Accurate Extraction of Schottky Barrier Height and Universality of Fermi Level De-pinning of van der Waals Contacts
Krishna Murali, Medha Dandu, Kenji Watanabe, Takashi Taniguchi, Kausik Majumdar
AAccurate Extraction of Schottky Barrier Heightand Universality of Fermi Level De-pinning of vander Waals Contacts
Krishna Murali, † , ‡ Medha Dandu, † , ‡ Kenji Watanabe, ¶ Takashi Taniguchi, § andKausik Majumdar ∗ , ‡ † Equal contribution ‡ Department of Electrical Communication Engineering, Indian Institute of Science,Bangalore 560012, India ¶ Research Center for Functional Materials, National Institute for Materials Science, 1-1Namiki, Tsukuba 305-044, Japan § International Center for Materials Nanoarchitectonics, National Institute for MaterialsScience, 1-1 Namiki, Tsukuba, 305-044 Japan
E-mail: [email protected]
Abstract
Due to Fermi level pinning (FLP), metal-semiconductor contact interfaces resultin a Schottky barrier height (SBH), which is usually difficult to tune. This makesit challenging to efficiently inject both electrons and holes using the same metal -an essential requirement for several applications, including light-emitting devices andcomplementary logic. Interestingly, modulating the SBH in the Schottky-Mott limitof de-pinned van der Waals (vdW) contacts becomes possible. However, accurateextraction of the SBH is essential to exploit such contacts to their full potential. In this a r X i v : . [ c ond - m a t . m e s - h a ll ] F e b ork, we propose a simple technique to accurately estimate the SBH at the vdW contactinterfaces by circumventing several ambiguities associated with SBH extraction. Usingthis technique on several vdW contacts, including metallic 2H-TaSe , semi-metallicgraphene, and degenerately doped semiconducting SnSe , we demonstrate that vdWcontacts exhibit a universal de-pinned nature. Superior ambipolar carrier injectionproperties of vdW contacts are demonstrated (with Au contact as a reference) in twoapplications, namely, (a) pulsed electroluminescence from monolayer WS using few-layer graphene (FLG) contact, and (b) efficient carrier injection to WS and WSe channels in both nFET and pFET modes using 2H-TaSe contact.
2o be able to minimize the fractional contribution of the metal-semiconductor contactresistance in the total device resistance is a key to achieve high-performance electronic de-vices.
Unfortunately, with device scaling, this fractional contribution increases due to areduction in the area of the contact footprint. The Schottky barrier height (SBH) at themetal-semiconductor interface plays a key role in determining the contact resistance. In anideal contact interface, the SBH ( φ B ) is given by the difference between the metal work func-tion ( W m ) and the semiconductor electron affinity ( χ s ), as predicted by the Schottky-Mottrule: φ B = W m − χ s (1)However, in a real scenario, the interfacial dipole, metal induced gap states (MIGS), andinterfacial defects lead to Fermi level pinning (FLP), which has plagued the developmentof ultra-low resistance contacts for several materials, including n-Ge, p-InGaAs, andp-MoS . In the context of the layered semiconductors, the possibility of ultra-clean, atomically smooth,dangling-bond-free surface, coupled with a relatively weak van der Waals (vdW) interactionat the interface reignited the hope to achieve a completely de-pinned contact close to theSchottky-Mott limit.
However, the estimated SBH value is often confounded by severalambiguities related to the extraction procedure. Given the scientific and technological im-portance, it is highly desirable to extract the “true” SBH values of the vdW contacts in anunambiguous manner. In this work, we propose a simple, yet powerful technique to achievethe same. We show that while the Au Fermi level is strongly pinned close to the conductionband edge of multi-layer WS , several types of vdW contacts including few-layer graphene(FLG), metallic 2H-TaSe , and degenerately doped semiconducting SnSe exhibit highlyde-pinned nature.Is a de-pinned contact always technologically desirable? Not necessarily in every situation.For example, in unipolar device applications, a contact strongly pinned close to the bandedge is desirable to reduce SBH and hence the contact resistance, as in the case of n-InAs, and n-MoS . Also, if the de-pinning is achieved by reduced interaction betweenthe metal and the semiconductor, it can adversely affect the tunneling efficiency of the car-riers through the interface, degrading the contact resistance, for example in the case of MIScontacts. Besides, a perfectly de-pinned contact may introduce additional device-to-devicevariability due to spurious change in doping.On the other hand, a de-pinned contact brings in an unprecedented advantage where ambipo-lar carrier injection is required. Two key applications are light-emitting diode (LED) andcomplementary metal oxide semiconductor field effect transistor (CMOSFET). In LED, onemust seamlessly inject both electron and hole to the active layer to achieve high efficiency.In CMOSFET, it is crucial to achieve a low-resistance contact for both n-type and p-typesource/drain for the n-FET and p-FET, respectively, using the same metal. In this work,we demonstrate efficient pulsed electroluminescence in 1L-WS by contrasting the carrierinjection from graphene and Au contacts. We also show strong ambipolar carrier injectionby 2H-TaSe contact in WSe and WS channel with an efficiency better than Au, strikingly,for both electrons and holes. Sources of ambiguity in SBH extraction:
