# Fileset

[Adv Funct Materials - 2024 - Yu - Reversible Polarity Control in 2D MoTe2 Field‐Effect Transistors for Complementary Logic.pdf](https://mdr.nims.go.jp/filesets/4f6d5441-450e-4ec8-8fec-ac7878dbb242/download)

## Creator

Byoung‐Soo Yu, Wonsik Kim, Jisu Jang, Je‐Jun Lee, Jung Pyo Hong, Namhee Kwon, Seunghwan Kim, Aelim Ha, Hong‐Kyu Kim, Jae‐Pyoung Ahn, Kwangsik Jeong, [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), Gunuk Wang, Jongtae Ahn, Soohyung Park, Do Kyung Hwang

## Rights

[Creative Commons BY-NC-ND Attribution-NonCommercial-NoDerivs 4.0 International](https://creativecommons.org/licenses/by-nc-nd/4.0/)

## Other metadata

[Reversible Polarity Control in 2D MoTe<sub>2</sub> Field‐Effect Transistors for Complementary Logic Gate Applications](https://mdr.nims.go.jp/datasets/5eb9b27e-effb-4a6e-a337-d173d6665738)

## Fulltext

Reversible Polarity Control in 2D MoTe2 Field‐Effect Transistors for Complementary Logic Gate ApplicationsRESEARCH ARTICLEwww.afm-journal.deReversible Polarity Control in 2D MoTe2 Field-EffectTransistors for Complementary Logic Gate ApplicationsByoung-Soo Yu, Wonsik Kim, Jisu Jang, Je-Jun Lee, Jung Pyo Hong, Namhee Kwon,Seunghwan Kim, Aelim Ha, Hong-Kyu Kim, Jae-Pyoung Ahn, Kwangsik Jeong,Takashi Taniguchi, Kenji Watanabe, Gunuk Wang, Jongtae Ahn,* Soohyung Park,*and Do Kyung Hwang*Precise control over polarity in field-effect transistors (FETs) plays a pivotalrole in the design and construction of complementarymetal–oxide–semiconductor (CMOS) logic circuits. In particular, achievingsuch precise polarity control in 2D semiconductors is crucial for the furtherdevelopment of advanced electronic applications beyond unit devices. Thispaper presents a systematic investigation on the reversible transition ofcarrier types in a 2D MoTe2 semiconductor under different annealingatmospheres. Photoemission spectroscopy and density functional theory(DFT) calculations demonstrate that annealing processes in vacuum and inambient air induce a modification in the density of states, resulting inalterations in p-type or n-type characteristics. These reversible changes areattributed to the physisorption and elimination of oxygen on the surface ofMoTe2. Furthermore, it is found that the device geometry affects the polarityof the transistor. By strategically manipulating both the annealing conditionsand the geometric configuration, the n- and p-type unipolar characteristics ofMoTe2 FETs are successfully modulated and ultimately demonstrating that thefunctionality of not only a complementary inverter with a high voltage gain of≈20, but also more complex logic circuits of NAND and NOR gates.B.-S. Yu, J. Jang, J.-J. Lee, J. P. Hong, D. K. HwangCenter for Opto-Electronic Materials and DevicesPost-Silicon Semiconductor InstituteKorea Institute of Science and Technology (KIST)Seoul 02792, Republic of KoreaE-mail: dkhwang@kist.re.krB.-S. Yu, J.-P. Ahn, S. Park, D. K. HwangDivision of Nanoscience & TechnologyKIST SchoolUniversity of Science and Technology (UST)Seoul 02792, Republic of KoreaE-mail: soohyung.park@kist.re.krThe ORCID identification number(s) for the author(s) of this articlecan be found under https://doi.org/10.1002/adfm.202404129© 2024 The Authors. Advanced Functional Materials published byWiley-VCH GmbH. This is an open access article under the terms of theCreative Commons Attribution-NonCommercial-NoDerivs License,which permits use and distribution in any medium, provided the originalwork is properly cited, the use is non-commercial and no modificationsor adaptations are made.DOI: 10.1002/adfm.2024041291. IntroductionPrecise control over the polarity of chargecarriers in semiconductors is crucial forconstructing modern electronic circuitsystems.[1–3] The conventional silicon semi-conductor has been widely employed invarious everyday electronic applicationsdue to its capability to easily modulatethe carrier type (n- or p-type) throughion implantation doping technology. 2Dsemiconductors have gained considerableattention as an emerging and significantclass of materials, owing to their fascinatingproperties.[4,5] 2D materials not only offeropportunities for fundamental studies butalso possess tremendous potential for ad-vancing nanoelectronic and nanophotonictechnologies.[3,4,6–8] Despite their merits,modulation of the polarities of 2D semi-conductors remains a significant challenge.This obstacle could impede the furtherdevelopment of more advanced deviceW. Kim, N. Kwon, S. Kim, A. Ha, H.-K. Kim, J.-P. Ahn, S. ParkAdvanced Analysis and Data CenterKorea Institute of Science and Technology (KIST)Seoul 02792, Republic of KoreaJ. P. Hong, G. Wang, D. K. HwangKU-KIST Graduate School of Converging Science and TechnologyKorea UniversitySeoul 02841, Republic of KoreaK. JeongDivision of AI Semiconductor ScienceYonsei UniversityWonju 26493, Republic of KoreaT. Taniguchi, K. WatanabeAdvanced Materials LaboratoryNational Institute for Materials ScienceTsukuba 305-0044, JapanJ. AhnDepartment of PhysicsChangwon National UniversityChangwon 51139, Republic of KoreaE-mail: jongtae@changwon.ac.krAdv. Funct. Mater. 2024, 34, 2404129 2404129 (1 of 9) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbHhttp://www.afm-journal.demailto:dkhwang@kist.re.krmailto:soohyung.park@kist.re.krhttps://doi.org/10.1002/adfm.202404129http://creativecommons.org/licenses/by-nc-nd/4.0/mailto:jongtae@changwon.ac.krhttp://crossmark.crossref.org/dialog/?doi=10.1002%2Fadfm.202404129&domain=pdf&date_stamp=2024-05-15www.advancedsciencenews.com www.afm-journal.deapplications, such as homogeneous complementary metal–oxide–semiconductor (CMOS) circuits.