Articles

https://doi.org/10.1038/s41928-022-00746-6

Interaction- and defect-free van der Waals

contacts between metals and two-dimensional
semiconductors
Gihyeon Kwon1, Yoon-Ho Choi1,2, Hyangsook Lee3, Hyeon-Sik Kim1, Jeahun Jeong1, Kwangsik Jeong 4
,
Min Baik1, Hoedon Kwon1, Jaemin Ahn3, Eunha Lee 3 ✉ and Mann-Ho Cho 1,5 ✉

High Schottky barrier heights at metal–semiconductor junctions due to Fermi-level pinning can degrade the performance of
electronic devices and increase their energy consumption. Van der Waals contacts between metals and two-dimensional semi-
conductors without Fermi-level pinning are theoretically possible, but have not been achieved due to the presence of interac-
tions such as interface defects and orbital overlap. Here we show that interaction- and defect-free van der Waals contacts can
be formed between a range of metals and two-dimensional semiconductors via a metal deposition process that uses a sele-
nium buffer layer. Our contacts obey the Schottky–Mott rule and have a Fermi-level pinning of –0.91. A comparison between
the van der Waals contacts and typical direct metal contacts reveals differences in interface gap distances, band bending and
electrical characteristics. Using gold van der Waals contacts, we fabricate p-type tungsten diselenide field-effect transistors
that exhibit stable operation with on/off ratio of 106, mobility of 155 cm2 (V s)–1, contact resistance of 1.25 kΩ μm and Schottky
barrier height of 60 meV.

A
s semiconductor technology has advanced, metal–semi- van der Waals (vdW) bonding with the metal7,8. However, in reality,
conductor junctions have become an increasingly impor- FLP remains a serious concern. The interfacial interaction between
tant aspect of electronic and optoelectronic devices. Charge evaporated metal atoms having a high thermal, or impinging, energy
carrier transport across metal–semiconductor junctions, and their and the 2D semiconductor is strong enough to break the vdW gap
related Schottky barrier heights (SBHs), determine device perfor- and surface covalent bonds. This can induce chemical interactions
mance1. In particular, contact resistance (Rc) between a metal and that lead to FLP caused by many mid-gap states at the interface9–15.
a semiconductor, which is due to the Schottky barrier, can limit the The undesirable interactions between 2D materials and metals
device lifetime and increase the energy consumption of integrated can be reduced during the deposition process by using a transferred
device systems2. metal contact method and an alloy metal contact method16,17. With
According to the Schottky–Mott rule the transferred metal contact method, it is difficult to transfer metal
contacts in a sufficiently clean environment; therefore, this approach
Φelectron = Φm − χ s , (1) is vulnerable to external influence. The methods are also difficult to
apply to industrially relevant large-substrate technologies. With the
alloy metal contact method, alloys rather than pure metals are used
Φhole = Eg + χ s − Φm , (2) and the range of suitable work-function alloyed metals is limited.
In this Article, we show that interaction- and defect-free vdW
where Φelectron and Φhole are the SBHs for electrons and holes, respec- metal contacts to 2D materials can be created by using selenium as a
tively; Φm, χs and Eg are the work function of the metal, electron buffer layer to reduce physical and chemical interactions. Se is easily
affinity and bandgap of the semiconductor, respectively. Thus, in evaporated at temperatures below approximately 150 °C in vacuum
theory, the SBH can be controlled through the work function of the due to its low cohesive energy18 (Supplementary Fig. 1 shows a com-
metal to create an ideal metal–semiconductor junction3. However, parison of the evaporation temperature of different relevant ele-
in experiments, the Schottky–Mott rule fails because of Fermi-level ments). Se also has a number of properties that make it well suited
pinning (FLP) induced by mid-gap states at the metal–semiconduc- for use as a buffer layer, including its cohesive energy, mass and
tor interface. chemical behaviour. To illustrate the capabilities of our approach,
For three-dimensional, or bulk, semiconductors, studies have we fabricate field-effect transistors (FETs) with vdW metal contacts
shown that many mid-gap states arise from interactions at metal– using different metals and 2D semiconductors.
semiconductor interfaces, such as orbital overlap, metal interdiffu-
sion, strain and defects4–6. Two-dimensional (2D) semiconductors Interface structure in vdW and direct contacts
without surface dangling bonds should, in contrast, be free from To fabricate a vdW metal contact between three-dimensional metal
FLP because of the filled covalent bonds at the surface and weak and 2D transition metal dichalcogenides (TMDs), a 50-nm-thick

Department of Physics, Yonsei University, Seoul, Republic of Korea. 2Yield Enhancement team, Foundry, Samsung Electronics, Hwaseong-si, Republic of
1

Korea. 3Analytical Engineering Group, Samsung Advanced Institute of Technology, Samsung Electronics, Suwon, Republic of Korea. 4Division of Physics
and Semiconductor Science, Dongguk University, Seoul, Republic of Korea. 5Department of System Semiconductor Engineering, Yonsei University, Seoul,
Republic of Korea. ✉e-mail: [email protected]; [email protected]

Nature Electronics | VOL 5 | April 2022 | 241–247 | www.nature.com/natureelectronics 241

Articles Nature Electronics

a b c
vdW metal contact vdW Au contact Direct Au contact

Se
W

Au Se deposition
5.3 Å 4.7 Å

Au eposition

Annealing in vaccum

vdW Au contact

e
Ideal vdW gap

Direct Au contact
Direct metal contact
f g

Orbital overlap Defect

vdW Au contact Direct Au contact

Fig. 1 | WSe2 with vdW and direct Au contact. a, Schematic of the sample cross-sections during the fabrication of vdW and direct Au contact on WSe2.
b, HAADF-STEM image of the cross-section of WSe2 with vdW Au contact. Scale bar, 1 nm. c–g, STEM images of WSe2 cross-section with direct Au
contact (c; scale bar, 1 nm), vdW Au contact (d; scale bar, 5 nm), direct Au contact (red boxes denote regions containing defects) (e; scale bar, 5 nm), vdW
Au contact (f; scale bar, 20 nm) and direct Au contact (g; scale bar, 20 nm) (the two linked STEM images are joined in d and e).

