Das Et Al 2012 High Performance Multilayer Mos2 Transistors With Scandium Contacts
Das Et Al 2012 High Performance Multilayer Mos2 Transistors With Scandium Contacts
pubs.acs.org/NanoLett
© 2012 American Chemical Society 100 dx.doi.org/10.1021/nl303583v | Nano Lett. 2013, 13, 100−105
Nano Letters Letter
gold (ΦM = 5.4 eV)-to-MoS2 contacts are Ohmic,8,15,25 our lithographic techniques were employed to pattern the source/
results clearly indicate the existence of Schottky barriers for drain contacts, and electron beam evaporation was used to
large work function metals like Ni and Pt. deposit the desired metal contacts (Sc, Ti, Ni, and Pt). The
We have also evaluated the field effect mobility of multilayer channel length for all of the devices is 5 μm, and the flake
MoS2 flakes for different metal contacts and different flake thicknesses range between 2 and 12 nm (3−18 mono layers).
thicknesses using a back-gated transistor geometry. As The deposited source/drain contact film thickness is 50 nm for
expected, the extracted mobility values were strongly depend- Ti, Ni, and Pt while to avoid damaging of the contacts in case of
ent on the metal-to-MoS2 contact interface; that is, the intrinsic scandium required a 30 nm Sc/20 nm Ni stack.
mobility is masked easily by the presence of the aforementioned Figure 2a shows the expected line-up of the metal Fermi level
Schottky barrier at the source and drain contacts. We have also with the electronic bands of MoS2 if only the difference of the
observed a nonmonotonic trend in the intrinsic field effect
mobility of MoS2 as a function of the MoS2 layer thickness. To
describe our experimental findings we have employed a model
including Thomas−Fermi charge screening and interlayer
coupling. The highest mobility values were obtained for a
finite layer thickness of around 10 nm. In this case the extracted
mobility values are 21, 90, 125, and 184 cm2/(V s) for Pt, Ni,
Ti ,and Sc contacts, respectively, clearly indicating that
improving the contact quality is essential to harvest the
intrinsic material properties.
A significant enhancement in the field effect mobility up to
700 cm2/(V s) is achieved by covering the top of the back-
gated MoS2 transistor with a thin layer of 15-nm-thick Al2O3.
While the mobility we find is substantially higher than
previously reported, the effect itself had been also observed
by other groups.8,10,12
Extraction of Schottky Barrier Height. Single-layer MoS2
consists of a stack of three hexagonally packed atomic layers of
sulfur−molybdenum−sulfur with strong intralayer covalent
bonding.16,17 In contrast, bulk MoS2 is a stack of single layers,
held together by weak van der Waals interlayer interaction and,
therefore, allows for micromechanical exfoliation of mono or Figure 2. (a) Expected (not correct) line-up of metal Fermi level with
few layerssimilar to the fabrication of graphene from the electronic bands of MoS2 flake if only the difference of the electron
affinity of MoS2 and the work function of the corresponding metal is
graphite. Figure 1a,b shows atomic force microscopy (AFM) considered. (b) The cartoon of expected transfer characteristics based
and optical images of a 3 nm (4 monolayer) thick MoS2 flake, on part a. (c) Transfer characteristics of back-gated 6-nm-thick MoS2
exfoliated using standard scotch tape technique on 100 nm transistor with Sc, Ti, Ni, and Pt metal contacts. The inset shows the
silicon dioxide (SiO2) substrate with underlying highly doped actual line-up based on the experimental data.
