = 0 V with negligible hysteresis for all temperatures, excluding The temperature dependence of the mobility is extracted from
the possibility of hysteresis influencing our conclusions related conductance curves in the Vtg = 1–5 V range, using the expression

NATURE MATERIALS | VOL 12 | SEPTEMBER 2013 | www.nature.com/naturematerials 817

© 2013 Macmillan Publishers Limited. All rights reserved.
ARTICLES NATURE MATERIALS DOI: 10.1038/NMAT3687

a 300 63 cm2 V s−1 at 240 K for n2D ∼ 1.35 × 1013 cm−2 . This is a distinct
6 Vds = 500 mV 1.5 T = 4.2 K Vtg = 0 V
20 K difference from devices fabricated in a single-gate configuration
T = 4.2 K 1.0
60 K
100 K
20 K 250 (Fig. 2c). We relate this behaviour to effective damping of Coulomb
5 0.5
30 K scattering on charged impurities due to the presence of the high-κ

Ids (μA)
50 K 0.0
80 K ¬0.5 200 dielectric and the metallic top gate that changes the dielectric
4 120 K environment of monolayer MoS2 (ref. 27). At low temperatures, the
σ (e 2/h)

¬1.0
160 K

G (μS)
200 K ¬1.5 150 influence of charged impurities on mobility is stronger for lower
3
240 K ¬0.4 ¬0.2 0.0 0.2 0.4 electron densities. For example, at 10 K for n2D ∼ 0.76 × 1013 cm−2
Vds (V)
2 100 we extracted a mobility of 132 cm2 V s−1 whereas for n2D ∼ 1.35 ×
1013 cm−2 we extracted a mobility of 184 cm2 V s−1 . In the phonon-
1 50 limited part between 100 and 300 K, the mobility can be fitted
to the expression µ ∼ T −γ , with the exponent γ being in the
0 0 0.55–0.78 range for electron concentrations n2D between ∼ 0.76 ×
¬4 ¬2 0 2 4 1013 cm−2 and 1.35×1013 cm−2 (Fig. 3c). For all of our double-gated
Vtg (V) monolayer devices we find this exponent to be between 0.3 and 0.78,
b 6 3 whereas for one double-layer device we find a value of 1.47. These
5 n2D = 1.5 × 1013 cm¬2 values for monolayer MoS2 are much smaller than the theoretically
4 2
3 predicted value of γ ≈ 1.52 (ref. 30) or bulk crystals (γ ≈ 2.6;
2 100 ref. 24). This indicates that in addition to the quenching of the
homopolar phonon mode, other mechanisms might influence the
6 mobility of monolayer MoS2 in dual-gated devices, for example,
1 5
σ (e 2/h)

G (μS)
4 phonon screening induced by the metallic top gate or a change
6 3 in the strength of electron–phonon coupling. Further theoretical
5
4 2 modelling could shed more light on these mechanisms.
3
Just as in the case of single-gated devices, we model the
2 n2D = 3.6 × 1012 cm¬2 10 temperature dependence of G in the insulating regime of our
6 double-gated devices with thermally activated behaviour (Fig. 4a).
0.1 5
4 Here, we observe that the activated behaviour fits our data very
60 100 140 180 220 well in the 100–250 K temperature range, with extracted activation
T (K)
energies Ea shown in Fig. 4b.
We have performed Hall-effect measurements on all MoS2
c 2
devices covered with a dielectric layer presented here to accurately
μ ~ T¬0.78 determine the mobility, density of charge carriers and the capacitive
coupling of MoS2 layers to control gate electrodes (bottom or
top gates). Figure 5a shows the transverse Hall resistance Rxy of
μ (cm2 V¬1 s¬1)