Au metal electrode was deposited on a mechanically exfoliated Moreover, in direct Au contact, a strained Au lattice existed at
WSe2 flake with a 10-nm-thick Se buffer layer. After the Au depo- the interfacial region on WSe2. In contrast, in the vdW Au contact,
sition process, Se was removed and environmental contamination no distortion of the Au lattice on WSe2 was observed at the inter-
was minimized by annealing at 150 °C. An Au contact fabricated face region20. The distorting effect of Au atoms with high thermal
using an interfacial controlled layer of Se (vdW Au contact (AuvdW)) energy could be partially observed in the structure of the WSe2
was compared with an Au contact fabricated using direct metalliza- layers (Fig. 1e, red box). The vdW contact, however, had a uni-
tion without the buffer layer (direct Au contact (Audirect)) (Fig. 1a). form and clean vdW interface and no distorted structure (Fig. 1d),
To investigate the atomic structure at the WSe2–Au interface, we resulting in the formation of the atomically flat Au–WSe2 interface
observed cross-sectional atomic-resolution images in the Au/WSe2 (Fig. 1f). Conversely, in direct Au contact without the buffer layer,
structure using high-angle annular dark-field (HAADF) scan- the Au–WSe2 interface was rough, with irregular strain induced by
ning transmission electron microscopy (STEM). A difference in strong interaction (Fig. 1g). As shown in Supplementary Fig. 2, we
the interface gap (nearest distance between W and Au atoms) was removed the two types of Au grown on WSe2 using a wet-etching
clearly observed: the contact distance between WSe2 and Au with solution (H2O + I2 + KI) (ref. 21). The interaction between Au and
the Se layer (vdW contact), namely, ~5.30 Å, was larger than that WSe2 in the case of vdW contact was weaker than that in the case
without the Se layer (direct contact), namely, ~4.70 Å (Fig. 1b–c). of direct contact, causing WSe2 and Au to be perfectly separated by
Theoretically, the distance between the W of WSe2 and Au in an the vdW gap.
ideal vdW contact can be calculated using the vdW diameters of the
W atom and Se atom8,16: Comparison between vdW and direct contacts for FLP
To investigate how the FLP effect depends on the metal contact
DW−Au = RvdW vdW
Au + RSe + DW−Se , (3) method in an electrical conduction process, we fabricated FETs
using two Au (50 nm) contact methods: direct contact and vdW
where DW–Au, DW–Se, RAuvdW and RSevdW are the nearest distance contact through electron-beam evaporation on the same few-layer
between a W atom and Au atom, distance between a W atom and WSe2 (~12 nm) (Fig. 2a). The gate-voltage–drain-current (VG–Ids)
Se atom of WSe2, vdW radius of Au, and vdW radius of Se, respec- transfer curve of the Au metal contact with WSe2 for a gate-voltage
tively. Given that DW–Se, RAuvdW and RSevdW are 1.69, 1.66 and 1.90 Å, sweep from +70 to −70 V (Fig. 2b) shows a clear difference between
respectively, the calculated value of DW–Au ≈ 5.25 Å, comparable with the FETs. The direct Au contact case shows n-type characteristics,
the value of DW–Au measured using a Se buffer layer (5.30 Å)19. The because numerous mid-gap states are formed near the conduc-
difference between the two values is <1%, indicating that an ideal tion band edge, resulting in FLP near the conduction band at the
vdW contact resulting from very weak interaction was formed at the Au–WSe2 interface. By contrast, the junction in the vdW con-
WSe2–Au interface in the presence of the Se layer. tact case follows the Schottky–Mott rule; the FLP-free ideal vdW

242 Nature Electronics | VOL 5 | April 2022 | 241–247 | www.nature.com/natureelectronics

Nature Electronics Articles
a b
10–4

Ids
10–5
Au WSe2 Au

10–6
hBN
SiO2
–7
10

Ids (A)
Back-gate
p-Si
vdW Au contact –8
10
Direct Au contact VG
–9
10
12 nm

10–10 Audirect
WSe2 AuvdW
5 µm
–40 0 40
Gate voltage (V)

c d
2.5 2.5
vdW Au contact vdW Au contact
Se 3d W 4f
Pristine WSe2 Pristine WSe2
Direct Au contact Direct Au contact
2.0 2.0

Intensity (a.u.)
Intensity (a.u.)