silicon. Figure 1c,d shows a 3-D sketch and an scanning
electron microscopy (SEM) image of a prototype back-gated
MoS2 transistor used for our contact study. Standard electron affinity of MoS2 and the work function of the
corresponding metal is considered. Electron injection should
occur for Sc and Ti, while Ni and Pt are expected to provide
access to the valence band and should thus enable hole
injection. Figure 2b illustrates the expected device character-
istics qualitatively under this assumption. On the other hand
Figure 2c shows the experimental transfer characteristics of
back gated MoS2 transistors with Sc, Ti, Ni, and Pt metal
contacts after adjusting for the threshold voltage shift. The
threshold voltages for Sc, Ti, Ni, and Pt contact devices were
found to be on average (sampled over 4−5 devices) −6.0, −1.0,
1.5, and 4.0 V, respectively. Threshold voltages are defined by
the deviation of the IDS versus VGS characteristics from their
exponential behavior in the subthreshold regime. Note that the
trend in the values of the threshold voltages is consistent with
the observed trend in on-currents.28 Interestingly, all of the
metal contacts exhibit n-type FET characteristics, which
indicate that the Fermi levels for all of these metals line-up
close to the conduction band edge of MoS2 as shown in the
inset of Figure 2c. Also, there is a clear trend of decreasing on-
state performance, that is, decreasing on-current for positive
Figure 1. (a) AFM image and (b) AFM height profile of a 3 nm thin VGS−VTH from Sc to Pt, suggesting a change in the charge
MoS2 flake. (c) SEM image and (d) 3-D cartoon of a prototype back- injection properties of the various contact metals. As will be
gated MoS2 transistor used for the study. discussed below these findings are consistent with Schottky
101 dx.doi.org/10.1021/nl303583v | Nano Lett. 2013, 13, 100−105
Nano Letters Letter
Figure 3. (a) Experimental transfer characteristics of a back-gated MoS2 transistor with Ni contacts at different operating temperatures. The insets
represent the energy band diagrams corresponding to the applied gate biases in three distinct regions: below flat band, below threshold, and above
threshold. (b) Arrhenius-type plot constructed using part a at different gate voltages and (c) extracted effective barrier height (ΦB) as a function of
applied gate voltages for Ni contact devices. (d) Extracted true Schottky barrier height ΦSB for different work function metals like Sc, Ti, Ni, and Pt.
The dotted line is a guide to the eyes.
barriers of varying heights formed at the metal-to-channel device, A is the Richardson’s constant, kB is the Boltzmann
interface, and a detailed temperature-dependent study has been constant, q is the electronic charge, T is the temperature, and
used to quantitatively analyze those barriers. VDS is the source to drain bias. VFB can now be readily
As described in greater detail in ref 29, tunneling through determined from ΦB (at the source-to-channel interface) as a
Schottky barrier at the source-to-channel interface is not only function of gate voltage. Figure 3c shows our findings for Ni
limiting the charge injection in the device on-state but also electrodes. For sufficiently negative gate voltages the effective
plays a critical role when evaluating the device off-state in the barrier height extracted from Figure 3b linearly responds to the
subthreshold region of the transistor. In fact it is the thermally gate voltage VGS (also see the Supporting Information). This is
assisted tunneling that impacts the inverse subthreshold slope a result of the fact that only Ithermionic determines the current
in Schottky barrier devices in general.30 The band bending flow through the device. It is when ΦB deviates from this linear
diagrams in Figure 3a illustrate the contributions of thermionic trend when Itunneling becomes relevant and VFB is reached. Note
emission current component (Ithermionic) and the thermally that the threshold voltage VTH that defines the device on-state
assisted tunneling current component (Itunneling) under different is larger than VFB. From Figure 3c, a true Schottky barrier
gate voltage conditions. Figure 3a shows the change in current height of 150 meV is extracted. Figure 3c also illustrates how
through the device with the gate voltage for different the wrong choice of gate voltageone that is larger than VFB
temperatures that are used to create the Arrhenius plot in can easily result in the extraction of a barrier height that is too
Figure 3b. When evaluating the current flow through the device small since thermally assisted tunneling adds another current
as a function of temperature, both contributionsIthermionic and component that is not included in the standard thermal
Itunnelingneed to be considered. The band diagrams also emission theory. It is this additional tunneling current that is
illustrate that the tunneling current can only be ignored when responsible for a perceived “Ohmic contact” at the interface
the gate voltage is below the flat band voltage VFB. If one can even for substantial Schottky barrier heights. The key is that an
identify VFB, the Arrhenius plot for this gate voltage will reveal ultrathin channel like MoS2 will always give rise to a short
the true Schottky barrier height ΦSB.31 To do so, the slope of tunneling distance and thus substantial tunneling currents as
the curves in Figure 3b in the high-temperature region is has been discussed in greater detail previously.32,33
analyzed assuming conventional thermionic emission theory. Figure 3d shows the extracted Schottky barrier heights for
Using the equation IDS = AT2 exp((qϕB)/(kBT)[1 − exp- the different metals used for our contact study as a function of
((qVDS)/(kBT))], the effective Schottky barrier height is their corresponding work functions. We find Schottky barrier
extracted. In this equation IDS is the current through the heights of 230, 150, 50, and 30 meV for Pt, Ni, Ti, and Sc,
102 dx.doi.org/10.1021/nl303583v | Nano Lett. 2013, 13, 100−105
Nano Letters Letter
respectively. In total more than 4−5 devices were characterized to drain bias (VDS) for low VDS values which is consistent with
per metal contact type under investigation. The extracted slope the fact that Pt exhibits the largest Schottky barrier height of
dΦSB/ dΦM of around 0.1 (note that this number is 230 meV. The current through Ni contacted devices, however,
approximately 0.27 in the case of silicon34) indicates that shows a linear dependence on VDS despite the presence of a
different from previous assumptions strong pinning at the sizable Schottky barrier of 150 meV at the Ni−MoS2 interface.