[1,9] Considerable efforthas been devoted to controlling the polarities of 2D tran-sistors, leading to various doping strategies such as solid-state diffusion,[10] charge transfer doping,[11] and substitutionaldoping.[12] However, these methods often involve complicatedfabrication processes and may result in physical and/or chem-ical damage, such as surface contamination and the creationof structural defects.[13,14] Recently, several studies have high-lighted the fundamental role of Schottky barriers at the metal–2D semiconductor junction in determining the carrier typeof 2D transistors.[9,15–17] Schottky barrier engineering allowsfor the simultaneous injection of both electrons and holesinto the 2D semiconductor, offering precise control over itspolarity.[18,19] But one challenging issue remains to be resolved:Fermi level pinning.[2,20–22] Moreover, while the aforementionedapproaches facilitate polarity transitions from n-type to p-type orvice versa,[5,23–25] an investigation into reversible transitions hasnot been systematically conducted to date.In the meantime, device geometry could be a critical factor indetermining the polarity of a 2D transistor because the gate fieldplays a key role in controlling the flow of charge carriers and in-fluencing their behavior in the 2D active channel.[26] Even underthe same doping conditions, the aspects of polarity transitionscan exhibit variation due to the influence of the gate field andother structural factors. The interplay between device geometry(i.e., gate field distribution) and intrinsic property changes of a2D semiconductor therefore contributes to the complex behaviorof charge carriers within a 2D transistor. Despite this fact, a sys-tematic investigation into the correlation among intrinsic prop-erty changes, device geometry effects, and the resulting polarityof a 2D transistor is still lacking.Here, we present a demonstration of a reversible polarity tran-sition between unipolar n- and p-type characteristics of a molyb-denum ditelluride (MoTe2) semiconductor, which is achievedthrough facile annealing processes under controlled environ-ments: air and vacuum conditions. This reversible polarity transi-tion in the MoTe2 layer can be correlated with oxygen physisorp-tion and desorption during air annealing and vacuum anneal-ing, respectively, with supporting evidence from photoemissionspectroscopy results and density functional theory (DFT) calcu-lations. In addition, it is found that the gate electric field at theMoTe2/contact electrode interface has a significant impact on thepolarity of the transistor. By combining optimized annealing con-ditions with appropriate device geometries in the design and fab-rication stages, MoTe2 field-effect transistors (FETs) exhibit re-versible unipolar n- and p-type characteristics. Based on this find-ing, we have successfully implemented a homogeneous comple-mentary logic inverter consisting of symmetric unipolar n- and p-type MoTe2 FETs, which achieved a high gain value of ≈20 at VDD= 5 V. In addition, this is also extended to construct more com-plex logic gate applications such as NAND and NOR. The pro-posed polarity control technique strengthens the potential of 2Delectronic devices in high-performance functional applications.2. Results and DiscussionIn order to understand the influence of air on the densityof states (DOS) of MoTe2, a series of ultraviolet photoelectronspectroscopy (UPS) and X-ray photoelectron spectroscopy (XPS)measurements were conducted in a step-by-step manner. First,2H-MoTe2 crystals were prepared by mechanical exfoliation un-der ultrahigh vacuum conditions of 10−9 torr, enabling the mea-surement of the intrinsic DOS of MoTe2. Subsequently, theMoTe2 samples underwent three annealing steps, each consist-ing of 90 min of annealing in air, 90 min of annealing undervacuum, and 30 min of annealing in air. Figure 1a–e shows rep-resentative UPS and XPS spectra of MoTe2 obtained at variousannealing steps conducted in a sequential order: as-exfoliated(black line), air annealing for 90 min (blue line), vacuum an-nealing for 90 min (red line), and subsequent air annealing for30 min (light blue line). Figure 1a,b displays the UPS spectra inthe secondary cutoff (SECO) and valence band regions, respec-tively, indicating the work function (WF) and the frontier orbitalof MoTe2. The bottommost line (black) shows that the WF of theexfoliated MoTe2 flake was measured to be 4.25 eV. The valencespectral features of the MoTe2 flake in Figure 1b show its intrin-sic DOS, having a valence band maximum (VBM) of 0.77 eV ob-tained from the onset of the spectrum. Based on the reportedoptical bandgap of ≈1.0 eV,[25,27,28] the conduction band mini-mum (CBM) of the exfoliated MoTe2 flake can be estimated tobe 0.23 eV. This finding reveals that the exfoliated MoTe2 flakeexhibits a strong n-type characteristic. It can be explained that, ingeneral, MoTe2 contains intrinsic Te vacancy defect sites due totheir low formation energy,[29] which generate new energy statesclose to the conduction band edge.[30] This can induce n-typeconduction in MoTe2 by leaving behind electrons (VTe∙∙ + 2e− =Te2).[31,32] As a result, MoTe2 may be highly sensitive to the ad-sorption of O2 or H2O molecules that are abundant in ambientatmosphere.