our main dual-gated monolayer device, which follows a linear

dependence on the magnetic field B. From the inverse slope of Rxy
100 we can directly determine the electron density n2D in the MoS2
9 n2D = 0.76 × 1013 cm¬2 channel. The variation of the electron density extracted from Rxy
8 as a function of the top-gate voltage Vtg is shown in Fig. 5b. The
0.96 × 1013 cm¬2
7 1.15 × 1013 cm¬2 μ ~ T ¬0.55 slope of this dependence gives directly the capacitance Ctg,Hall =
6
1.35 × 1013 cm¬2 3.17 × 10−7 F cm−2 used in calculation of the field-effect mobility
(Fig. 3c). We also directly measure the capacitive coupling between
1 10 100 the channel and the bottom gate in devices where the MoS2 channel
T (K) is covered with a dielectric layer and in devices with disconnected
top gates and compare them with the geometric capacitance per
Figure 3 | Electron transport in dual-gated monolayer MoS2 . a, G and σ for unit area calculated using the parallel-plate capacitance model
different values of Vtg and T. For low values of Vtg , σ decreases with T. Cgeom = ε0 εr /dox,bottom , where dox,bottom is the thickness of the
Above Vtg ∼ 1–2 V, monolayer MoS2 enters a metallic state, with increasing bottom-gate oxide10 . We find that encapsulation in a dielectric
σ as T is decreased. The inset shows double Ids –Vds sweeps. can increase the capacitive coupling from Cgeom by a factor of 2.4,
b, T-dependence of σ for different values of n2D . c, µ is practically similarly to graphene devices45 , and disconnecting the top gate
independent of T under 30 K, indicating screening of charged impurities increases the capacitive coupling by a factor of 53 (Supplementary
due to deposition of the top-gate dielectric. Above ∼100 K, µ decreases Fig. S4). These measurements prove that the capacitance can
owing to phonon scattering and follows a T −γ dependence with be underestimated in a complicated dielectric environment, both
γ = 0.55–0.78. The strongly reduced value of the exponent γ with respect in the case of disconnected top gates10–12 and encapsulation44,46 ,
to the single-gated device (γ = 1.4) is indicative of phonon mode resulting in mobility values that are probably overestimated. To
quenching. Error bars are estimated on the basis of uncertainties in accurately measure the field-effect mobility of FETs based on 2D
determining the voltage drop across the channel. materials one needs to measure the actual capacitance using either
cyclic voltammetry28 or Hall-effect measurements as outlined here.
for field-effect mobility µ = [dG/dVbg ] × [L12 /(WCtg,Hall )], with In conclusion, we have performed conductance and mobility
capacitance Ctg,Hall extracted from Hall-effect measurements. For measurements on monolayer MoS2 FETs in single- and dual-
all dual-gate devices that we have characterized, we observe gate configurations. Using a top gate and solid-state dielectrics,
a monotonous increase of the mobility as the temperature is we were able to tune the charge carrier density to more than
decreased with a saturation at low temperatures. Figure 3c shows n2D ∼ 3.6 × 1013 cm−2 , inducing the transition from the insulating
the temperature dependence of mobility for the main device to the metallic phase in monolayer MoS2 . A quantum critical point,
presented here. The mobility at 4 K is 174 cm2 V s−1 , reaching separating the metallic phase, stabilized by electronic interactions,

818 NATURE MATERIALS | VOL 12 | SEPTEMBER 2013 | www.nature.com/naturematerials

© 2013 Macmillan Publishers Limited. All rights reserved.
NATURE MATERIALS DOI: 10.1038/NMAT3687 ARTICLES
a 10¬5 b
80
Vtg = ¬1.4 V

10¬6
60

10¬7

Ea (meV)
G (S)

40
10¬8

20
10¬9

Vtg = ¬3.4 V
10¬10 0
0.004 0.006 0.008 0.010 0.012 ¬3.0 ¬2.5 ¬2.0 ¬1.5
1/T (K¬1) Vtg (V)

Figure 4 | Ea for the top-gated monolayer MoS2 in the insulating regime. a, Arrhenius plot of the conductance of monolayer MoS2 covered with HfO2 , in
the insulating regime. b, Dependence of Ea on Vtg .

a 500 b
1.6
Vtg = 3 V Ctg-Hall /e = 1.98 × 1012 F/C × cm¬2
4V
400 5V 1.5

n2D (1013 cm¬2) 1.4

300
Rxy (Ω)

1.3

200
1.2

100 1.1 n2D = n0 + Ctg-Hall × Vtg /e

n0 = 5.6 × 1012 cm¬2
1.0
0
0 2 4 6 8 10 2.0 2.5 3.0 3.5 4.0 4.5 5.0
B (T) Vtg (V)