1.5 1.5

1.0 1.0

0.5 0.5

0 0
58 57 56 55 54 53 38 36 34 32 30
Binding energy (eV) Binding energy (eV)

Fig. 2 | FET properties of WSe2 with vdW and direct Au contact. a, Schematic and optical images of WSe2 (yellow dotted line) FETs fabricated with direct
Au contact (black dashed line) and vdW Au contact (red dashed line). b, Transfer curve of WSe2 FET with direct Au contact (Audirect) and vdW Au contact
(AuvdW). For both contacts, channel length L = 1.6 μm, channel width W = 5.0 μm and drain voltage Vds = 1 V. c,d, XPS spectra acquired to confirm interaction
between WSe2 and Au with vdW Au contact and direct Au contact. XPS peaks of pristine WSe2 are observed: W 4f5/2,7/2 (35.0 eV, 32.8 eV; d) and Se
3d3/2,5/2 (56.0 eV, 55.1 eV; c). The high binding shift of W 4f peaks and Se 3d peaks with the direct Au contact is 0.1 eV, and the low binding shift of W 4f
peaks and Se 3d peaks with the vdW Au contact is −0.1 eV.

contact enables Au (Φm = 5.3 eV) with a high work function to To investigate the annealing effect of direct Au contact, we pre-
form a low (high) SBH for holes (electrons) between WSe2 and Au. pared—in the same way as in the vdW metal case—a device con-
Consequently, the transfer curve for the WSe2 FET forming a vdW taining WSe2 in direct contact with Au that was processed with
contact with Au shows p-type characteristics with a large on/off annealing in vacuum. The transfer curve of this device still shows
ratio (>106) and high mobility (~135 cm2 (V s)–1). n-type behaviour due to FLP, except for a slight increase in the accu-
To confirm the change in chemical state caused by the WSe2 mulation current (Supplementary Fig. 7).
layer during Au deposition (1 nm), we acquired the W 4f and Se The effectiveness of the vdW contact method was also investigated
3d X-ray photoelectron spectroscopy (XPS) data of WSe2 for the using another metal deposition process: sputtering (Supplementary
vdW Au contact, direct Au contact and pristine WSe2 (Fig. 2c,d). Fig. 8a). Although sputtered metal atoms commonly have a kinetic
The overall core-level binding energy shift caused by the change energy of up to tens of electronvolts, which is considerably higher
in Fermi level indicates that the generation of interfacial charged than the evaporation thermal energy of metal atoms28, a vdW con-
states is dependent on the contact method. In the direct Au con- tact with Au fabricated using sputtering still prevented FLP, as evi-
tact case, the core-level spectra of W 4f and Se 3d were shifted to denced by the p-type characteristics of its transfer curve. As shown
a higher binding energy (+0.1 eV), reflecting n doping compared in Supplementary Fig. 8b–d, we fabricated metal (Au)–semicon-
with pristine WSe2. By contrast, in the vdW Au contact case, the ductor (WSe2)–metal (Au) Schottky diodes. The transfer curves of
spectra were shifted to a lower binding energy (−0.1 eV), revealing each type of FET indicated a large on/off ratio (>105), and the out-
p doping compared with pristine WSe2. Supplementary Fig. 3 con- put curves of the Schottky diodes using Au electrodes with/without
tains a detailed explanation of the binding shift of doped semicon- FLP indicated a large rectification ratio (~104–105), demonstrating
ductors22–26. As shown in Supplementary Fig. 4, the XPS data reveal the satisfactory performance of the Schottky diodes29.
that the Se buffer layer was completely removed after annealing, To understand the electronic band structure at the Au–WSe2
even at 150 °C. We also performed Raman spectroscopy to confirm interface, we performed density functional theory (DFT) calcula-
that the Se buffer layer does not react with various TMDs, because tions based on the structural data shown in Fig. 1. As shown in Fig. 3a
the vaporization temperature is only 150 °C (Supplementary and Supplementary Fig. 9, in the case of WSe2 with direct Au con-
Fig. 5 and 6)27. tact, we observed that new states were formed in the mid-bandgap

Nature Electronics | VOL 5 | April 2022 | 241–247 | www.nature.com/natureelectronics 243

Articles Nature Electronics

a b
Direct Au contact on WSe2 vdW Au contact on WSe2
1.0 1.0

0.5 4.36 Å 0.5 5.10 Å

Ec Ec
Energy (eV)

Energy (eV)
Ef Ef Ef Ef
0 Au Mid-gap states Ev 0 Au Ev
WSe2 WSe2

–0.5 –0.5
Mid-gap states W + Se W + Se
Se Se
W W
–1.0 –1.0
0 0.5 1.0 1.5 2.0 2.5 3.0 0 0.5 1.0 1.5 2.0 2.5 3.0
DOS (eV–1) DOS (eV–1)

Fig. 3 | DFT results for WSe2 with vdW and direct Au contact. a, Density of W and Se states in WSe2 with direct Au contact and band diagram of WSe2–
Au interface. b, Density of W and Se states in WSe2 with vdW Au contact and band diagram of WSe2–Au interface. Stable point of Au, optimized by
adjusting the vdW gap between WSe2 surface and Au, is larger (5.10 Å) than in the direct contact case (4.36 Å); this is consistent with the observed STEM
data. DOS is the density of states of the total orbitals combined with those of the s, p and d orbitals. Ec is the energy of conduction band edge, Ev is the
energy of valence band edge, and Ef is the energy of Fermi level.

a b c
10–5 WSe2–Ag WSe2–Co WSe2–Pt
–5
10 10–5
10–6
10–6 10–6
–7
10 10–7 10–7
Ids (A)

Ids (A)

Ids (A)
–8
10 10–8
10–8
–9
10–9 10
10–9
10 –10 10–10
10–10
–11
Agdirect 10–11 Codirect Ptdirect
10
AgvdW CovdW 10–11 PtvdW
10–12 10–12
–40 –20 0 20 40 –40 –20 0 20 40 –40 –20 0 20 40
Gate voltage (V) Gate voltage (V) Gate voltage (V)

d e f
MoS2–Au 10–4
10–5 WS2–Au MoSe2–Au
10–5
10–5
10–6
10–6
10–6
10–7
–7
10
10–7
Ids (A)