[23,33]After annealing in ambient conditions (blue lines), newpeaks emerge in the XPS spectra of Te 3d, Mo 3d, and O 1s(Figure 1c–e, respectively), providing direct evidence of oxida-tion. Simultaneously, the WF of the MoTe2 flake increased to4.89 eV, while the valence band features broadened with a shiftin onset to 0.37 eV. This suggests that surface oxidation changesthe carrier type of MoTe2 from strong n-type to p-type. Next,we performed vacuum annealing to remove the adsorbed O2or H2O molecules, which was mostly successful as shown withthe red lines in Figure 1a–e. A comparison of the UPS spectra(Figure 1a,b) obtained from as-exfoliated and vacuum-annealedsamples reveals that, except for a very slight spectrum broad-ening, the spectra are nearly identical. Upon closer inspectionof the Mo 3d spectra, the Mo─O interaction (yellow peak) isobserved to be partially irreversible even after vacuum annealing.It becomes apparent that while the dominant contribution isfrom reversible oxidation, only a very small contribution from ir-reversible oxidation occurs simultaneously. In contrast, the Te 3dspectra shows fully reversible oxidation/deoxidation behaviors.The changes observed in the O 1s core level spectra (Figure 1e)support above two contributions of oxidation/deoxidationprocesses.Additionally, we conducted identical experiments in controlledenvironments with H2O and O2 to determine the relative signifi-cance of each and to elucidate their respective roles based on theobserved phenomena described above. For this, deionized (DI)water was drop-casted onto the clean surface of MoTe2, while an-nealing itself was carried out in a pure O2 atmosphere (99.99%,Adv. Funct. Mater. 2024, 34, 2404129 2404129 (2 of 9) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 2024, 41, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202404129 by National Institute For, Wiley Online Library on [19/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.deFigure 1. Representative a) secondary cutoff and b) valence region UPS and XPS spectra of c) Te 3d, d) Mo 3d, and e) O 1s of MoTe2 obtained at variousannealing steps conducted in the following sequential order: as-exfoliated (black line), air annealing for 90 min (blue line), vacuum annealing for 90 min(red line), and subsequent air annealing for 30 min (light blue line). f) Summarized work function and valence band maximum as a function of therepeated air and vacuum annealing steps. g) Energy level of MoTe2 as-exfoliated (black), 90 min air annealing (blue), and 90 min vacuum annealing(red).1 atm). As depicted in the UPS spectra (Figure S1a,b, SupportingInformation), annealing in an O2 atmosphere resulted in a pro-nounced polarity shift from n-type to p-type, closely resemblingthe changes observed with air annealing. However, H2O had neg-ligible influence on the polarity of MoTe2, indicating that H2Omolecules are not the predominant factors contributing to the ox-idation observed previously. Similarly, XPS spectra obtained fromsamples annealed in an O2 atmosphere clearly revealed evidenceof oxidation, whereas no such evidence was identified followingdrop-casting of DI water (Figure S2a–c, Supporting Information).Therefore, we can conclude that O2 molecules are the primaryfactors responsible for the oxidation process on the surface ofMoTe2.From the spectral information, we can establish the followinghypothesis regarding surface oxidation. O2 molecules tend to eas-ily dissociate into two oxygen atoms after adsorption on top of theMoTe2 layer.[34] Afterward, the oxygen atoms can either be located1) on the top of chalcogen Te atoms or 2) in Te vacancy sites (oredges), leading to the formation of Mo─O bonds in the MoTe2structure.[23,29,35] The atomic configuration of these two oxygenatom locations is discussed in more detail below using DFTcalculations.Figure 1f summarizes the WF and VBM as a function of re-peated air and vacuum annealing steps as measured in the follow-ing manner. When MoTe2 flake is annealed in air, the VBM grad-ually shifts toward the Fermi energy (EF) and the WF increasesover the annealing time. However, when annealed in vacuum,the opposite trend is observed. In short, annealing in air and vac-uum can reversibly control the polarity transition, i.e., changein the carrier type of MoTe2 between n- and p-type, except forthe small contribution of Mo─O interaction. The energy levelsof MoTe2 measured during air and vacuum annealing are plottedin Figure 1g.We then performed DFT calculations to confirm the polaritytransition through oxygen adsorption/desorption and investigatethe impact of the atomic structure of two distinct oxygen adsorp-tions (on the top of chalcogen Te atoms and in Te vacancy sites) onthe changes in the electronic structure of the MoTe2 layer, as sug-gested by the XPS spectra. Figure 2a,b depicts the energy bandstructure and atomic structure of MoTe2 with and without a Te va-cancy, which are well-matched with the previously reported bandstructure.[29,30,36] In Figure 2b, we observe weak localized states(red) induced by a Te vacancy near the CBM. These states can actas electron donor states, leading to n-type properties. From thisAdv. Funct. Mater. 