metal/MoS2 interface determines the line-up between the metal As discussed above, thermally assisted tunneling is responsible
Fermi level and the conduction band of MoS2. It also explains for this finite temperature effect, and consequently a linear
why only n-type characteristics are obtained in all reported relationship between current and drain voltage does not
cases27 even if large work function metals had been used. Note necessarily indicate an Ohmic contact.33 The increasing
that if the chemical reactivity, deposition-induced mechanical magnitude of the current from 0.1 to 0.7 to 1.0 to 4.8 μA/
deformation, or wettability of the metals was to influence the μm at a VDS of 1 V for Pt, Ni, Ti, and Sc contacted devices,
Schottky barrier height analysis significantly, the linear trend in respectively, is consistent with the fact that smaller Schottky
Figure 3d would be obscured. barrier heights at the metal-to-MoS2 interface result in better
Impact of Metal Contact. Next we turn our attention to carrier injection and hence smaller contact resistance values.
the impact of the Schottky barrier heights on the device Figure 4 also shows the transfer characteristics of 10-nm-
characteristics. The inset of Figure 4 shows the output thick MoS2 transistors with Sc, Ti, Ni, and Pt metal contacts for
VDS = 0.2 V at room temperature. The transconductance gm
(defined as gm = dIDS/dVGS) normalized by the width (gm/W)
shows a monotonic increase from 0.03, 0.14, 0.20 to 0.28 μS/
μm from Pt, Ni, Ti to Sc contacted devices. Field effect mobility
values of the ∼10 nm MoS2 flakes were extracted using the
conventional equation of gm = μnCOX(W/L)VDS at low VDS =
0.2 V (where μn is the field effect mobility and W and L are the
channel width and the channel length, respectively, COX = εOX/
dOX, where εOX is the dielectric constant and dOX is the
thickness of the gate oxide, dOX = 100 nm and for SiO2, εOX = 3
× 10−11, which gives COX ∼ 3 × 10−4 F/m2, and finally L ∼ 5
μm). The extracted effective field-effect mobility values were
found to be 21, 36, 125, and 184 cm2/(V s) for Pt, Ni, Ti, and
Sc contacted devices respectively, clearly highlighting how even
Figure 4. Transfer characteristics of 10 nm thin MoS2 back-gated small Schottky barriers can impact the mobility extraction
transistors with Sc, Ti, Ni, and Pt metal contacts at 300 K for VDS = 0.2
V. The inset shows the output characteristics of the corresponding significantly. Note that the maximum values of gm were used to
devices for a gate voltage overdrive of 4.0−5.0 V. extract the mobility values. The above findings clearly indicate
the importance of proper contact formation to harvest the
intrinsic properties of novel nano materials and explain the
characteristics of a back-gated ∼10-nm-thick MoS2 transistors extremely high mobility values reported below by the use of Sc
at 300 K for different metal contacts for the same gate voltage contacts. Also note that the mobility values extracted by
overdrive (VGS − VTH = 4 V) which corresponds to a carrier Radisavljevic et al.,8 for monolayer MoS2 and Wang et al.,12 for
concentration of ∼1.4 × 1012/cm2. The current through the Pt bilayer MoS2 using a back-gated device geometry are similar to
contact device shows an exponential dependence on the source ours prior to the deposition of a high-k dielectric layer and are
Figure 5. (a) Transfer characteristics of 20, 10, 6, 4, and 2 nm thick MoS2 field effect transistor with Sc as the source/drain contacts at room
temperature for VDS = 0.2 V. The inset shows the output characteristics for the same flakes for a gate overdrive voltage of 2 V. (b) The extracted
effective field effect mobility as a function of the MoS2 layer thickness. The dotted line is a fit to the experimental data using a model (inset) that
includes Thomas−Fermi screening with λ = 3.0 nm and interlayer conductivity with σint = 1 mS. The solid line is the simulated field effect mobility
without any interlayer resistance.