Ids (A)

Ids (A)

10–8
–8
10
10–8
10–9
10–9 10–9
10–10
10–10 Audirect Audirect 10–10 Audirect
10–11
AuvdW AuvdW AuvdW
10–11 10–11
–40 –20 0 20 40 –40 0 40 –40 0 40
Gate voltage (V) Gate voltage (V) Gate voltage (V)

Fig. 4 | Properties of FETs made of various TMDs with vdW and direct metal contact. a, Transfer curve of WSe2 FET with direct Ag contact (Agdirect) and
vdW Ag contact (AgvdW). For both contacts, L = 2.0 μm, W = 2.5 μm and Vds = 1 V. b, Transfer curve of the WSe2 FET with direct Co contact (Codirect) and
vdW Co contact (CovdW). For both contacts, L = 2 μm, W = 8 μm and Vds = 1 V. c, Transfer curve of WSe2 FET with direct Pt contact (Ptdirect) and vdW Pt
contact (PtvdW). For both contacts, L = 2 μm, W = 2 μm and Vds = 1 V. d, Transfer curve of MoS2 FET with direct Au contact (L = 2 μm, W = 3 μm and Vds = 1 V)
and vdW Au contact (L = 2 μm, W = 3 μm and Vds = 1 V). e, Transfer curve of WS2 FET with direct and vdW Au contacts (L = 1.5 μm, W = 5.0 μm and Vds = 1 V
for both cases). f, Transfer curve of MoSe2 FET with direct Au contact (L = 1 μm, W = 6 μm, and Vds = 1 V) and vdW Au contact (L = 2 μm, W = 12 μm and
Vds = 1 V). Here L is the channel length, W is the channel width and Vds is the drain voltage.

region of WSe2 (ref. 30). From the change in potential energy accord- downward bending of the interfacial band structure31. Additionally,
ing to the z-axis position (Supplementary Fig. 10a,b), we confirmed the change in the total charge density of electrons indicated charge
that the reduced potential energy of WSe2 at the interface caused the exchange at the Au–WSe2 interface. Because the electron charges

244 Nature Electronics | VOL 5 | April 2022 | 241–247 | www.nature.com/natureelectronics

Nature Electronics Articles
a b c
4 WSe2–AgvdW (VG = +60 V) WSe2–CovdW (VG = –60 V)
1.2 Agdirect
0.6
AgvdW Ptdirect 3
Codirect

Ids (µA)

Ids (µA)
1.0 Audirect 0.4
2
Sdirect ≈ –0.11

0.8 1 0.2

eV
0
.3
SBHhole (eV)

–4

eV

eV
0 0

0
.0

.1
0.6 0 0.1 0.2 0.3 0.4 0.5 0 0.4 0.8 1.2

eV
–5

–4
SvdW ≈ –0.91

eV
0

5
.3

.3
eV
Drain voltage (V) Drain voltage (V)

–5

–5
0
.6
–5
0.4 d e
WSe2–AgvdW (VG = –60 V) WSe2–Pt vdW (VG = –60 V)
Ag Co Au Pt WSe2 12
30 10
0.2 CovdW
8

Ids (µA)
Ids (µA)
PtvdW 20
AuvdW 6
0
4.4 4.8 5.2 5.6 4
10
Work function (eV) 2

0 0
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
Drain voltage (V) Drain voltage (V)

Fig. 5 | SBH between WSe2 and various vdW and direct metal contacts, and output curves of vdW metal contacts. a, SBH of hole (SBHhole) between bulk
WSe2 (conduction band edge is −4.10 eV and valence band edge is −5.35 eV) and AgvdW, CovdW, AuvdW, PtvdW, Agdirect, Codirect, Audirect and Ptdirect. The slope
of vdW contact is SvdW = −0.91 and the slope of direct contact is Sdirect = −0.11. Flat-band Schottky barrier height between WSe2 and various direct metal
contacts (Ag, Co, Au and Pt) through the thermionic emission effect (equation (4)). b, Output curve of WSe2 FET with vdW Ag contact (AgvdW) (L = 2 μm
and W = 5 μm). c, Output curve of WSe2 FET with vdW Co contact (CovdW) (L = 2 μm and W = 4 μm). d, Output curve of WSe2 FET with vdW Au contact
(AuvdW) (L = 0.8 μm and W = 2.0 μm). e, Output curve of WSe2 FET with vdW Pt contact (PtvdW) (L = 2 μm and W = 2 μm). Here L is the channel length and
W is the channel width.