2024, 34, 2404129 2404129 (3 of 9) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 2024, 41, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202404129 by National Institute For, Wiley Online Library on [19/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.deFigure 2. Calculated electronic band structure along with top and side views of the atomic structure for a) intrinsic MoTe2, b) Te-defective MoTe2, c)MoTe2 with oxygen physisorption, and d) oxygen-incorporated Te-defective MoTe2. Red dashed lines represent defects, oxygen physisorption, or oxygen-incorporated Te-defective states in the electronic band structures.finding, we can explain the n-type characteristics observed in theUPS measurement results of pristine MoTe2 by the presence ofnaturally occurring defects in pristine MoTe2.Upon oxygen adsorption, localized states induced by oxygencomponents appear within the bandgap in both cases, as shownin red in Figure 2c,d. First, in the case of MoTe2 without Te de-fects, the bottom of Figure 2c illustrates the most stable atomicposition of oxygen, corresponding to physisorption. In this case,the gap state is very close to the VBM, and it therefore operatesas an electron acceptor state, leading to MoTe2 exhibiting p-typecharacteristics. In contrast, when oxygen is in a Te vacancy site,the electronic structure (band structure and bandgap) undergoessignificant changes due to strong Mo─O chemical bonding, andthis is accompanied by the appearance of oxygen-induced statesnear the CBM (red). This barely induces p-type characteristics,unlike the physisorption of oxygen on top of defect-free MoTe2.In summary, in pristine MoTe2, the physisorption and desorptionof oxygen are reversible processes, resulting in a reversible shiftbetween n- and p-type characteristics. On the other hand, whenoxygen is located in a defect site, its lower probability of beingthermally separated leads to irreversible changes in the electronicstructure of MoTe2.One of the most efficient methods to examine phase changesin 2D materials is Raman spectroscopy. Here, the Raman spec-trum of as-exfoliated 2H-MoTe2 flake indicates three obviouspeaks corresponding to the out-of-plane A1g mode, in-plane E12gmode, and bulk in-active B12g mode, at 173, 233, and 289 cm−1,respectively (Figure S3a, Supporting Information). The observedpeaks in the MoTe2 flakes remain practically unchanged after airannealing, indicating that the original semiconducting 2H phasein the bottom layers is entirely maintained without structuraldamage or major compositional changes in the MoTe2 layer.[37]Furthermore, the Raman peak of h-BN in the vertically stackedh-BN/MoTe2 structure remained unshifted upon air annealing,indicating that the interfacial properties were not chemically al-tered by oxygen adsorption (Figure S3b, Supporting Informa-tion). This result supports that the carrier type of MoTe2 can bereversibly controlled by repetitive annealing processes, as shownin the UPS and XPS analysis of Figure 1.After conducting fundamental characterizations using theabove-mentioned photoemission spectroscopy and theoreticalcalculations, MoTe2 FETs were fabricated to examine the possi-bility of reversible polarity transitions in actual devices. At first,we used the conventional bottom-gate top-contact geometry, inwhich a gate electrode is overlapped with source/drain (S/D) con-tact electrodes as depicted in Figure 3a,b. The thicknesses of theMoTe2 channel and h-BN dielectric were measured to be ≈16 and10 nm, respectively, from the atomic force microscopy (AFM)profile (Figure S4a,b, Supporting Information). Figure 3c dis-plays the transfer characteristics of the MoTe2 FET as-fabricated,after vacuum annealing, and after air annealing; the output char-acteristics are shown in Figure S4c,d (Supporting Information).Interestingly, the as-fabricated MoTe2 FET exhibited p-type be-havior, in contrast to the n-type behavior observed in the as-exfoliated MoTe2 used for spectroscopic analysis in Figure 1. Thisp-type conduction can be attributed to the adsorption of oxygenatoms during device fabrication under an air atmosphere. Theas-fabricated device is consistent with the results of the air an-nealing steps in the spectroscopy analysis. It is worth noting thatambipolar behaviors were observed after further vacuum anneal-ing and air annealing, rather than unipolar carrier-type transi-tions. After vacuum annealing, n-dominant ambipolar behaviorwas observed, where the electron current was one order of mag-nitude higher than the hole current. On the other hand, a p-dominant ambipolar behavior was achieved after air annealing,with the hole current dominating over the electron current. WeAdv. Funct. Mater. 2024, 34, 2404129 2404129 (4 of 9) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 2024, 41, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202404129 by National Institute For, Wiley Online Library on [19/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.deFigure 3. a) Conventional bottom-gate top-contact device geometry and schematic illustration of the atomic structure around the MoTe2/Pt contactarea. b) Optical microscopy image of the MoTe2 FET. c) Transfer characteristics of the MoTe2 FET as-fabricated (drain bias, VDS = −1 V), after vacuumannealing (VDS = 1 V), and after air annealing (VDS = −1 V). d) Energy band diagrams of the MoTe2/Pt contact at equilibrium and under positive andnegative VGS.hypothesize that charge injection from the MoTe2/Pt contact,modulated by gate bias (VGS), is correlated with the observed am-bipolar behaviors. Figure 3d exhibits energy band diagrams ofthe MoTe2/Pt contact at equilibrium and under positive and neg-ative VGS. Considering the Fermi level pinning effect observed inpreviously reported research, the energy level of the Pt electrodeis determined to be slightly above the midgap level of MoTe2.