■
eliminated.
High-k for High Performance. Finally, to prove that the
above-reported mobility values are on par or even exceed ASSOCIATED CONTENT
previously reported experimental data, we followed an approach *
S Supporting Information
reported before that has shown to result in a substantial Extraction of Schottky barrier height: Sc, Ti, Ni, and Pt contact.
improvement in mobility after atomic layer deposition (ALD) This material is available free of charge via the Internet at
of high-k materials onto MoS2.8,11,12,18 We have combined this http://pubs.acs.org.
treatment with our findings about the ideal contact and layer
thickness from above by depositing 15 nm of Al2O3 on top of
our back-gated transistors. Figure 6 shows the measured output
■ AUTHOR INFORMATION
Notes
The authors declare no competing financial interest.
■
and transfer characteristics of a 10 nm MoS2 FET with
scandium contacts and ALD layer. The effective field-effect
mobility of the MoS2 flake extracted assuming a parallel plate ACKNOWLEDGMENTS
capacitor arrangement (ignoring any possible impact of fringing We would like to acknowledge Dave Lubelski for helping with
fields and the top dielectric) is increased by a factor of 3.8, from the Sc deposition and Ali Razavieh for ALD deposition. This
184 to 700 cm2/(V s). As mentioned above, this high value is a work was supported by the Nanotechnology Research Initiative
104 dx.doi.org/10.1021/nl303583v | Nano Lett. 2013, 13, 100−105
Nano Letters Letter
(NRI) through a supplement to the Network for Computa- (21) Eda, G.; Yamaguchi, H.; Voiry, D.; Fujita, T.; Chen, M.;
tional Nanotechnology (NCN), which is supported by National Chhowalla, M. Photoluminescence from Chemically Exfoliated MoS2.
Science Foundation (NSF) under grant no. EEC-0634750. Nano Lett. 2011, 11, 5111−16.
■
(22) Li, H.; et al. Fabrication of Single- and Multilayer MoS2 Film-
Based Field-Effect Transistors for Sensing NO at Room Temperature.
REFERENCES Small 2012, 8, 63−67.
(1) Novoselov, K. S.; Jiang, D.; Schedin, F.; Booth, T. J.; Khotkevich, (23) Xie, R.; Phung, T. H.; Yu, M.; Zhu, C. Effective Surface
V. V.; Morozov, S. V.; Geim, A. K. Two-dimensional atomic crystals. Passivation by Novel SiH4−NH3 Treatment and BTI Characteristics
Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 10451−53. on Interface-Engineered High-Mobility HfO2-Gated Ge pMOSFETs.
(2) Kim, S.; Nah, J.; Jo, I.; Shahjerdi, D.; Colombo, L.; Yao, Z.; IEEE Trans. Electron Devices 2010, 57, 1399−1407.
Tutuc, E.; Banerjee, S. K. Realization of a high mobility dual-gated (24) Martens, K.; Chui, C. O.; Brammertz, G.; de Jaeger, B.; Kuzum,
graphene field-effect transistor with Al2O3 dielectric. Appl. Phys. Lett. D.; Meuris, M.; Heyns, M.; Krishnamohan, T.; Saraswat, K.; Maes, H.