were supplied from Au to WSe2, the WSe2 at the interfacial region Ids = AA∗ T2 exp (−ΦSBH /kT) , (4)
became n doped, which is consistent with the XPS and electrical
conduction results (Fig. 2)32,33. The interface total charge density of where ΦSBH, A, A* and k are the device SBH, device junction area,
direct Au contact was 0.1 e Å–3. Furthermore, through the total charge Richardson constant and Boltzmann constant, respectively, we
density difference, we confirmed that WSe2 with direct Au contact could extract a flat-band hole SBH (Φhole = 60 meV) between the
had considerable electron exchange, whereas WSe2 with vdW Au five-layer WSe2 and AuvdW in the drain current Ids given by the trans-
contact had relatively less; this indicates that the former forms a fer curve at the measurement temperature T (140–350 K)9,16. The
very strong interaction compared with the latter34. Conversely, as notable low hole SBH explains the p-type ohmic behaviour, low
shown in Fig. 3b, the density of states (DOS) of WSe2 formed in a contact resistance and excellent FET performance of WSe2 with
vdW contact with Au exhibits fewer new mid-gap states; the large vdW Au contact. As shown in Supplementary Fig. 14, which com-
interface gap between WSe2 and Au prevents orbital overlap, unlike pares the contact resistance of p-type WSe2 in various studies, our
that of WSe2 with the direct Au contact. As shown in Supplementary result is the lowest using a pure metal contact without doping or a
Fig. 10c,d, the increase in the potential energy of WSe2 in the forma- polymer gate insulator5,16,36–45.
tion of the vdW contact with Au causes upward band bending at To confirm the stability of the WSe2 FET with vdW Au contact,
the interface; this is consistent with the prediction of the Schottky– we measured the transfer curves of such FETs after fabrication and
Mott rule. The total charge density at the WSe2–Au interface indi- after three months. We confirmed that WSe2 FET with vdW Au con-
cates that WSe2 at the interface region becomes p doped because of tact has good stability, and that the vdW interface is maintained for
the electron charge supplied from WSe2 to Au. The interface total a long time, at least three months (Supplementary Fig. 15).
charge density of the vdW Au contact was 0.04 e Å–3. The role of the
Se buffer layer in the control of FLP according to theory and experi- Applicability to other materials
ment is shown via the schematic in Supplementary Fig. 11. To verify the usefulness of the vdW contact method using the Se
buffer layer, we investigated the charge-transport characteris-
Contact resistance of WSe2 with vdW Au contact tics of FETs fabricated using various combinations of metals and
As shown in Supplementary Fig. 12, the contact resistance (Rc) 2D semiconductors—WSe2–Ag, WSe2–Co, WSe2–Pt, MoS2–Au,
between WSe2 and Au in the vdW contact was extracted from the WS2–Au and MoSe2–Au—according to the contact method for
output curve for different channel lengths at a back-gate voltage WSe2 FETs (Fig. 4a–c). The work functions of Ag (Φm = 4.3 eV), Co
of −50 V. The vdW contact exhibited a very low contact resistance (Φm = 5.0 eV) and Pt (Φm = 5.6 eV) differ considerably from each
(2Rc = 2.5 kΩ μm) through transmission-line measurement35. An other and from that of Au (Φm = 5.3 eV)16,46.
ultrathin WSe2 FET with vdW Au contact was fabricated to investi- According to the Schottky–Mott rule, the n-type character-
gate the hole SBH between WSe2 and Au (Supplementary Fig. 13). istic should be dominant in WSe2 FETs with low-work-function
Using the formula for the thermionic emission effect metal contacts (because of the low electron SBH), and the p-type

Nature Electronics | VOL 5 | April 2022 | 241–247 | www.nature.com/natureelectronics 245