The back-gated FET (Figure 1a) is a popularstructure for the extraction of SBH ( φ B ) in layered semiconductors. The effec-tive barrier height ( φ B ) is computed at different values of gate voltage (V g ) using Arrheniusequation. From the φ B versus V g plot, the knee-point is identified as the flat-band conditionwhere the transport mechanism switches from purely thermionic to tunneling or thermallyassisted tunneling process and thus corresponds to the SBH of that interface. Such a quali-tative estimation of the flat-band condition and the nature of carrier transport in the lateraldevice structure lead to several ambiguities: (1) In a typical field effect device, the bandedge position linearly depends on V g in the sub-threshold regime. On the other hand, abovethe threshold voltage, the relation becomes sub-linear due to screening. Figure 1(b) showsthe simulated values of the position of the conduction band edge (with respect to the Fermi4evel) for an ultra-thin channel as a function of V g for varying channel doping. The resultsare obtained from coupled 1D Poisson-Schrodinger equations and, hence, independent of theexplicit carrier injection mechanism from the source. This suggests that the knee-point doesnot necessarily reflect the flat-band condition for the source injection, and can be confoundedby the gate electrostatics. For example, at low SBH, the knee point can occur when flat-band is not reached (panel A in Figure 1c). On the other hand, for large SBH, the tunnelingcomponent can be low in the total current, and the extracted φ B can still be well in theexponential region, even beyond the flat-band condition (panel B in Figure 1c). This willpush the knee-point beyond the true flat-band condition. (2) The source and drain contactsin the lateral architecture, as shown in Figure 1a, act as a back-to-back diode. Further,the mixed-dimensional electron transport in the source contact itself must be carefullyaddressed since this is not directly accounted for in the standard Arrhenius equation. (3)The change in the channel resistance with temperature due to a modulation in the carriermobility further results in a significant ambiguity. This is particularly problematic for lowSBH, where the channel resistance forms a large fraction of the total resistance. These am-biguities result in large variability in the extracted SBH, as depicted in Figure 1d where wesummarize the reported SBH values for various contacts to MoS . Proposed Method:
To avoid these ambiguities, we propose a test structure where a multi-layer TMDC channel is vertically sandwiched between two asymmetric metal contacts (Fig-ure 2a), giving rise to a built-in electric field in the multi-layer (Figure 2b). This techniqueprovides an estimation of the SBH of both the contact interfaces simultaneously using threesteps: (i) identification of the flat-band condition from the open-circuit voltage (V oc ) underphotoexcitation; ii) extraction of SBH of one of the contacts ( φ B ) from temperature ( T )dependent current measurement at the flat-band condition using Arrhenius plot; iii) finally,extraction of the SBH of the other contact interface using φ B = φ B + | V oc | . Note that,the Arrhenius equation remains directly applicable by enforcing a one-dimensional charge5ransport using the vertical structure instead of a lateral one. The channel length of thevertical device is only the thickness of the layered material. Thus, the total resistance isdominated by the source injection, eliminating any effect due to temperature-induced mo-bility variation.First, we explain how to identify the true flat-band condition. As the vertical device usesasymmetric contacts, light illumination on the device produces a finite short circuit current(I sc ) due to the presence of a built-in field in the active layer. The sign of I sc depends onthe direction of the built-in field, which, in turn, is decided by the relative values of φ B and φ B . Under light illumination, the net current is given by I net = I dark + I ph where I dark isthe dark current and I ph is the photocurrent. The open-circuit voltage (V oc ) is identified asthe voltage at which I net = 0, that is I ph = − I dark . It corresponds to the correct flat-band(zero field) condition only when I ph = 0, which in turn implies I dark = 0 (Figure 2c). Weobserve a significant reduction in the magnitude of V oc with an increase in T (See Support-ing Figure S1). This is attributed to a temperature-induced enhancement in I dark due tothermal injection over the barrier and thermally activated defects. For I dark (cid:54) = 0, I net underillumination becomes zero at a finite opposite I ph , thus V oc does not correspond to the trueflat-band condition (Figure 2d). To ensure a small I dark , we thus measure the V oc at a lowtemperature ( T = 6 . (thickness ∼ -FLG (D1) (ii) Au-WS -TaSe (D2) and (iii)FLG-WS -TaSe (D3) (see Methods in Supporting Information Note 1 for sample prepa-6ation and measurement details). The choice of the stacks allows us to estimate the SBHof each interface from two independent measurements. The temperature-dependent I dark characteristics (both forward and reverse sweeps) of representative samples (see SupportingFigure S2 for optical images and Raman characterization) from each stack are illustratedin Figure 2e-g. In the same plots, we also show the current under light illumination at6.7 K (red dashed trace) and the corresponding V oc values. In devices D1 and D2, the Aucontact is biased, whereas in D3, the FLG contact is biased. The sign of V oc suggests thatthe photogenerated electrons in WS are transferred to Au contact, and the correspondingequilibrium band diagram will have a bending as displayed in Figure 2b. Note that I dark at 6.7 K is negligible for all the devices at V oc , leading to an accurate estimation of theflat-band condition.We now measure the device dark current at different T keeping the bias same as in the low-temperature-flat-band condition, and the SBH is deduced from the slope of the Arrheniusplot, as shown in Figures 2h-j. From Figures 2h and 2i, we estimate φ B (Au) to be 0.27 and0.29 eV, respectively. Using Figures 2e and 2h, φ B (FLG) = φ B (Au) + | V oc | = (0.29 + 0.13)eV = 0.42 eV. Separately, φ B (FLG) is estimated to be 0.35 eV from Figure 2j. Similarly, φ B (TaSe ) is estimated as 0.765 and 0.895 eV from Figures 2f-i and 2g-j, respectively.Figure 3 summarizes the extracted SBH values of Au, FLG, and TaSe contacts to WS and its relationship with work function (W m ) of the corresponding contact material. Weuse the work function values of FLG from the transfer characteristics we obtained in a topgated structure, which suggests the intrinsic nature of FLG in our experiment(See Support-ing information Figure S3). Thus, we use a work function of 4.5 eV. For SnSe and Au weextracted the work function values from Kelvin Probe Force Microscopy (KPFM) measure-ments, whereas for TaSe , we use values from literature. The notation S in this figurewhich indicates the degree of the FLP, is expressed as: S = | d ( φ B ) d ( W m ) | . (2)7 = 1 corresponds to the ideal Schottky-Mott limit of de-pinning, while S = 0 indicates acompletely pinned Fermi level. Figure 3 shows that FLG exhibits a completely de-pinnedcontact ( S = 1), in good agreement with other recent works. On the other hand, Au ex-hibits a high degree of pinning ( S = 0 .