2009, 94, 062107. E.; Groeseneken, G. On the Correct Extraction of Interface Trap
(3) Lipp, A.; Schwetz, K. A.; Hunold, K. Hexagonal Boron Nitride: Density of MOS Devices with High-Mobility Semiconductor
Fabrication, Properties and Applications. J. Eur. Ceram. Soc. 1989, 5, Substrates. IEEE Trans. Electron Devices 2008, 55, 547−56.
3−9. (25) Neal, A. T.; Liu, H.; Gu, J. J.; Ye, P. D. Metal Contact to MoS2.
(4) Avouris, P. Graphene: Electronic and Photonic Properties and Proc. DRC 2012, 65−66.
Devices. Nano Lett. 2010, 10, 4285−94. (26) Popov, I.; Seifert, G.; Tománek, D. Designing electrical contacts
(5) Zhang, H.; Liu, C.; Qi, X.; Dai, X.; Fang, Z.; Zhang, S. to MoS2 monolayers: A computational study. Phys. Rev. Lett. 2012,
Topological insulators in Bi2Se3, Bi2Te3, and Sb2Te3 and with a single 108, 156802.
dirac cone on the surface. Nat. Phys. 2009, 5, 438−442. (27) We note at this point that an article by Fontana et al.38 found p-
(6) Joensen, P.; Frindt, R. F.; Morrison, S. R. Single-layer MoS2. type conduction with Pd contacts. This data set is the only one we are
Mater. Res. Bull. 1986, 21, 457−461. aware of that does not follow the trends published by other authors
(7) Wilson, J. A.; et al. The transition metal di-chalcogenides using large work function metal contacts and observed by us as
discussion and interpretation of the observed optical, electrical and described here.
structural properties. Adv. Phys. 1969, 18, 193−335. (28) For scaled gate oxides, the threshold voltage of a metal
(8) Radisavljevic, B.; Radenovic, A.; Brivio, J.; Giacometti, V.; Kis, A. contacted field-effect transistor is in general expected to be
Single-layer MoS2 transistors. Nat. Nanotechnol. 2011, 6, 147−150. independent of the contact metal used. However, the relatively thick
(9) Hwang, W. S.; Remskar, M.; Yan, R.; Protasenko, V.; Tahy, K.; back gate oxide used in our study results in the contact Schottky
Chae, S. D.; Zhao, P.; Konar, A.; Xing, H. G.; Seabaugh, A.; Jena, D. barrier height impacting the actual threshold voltage value due to a
Transistors with chemically synthesized layered semiconductor WS2 transmission probability substantially smaller than unity.
exhibiting 105 room temperature modulation and am-bipolar behavior. (29) Appenzeller, J.; Radisavljević, M.; Knoch, J.; Avouris, Ph.
Appl. Phys. Lett. 2012, 101, 013107. Tunneling Versus Thermionic Emission in One-Dimensional Semi-
(10) Fang, H.; Chuang, S.; Chang, T. C.; Takei, K.; Takahashi, T.; conductors. Phys. Rev. Lett. 2004, 92, 048301.
Javey, A. High-Performance Single Layered WSe2 p-FETs with (30) Appenzeller, J.; Knoch, J.; Derycke, V.; Martel, R.; Wind, S.;
Chemically Doped Contacts. Nano Lett. 2012, 12, 3788−92. Avouris, Ph. Field-Modulated Carrier Transport in Carbon Nanotube
(11) Radisavljevic, B.; Whitwick, M. B.; Kis, A. Integrated Circuits Transistors. Phys. Rev. Lett. 2002, 89, 126801.
and Logic Operations Based on Single-Layer MoS2. ACS Nano 2011, (31) Note that our Schottky barrier analysis does not take into
5, 9934−38. account any temperature dependent scattering in the MoS2 channel.