Articles Nature Electronics

with high-work-function metal contacts (because of the low hole than transfer and alloy metal contact fabrication methods, and
SBH). However, the transfer curves of all the WSe2 FETs with direct could be of use in the large-scale fabrication of devices.
metal contact exhibited n-type unipolar characteristics owing to
the strong FLP. In contrast, the transfer curves of WSe2 FETs with Methods
vdW metal contact exhibited the typical behaviour expected from Device fabrication. Mechanically exfoliated hexagonal boron nitride (hBN,
the work function of the metal. Thus, we observed that the WSe2 purchased from HQ Graphene) was stacked on a cleaned SiO2 substrate (285 nm)
on heavily doped Si using the polydimethylsiloxane stamp method, where hBN/
FET using the vdW contact with Ag exhibited clear n-type unipolar SiO2 and Si were used as the back-gate insulator and electrode, respectively. Next,
behaviour, whereas those with Co and Pt exhibited p-type domi- mechanically exfoliated TMDs (WSe2, WS2 and MoSe2 purchased from HQ
nant ambipolar and strong p-type unipolar behaviour, respectively, Graphene) were stacked on hBN using the polydimethylsiloxane stamp method.
as expected from the work functions and SBH. Therefore, we can For metal electrode fabrication, polymethyl methacrylate (PMMA) was spin coated
on the TMD/hBN stack, and a sample with PMMA was baked at 180 °C for 10 min.
adjust the type of FET by using the existence or non-existence of
Subsequently, electron-beam lithography was performed to pattern the electrode.
FLP or by changing the work function of the metal. In the case of direct metal contacts, 50-nm-thick Au (Co and Pt) and 100-nm-thick
We verified the usefulness of the vdW metal contact method Ag layers were deposited by electron-beam evaporation in vacuum (<10−6 torr). In
with several other TMDs (MoS2, MoSe2 and WS2) as the channel the case of vdW metal contacts, before metal deposition, a 10-nm-thick Se layer was
material (Fig. 4d–f). Despite Au’s high work function, the transfer deposited by thermal evaporation in high vacuum (<10−9 torr), after which the metal
was deposited by electron-beam evaporation in vacuum (<10−6 torr). Both direct and
curves of all the TMD FETs with direct Au contact showed n-type vdW metal contacts were soaked in acetone for removing PMMA. Next, devices with
unipolar behaviour. By contrast, the transfer curves of the MoSe2 vdW metal contacts were annealed at 150 °C in high vacuum (<10−9 torr) for 4 h to
and WS2 FETs using a vdW contact with Au showed p-type domi- remove the Se buffer layer. To exclude any differences in the channel properties, both
nant ambipolar behaviour. This is because the MoSe2 and WS2 work direct and vdW metal contact methods were used for fabrication on the same flakes.
functions (including electron affinity) are higher than that of WSe2, For samples used in XPS, the thickness of the Au layer was 1 nm.
indicating that the electron SBH between MoSe2 and WS2 is a little Spectroscopic measurements. To obtain the XPS spectra, we used a
lower than with WSe2 and its p-type unipolar behaviour. Further, high-resolution XPS instrument (PHI, VersaProbe). All the XPS spectra were
in the case of MoS2, which is an n-type TMD with a very high calibrated using the C1s reference peak at 284.8 eV. The Raman spectra were
work function, it was not possible to operate a p-type FET using a obtained using a micro-Raman spectrometer (HORIBA LabRAM ARAMIS) with a
vdW Au contact. However, we confirmed that FLP was controlled 532-nm laser and a grating of 1,800 g mm–1. All the Raman spectra were calibrated
using the Si reference peak at 520.4 cm−1.
through a threshold-voltage (Vth) difference of approximately
+30 V between direct contact and vdW contact. The n-type char- STEM. Cross-sectional transmission electron microscopy samples were prepared
acteristic became stronger as the TMD work function increased using a dual-beam focused ion beam system (Helios 450F1, Thermo Fisher
(ΦWSe2 < ΦMoSe2 ≒ ΦWS2 < ΦMoS2)47. Overall, the transfer curves of Scientific). A 300 kV aberration-corrected STEM system (Titan3 G2 60-300,
FETs based on various TMDs indicate that FLP between any type Thermo Fisher Scientific) equipped with double Cs correctors was used to acquire
the HAADF-STEM images of the atomic structure at the Au–WSe2 interface.
of TMD and any metal can be successfully controlled by the vdW
metal contact. In addition, as shown in Supplementary Fig. 16, we Electrical measurements. Current–voltage (I–V) measurements of all the
confirmed that the vdW contact using the Se buffer layer method electrical devices were performed with a Keithley 6400 system at room temperature
is enough to apply in monolayer 2D materials through the strong (~300 K). Low-temperature I–V measurements for determining the SBH were
p-type monolayer WSe2 FET with vdW Au contact. performed in vacuum.
The linear mobility of p-type FETs was calculated as
In Supplementary Figs. 17 and 18, ΦSBH according to the contact
method between various metals and WSe2 is obtained. The deriva- μ = −L/W × gm /(Vds × Cox ), (6)
tive of this quantity with respect to metal work function Φm is the S where L is the channel length, W is the channel width, Vds is the source–drain
factor (FLP strength)9,16,48: voltage, Cox is the capacitance of the back-gate oxide and gm is the transconductance
(gm = ∂Ids/∂VG). The back-gate voltage (VG) of all the devices in our study was
applied through an hBN (~30 nm)/SiO2 (285 nm) (Cox = 1.1 × 10−8 F cm−2) dielectric
S = dΦSBH /dΦm , (5) stack.
Through the output curve, we measured the total resistance (Rtot) of FETs with
If S = 1 (about SBH of electron) and S = −1 (about SBH of hole), different channel lengths and the same channel width and sheet resistance (Rs). The
the Schottky–Mott rule is realized. As shown in Fig. 5a, the S factor of contact resistance (2Rc) was extracted from the y intercept of the Rtot–L graph from
the vdW contact is SvdW = −0.91, which is close to −1.00. In contrast, the transmission-line measurement as
the S factor of direct contact is −0.11, which is very flat, indicating Rtot = 2Rc + Rch = 2Rc + L/WRs , (7)
that SBH is insensitive to the metal work function owing to strong where Rtot is the total resistance; Rc and Rch are the contact and channel resistances,
FLP. Figure 5b–e shows the shape of the output curves in vdW con- respectively; L is the channel length; W is the channel width; and Rs is the sheet
tact. Low-SBH contact with Ag, Au and Pt produces a linear curve, resistance of the channel.
implying ohmic contact, whereas a somewhat-higher-SBH contact
with Co produces a nonlinear shape, implying a Schottky contact. DFT calculations. To analyse the band structure of the five-layer WSe2 according
to the distance from Au, we conducted DFT calculations using the Vienna ab initio
Thus, the experimental SBH effectively explains the electrical prop- simulation package. Models of the vdW and direct Au contacts on WSe2 were
erties. With FLP successfully eliminated by the vdW metal contact, developed by optimizing the geometry of the two contact models such that the
the SBH in FETs using TMDs can be easily adjusted by appropri- nearest distance between the W and Au atoms for the vdW Au contact was 5.10 Å
ately selecting a metal with the desired work function. and that for the direct Au contact was 4.36 Å. For all the calculations, we used the
optB86b-vdW functional. Geometry optimization was performed until the energy
converged to less than 0.01 eV Å–1 with a 9 × 9 × 1 k-point grid such that the k
Conclusions spacing was less than 0.25 Å–1; the cut-off energy was set to 350 eV. The electronic
We have reported a method for fabricating defect-free vdW contacts structure was obtained with an 11 × 11 × 1 k-point grid to make the k spacing less
with weak interaction between pure metals and 2D semiconductors than 0.20 Å−1; again, the cut-off energy was set to 350 eV. After optimizing the
without FLP. Through theoretical and experimental microstructural geometry of each unit cell, we compared the DOS of all the WSe2 band structures,
analysis, we revealed that the effect of FLP depends on the inter- except for that of Au. To understand band bending in the band diagram at the
WSe2–Au interface, we calculated the potential energy and total charge density of
face gap between the 2D semiconductor and metal, helping to fur- the electrons according to the position of the z axis.
ther understand the cause of FLP. To illustrate the potential of our
approach, we fabricated FETs with vdW metal contacts using differ- Data availability
ent metals and 2D semiconductors. Creating vdW contacts using an The data that support the findings of this study are available from the
Se buffer layer could provide a more accurate way to control SBHs corresponding authors upon reasonable request.