25) close to the conduction band edge. The extractedaverage S for TaSe is ∼ and hence has the potential to be a superior ambipolar contact forCMOS applications, as discussed later. Results from a few other devices are summarized inSupporting Figure S4. The relatively large variability in the extracted SBH values could beintrinsically related to the vdW contact interfaces’ de-pinned nature due to spurious changesin doping in WS , interface quality, and doping in the contact (for FLG).To further validate this method, we fabricated a separate TaSe -WS -SnSe stack whereSnSe is a degenerately n-doped, highly conductive layered semiconductor. The extractedSBH value of SnSe is 0.71 eV (See Supporting Figure S4), which corresponds to S = 0 . oc being extracted at low temperature can lead to a minor inaccuracy ( <
40 meV)in the extracted value of the SBH at room temperature due to a reduction in the bandgapfrom low temperature to room temperature (∆ E g ∼
80 meV as supported by temperature-dependent photoluminescence spectra in Supporting Figure S5). Also, one must ensure thattunneling induced dark current is negligible at V oc for the accurate estimation of the flat-band condition. SBH, being an interface property, is not expected to change significantlybetween few-layer and bulk materials. So, our estimated SBH using thicker TMDs should8e equally applicable for few layers as well. However, ultra-thin (for example, a monolayer)samples will require additional caution due to possible tunnelling leakage. Pulsed electroluminescence - ambipolar versus unipolar contact:
To further cor-roborate the values of SBH extracted for Au and FLG with WS , we prepare 1L-WS /Auand 1L-WS /FLG devices to demonstrate pulsed electroluminescence (EL). The schematicof the fabricated EL devices (optical images in Supporting Figure S6) are shown in Figure 4a(top panel: 1L-WS /Au, bottom panel: 1L-WS /FLG). This is a two-terminal device wherethe source contact is grounded, and an ac voltage pulse is applied to the gate electrode.During the measurement, a 2 MHz voltage pulse with a peak-to-peak (-V p to +V p ) voltageof 5 V (rise/fall time 4/4 ns) is applied symmetrically about a dc offset voltage ( V ). Aselectrons and holes are injected from the source contact to the WS layer alternatively, ELis produced at the transition edges near the source contact. The role of V in the ELintensity is schematically explained in Figure 4b for the case of a metal-contact with Fermilevel pinned close to the conduction band. For V = 0 V (top panel), when V g is at -V p ,the WS layer is p-doped. The switching of the applied V g from -V p to +V p induces largetunneling of electrons from the source contact to WS . During the transition edge, some ofthe remaining holes in the valance band recombine with the incoming electrons, giving rise toan EL signal. The opposite mechanism generates another EL pulse when V g switches from+V p to -V p . The measured steady-state EL intensity is the sum of the EL output arisingfrom all such rising and falling transition edges.As the initial position of metal Fermi level is near the conduction band edge of WS , thelimited availability of hole density limits the overall EL output. With V <
0, the availablehole density for recombination increases, in turn increasing the EL intensity, and reaches amaximum (middle panel of Figure 4b) at V = V max . However, with further negative V ,the injected electron density is reduced, eventually limiting the EL intensity (bottom panel).We thus expect a non-monotonic EL output as a function of V , and V = V max can be used9s a probe to validate the relative SBH of Au and FLG qualitatively. Figure 4c-e shows theEL spectra at different V , with an exciton peak around 2.01 eV at 300 K. Figure 4f depictsthe dependency of the peak intensity with V for Au and FLG contacts. While for the FLGcontact, the peak intensity occurs at a small negative V ( V max ∼ − . V max < − TaSe as an ambipolar vdW contact for CMOS: The estimated SBH indicates thatTaSe , a prospective candidate for realizing weakly pinned vdW contact, aligns close to themid-gap of WS as shown in Figure 3. This finding implies the possibility of using TaSe to achieve pronounced ambipolar carrier injection, which allows the design optimization ofTMD transistors for CMOS technology. In order to demonstrate this, we fabricate top-gated lateral few-layer WS transistor with asymmetric contacts, namely TaSe and Au,as depicted in Figure 5a (see Supporting Figure S7 in for sample details). The hBN layerbeneath the channel layer minimizes any detrimental substrate effect on the electrical charac-teristics. Using the transfer characteristics at different biasing configurations, we infer thecarrier injection efficiency of TaSe as a source contact with Au as a control. For an apple-to-apple comparison, we compare the drain current ( I ds ) in the MOSFET mode ( V gs > V ds > V gs < V ds < as thesource as illustrated in Figure 5b. Further, to avoid the potential barrier at the drain endon the source injection efficiency, we compare the carrier injection level of one contact withthe other at a relatively higher V ds .With Au (TaSe ) as the source contact, the colored regions of panels I and II (III and IV)in Figure 5c depict the electron and hole injection characteristics of Au (TaSe ) respectively.The corresponding output characteristics are provided in Supporting Figure S7. TaSe ex-hibits ambipolar characteristics, while Au predominantly shows n-type behavior. The hole10urrent in the p-FET mode is 2x10 -fold higher when TaSe is used as a source comparedwith Au for an overdrive voltage ( | V gs − V min | ) of 3 V at | V ds | = 1 V (indicated by blacksquares). Also, TaSe exhibits enhanced electron injection compared to Au with 5-fold highercurrent in the n-FET mode at an overdrive of 5 V and | V ds | = 1 V.We also demonstrate the ambipolar injection nature of TaSe using a few-layer WSe as thechannel (see Supporting Figure S8 for sample details and characterization). The transfercharacteristics at different biasing configurations are shown in Figure 5d. Here, we find thatAu shows ambipolar injection into WSe with a slightly higher hole current than the electroncurrent. These observations agree with the previous results seen from transferred Au con-tacts. Compared to Au contact, TaSe injection exhibits a 5-fold higher electron currentin the n-FET mode and similar hole current in the p-FET mode, for an overdrive voltage of4 V. We also illustrate ambipolar characteristics of TaSe with a monolayer WSe channel,where TaSe exhibits improved electron and hole injection than Au (see Supporting FigureS9).The striking fact that TaSe as a source can inject both types of carriers (electrons for n-FET and holes for p-FET) more efficiently than Au suggests a weakly pinned nature ofTaSe contact leading to a possible modulation of the SBH by the gate voltage. We notethat TaSe is susceptible to oxidation which exhibits a-Se peak at ∼ cm − in the Ramanscattering of partially or completely oxidized flakes. However, during our device fabrica-tion, we minimize the exposure of TaSe to ambience by quick successive transfer of channellayer which avoids surface oxidation at TMD/TaSe interface as discussed in Supportinginformation Note 2. Hence the role of interface TaO x is negligible on the depinning of TaSe contact. In summary, we demonstrated a simple technique to accurately estimate the Schottky barrierheight at a vdW contact interface. We applied the technique to demonstrate that severaltypes of layered contact materials, including semiconducting SnSe , semi-metallic few-layer11raphene, and metallic TaSe exhibit a universal de-pinned nature compared to bulk metalslike Au. We exploited the ambipolar carrier injection from such de-pinned vdW contact todemonstrate two key applications: efficient pulsed electroluminescence and improved carrierinjection for both pFET and nFET. The proposed SBH extraction technique and the uni-versality of the de-pinned nature of vdW contacts should have intriguing prospects towardsfuture nano-electronic devices. SUPPORTING INFORMATION
Supporting information is available.
ACKNOWLEDGMENTS
This work was supported in part by the support of a grant from Indian Space Research Or-ganization (ISRO), a grant from MHRD under STARS, grants under Ramanujan Fellowshipand Nano Mission from the Department of Science and Technology (DST), Government ofIndia, and support from MHRD, MeitY and DST Nano Mission through NNetRA. K.W. andT.T. acknowledge support from the Elemental Strategy Initiative conducted by the MEXT,Japan, Grant Number JPMXP0112101001, JSPS KAKENHI Grant Numbers JP20H00354and the CREST(JPMJCR15F3), JST.
NOTES
The authors declare no competing financial or non-financial interest.
Data Availability
The data is available on reasonable request from the corresponding author.12 eferences (1) Ng, K. K.; Lynch, W. T. The Impact of Intrinsic Series Resistance on MOSFET Scaling.