Figure 5 | Hall-effect measurements in the dual-gated monolayer MoS2 device. a, Hall resistance Rxy versus B for different positive values of Vtg .
b, Electron concentration n2D extracted from Rxy or different values of Vtg . From the slope of the red solid line we calculate the capacitance per unit area
Ctg,Hall of the top-gate MoS2 device. The residual doping of the MoS2 channel is no = 5.6 × 1012 cm−2 . All measurements are performed at T = 4 K with a
grounded back-gate electrode.

from the insulating phase, where disorder prevails over the ALD was performed in a Beneq system using the reaction of H2 O with
electronic interactions, has been identified. This transition point tetrakis(ethyl-methylamido)hafnium. Electrical characterization was carried
is in good agreement with theory and shows that monolayer out using National Instruments DAQ cards, SR570 current preamplifiers,
SR560 low noise voltage preamplifiers, and an Oxford Instruments Heliox
MoS2 could be an interesting new material system for investigating cryo-magnetic system.
low-dimensional correlated electron behaviour. The MIT could
also be used for new types of switch, especially fast optoelectronic
Received 11 January 2013; accepted 15 May 2013; published online
switches based on differences in optical transmission in metallic and
23 June 2013
insulating states47 . In addition to allowing high charge densities,
the high-κ HfO2 used as the top-gate dielectric also changes the
dielectric environment and effectively screens Coulomb scattering, References
1. Wang, Q. H., Kalantar-Zadeh, K., Kis, A., Coleman, J. N. & Strano, M. S.
which results in mobility improvement in dual-gate devices. Electronics and optoelectronics of two-dimensional transition metal
Furthermore, the presence of the top-gate dielectric and metal dichalcogenides. Nature Nanotech. 7, 699–712 (2012).
electrode results in a quenching of the homopolar mode, which 2. Novoselov, K. S. et al. Two-dimensional atomic crystals. Proc. Natl Acad. Sci.
is polarized in the direction normal to the layer, leading to a USA 102, 10451–10453 (2005).
strong decrease of the mobility exponent γ in µ ∼ T −γ . Our 3. Coleman, J. N. et al. Two-dimensional nanosheets produced by liquid
exfoliation of layered materials. Science 331, 568–571 (2011).
results provide a new picture of the mobility issue in different 4. Smith, R. J. et al. Large-scale exfoliation of inorganic layered compounds in
configurations of MoS2 devices, which should shed new light on aqueous surfactant solutions. Adv. Mater. 23, 3944–3948 (2011).
the directions for further improvements in device quality and 5. Liu, K-K. et al. Growth of large-area and highly crystalline MoS2 thin layers on
characterization techniques. insulating substrates. Nano Lett. 12, 1538–1544 (2012).
6. Zhan, Y., Liu, Z., Najmaei, S., Ajayan, P. M. & Lou, J. Large-area vapor-phase
growth and characterization of MoS2 atomic layers on a SiO2 substrate. Small
Methods 8, 966–971 (2012).
MoS2 flakes were exfoliated from molybdenite crystals (SPI Supplies Brand 7. Bertolazzi, S., Brivio, J. & Kis, A. Stretching and breaking of ultrathin MoS2 .
Moly Disulphide) by the Scotch-tape micromechanical cleavage technique. ACS Nano. 5, 9703–9709 (2011).