246 Nature Electronics | VOL 5 | April 2022 | 241–247 | www.nature.com/natureelectronics

Nature Electronics Articles
Received: 30 August 2021; Accepted: 14 March 2022; 33. Gao, J. & Gupta, M. Titanium disulfide as Schottky/ohmic contact
Published online: 21 April 2022 for monolayer molybdenum disulfide. npj 2D Mater. Appl. 4,
26 (2020).
34. Min, K. A., Park, J., Wallace, R. M., Cho, K. & Hong, S. Reduction of Fermi
References level pinning at Au–MoS2 interfaces by atomic passivation on Au surface.
1. Tung, R. T. The physics and chemistry of the Schottky barrier height. Appl. 2D Mater. 4, 015019 (2017).
Phys. Rev. 1, 011304 (2014). 35. English, C. D., Shine, G., Dorgan, V. E., Saraswat, K. C. & Pop, E. Improved
2. Allain, A., Kang, J., Banerjee, K. & Kis, A. Electrical contacts to contacts to MoS2 transistors by ultra-high vacuum metal deposition. Nano
two-dimensional semiconductors. Nat. Mater. 14, 1195–1205 (2015). Lett. 16, 3824–3830 (2016).
3. Schottky, W. Halbleitertheorie der Sperrschicht. Naturwissenschaften 26, 36. Zhang, R., Drysdale, D., Koutsos, V. & Cheung, R. Controlled layer
843 (1938). thinning and p-type doping of WSe2 by vapor XeF2. Adv. Funct. Mater. 27,
4. Hasegawa, H. & Sawada, T. On the electrical properties of compound 1702455 (2017).
semiconductor interfaces in metal/insulator/ semiconductor structures and 37. Shokouh, S. H. H. et al. High-performance, air-stable, top-gate, p-channel
the possible origin of interface states. Thin Solid Films 103, 119–140 (1983). WSe2 field-effect transistor with fluoropolymer buffer layer. Adv. Funct. Mater.
5. Kong, L. et al. Doping-free complementary WSe2 circuit via van der Waals 25, 7208–7214 (2015).
metal integration. Nat. Commun. 11, 1866 (2020). 38. Fang, H. et al. High-performance single layered WSe2 p-FETs with chemically
6. Wang, Y. et al. Does p-type ohmic contact exist in WSe2–metal interfaces? doped contacts. Nano Lett. 12, 3788–3792 (2012).
Nanoscale 8, 1179–1191 (2016). 39. Chuang, H. J. et al. High mobility WSe2 p- and n-type field-effect transistors
7. Liu, Y., Stradins, P. & Wei, S. H. Van der Waals metal-semiconductor contacted by highly doped graphene for low-resistance contacts. Nano Lett.
junction: weak Fermi level pinning enables effective tuning of Schottky 14, 3594–3601 (2014).
barrier. Sci. Adv. 2, e1600069 (2016). 40. Allain, A. & Kis, A. Electron and hole mobilities in single WSe2. ACS Nano 8,
8. Liu, Y., Huang, Y. & Duan, X. Van der Waals integration before and beyond 7180–7185 (2014).
two-dimensional materials. Nature 567, 323–333 (2019). 41. Movva, H. C. P. High-mobility holes in dual-gated WSe2 field-effect
9. Kim, C. et al. Fermi level pinning at electrical metal contacts of monolayer transistors. ACS Nano 9, 10402–10410 (2015).
molybdenum dichalcogenides. ACS Nano 11, 1588–1596 (2017). 42. Yamamoto, M., Nakaharai, S., Ueno, K. & Tsukagoshi, K. Self-limiting oxides
10. Jang, J. et al. Clean interface contact using a ZnO interlayer for on WSe2 as controlled surface acceptors and low-resistance hole contacts.
low-contact-resistance MoS2 transistors. ACS Appl. Mater. Interfaces 12, Nano Lett. 16, 2720–2727 (2016).
5031–5039 (2020). 43. Xu, S. et al. Universal low-temperature ohmic contacts for quantum transport
11. Mleczko, M. J. et al. Contact engineering high-performance n-type MoTe2 in transition metal dichalcogenides. 2D Mater. 3, 021007 (2016).
transistors. Nano Lett. 19, 6352–6362 (2019). 44. Chen, C. H. et al. Hole mobility enhancement and p-doping in monolayer
12. Stanford, M. G. et al. High conduction hopping behavior induced in WSe2 by gold decoration. 2D Mater. 1, 034001 (2014).
transition metal dichalcogenides by percolating defect networks: toward 45. Chuang, H. J. et al. Low-resistance 2D/2D ohmic contacts: a universal
atomically thin circuits. Adv. Funct. Mater. 27, 1702829 (2017). approach to high-performance WSe2, MoS2, and MoSe2 transistors. Nano Lett.
13. Zhou, W. et al. Intrinsic structural defects in monolayer molybdenum 16, 1896–1902 (2016).
disulfide. Nano Lett. 13, 2615–2622 (2013). 46. Smyth, C. M., Addou, R., McDonnell, S., Hinkle, C. L. & Wallace, R. M.
14. Santosh, K. C., Longo, R. C., Addou, R., Wallace, R. M. & Cho, K. Impact of WSe2-contact metal interface chemistry and band alignment under high
intrinsic atomic defects on the electronic structure of MoS2 monolayers. vacuum and ultra high vacuum deposition conditions. 2D Mater. 4,
Nanotechnology 25, 375703 (2014). 025084 (2017)
15. Hong, J. et al. Exploring atomic defects in molybdenum disulphide 47. Kim, H. G. & Choi, H. J. Thickness dependence of work function, ionization