IEEE Transactions on Electron Devices , , 503–511.(2) Giubileo, F.; Di Bartolomeo, A. The role of contact resistance in graphene field-effectdevices. Progress in Surface Science , , 143–175.(3) Taur, Y.; Ning, T. H. Fundamentals of Modern VLSI Devices ; Cambridge UniversityPress, 2013.(4) Schulman, D. S.; Arnold, A. J.; Das, S. Contact engineering for 2D materials anddevices.
Chemical Society Reviews , , 3037–3058.(5) Sze, S.; Ng, K. K. Physics of Semiconductor Devices ; pp 141–142.(6) Allain, A.; Kang, J.; Banerjee, K.; Kis, A. Electrical contacts to two-dimensional semi-conductors.
Nature Materials
ACS Nano , , 1588–1596.(8) Wang, Q.; Shao, Y.; Gong, P.; Shi, X. Metal-2D multilayered semiconductor junctions:layer-number dependent Fermi-level pinning. Journal of Materials Chemistry C , , 3113–3119.(9) Gallacher, K.; Velha, P.; Paul, D. J.; MacLaren, I.; Myronov, M.; Leadley, D. R. Ohmiccontacts to n-type germanium with low specific contact resistivity. Applied PhysicsLetters , , 022113.(10) Nishimura, T.; Kita, K.; Toriumi, A. Evidence for strong Fermi-level pinning due tometal-induced gap states at metal/germanium interface. Applied Physics Letters , , 123123. 1311) Lin, J. C.; Yu, S. Y.; Mohney, S. E. Characterization of low-resistance ohmic contactsto n- and p-type InGaAs. Journal of Applied Physics , , 044504.(12) Kim, H.; Kim, I. PtTiPtAu and PdTiPtAu ohmic contacts to p-InGaAs. Compoundsemiconductor 1997. Proceedings of the IEEE Twenty-Fourth International Symposiumon Compound Semiconductors, San Diego, CA, USA. 1997; pp 175–178.(13) Zhang, S.; Le, S. T.; Richter, C. A.; Hacker, C. A. Improved contacts to p-type MoS transistors by charge-transfer doping and contact engineering. Applied Physics Letters , , 073106.(14) Zhang, Y. J.; Ye, J. T.; Yomogida, Y.; Takenobu, T.; Iwasa, Y. Formation of a stablep-n junction in a liquid-gated MoS ambipolar transistor. Nano Letters , , 3023–3028.(15) Chuang, S.; Battaglia, C.; Azcatl, A.; McDonnell, S.; Kang, J. S.; Yin, X.; Tosun, M.;Kapadia, R.; Fang, H.; Wallace, R. M.; Javey, A. MoS P-type transistors and diodesenabled by high work function MoO x contacts. Nano Letters , , 1337–1342.(16) Nipane, A.; Karmakar, D.; Kaushik, N.; Karande, S.; Lodha, S. Few-Layer MoS p-TypeDevices Enabled by Selective Doping Using Low Energy Phosphorus Implantation. ACSNano , , 2128–2137.(17) Liu, X.; Qu, D.; Ryu, J.; Ahmed, F.; Yang, Z.; Lee, D.; Yoo, W. J. P-Type PolarTransition of Chemically Doped Multilayer MoS Transistor.
Advanced Materials , , 2345–2351.(18) Liu, Y.; Guo, J.; Zhu, E.; Liao, L.; Lee, S. J.; Ding, M.; Shakir, I.; Gambin, V.;Huang, Y.; Duan, X. Approaching the Schottky-Mott limit in van der Waals metal-semiconductor junctions. Nature , , 696–700.1419) Song, S. et al. Wafer-scale production of patterned transition metal ditelluride layersfor two-dimensional metal–semiconductor contacts at the Schottky–Mott limit. NatureElectronics , , 207–215.(20) Mahajan, M.; Kallatt, S.; Dandu, M.; Sharma, N.; Gupta, S.; Majumdar, K. Light emis-sion from the layered metal 2H-TaSe and its potential applications. CommunicationsPhysics , , 88.(21) Krishna, M.; Kallatt, S.; Majumdar, K. Substrate effects in high gain , low operatingvoltage SnSe photoconductor. Nanotechnology , , 035205.(22) Singisetti, U.; Wistey, M. A.; Zimmerman, J. D.; Thibeault, B. J.; Rodwell, M. J.;Gossard, A. C.; Bank, S. R. Ultralow resistance in situ Ohmic contacts to InGaAs/InP. Applied Physics Letters , , 183502.(23) Shiraishi, Y.; Furuhata, N.; Okamoto, A. Infulence of metal/n-InAs/interlayer/n-GaAsstructure on nonalloyed ohmic conytact resistance. Journal of Applied Physics , , 5099.(24) Li, M.; Shin, G.; Lee, J.; Oh, J.; Lee, H. D. Low contact resistance of NiGe/ p -Ge byindium segregation during Ni germanidation. AIP Advances , , 065312.(25) Spann, J. Y.; Anderson, R. A.; Thornton, T. J.; Harris, G.; Thomas, S. G.; Tracy, C.Characterization of nickel germanide thin films for use as contacts to p-channel germa-nium MOSFETs. IEEE Electron Device Letters , , 151–153.(26) Das, S.; Chen, H. Y.; Penumatcha, A. V.; Appenzeller, J. High performance multilayerMoS transistors with scandium contacts. Nano Letters , , 100–105.(27) Chen, J. R.; Odenthal, P. M.; Swartz, A. G.; Floyd, G. C.; Wen, H.; Luo, K. Y.;Kawakami, R. K. Control of Schottky barriers in single layer MoS transistors withferromagnetic contacts. Nano Letters , , 3106–3110.1528) Yu, H.; Schaekers, M.; Demuynck, S.; Barla, K.; Mocuta, A.; Horiguchi, N.; Collaert, N.;Thean, A. V. Y.; De Meyer, K. MIS or MS? Source/drain contact scheme evaluation for7 nm Si CMOS technology and beyond. 2016 16th International Workshop on JunctionTechnology, IWJT, Shanghai. 2016; pp 19–24.(29) Gupta, S.; Rortais, F.; Ohshima, R.; Ando, Y.; Endo, T.; Miyata, Y.; Shiraishi, M.Monolayer MoS field effect transistor with low Schottky barrier height with ferromag-netic metal contacts. Scientific Reports , , 1–7.(30) Kaushik, N.; Nipane, A.; Basheer, F.; Dubey, S.; Grover, S.; Deshmukh, M. M.;Lodha, S. Schottky barrier heights for Au and Pd contacts to MoS . Applied PhysicsLetters , , 113505.(31) Das, S.; Prakash, A.; Salazar, R.; Appenzeller, J. Toward low-power electronics: Tun-neling phenomena in transition metal dichalcogenides. ACS Nano , , 1681–1689.(32) Kwon, J.; Lee, J. y.; Yu, Y. j.; Lee, C. h.; Cui, X.; Hone, J.; Lee, G. h. Thickness-dependent Schottky barrier height of MoS field-effect transistors. Nanoscale , ,6151–6157.(33) Dankert, A.; Langouche, L.; Kamalakar, M. V.; Dash, S. P. High-performance molyb-denum disulfide field-effect transistors with spin tunnel contacts. ACS Nano , ,476–482.(34) Mahajan, M.; Murali, K.; Kawatra, N.; Majumdar, K. Gate-Controlled Large Resis-tance Switching Driven by Charge-Density Wave in 1T-TaS /2H-MoS Heterojunctions.