[38]After vacuum annealing (n-type MoTe2), the contact pinning re-sulted in a small downward band bending at equilibrium. When apositive VGS (gate–source voltage) is applied, the downward bandbending allows electrons to undergo both thermionic emissionand tunneling, which, in turn, contribute to a high electron cur-rent of 1.63 × 10−5 A at VGS = +5 V. On the other hand, the up-ward band bending caused by a negative VGS hinders electroninjection but allows for the possibility of hole conduction. Due toa larger barrier width, a relatively lower but still considerable holecurrent of 3.08× 10−6 A at−5 V was observed. After air annealing,an opposite trend becomes apparent. The p-doping effect leadsto upward band bending in the equilibrium state. Under posi-tive VGS, only weak electron injection is observed because of thelarge barrier. In contrast, the large upward bending by negativeVGS facilitates hole injection, resulting in a high hole current of3.96 × 10−5 A at −5 V. In order to confirm reproducibility, fivemore batches with the same geometry were fabricated and exam-ined for individual device characteristics (Figure S5a–c, Support-ing Information). Similar n-dominant and p-dominant ambipo-lar characteristics were observed with current ratios (I5V/I−5 V) of≈5 × 100 and 4 × 10−2 after vacuum annealing and air annealing,respectively. These results indicate that a proper device geometryis essential to achieve polarity transitions for unipolar electron orhole conduction.After observing the results with the conventional geometry, wenow propose utilizing a locally aligned bottom-gate top-contactgeometry, wherein the gate electrode does not overlap with theS/D contact electrodes as depicted in Figure 4a,b. In this con-figuration, VGS specifically influences the semiconducting layerthat is in contact with the gate electrode, and consequently, the in-trinsic regions of MoTe2 near the S/D contact, unaffected by VGSmodulation, would play a crucial role in the polarity transition.The line width of the locally aligned bottom-gate was measured tobe ≈6 μm (Figure 4b), and the thicknesses of the MoTe2 channeland h-BN dielectric were estimated to be ≈18 and 10 nm, respec-tively (Figure S6a,b, Supporting Information). Figure 4c plots thetransfer characteristics of the MoTe2 FET with locally alignedbottom-gate top-contact geometry as-fabricated, after vacuumannealing, and after air annealing. The output characteristicsAdv. Funct. Mater. 2024, 34, 2404129 2404129 (5 of 9) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 2024, 41, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202404129 by National Institute For, Wiley Online Library on [19/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.deFigure 4. a) Locally aligned bottom-gate top-contact and d) bottom-gate bottom-contact device geometries and corresponding schematic illustrationsof the atomic structure around the MoTe2/Pt contact area. b,e) Optical microscopy images of the MoTe2 FETs in (a) and (d). Transfer characteristics ofc) locally aligned bottom-gate top-contact and f) bottom-gate bottom-contact MoTe2 FETs as-fabricated (VDS = −1 V), after vacuum annealing (VDS =1 V), and after air annealing (VDS = −1 V). g) Energy band diagrams of the MoTe2/Pt contact at equilibrium and under positive and negative VGS.are shown in Figure S6c,d (Supporting Information). As ex-pected, the as-fabricated device showed p-type behavior and a re-duced off current on the order of 10−10 A due to the presenceof relatively high-resistance intrinsic regions. More importantly,unlike the previous conventional bottom-gate top-contact device,the MoTe2 FET with locally aligned bottom-gate top-contact ge-ometry clearly exhibited reversible unipolar transitions from p-type to n-type through vacuum annealing and from n-type to p-type through air annealing. Five more device batches were alsoinvestigated (Figure S7a–c, Supporting Information), and clearpolarity transitions between unipolar n- and p-type conduction(I5V/I−5V: ≈1 × 103 in vacuum annealing and 5 × 10−4 in air an-nealing) were achieved. An alternative device architecture thatcould also lead to a similar polarity transition is bottom-gatebottom-contact geometry, in which case the MoTe2 regions in thecontact areas are barely affected by VGS due to the S/D electrodes,resulting in these edges acting as intrinsic regions. Figure 4d,eshows a cross-sectional view and optical microscope image ofAdv. Funct. Mater. 2024, 34, 2404129 2404129 (6 of 9) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 2024, 41, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202404129 by National Institute For, Wiley Online Library on [19/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.deFigure 5. a) Cross-sectional view of the complementary inverter circuit design composed of n- and p-type MoTe2 FETs with locally aligned bottom-gatetop-contact geometry. The n-FET and p-FET were obtained by vacuum annealing and air annealing, respectively. b) Transfer characteristics and opticalmicroscopy images (insets, scale bar 20 μm) of the n-FET with Al2O3 passivation film and the p-FET. c) Output voltage (left axis) and voltage gain (rightaxis) of the inverter as a function of VDD from 2 to 5 V with a step of 1 V. d) Circuit diagrams of NAND and NOR gates consisting of four MoTe2 FETs(two p-type transistors and two n-type transistors). Output voltages of e) NAND and f) NOR gates as a function of input states (VIN A, VIN B) at VDD =4 V. A VIN of 0 and 8 V was used as input ‘0′ and ‘1′, respectively.a MoTe2 FET with bottom-gate bottom-contact geometry. Mul-tilayer MoTe2 with a thickness