NATURE MATERIALS | VOL 12 | SEPTEMBER 2013 | www.nature.com/naturematerials 819

© 2013 Macmillan Publishers Limited. All rights reserved.
ARTICLES NATURE MATERIALS DOI: 10.1038/NMAT3687

8. Brivio, J., Alexander, D. T. L. & Kis, A. Ripples and layers in ultrathin MoS2 35. Punnoose, A. & Finkel’stein, A. M. Metal-insulator transition in disordered
membranes. Nano Lett. 11, 5148–5153 (2011). two-dimensional electron systems. Science 310, 289–291 (2005).
9. Lembke, D. & Kis, A. Breakdown of high-performance monolayer MoS2 36. Ho, L. H., Micolich, A. P., Hamilton, A. R. & Sushkov, O. P. Ground-plane
transistors. ACS Nano. 6, 10070–10075 (2012). screening of Coulomb interactions in two-dimensional systems: How
10. Radisavljevic, B., Radenovic, A., Brivio, J., Giacometti, V. & Kis, A. Single-layer effectively can one two-dimensional system screen interactions in another.
MoS2 transistors. Nature Nanotech. 6, 147–150 (2011). Phys. Rev. B 80, 155412 (2009).
11. Radisavljevic, B., Whitwick, M. B. & Kis, A. Integrated circuits 37. Salmani-Jelodar, M., Yaohua, T. & Klimeck, G. Single layer MoS2 band
and logic operations based on single-layer MoS2 . ACS Nano. 5, structure and transport. Proc. Int. Semiconductor Device Research Symp.
9934–9938 (2011). (ISDRS) 1–2 (2011).
12. Radisavljevic, B., Whitwick, M. B. & Kis, A. Small-signal amplifier based on 38. Ghosh, A. et al. Electron assisted variable range hopping in strongly correlated
single-layer MoS2 . Appl. Phys. Lett. 101, 043103 (2012). 2D electron systems. Phys. Status Solidi B 230, 211–216 (2002).
13. Ye, J. T. et al. Superconducting dome in a gate-tuned band insulator. Science 39. Okamoto, T., Hosoya, K., Kawaji, S. & Yagi, A. Spin degree of freedom in a
338, 1193–1196 (2012). two-dimensional electron liquid. Phys. Rev. Lett. 82, 3875–3878 (1999).
14. Lebegue, S. & Eriksson, O. Electronic structure of two-dimensional crystals 40. Dobrosavljevic, V., Trivedi, N. & Valles, J. M. Conductor-Insulator Quantum
from ab initio theory. Phys. Rev. B 79, 115409 (2009). Phase Transitions (Oxford Univ. Press, 2012).
15. Kuc, A., Zibouche, N. & Heine, T. Influence of quantum confinement on 41. Gurvitch, M. Ioffe-Regel criterion and resistivity of metals. Phys. Rev. B 24,
the electronic structure of the transition metal sulfide TS2 . Phys. Rev. B 83, 7404–7407 (1981).
245213 (2011). 42. Mark, R. G., Adkins, C. J., Haim, B. & Ralph, R. Experimental study of the
16. Splendiani, A. et al. Emerging photoluminescence in monolayer MoS2 . Ioffe-Regel criterion for amorphous indium oxide films. J. Phys. Condens.
Nano Lett. 10, 1271–1275 (2010). Matter 10, 809–819 (1998).
17. Mak, K. F., Lee, C., Hone, J., Shan, J. & Heinz, T. F. Atomically thin MoS2 : A 43. Lee, P. A. & Ramakrishnan, T. V. Disordered electronic systems. Rev. Mod.
new direct-gap semiconductor. Phys. Rev. Lett. 105, 136805 (2010). Phys. 57, 287–337 (1985).
18. Mak, K. F., He, K., Shan, J. & Heinz, T. F. Control of valley polarization in 44. Das, S., Chen, H-Y., Penumatcha, A. V. & Appenzeller, J. High performance
monolayer MoS2 by optical helicity. Nature Nanotech. 7, 494–498 (2012). multilayer MoS2 transistors with scandium contacts. Nano Lett. (2012).
19. Zeng, H., Dai, J., Yao, W., Xiao, D. & Cui, X. Valley polarization in MoS2 45. Xia, J. L., Chen, F., Wiktor, P., Ferry, D. K. & Tao, N. J. Effect of top
monolayers by optical pumping. Nature Nanotech. 7, 490–493 (2012). dielectric medium on gate capacitance of graphene field effect transistors:
20. Cao, T. et al. Valley-selective circular dichroism of monolayer molybdenum Implications in mobility measurements and sensor applications. Nano Lett. 10,