(12) Wang, H.; Yu, L.; Lee, Y.-H.; Shi, Y.; Hsu, A.; Chin, M. L.; Li, However, because of the fact that typical scattering rates exhibit a
L.-J.; Dubey, M.; Kong, J.; Palacios, T. Integrated Circuits Based on power law dependence on temperature and not an exponential
dependence as the Schottky barrier transmission probability, the slope
Bilayer MoS2 Transistors. Nano Lett. 2012, 12, 4674−80.
of the Arrhenius type plot is believed to be a fairly accurate measure of
(13) Bolotin, K. I.; Sikes, K. J.; Jiang, Z.; Klima, M.; Fudenberg, G.;
the Schottky barrier heights.
Hone, J.; Kim, P.; Stormer, H. L. Ultrahigh electron mobility in
(32) Appenzeller, J.; Knoch, J.; Bjork, M. T.; Riel, H.; Schmid, H.;
suspended graphene. Solid State Commun. 2008, 146, 351−55.
Riess, W. Towards nanowire electronics. IEEE Trans. Electron Devices
(14) Yoon, Y.; Ganapathi, K.; Salahuddin, S. How Good Can
2008, 54, 2827−45.
Monolayer MoS2 Transistors Be? Nano Lett. 2011, 11, 3768−73.
(33) Heinze, S.; Tersoff, J.; Martel, R.; Derycke, V.; Appenzeller, J.;
(15) Chang, C.; Zhang, W.; Lee, Y.-H.; Lin, Y.-C.; Chang, M.-T.; Su,
Avouris, Ph. Carbon Nanotubes as Schottky barrier transistors. Phys.
C.-Y.; Chang, C.-S.; Li, H.; Shi, Y.; Zhang, H.; Lai, C.-S.; Li, L.-J.
Rev. Lett. 2002, 89, 106801.
Growth of Large-Area and Highly Crystalline MoS2 Thin Layers on
(34) Sze, S. M. Semiconductor Devices, 2nd ed.; John Wiley & Sons:
Insulating Substrates. Nano Lett. 2012, 12, 1538−44.
New York.
(16) Han, S. W.; Kwon, H.; Kim, S. K.; Ryu, S.; Yun, W. S.; Kim, D.
(35) Note that even in the case of small Schottky barriers the term
H.; Hwang, J. H.; Kang, J.-S.; Baik, J.; Shin, H. J.; Hong, S. C. Band-gap
“Ohmic contacts” is ONLY justified if gating does not include the
transition induced by interlayer van der Waals interaction in MoS2.
metal electrode-to-channel interface.
Phys. Rev. B 2011, 84, 045409.
(36) Sui, Y.; Appenzeller, J. Screening and Interlayer Coupling in
(17) Salmani, J. M.; Tan, Y.; Klimeck, G. Single Layer MoS2 Band
multilayer graphene field effect transistors. Nano Lett. 2009, 9, 2973−
Structure and Transport. Int. Semiconductor Device Res. Symp. 2011, pp
77.
1−2; DOI: 10.1109/ISDRS.2011.6135408.
(37) Any insulating substrate (gate oxides) has potential variations
(18) Liu, H.; Ye, P. D. MoS2 Dual-Gate MOSFET with Atomic-
due to fixed charges that reside in them. Thicker the layer thickness
Layer-Deposited Al2O3 as Top-Gate Dielectric. IEEE Trans. Electron
higher is the potential variation.
Devices 2012, 33, 546−48. (38) Fontana, M.; Deppe, T.; Boyd, A. K.; Rinzan, M.; Liu, A. Y.;
(19) Qiu, H.; Pan, L.; Yao, Z.; Li, J.; Shi, Y.; Wang, X. Electrical Paranjape, M.; Barbara, P. Photovoltaic effect in gated MoS2 Schottky
characterization of back-gated bi-layer MoS2 field-effect transistors and junctions. arXiv:1206.6125.
the effect of ambient on their performances. Appl. Phys. Lett. 2012, (39) Lembke, D.; Kis, A. Breakdown of High Performance
100, 123104. Monolayer MoS2 Transistors. ACS Nano 2012, 6, 10070−75.
(20) Lee, K.; Kim, H.-Y.; Lotya, M.; Coleman, J. N.; Kim, G.-T.;
Duesberg, G. S. Electrical Characteristics of Molybdenum Disulfide
Flakes Produced by Liquid Exfoliation. Adv. Mater. 2011, 23, 4178−
82.