of ≈26 nm and h-BN dielectricof ≈16 nm were used (Figure S8a,b, with output characteris-tics shown in Figure S8c,d, Supporting Information). Based onthe transfer characteristics (Figure 4f), consistent behaviors wereclearly observed in the following order: initial p-type behavior dur-ing fabrication, n-type transition during vacuum annealing, andp-type transition during air annealing. Five additional batchesalso showed clear majority carrier types as electrons (vacuum an-nealing) and holes (air annealing), with I5V/I−5 V ratios of ≈9 ×102 and 1 × 10−3, respectively (Figure S9a–c, Supporting Infor-mation).These consistent results obtained from MoTe2 FETs with bothlocally aligned bottom-gate top-contact and bottom-gate bottom-contact geometries confirm that the intrinsic regions unmodu-lated by VGS are a critical factor in determining the polarity tran-sition. The energy band diagrams of the MoTe2/Pt contact, as de-picted in Figure 4g, explain the underlying physical mechanism.In the equilibrium state, the band alignments are similar to thosein the conventional bottom-gate top-contact device. However, theintrinsic regions of MoTe2 near the metal–semiconductor junc-tion retain their initial band structure despite the applied VGS. Inthe case of vacuum-annealed devices, electrons can be injectedand transported under positive VGS, while hole injection can-not occur under negative VGS due to the high barrier height andwidth induced by the intrinsic region. As a result, unipolar n-type characteristics were clearly observed in the transfer curves(Figure 4c,f). On the other hand, after air annealing, only holeconduction is possible under negative VGS, resulting in unipolarp-type characteristics.Furthermore, we also investigated the key molecules that in-duce oxidation and the p-type doping effect in actual MoTe2 de-vices. In order to compare with that from air annealing, O2annealing was carried out after vacuum annealing using lo-cally aligned bottom-gate top-contact geometry. The MoTe2 FETsclearly presented the polarity transition from n-type to p-typethrough O2 annealing, indicating that the O2 atmosphere causesa nearly identical polarity change compared with air atmosphere(Figure S10a,b, Supporting Information). From these device char-acteristics, it was found that the p-type transition of MoTe2 FETscan also be mainly attributed to O2 molecules.By combining the annealing conditions with the above de-vice geometry (locally aligned bottom-gate top-contact), we nowdemonstrate a homogeneous CMOS logic gate application. First,an inverter circuit, the simplest logic gate, was constructed asshown in Figure 5a. In order to prevent p-type doping effects inthe n-FET by adsorption of oxygen atoms onto the MoTe2 layerbefore device measurement, a thin (10 nm) aluminum oxide(Al2O3) layer was deposited on the device as a passivation layervia atomic layer deposition (ALD). Figure 5b exhibits unipolar n-and p-type transfer characteristics. Interestingly, the on currentof the n-FET increased by approximately one order of magnitudeafter the Al2O3 deposition compared to the vacuum-annealed de-vice (Figure 4c), rising from the order of 10−7 to 10−6 A. Thisenhancement is due to the Al2O3 ALD process, which promotesn-type conduction.[23,39,40] Figure 5c displays the voltage transferAdv. Funct. Mater. 2024, 34, 2404129 2404129 (7 of 9) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 2024, 41, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202404129 by National Institute For, Wiley Online Library on [19/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.decharacteristics and voltage gain curves of the homogeneous com-plementary inverter circuit at various supply voltages (VDD). Theinverter exhibits a distinct logic output with a sharp voltage tran-sition, attaining a high gain of ≈20 at a VDD of 5 V. Then for amore complex application, we constructed NAND and NOR logiccircuits by integrating four MoTe2 FETs, comprising two p-typetransistors and two n-type transistors, as illustrated in the circuitdiagram of Figure 5d. Optical images and transfer characteristicsof the four transistors are respectively provided in Figure S11a,b(Supporting Information). As depicted in Figure 5e,f, the outputvoltages demonstrate that the MoTe2 FETs facilitate the success-ful operation of NAND and NOR gates. The p-type doping effectunder air exposure for more than one day weakened the n-typepolarity (Figure S12, Supporting Information), indicating the ne-cessity for further studies to achieve long-term stability of n-FETs.While efforts should be made to enhance the stability of n-typecharacteristics, our straightforward and efficient annealing strat-egy for n- or p-doping allows for precise control of the polarity inMoTe2 devices, enabling the implementation of not only a high-performance homogeneous CMOS inverter but also more com-plex NAND and NOR logic gates.3. ConclusionWe have controlled the polarity transition between unipolar n-and p-type characteristics of a MoTe2 semiconductor through theuse of air annealing and vacuum annealing. The reversible polar-ity transition in the MoTe2 layer can be attributed to the adsorp-tion and desorption of oxygen atoms during the annealing steps,as supported by photoemission spectroscopy results and DFT cal-culations. In addition, we identified the critical role played by theintrinsic regions of MoTe2 near the S/D contact, which remainedunmodulated by VGS, in the polarity transition of the devices.By employing both locally