monolayers. Nat. Commun. 6, 6293 (2015). energy, and electron affinity of Mo and W dichalcogenides from DFT and
16. Liu, Y. et al. Approaching the Schottky–Mott limit in van der Waals GW calculations. Phys. Rev. B 103, 085404 (2021).
metal–semiconductor junctions. Nature 557, 696–700 (2018). 48. Michaelson, H. B. The work function of the elements and its periodicity.
17. Wang, Y. et al. Van der Waals contacts between three-dimensional metals and J. Appl. Phys. 48, 4729 (1977).
two-dimensional semiconductors. Nature 568, 70–74 (2019).
18. Mattox, D. M. The Foundations of Vacuum Coating Technology
(Springer, 2003). Acknowledgements
19. Bondi, A. Van der Waals volumes and radii. J. Phys. Chem. 68, This work was partially supported by an Industry–Academy joint research programme
441–451 (1964). between Samsung Electronics and Yonsei University. This work was supported by the
20. Bitla, Y. & Chu, Y. H. Van der Waals oxide heteroepitaxy for soft transparent National Research Foundation of Korea (NRF) grant funded by the Korean government
electronics. Nanoscale 12, 18523–18544 (2020). (grant no. 2017R1A5A1014862; SRC programme, vdWMRC center). This work
21. Desai, S. B. et al. Gold-mediated exfoliation of ultralarge was supported by the NRF grant funded by the Government of Korea (MSIP) (no.
optoelectronically-perfect monolayers. Adv. Mater. 28, 4053–4058 (2016). 2021M3H4A1A03052566).
22. Addou, R. & Wallace, R. M. Surface analysis of WSe2 crystals: spatial and
electronic variability. ACS Appl. Mater. Interfaces 8, 26400–26406 (2016). Author contributions
23. Sezen, H. & Suzer, S. Communication: enhancement of dopant dependent G.K. conceived the research and designed the experiments with the help of M.-H.C.
X-ray photoelectron spectroscopy peak shifts of Si by surface photovoltage. G.K., Y.-H.C., H. Kim and J.J. contributed to the device fabrication and setting the
J. Chem. Phys. 135, 141102 (2011). equipment. G.K., Y.-H.C., H. Kim, J.J., M.B. and H. Kwon performed the experiments
24. Smyth, C. M., Addou, R., McDonnell, S., Hinkle, C. L. & Wallace, R. M. and data analysis. K.J. contributed to the DFT calculations. H.L., J.A. and E.L. made
Contact metal–MoS2 interfacial reactions and potential implications on a significant contribution to the STEM measurement and analysis. The manuscript
MoS2-based device performance. J. Phys. Chem. C 120, 14719–14729 (2016). was written by G.K. and revised by M.-H.C. All the experiments and analyses were
25. Qu, D. et al. Carrier-type modulation and mobility improvement of thin supervised by M.-H.C.
MoTe2. Adv. Mater. 29, 1606433 (2017).
26. Shen, P. C. et al. Ultralow contact resistance between semimetal and
monolayer semiconductors. Nature 593, 211–217 (2021).
Competing interests
The authors declare no competing interests.
27. Huang, J. et al. Stable electrical performance observed in large-scale monolayer
WSe2(1-x)S2x with tunable band gap. Nanotechnology 27, 13LT01 (2016).
28. Ramos, R., Cunge, G., Touzeau, M. & Sadeghi, N. Measured velocity Additional information
distribution of sputtered Al atoms perpendicular and parallel to the target. Supplementary information The online version contains supplementary material
J. Phys. D 41, 152003 (2008). available at https://doi.org/10.1038/s41928-022-00746-6.
29. Lim, J. Y. et al. Homogeneous 2D MoTe2 p–n junctions and CMOS inverters
formed by atomic-layer-deposition-induced doping. Adv. Mater. 29, Correspondence and requests for materials should be addressed to
1701798 (2017). Eunha Lee or Mann-Ho Cho.
30. Popov, I., Seifert, G. & Tománek, D. Designing electrical contacts to MoS2 Peer review information Nature Electronics thanks the anonymous reviewers for their
monolayers: a computational study. Phys. Rev. Lett. 108, 156802 (2012). contribution to the peer review of this work.
31. Fan, Z. Q. et al. In-plane Schottky-barrier field-effect transistors based on Reprints and permissions information is available at www.nature.com/reprints.
1T/2H heterojunctions of transition-metal dichalcogenides. Phys. Rev. B 96,
165402 (2017). Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
32. Zhong, H. et al. Interfacial properties of monolayer and bilayer MoS2 contacts published maps and institutional affiliations.
with metals: beyond the energy band calculations. Sci. Rep. 6, 21786 (2016). © The Author(s), under exclusive licence to Springer Nature Limited 2022

Nature Electronics | VOL 5 | April 2022 | 241–247 | www.nature.com/natureelectronics 247