Physical Review Applied , , 024031.(35) Abraham, M.; Mohney, S. E. Annealed Ag contacts to MoS field-effect transistors. Journal of Applied Physics , , 115306.1636) Zheng, S.; Lu, H.; Liu, D.; Robertson, J. Insertion of an ultrathin Al2O3 interfaciallayer for Schottky barrier height reduction in WS field-effect transistors. Nanoscale , , 4811–4821.(37) Gourmelon, E.; Bern`ede, J. C.; Pouzet, J.; Marsillac, S. Textured MoS thin filmsobtained on tungsten: Electrical properties of the W/MoS contact. Journal of AppliedPhysics , , 1182–1186.(38) Somvanshi, D.; Kallatt, S.; Venkatesh, C.; Nair, S.; Gupta, G.; Anthony, J. K.; Kar-makar, D.; Majumdar, K. Nature of carrier injection in metal/2D-semiconductor inter-face and its implications for the limits of contact resistance. Physical Review B , , 205423.(39) Kang, J.; Liu, W.; Sarkar, D.; Jena, D.; Banerjee, K. Computational study of metalcontacts to monolayer transition-metal dichalcogenide semiconductors. Physical ReviewX , , 031005.(40) Zahin, A. Schottky Barrier Heights at Two-Dimensional Metallic and SemiconductingTransition-Metal Dichalcogenide Interfaces. Ph.D. thesis, 2017.(41) Tsoutsou, D.; Aretouli, K. E.; Tsipas, P.; Marquez-Velasco, J.; Xenogiannopoulou, E.;Kelaidis, N.; Aminalragia Giamini, S.; Dimoulas, A. Epitaxial 2D MoSe (HfSe ) Semi-conductor/2D TaSe Metal van der Waals Heterostructures.
ACS Applied Materialsand Interfaces , , 1836–1841.(42) Tsai, H. S.; Liu, F. W.; Liou, J. W.; Chi, C. C.; Tang, S. Y.; Wang, C.; Ouyang, H.;Chueh, Y. L.; Liu, C.; Zhou, S.; Woon, W. Y. Direct Synthesis of Large-Scale MultilayerTaSe2 on SiO2/Si Using Ion Beam Technology. ACS Omega , , 17536–17541.(43) Yu, L.; Lee, Y. H.; Ling, X.; Santos, E. J.; Shin, Y. C.; Lin, Y.; Dubey, M.; Kaxiras, E.;Kong, J.; Wang, H.; Palacios, T. Graphene/MoS Hybrid technology for large-scaletwo-dimensional electronics.
Nano Letters , , 3055–3063.1744) Chuang, H. J.; Tan, X.; Ghimire, N. J.; Perera, M. M.; Chamlagain, B.; Cheng, M.M. C.; Yan, J.; Mandrus, D.; Tom´anek, D.; Zhou, Z. High mobility WSe p - And n -Field-effect transistors contacted by highly doped graphene for low-resistance contacts. Nano Letters , , 3594–3601.(45) Leong, W. S.; Luo, X.; Li, Y.; Khoo, K. H.; Quek, S. Y.; Thong, J. T. Low resis-tance metal contacts to MoS devices with nickel-etched-graphene electrodes. ACSNano , , 869–877.(46) Farmanbar, M.; Brocks, G. Ohmic Contacts to 2D Semiconductors through van derWaals Bonding. Advanced Electronic Materials , , 1500405.(47) Lien, D. h.; Iii, J. W. A.; Han, K. Large-area and bright pulsed electroluminescence inmonolayer semiconductors. Nature Communications , , 1229.(48) Paur, M.; Molina-mendoza, A. J.; Mueller, T. Electroluminescence from multi-particleexciton complexes in transition metal dichalcogenide semiconductors. Nature Commu-nications , , 1709.(49) Ren, Y.; Yang, X.; Zhou, L.; Mao, J.-Y.; Han, S.-T.; Zhou, Y. Recent Advances in Am-bipolar Transistors for Functional Applications. Advanced Functional Materials , , 1902105.(50) Wang, F.; Tu, B.; He, P.; Wang, Z.; Yin, L.; Cheng, R.; Wang, J.; Fang, Q.; He, J.Uncovering the Conduction Behavior of van der Waals Ambipolar Semiconductors. Advanced Materials , , 1805317.(51) Kong, L.; Zhang, X.; Tao, Q.; Zhang, M.; Dang, W.; Li, Z.; Feng, L.; Liao, L.; Duan, X.;Liu, Y. Doping-free complementary WSe circuit via van der Waals metal integration. Nature communications , , 1–7. 1852) Liu, Y.; Guo, J.; Zhu, E.; Liao, L.; Lee, S.-J.; Ding, M.; Shakir, I.; Gambin, V.;Huang, Y.; Duan, X. Approaching the Schottky–Mott limit in van der Waals metal–semiconductor junctions. Nature , , 696–700.(53) Liu, Y.; Stradins, P.; Wei, S.-H. Van der Waals metal-semiconductor junction: WeakFermi level pinning enables effective tuning of Schottky barrier. Science advances , , e1600069.(54) Cartamil-Bueno, S. J.; Steeneken, P. G.; Tichelaar, F. D.; Navarro-Moratalla, E.;Venstra, W. J.; van Leeuwen, R.; Coronado, E.; van der Zant, H. S.; Steele, G. A.;Castellanos-Gomez, A. High-quality-factor tantalum oxide nanomechanical resonatorsby laser oxidation of TaSe 2. Nano Research , , 2842–2849.(55) Sun, L.; Chen, C.; Zhang, Q.; Sohrt, C.; Zhao, T.; Xu, G.; Wang, J.; Wang, D.;Rossnagel, K.; Gu, L., et al. Suppression of the Charge Density Wave State in Two-Dimensional 1T-TiSe2 by Atmospheric Oxidation. Angewandte Chemie InternationalEdition , , 8981–8985.(56) Li, X.-X.; Fan, Z.-Q.; Liu, P.-Z.; Chen, M.-L.; Liu, X.; Jia, C.-K.; Sun, D.-M.; Jiang, X.-W.; Han, Z.; Bouchiat, V., et al. Gate-controlled reversible rectifying behaviour intunnel contacted atomically-thin MoS 2 transistor. Nature communications , ,1–7. 19 (a) (b) Metal
Metal
TMD T r u e f l a t - b a n d i n A Gate Voltage (V) B a rr i e r H e i g h t ( e V ) Work Function (eV)
AuPd PtNiTi Sc Al Ag Py CuCoW (d) T r u e f l a t - b a n d i n B S B H ( e V ) AB (c) C o n d u c t i o n B a n d E dg e ( e V ) Gate Voltage (V)
Knee point Varying doping
Figure 1:
Sources of ambiguities in SBH extraction: (a) Schematic diagram of a back-gated lateral structure used for SBH extraction. The arrows indicate mixed dimensionalcarrier transport in the contact region. (b) The position of the conduction band edge (withrespect to the Fermi level) as a function of V g for different channel doping, as obtained fromthe solution of coupled Poisson-Schrodinger equations. The channel thickness is taken as 10nm. (c) A representative graph to illustrate the deviation of actual flat-band condition fromthe knee point (Green open circle) for cases of small SBH (A) and large SBH (B). (d) Abenchmarking plot illustrating reported SBH values of different metals on MoS , indicatinga large variation. 20 q/kT (V -1 )
38 40 42 44 46-41-39-37 l n ( I d / T ) ϕ B(Au)0 = 0.29 eV ϕ 𝐵(𝐹𝐿𝐺) = 0.42 eV ϕ 𝐵(𝐴𝑢)0 = 0.27 eV ϕ 𝐵(𝑇𝑎𝑆𝑒 ) = 0.765 eV
38 42 46 50 ϕ 𝐵(𝐹𝐿𝐺) = 0.35 eV ϕ 𝐵(𝑇𝑎𝑆𝑒 ) = 0.895 eV (h) (i) (j) -1.0 -0.5 0.0 0.5 1.0 V d (V)
225 K 250 K 275 K 295 K -1.0 -0.5 0.0 0.5 1.010 -17 -14 -11 -8 I d ( A ) V oc = -0.13V V oc = -0.545VV oc = -0.495VAu-WS -FLG Au-WS -TaSe FLG-WS -TaSe (e) (f) (g)(a) M1 M2 +V=0,I=I sc (b) - (c) (d) M1M2 WS +T=6.7 K- +V =V oc ϕ B10 ϕ B20 - 𝐼 𝑝ℎ = 𝐼 𝑑𝑎𝑟𝑘 = T=295 K + +- V = V oc - 𝐼 𝑝ℎ = −𝐼 𝑑𝑎𝑟𝑘 -1.0 -0.5 0.0 0.5 1.0
225 K 265 K 235 K 275 K 245 K 285 K 255 K 295 K
D1 D2 D3
Figure 2:
Proposed SBH extraction method: (a) Schematic of the proposed vertical teststructure with asymmetric contacts. (b)-(d) Schematic representation of band diagrams (b)in equilibrium showing built-in electric field in the multi-layer, (c) ideal flat-band condition at V oc (at low temperature where I dark is suppressed), and (d) deviation from flat-band conditionat V oc at higher temperatures due to non-zero I dark . (e)-(g) Current-Voltage characteristicsunder dark condition at different temperatures for the three stacks, namely, Au-WS -FLG(D1), Au-WS -TaSe (D2), and FLG-WS -TaSe (D3). The black and the red dashed tracesindicate current with and without light, respectively, at T = 6 . V oc value at 6 . I dark at this temperature. (h)-(j) The corresponding Arrhenius plots for D1-D3along with linear fits (dashed lines) to deduce the SBH of one of the contacts. The SBH ofthe other contact is obtained by adding the V oc ( T = 6 . .0 4.4 4.8 5.20.00.30.60.9 FLG SnSe TaSe Au S B H ( e V ) Work Function (eV)
Figure 3:
Fermi level de-pinning with vdW contacts:
Summary of the extracted SBHversus metal work function Au, FLG, TaSe , and SnSe , when contacted to WS . The dashedlines correspond to the different pinning factors ( S ). The vdW contacts exhibit significantlyde-pinned nature compared to Au. 22 .95 2.00 2.050.0-0.5-1.0-1.5-2.0 Energy (eV) O ff s e t V o l t a g e ( V ) -2.0 -1.5 -1.0 -0.5 0.00.010.11 Offset Voltage (V)
Au FLG N o r m a l i z e d E L I n t e n s i t y Energy (eV)
Energy (eV)
FLGAu (c)(d) (e) (f)(a) (b) V = 0 VV = V max V < V max WS hBNFLG Figure 4:
Pulsed Electroluminescence - unipolar versus ambipolar injection: (a)Schematic diagram of the pulsed electroluminescence devices with hBN as gate dielectric.Top panel shows the device with Au as source contact and bottom panel shows the devicewith FLG as source contacts. A pulse train (5 V peak-to-peak) riding on a dc offset voltage( V ) is applied between gate and source electrode, and light emission occurs near the sourcecontact edge. (b) Principle of operation and the role of V in EL intensity modulation. Thetop, middle, and bottom rows correspond to V = 0, V = V max (maximum EL intensity),and V < V max for the case where Fermi level of the source is pinned close to the conductionband edge. (d) The colour plot of the measured EL intensity from 1L-WS as a function ofemission energy and V for Au and FLG as the source contacts. V max occurs for FLG at amuch lower voltage than for Au, indicating relative values of SBH for these two contacts.(e)-(f) Individual EL spectra of Figure (d) for different V for (e) Au and (f) FLG contacteddevice. (f) Normalized peak EL intensity plotted in log scale as a function of V for the twodevices. 23 ds V gs (a) (b) Au TaSe hBN SiO Si WS hBN Gr -4 -2 0 2 4 10 -14 -12 -10 -8 -6 -4 -2 0 2 4 10 -14 -12 -10 -8 -6 -4 -2 0 2 4 TaSe n-FET Au n-FET (d) II I IIIIV Au p-FET TaSe p-FET -4 -2 0 2 410 -14 -12 -10 -8 -6 -14 -12 -10 -8 -6 I d s ( A ) V gs (V) II IV Au n-FET TaSe n-FETAu p-FET TaSe p-FET I III (c) WS S D SSDDS TaSe n-FET + +_ D _ V ds > 0V gs > 0 V ds < 0V gs < 0 Au p-FETAu n-FET TaSe p-FET V gs (V) - 0.1 V - 1 V Au TaSe Figure 5:
Ambipolar injection from TaSe : (a) Schematic of top-gated lateral FETwith asymmetric contacts, Au and TaSe . The biasing configuration shown here is for TaSe electron injection. (b) Band diagrams of carrier injection into the channel with Au andTaSe as source in the MOSFET mode ( V gs > V ds > V gs < V ds < FET in log scale.Colored regions in orange and green highlight the electron and hole injection characteristicsof Au (TaSe ) source in panels I and II (III and IV), respectively. Black squares indicatethe current levels at | V ds | = 1 V for a gate overdrive of 5 V and 3 V in n-FET and p-FETmodes, respectively. (d) Transfer characteristics of top gated few layer WSe FET in logscale. Black squares highlight the current levels at | V ds | = 1 V for gate overdrive of 4 V.24 upporting Notes and FiguresNote 1Methods Sample preparation for vertical devices:
The vertical devices used for the extraction ofSchottky Barrier Height (SBH) are fabricated on a Si/SiO substrate with pre-patterned Aucontacts. Pre-deposited Au will act as one of the metal contact (M1) for devices D1 and D2.Multilayer WS (40–60 nm) is identified on PDMS by optical contrast and then dry trans-ferred to Au lines. Finally, the other contact, FLG/TaSe for device D1/D2 is transferredprecisely touching Au–WS portion in one end and another Au line in other end. In the caseof D3, FLG is transferred first on one of the Au lines, followed by the transfer of multilayerWS in such a way that WS touches only FLG and thus avoiding any contact with Au line.To conclude, TaSe is transferred precisely on the Au-WS -FLG junction without touchingeither Au or FLG. Fabricating devices on pre-patterned substrate makes the whole processmore efficient in terms of quality, time consumption and cost. Sample preparation for EL devices:
The devices for pulsed electro luminescent mea-surements are fabricated on a Si/SiO substrate with pre-patterned Au contacts. For thefabrication of device with Au as source contact, monolayer WS is mechanically exfoliatedand identified using optical contrast on PDMS. Then, selected flake is dry transferred di-rectly to one of the pre-patterned Au contacts. The thickness of the WS is confirmed withPL measurement. Then WS layer is capped with hBN and which also acts as gate dielectric.For gating the WS channel, few layer graphene is used as gate electrode which is connectedto another Au contact. For the device with FLG as source contact, we use back gated struc-ture. So, at first, hBN is transferred to pre-patterned Au contact covering full width of thecontact. Then monolayer WS is transferred without touching Au contact. Finally, few layergraphene is transferred by ensuring a proper overlap with monolayer WS . Here, bottomAu electrode will act as gate contact and FLG acts as source contact. For both the devices,2ate pulse is applied between gate electrode and source contact and EL spectra is obtainedfrom the WS region which is close to the source contact. Sample preparation for Lateral FET devices:
The top-gated lateral FET devices arefabricated using pre-patterned Au contacts on Si/SiO substrate. In order to avoid any ef-fect of the substrate on the SBH of contacts and carrier injection characteristics, hBN ( ∼ ( ∼
20 nm) is transferred touching one of the Au contact which is followed bytranser of few-layer TMD ( ∼
10 nm) to make contact with both Au and TaSe . For gatingthe TMD in both the channel and contact regions, hBN ( ∼ Electrical measurements:
All the electrical measurements are performed using Keithley4200A-SCS Parameter Analyzer under vacuum level of − torr.3
100 200 300-0.3-0.2-0.10.0 V o c ( V ) Temperature (K) -1.0 -0.5 0.0 0.5 1.010 -14 -12 -10 -8 -6 I d ( A ) V d (V)
50 K 100 K 200 K 125 K 225 K 150 K 250 K 175 K 295 K
Figure S1: (a) Reduction in the magnitude of V oc observed for Au-WS -FLG stack withchange in temperature resulting from the increase in the dark current.4
00 400 60001k2k3k A WS E A
200 400 60005001k
TaSe A E E A FLG
G band 2D band I n t e n s i t y ( c o u n t s ) Raman Shift (cm -1 ) (d)
200 400 60002k4k6k
Au-WS -TaSe Au-WS -FLG FLG-WS -TaSe
200 400 60001k2k WS -TaSe I n t e n s i t y ( c o u n t s ) (e) (f) (g)(h) (i) (j)
200 400 60001k2k
200 400 60001k2k (a) D1 (b) D2 (c) D3 FLGWS TaSe Figure S2: (a)-(c) Optical images of device (a) D1 (b) D2 and (c) D3. Scale bar is 5 µ m. (d)-(j) Raman spectra at (d) Au-WS -FLG junction (e) Au-WS -TaSe junction (f)FLG-WS -TaSe junction (g) WS -TaSe junction (h) isolated WS portion (i)isolated FLGportion and (j) isolated TaSe portion. -4 -2 0 2 40.3100.3120.3140.3160.3180.320 I d ( m A ) V g (V) V d = 0.5 V Figure S3: Transfer characteristics of FLG obtained in a top gated structure, where hBNis used as the top-gate dielectric. The minimum current occurs close to V g = 0 , suggestingnegligible amount of doping in the FLG flake. We thus take the work function of FLG as4.5 eV. 5 -1.0 -0.5 0.0 0.5 1.010 -16 -14 -12
290 K 300 K 305 K 310 K -1.0 -0.5 0.0 0.5 1.010 -16 -13 -10 -7 Au-WS -FLG Au-WS -TaSe -1.0 -0.5 0.0 0.5 1.0 -16 -13 -10 -7 I d ( A ) V d (V) V oc = -0.12 V V oc = -0.5 V
40 44 48 52-42-40-38
40 44 48 52-42-40-38 φ B(Au)0 = 0.37 eV 𝜑 𝐵(𝐹𝐿𝐺)0 = 0.49 eV φ B(Au)0 = 0.2 eV 𝜑 𝐵(𝑇𝑎𝑆𝑒 )0 = 0.7 eV l n ( I d / T ) q/kT (V -1 ) FLGWS TaSe (a) (b) (c)(d) (e) (f) (g) (i)(h)V d (V) V d (V)q/kT (V -1 ) q/kT (V -1 ) φ B(TaSe )0 = 0.62 eV 𝜑 𝐵(𝑆𝑛𝑆𝑒 )0 = 0.72 eV SnSe -WS -TaSe SnSe V oc = 0.1 V Figure S4: (a)-(c) Optical images of another set of vertical devices namely (a) Au-WS -FLGstack (DS1) (b) Au-WS -TaSe stack (DS2) (c) SnSe -WS -TaSe (DS3). Scale bar is 5 µ m.(d)-(f) Current-Voltage characteristics under dark condition at different temperatures forthe devices (d) DS1 (e) DS2 (f) DS3. The black and the red dashed traces indicate currentwith and without light, respectively, at T = 6 . K. The V oc value at . K in each structureis indicated in the inset. (g)–(i) The corresponding Arrhenius plots for DS1-DS3 along withlinear fits (dashed lines) to deduce the SBH of one of the contacts. The SBH of other contactis calculated by adding | V oc | to the first one.6 .90 1.95 2.00 2.05 2.10 2.15 Temperature (K) E n e r g y ( e V ) Energy (eV) P L I n t e n s i t y ( a . u ) (a) (b)