aligned bottom-gate top-contact andbottom-gate bottom-contact geometries along with appropriateannealing conditions, we achieved reversible unipolar n- and p-type characteristics in MoTe2 FETs. This achievement was furthervalidated through the successful operation of a high-performancehomogeneous CMOS logic inverter and more complex NANDand NOR logic gates leveraging unipolar p- and n-type MoTe2FETs. We believe that our proposed strategy provides an effectivemeans for precise polarity control of 2D semiconductors, thuspaving the way for the advancement of 2D nanoelectronics.4. Experimental SectionCharacterization of MoTe2 Samples: XPS and UPS experiments wereconducted using a PHI VersaProbe instrument to analyze the electronicstructure of multilayer MoTe2 flakes after each vacuum, air, and O2 an-nealing. For XPS, a micro-focused and monochromatic Al Ka (1486.6 eV)discharging lamp was utilized as the excitation light source, while a HeI (21.22 eV) discharging lamp was used for UPS. The energy referencesof both μ-XPS and UPS were calibrated with respect to the Fermi level ofAu (111) single crystal. To ensure accurate measurements, a −15 V po-tential was applied to the samples to determine the reliable secondaryelectron cutoff. The thicknesses of the h-BN and MoTe2 nanoflakes weremeasured with a Park Systems XE 100 AFM system. Additionally, Ramanspectroscopy (LabRAM HR Evolution, Horiba Jovin Yvon) was performedwith a laser excitation wavelength of 532 nm to investigate the scatteringfeatures of the MoTe2 layers before and after air annealing.Density Functional Theory (DFT): Model super cells of MoTe2 used inthis work were expanded to a large volume of 14.2 × 14.2 × 61.3 Å3 (4 × 4,4th monolayer) with a 30 Å vacuum slab. The projector augmented wave(PAW) method as implanted in the Vienna Ab initio Simulation Package(VASP) was used for calculations. Standard spin-polarized calculations(collinear) were performed. The electron wave functions were expandedup to an energy cutoff of 500 eV and Monkhorst-Pack 4× 4×1 k-point sam-pling was used. To obtain the band structure of the supercell, band unfold-ing was performed based on the effective band structure (EBS) method.[41]The convergence criteria for structure optimization and energy calculationwere set to accurate quality with SCF tolerance and an energy of 1.0 ×10−5 eV atom−1. The calculated band structures were rigidly shifted toalign with the Fermi level and valence band maximum of MoTe2 to ensurea clear comparison.Device Fabrication and Characterization: Ti (1 nm)/Au (10 nm) bottomgate electrodes were deposited on Si/SiO2 substrate using direct current(DC) magnetron sputtering and patterned through the photolithographicliftoff process. The top- and bottom-contact devices had gate lengths of20 μm, while the locally aligned bottom-gate top-contact device had agate length of 6 μm. Subsequently, h-BN nanoflakes were used as gatedielectrics, mechanically exfoliated using a poly(dimethylsiloxane) stamp.They were then transferred onto the pre-deposited gate electrodes in thethree different device geometries. For the bottom-gate top-contact devices,MoTe2 nanoflakes were exfoliated and transferred onto the pre-depositedh-BN flakes. Pt (20 nm) was deposited using DC magnetron sputteringas S/D electrodes and patterned using photolithography. For the bottom-gate bottom-contact devices, the S/D electrodes were formed on the h-BN dielectric, and the MoTe2 semiconducting layer was prepared usingthe above-mentioned method. All electrical measurements of the MoTe2FETs were performed using a semiconductor parameter analyzer (4200-SCS Keithley). Vacuum annealing of the MoTe2 devices was conducted at200 °C for 30 min in a vacuum chamber, followed by measurement in vac-uum (≈10−2 torr) at room temperature. Air annealing of the MoTe2 deviceswas conducted at 200 °C for 10 min on a hot plate, followed by measure-ment in ambient air at room temperature. The MoTe2 devices underwentO2 annealing at 200 °C for 10 min with a flow rate of 20 sccm, followedby measurement in ambient air at room temperature. The electrical char-acteristics of the homogenous MoTe2 CMOS inverter and NAND/NORgates were measured using a semiconductor parameter analyzer (Agilent4156 B) in a dark shield box.Supporting InformationSupporting Information is available from the Wiley Online Library or fromthe author.AcknowledgementsB.-S.Y. and W.K. contributed equally to this work. D.K.H. and S.P. ac-knowledge financial support from the Korea Institute of Science andTechnology (KIST) Institution Program (Grant nos. 2E32942, 2V09840-23-P024, and 2V09703), KU-KIST School project, the Institute of in-formation & communications Technology Planning & Evaluation (IITP)(Grant. no.2020-0-00841), and the National Research Foundation of Ko-rea (NRF) (Grant nos. 2023R1A2C2003985, 2021M3H4A6A02050353, and2022M3D1A2095315). J.A. acknowledges financial support from First Re-search Fund Program in Life at Changwon National University in 2023.The authors thank the Pohang Accelerator Laboratory for allocating syn-chrotron radiation beam time (PAL 3C and 4A1, Republic of Korea).Conflict of InterestThe authors declare no conflict of interest.Adv. Funct. Mater. 