disulphide. Nature Commun. 3, 887 (2012). 5060–5064 (2010).
21. Feng, W. et al. Intrinsic spin Hall effect in monolayers of group-VI 46. Bao, W., Cai, X., Kim, D., Sridhara, K. & Fuhrer, M. S. High mobility ambipolar
dichalcogenides: A first-principles study. Phys. Rev. B 86, 165108 (2012). MoS2 field-effect transistors: substrate and dielectric effects. Appl. Phys. Lett.
22. Yoon, Y., Ganapathi, K. & Salahuddin, S. How good can monolayer MoS2 102, 042104 (2013).
transistors be? Nano Lett. 11, 3768–3773 (2011). 47. Chudnovskiy, F., Luryi, S. & Spivak, B. in Future Trends in Microelectronics:
23. Alam, K. & Lake, R. K. Monolayer MoS2 transistors beyond the technology The Nano Millennium (eds Luriy, S., Xu, J.M. & Zaslavsky, A.) Part II, 148–155
road map. Elect. Dev. IEEE Trans. 59, 3250–3254 (2012). (Wiley, 2002).
24. Fivaz, R. & Mooser, E. Mobility of charge carriers in semiconducting layer
structures. Phys. Rev. 163, 743–755 (1967). Acknowledgements
25. Ghatak, S., Pal, A. N. & Ghosh, A. Nature of electronic states in atomically thin We thank W. Escoffier (LNCMI CNRS), B. Raquet (LNCMI CNRS) and S. Bertolazzi
MoS2 field-effect transistors. ACS Nano. 5, 7707–7712 (2011). (EPFL) for useful discussions as well as J-S. Heron (EPFL) for technical support. Device
26. Jena, D. & Konar, A. Enhancement of carrier mobility in semiconductor fabrication was carried out in part in the EPFL Center for Micro/Nanotechnology
nanostructures by dielectric engineering. Phys. Rev. Lett. 98, 136805 (2007). (CMI). We thank Z. Benes (CMI) for technical support with electron-beam lithography
27. Han, L. & Ye, P. D. MoS2 dual-gate MOSFET with atomic-layer-deposited and A. Radenovic and M. Whitwick (EPFL) for support with ALD deposition. This
Al2 O3 as top-gate dielectric. Elect. Dev. Lett. IEEE 33, 546–548 (2012). work was financially supported by ERC grant no. 240076, FLATRONICS: Electronic
28. Kim, S. et al. High-mobility and low-power thin-film transistors based on devices based on nanolayers.
multilayer MoS2 crystals. Nature Commun. 3, 1011 (2012).
29. Ming-Wei, L. et al. Mobility enhancement and highly efficient gating
of monolayer MoS2 transistors with polymer electrolyte. J. Phys. D 45,
Author contributions
345102 (2012). B.R. worked on device fabrication and performed the measurements. A.K. designed
30. Kaasbjerg, K., Thygesen, K. S. & Jacobsen, K. W. Phonon-limited mobility in the experiment and initiated the research. B.R. and A.K. analysed the results and
n-type single-layer MoS2 from first principles. Phys. Rev. B 85, 115317 (2012). wrote the manuscript.
31. Sze, S. M. & Ng, K. K. Physics of Semiconductor Devices (Wiley, 2007).
32. Mott, N. F. Metal-Insulator Transitions (Barnes Noble Books, Taylor Francis, Additional information
1974). Supplementary information is available in the online version of the paper. Reprints and
33. Abrahams, E., Anderson, P. W., Licciardello, D. C. & Ramakrishnan, T. V. permissions information is available online at www.nature.com/reprints. Correspondence
Scaling theory of localization: Absence of quantum diffusion in two dimensions. and requests for materials should be addressed to A.K.
Phys. Rev. Lett. 42, 673–676 (1979).
34. Evers, F. & Mirlin, A. D. Anderson transitions. Rev. Mod. Phys. 80, Competing financial interests
1355–1417 (2008). The authors declare no competing financial interests.

820 NATURE MATERIALS | VOL 12 | SEPTEMBER 2013 | www.nature.com/naturematerials

© 2013 Macmillan Publishers Limited. All rights reserved.