250 K (x4)
285 K (x3.5) XT Figure S5: (a) Temperature dependent photoluminescence spectra of monolayer WS whereX and T are exciton and trion emission peaks respectively. (b) Red shift of exciton peak (X)with increase in the temperature. WS GrhBN Source contact : FLG Source contact : Au (a) (b)
Figure S6: (a)-(b) Optical image of the EL devices where (a) FLG as source contact (b) Auas source contact. Scale bar is 5 µ m. 7 TaSe : Drain , Au : Source -2 -1 0 1 2-3 30.20.40.60.0-2 -1 0 1 2-3 31020030 -2 -1 0 1 2-3 3120-2 -1 0 1 2-3 351015 Gr hBN (Top) TaSe hBN (Bottom)WS
100 200 300 400 500 600 R a m a n i n t e n s i t y ( c o u n t s ) Raman shift (cm -1 ) V ds (V) I d s ( μ A ) V ds (V) I d s ( μ A ) (a) (b) (c)(d) Au : Drain , TaSe : Source - 1 V- 2 V- 3 V- 4 V- 5 V hBN GrWS TaSe G 2DA E E E A Au n-FET Au p-FETTaSe p-FETTaSe n-FET Figure S7: Top gated lateral few-layer WS FET with Au and TaSe contacts (a) Opticalimage of the device fabricated on pre-patterned Au contacts. Colored dotted lines highlightindividual layers. Scale bar is 5 µ m. (b) Raman characterization of WS (in the channelregion), TaSe , hBN and Graphene multilayers respectively. (c) Output characteristics ofFET (in linear scale) with Au as source and TaSe as drain under positive (left panel) andnegative (right panel) gate bias. (d) Output characteristics of FET (in linear scale) withTaSe as source and Au as drain under different gate bias.8
00 150 200 250 300300400500
TaSe hBN Gr
100 200 300 400 500 600
WSe WSe GrhBN (Top)TaSe hBN(Bottom) R a m a n i n t e n s i t y ( c o u n t s ) Raman shift (cm -1 ) (a) (c) A E E E G A E V ds V gs Au TaSe hBN SiO Si WSe hBN Gr (b) Figure S8: Top gated lateral FET with monolayer WSe channel and asymmetric contacts(a) Optical image of the device highlighting regions of different layers along with monolayerand multilayer WSe . (b) Transfer characteristics of this FET (in log scale) at differentbiasing configurations. Colored regions in orange and green highlight the electron and holeinjection characteristics of Au (I and II) and TaSe (III and IV) respectively.9 -14 -12 -10 -8 -6 -4 -2 0 2 4 -14 -12 -10 -8 -6 -4 WSe - 1 LGrhBN (Top)TaSe hBN(Bottom) WSe - M LTaSe n-FET Au n-FET II I IIIIV Au p-FET TaSe p-FET- 1 V- 0.1 V0.5 V1 V0.1 V - 0.5 V - 3 V3 V I d s ( A ) V gs (V) (a)(b) Figure S9: Top gated lateral few-layer WSe FET with Au and TaSe contacts (a) Opticalimage of the device showing the different layers. Scale bar is 5 µ m. (b) Schematic of top-gatedlateral FET with asymmetric contacts, Au and TaSe . The biasing configuration shown hereis for TaSe electron injection. (c) Raman characterization of WSe (in the channel region),TaSe , hBN and graphene multilayers respectively.10
00 150 200 250 300250300350400450500550 TaSe /hBN TaSe /air R a m a n i n t e n s i t y ( c o u n t s ) Raman shift (cm -1 ) a-Se -1.0 -0.5 0.0 0.5 1.0-0.006-0.004-0.0020.0000.0020.0040.006 I d ( A ) V d (V) (a) (b) Figure S10: Stability of TaSe (a) Raman characterization of air exposed and unexposedTaSe portions which shows absence of oxidation in encapuslated TaSe . (b) Linear I-Vcharacteristics of TaSe . Note 2
It has been been reported in literature that TaSe is susceptible to oxidation at a relativelyslower rate and a-Se peak at ∼ cm − is observed in the Raman scattering of partiallyor completely oxidized flakes. However, in the FET devices which are fabricated on pre-patterned electrodes, we transfer the channel layer, WSe or WS , within a few minutes afterthe exfoliation and transfer of TaSe . This minimizes the exposure of TaSe to ambienceand significantly avoids surface oxidation of TaSe at the contact interface to WSe or WS .Further, we encapsulate the FET devices with hBN layer at the top which isolates TaSe from ambience.We find a-Se peak in the Raman spectrum of TaSe only when the TaSe flake is exposedto air. However, such peak is found to be absent in the TaSe portion encapsulated by hBN(see Figure S8a). These spectra were taken after three months of device fabrication and avery weak a-Se Raman peak at the air exposed TaSe portion supports previous reports ofa slow rate of oxidation of TaSe . Moreover, any presence of surface oxide layer is usuallyreflected in I-V characteristics through a breakdown of the oxide layer. We find that TaSe is highly conducting with linear I-V characteristics (see Figure S8b) which further indicates11he absence of any surface oxide layer. Hence the role of interface TaO x is negligible on thedepinning of TaSe contact. References (1) Cartamil-Bueno, S. J.; Steeneken, P. G.; Tichelaar, F. D.; Navarro-Moratalla, E.; Ven-stra, W. J.; van Leeuwen, R.; Coronado, E.; van der Zant, H. S.; Steele, G. A.;Castellanos-Gomez, A. High-quality-factor tantalum oxide nanomechanical resonatorsby laser oxidation of TaSe 2.
Nano Research , , 2842–2849.(2) Sun, L.; Chen, C.; Zhang, Q.; Sohrt, C.; Zhao, T.; Xu, G.; Wang, J.; Wang, D.;Rossnagel, K.; Gu, L., et al. Suppression of the Charge Density Wave State in Two-Dimensional 1T-TiSe2 by Atmospheric Oxidation. Angewandte Chemie InternationalEdition ,56