2024, 34, 2404129 2404129 (8 of 9) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 2024, 41, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202404129 by National Institute For, Wiley Online Library on [19/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.deData Availability StatementThe data that support the findings of this study are available from the cor-responding author upon reasonable request.Keywordsfield-effect transistor, logic application, MoTe2, reversible polarity controlReceived: March 8, 2024Revised: April 27, 2024Published online: May 15, 2024[1] S. Y. Seo, G. Moon, O. F. N. Okello, M. Y. Park, C. Han, S. Cha, H.Choi, H. W. Yeom, S. Y. Choi, J. Park, M. H. Jo, Nat. Electron. 2021, 4,38.[2] J. Jang, H. S. Ra, J. Ahn, T. W. Kim, S. H. Song, S. Park, T. Taniguch,K. Watanabe, K. Lee, D. K. Hwang, Adv. Mater. 2022, 34, 2109899.[3] J. E. Seo, T. Das, E. Park, D. Seo, J. Y. Kwak, J. W. Chang, ACS Appl.Mater. Interfaces 2021, 13, 43480.[4] Q. H. Wang, K. Kalantar-Zadeh, A. Kis, J. N. Coleman, M. S. Strano,Nat. Nanotechnol. 2012, 7, 699.[5] X. Liu, A. Islam, J. Guo, P. X. L. Feng, ACS Nano 2020, 14, 1457.[6] M. Chhowalla, H. S. Shin, G. Eda, L. J. Li, K. P. Loh, H. Zhang, Nat.Chem. 2013, 5, 263.[7] K. F. Mak, J. Shan, Nat. Photonics 2016, 10, 216.[8] J. Ahn, K. Ko, J. H. Kyhm, H. S. Ra, H. Bae, S. Hong, D. Y. Kim, J. Jang,T. W. Kim, S. Choi, J. H. Kang, N. Kwon, S. Park, B. K. Ju, T. C. Poon,M. C. Park, S. Im, D. K. Hwang, ACS Nano 2021, 15, 17917.[9] L. G. Kong, X. D. Zhang, Q. Y. Tao, M. L. Zhang, W. Q. Dang, Z. W. Li,L. P. Feng, L. Liao, X. F. Duan, Y. Liu, Nat. Commun. 2020, 11, 1866.[10] A. Eftekhari, J. Mater. Chem. A 2017, 5, 18299.[11] S. Y. Zhang, H. M. Hill, K. Moudgil, C. A. Richter, A. R. H. Walker, S.Barlow, S. R. Marder, C. A. Hacker, S. J. Pookpanratana, Adv. Mater.2018, 30, 1802991.[12] J. Gao, Y. D. Kim, L. B. Liang, J. C. Idrobo, P. Chow, J. W. Tan, B. C. Li,L. Li, B. G. Sumpter, T. M. Lu, V. Meunier, J. Hone, N. Koratkar, Adv.Mater. 2016, 28, 9735.[13] V. P. Pham, G. Y. Yeom, Adv. Mater. 2016, 28, 9024.[14] G. Gawlik, P. Ciepielewski, J. M. Baranowski, Appl. Sci. 2019, 9, 544.[15] P. C. Shen, C. Su, Y. X. Lin, A. S. Chou, C. C. Cheng, J. H. Park, M. H.Chiu, A. Y. Lu, H. L. Tang, M. M. Tavakoli, G. Pitner, X. Ji, Z. Y. Cai,N. N. Mao, J. T. Wang, V. C. Tung, J. Li, J. Bokor, A. Zettl, C. I. Wu, T.Palacios, L. J. Li, J. Kong, Nature 2021, 593, 211.[16] S. Das, H. Y. Chen, A. V. Penumatcha, J. Appenzeller, Nano Lett 2013,13, 100.[17] R. T. Tung, Appl. Phys. Rev. 2014, 1, 011304.[18] T. D. Ngo, Z. Yang, M. Lee, F. Ali, I. Moon, D. G. Kim, T. Taniguchi,K. Watanabe, K. Y. Lee, W. J. Yoo, Adv. Electron. Mater. 2021, 7,2001212.[19] Z. Yang, C. Kim, K. Y. Lee, M. Lee, S. Appalakondaiah, C. H. Ra, K.Watanabe, T. Taniguchi, K. Cho, E. Hwang, J. Hone, W. J. Yoo, Adv.Mater. 2019, 31, 1808231.[20] Y. Liu, J. Guo, E. B. Zhu, L. Liao, S. J. Lee, M. N. Ding, I. Shakir, V.Gambin, Y. Huang, X. F. Duan, Nature 2018, 557, 696.[21] S. Chuang, C. Battaglia, A. Azcatl, S. McDonnell, J. S. Kang, X. T. Yin,M. Tosun, R. Kapadia, H. Fang, R. M. Wallace, A. Javey, Nano Lett.2014, 14, 1337.[22] Y. Y. Liu, P. Stradins, S. H. Wei, Sci. Adv. 2016, 2, 1600069.[23] D. S. Qu, X. C. Liu, M. Huang, C. Lee, F. Ahmed, H. Kim, R. S. Ruoff,J. Hone, W. J. Yoo, Adv. Mater. 2017, 29, 1606433.[24] M. Yamamoto, S. Nakaharai, K. Ueno, K. Tsukagoshi, Nano Lett.2016, 16, 2720.[25] Y. J. Park, A. K. Katiyar, A. T. Hoang, J. H. Ahn, Small 2019, 15,1901772.[26] G. V. Resta, S. Sutar, Y. Balaji, D. Lin, P. Raghavan, I. Radu, F. Catthoor,A. Thean, P. E. Gaillardon, G. de Micheli, Sci. Rep. 2016, 6, 29448.[27] J. Ahn, J. H. Kyhm, H. K. Kang, N. Kwon, H. K. Kim, S. Park, D. K.Hwang, ACS Photonics 2021, 8, 2650.[28] X. J. Liu, H. Yang, J. He, H. X. Liu, L. D. Song, L. Li, J. Luo, Small 2018,14, 1704049.[29] H. S. Liu, N. N. Han, J. J. Zhao, RSC Adv. 2015, 5, 17572.[30] X. C. Liu, D. S. Qu, L. Wang, M. Huang, Y. H. Yuan, P. Chen, Y. Y. Qu,J. Sun, W. J. Yoo, Adv. Funct. Mater. 2020, 30, 2004880.[31] J. K. Liu, Y. Y. Wang, X. Y. Xiao, K. A. Zhang, N. Guo, Y. Jia, S. Y. Zhou,Y. Wu, Q. Q. Li, L. Xiao, Nanoscale Res. Lett. 2018, 13, 291.[32] L. Yang, H. Wu, W. F. Zhang, X. Lou, Z. J. Xie, X. Yu, Y. Liu, H. X. Chang,Adv. Electron. Mater. 2019, 5, 1900552.[33] M. S. Choi, M. Lee, T. D. Ngo, J. Hone, W. J. Yoo, Adv. Electron. Mater.2021, 7, 2100449.[34] Y. Guo, S. Zhou, J. J. Zhao, ChemNanoMat 2020, 6, 838.[35] B. Chen, H. Sahin, A. Suslu, L. Ding, M. I. Bertoni, F. M. Peeters, S.Tongay, ACS Nano 2015, 9, 5326.[36] L. Lin, Z. Y. Feng, Z. Z. Dong, C. C. Hu, L. H. Han, H. L. Tao, Phys. E:Low-Dimens. Syst. Nanostructures 2023, 145, 115489.[37] Y. X. Ke, D. Y. Qi, C. Han, J. D. Liu, J. Q. Zhu, Y. J. Xiang, W. J. Zhang,ACS Appl. Electron. Mater. 2020, 2, 920.[38] M. J. Mleczko, A. C. Yu, C. M. Smyth, V. Chen, Y. C. Shin, S. Chatterjee,Y. C. Tsai, Y. Nishi, R. M. Wallace, E. Pop, Nano Lett. 2019, 19,6352.[39] J. Y. Lim, A. Pezeshki, S. Oh, J. S. Kim, Y. T. Lee, S. Yu, D. K. Hwang,G. H. Lee, H. J. Choi, S. Im, Adv. Mater. 2017, 29, 1701798.[40] H. Liu, A. T. Neal, M. W. Si, Y. C. Du, P. D. Ye, IEEE Electron DeviceLett. 2014, 35, 795.[41] V. Popescu, A. Zunger, Phys. Rev. B 2012, 85, 085201.Adv. Funct. Mater. 2024, 34, 2404129 2404129 (9 of 9) © 2024 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 2024, 41, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202404129 by National Institute For, Wiley Online Library on [19/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.de Reversible Polarity Control in 2D MoTe2 Field-Effect Transistors for Complementary Logic Gate Applications 1. Introduction 2. Results and Discussion 3. Conclusion 4. Experimental Section Supporting Information Acknowledgements Conflict of Interest Data Availability Statement Keywords