# Fileset

[AMI_MoS2.pdf](https://mdr.nims.go.jp/filesets/891e2e02-c3c1-4470-bdfc-8eabb67cb199/download)

## Creator

[Yen‐Ju Wu](https://orcid.org/0000-0003-2647-3407), Yann‐Wen Lan, Shih‐Chieh Hsu, Chen‐Hao Yeh, Yu‐Seng Ku, Jyh‐Chiang Jiang, [Yibin Xu](https://orcid.org/0000-0001-8600-8748)

## Rights

[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[Tuning Interfacial Thermal and Electrical Conductance across a Metal/MoS2 Monolayer through N‐Methyl‐2‐pyrrolidone Wet Cleaning](https://mdr.nims.go.jp/datasets/b06821a7-8def-4834-b901-a74b467eaa27)

## Fulltext

Tuning Interfacial Thermal and Electrical Conductance across a Metal/MoS2 Monolayer through N‐Methyl‐2‐pyrrolidone Wet Cleaningwww.advmatinterfaces.de2000364  (1 of 9) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, WeinheimFull PaperTuning Interfacial Thermal and Electrical Conductance across a Metal/MoS2 Monolayer through N-Methyl-2-pyrrolidone Wet CleaningYen-Ju Wu, Yann-Wen Lan,* Shih-Chieh Hsu, Chen-Hao Yeh, Yu-Seng Ku, Jyh-Chiang Jiang, and Yibin Xu*DOI: 10.1002/admi.2020003641. Introduction2D materials have become attractive for overcoming the bottleneck that causes effi-ciency degradation in nanoscale electronic devices such as thin-film transistors,[1] supercapacitors,[2] and photodetectors[3,4] as a result of reducing the particle size of bulk materials to the nanoscale. Com-pared with conventional bulk semicon-ductors such as Si, 2D materials have relatively high stability, dangling-bond-free surfaces, and ultrathin atomic-scale thickness. Here, the interface of Au and monolayer MoS2 was chosen for investiga-tion. Au is broadly used for electrodes with high stability under ambient conditions, whereas MoS2 is a 2D material in the tran-sition metal dichalcogenide (TMD) family. Because of its abundance, nontoxicity, and excellent electrical properties,[1,5] MoS2 has been widely studied. It can also be syn-thetized into large-area thin films. The high electrical contact resistance at the metal/2D material interface can be tuned Extensive effort is dedicated to developing 2D materials as an alternative to Si-based semiconductor technology. As the size decreases, heat dissipation at various interfaces becomes increasingly important in controlling device performance. On the other hand, the high interfacial thermal resistances can be applied for thermoelectric devices or thermal insulators by achieving ultralow thermal conductivity via nanostructuring. Here, it is found that a) the thermal and electrical conductance of the Au/MoS2 monolayer interface can be tuned by changing the interfacial chemical properties through N-methyl-2-pyrrolidone (NMP) wet cleaning while preserving the MoS2 structure; b) the effectiveness of the NMP cleaning process, which removes surface adsorb-ates, exhibits a temperature dependence; c) experimental results demonstrate that adequate oxygen adsorbates at the Au/MoS2 interface significantly improve the thermal and electrical conductance, in agreement with the simulation results. The interfacial thermal conductance increases by 339.87% when oxygen adsorbates are partially removed and decreases by 74.37% as a sharp interface exists (oxygen adsorbates are removed) compared to the as-deposited interfaces. The electrical conductance shows up to 2 order increase after NMP cleaning. This finding can both enhance the heat dissipation in functional devices and provide new options for the interface design of thermal insulating thin films.Dr. Y.-J. WuInternational Center for Young Scientists (ICYS)National Institute for Materials Science (NIMS)1-2-1 Sengen, Tsukuba, Ibaraki 305-0047, JapanDr. Y.-J. Wu, Dr. Y. XuCenter for Materials research by Information Integration (CMI2)Research and Service Division of Materials Data  and Integrated System (MaDIS)National Institute for Materials Science (NIMS)1-1 Namiki, Tsukuba, Ibaraki 305-0044, JapanE-mail: Xu.Yibin@nims.go.jpThe ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/admi.202000364.Prof. Y.-W. Lan, Y.-S. KuDepartment of PhysicsNational Taiwan Normal University88, Sec.4, Ting-Chou Rd., Taipei 11677, TaiwanE-mail: ywlan@ntnu.edu.twProf. S.-C. HsuDepartment of Chemical and materials EngineeringTamkang UniversityNo.151, Yingzhuan Rd., Tamsui Dist., New Taipei City 25137, TaiwanProf. C.-H. YehDepartment of Materials Science and EngineeringFeng Chia UniversityNo. 100, Wenhwa Rd., Seatwen, Taichung 40724, TaiwanProf. C.-H. YehFirst-Principles Simulation GroupNano-Theory FieldInternational Center for Materials Nanoarchitectonics (WPI-MANA)National Institute for Materials Science (NIMS)1-1 Namiki, Tsukuba, Ibaraki 305-0044, JapanProf. C.-H. Yeh, Prof. J.-C. JiangDepartment of Chemical EngineeringNational Taiwan University of Science and TechnologyNo.43, Keelung Rd., Sec.4, Da’an Dist., Taipei 10607, Taiwan© 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and repro-duction in any medium, provided the original work is properly cited.The copyright line for this article was changed on 5 June 2020 after original online publication.Adv. Mater. Interfaces 2020, 2000364http://creativecommons.org/licenses/by/4.0/http://creativecommons.org/licenses/by/4.0/http://crossmark.crossref.org/dialog/?doi=10.1002%2Fadmi.202000364&domain=pdf&date_stamp=2020-05-27www.advancedsciencenews.comwww.advmatinterfaces.de2000364  (2 of 9) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheimthrough phase or doping engineering.[6–9] An ultrahigh vacuum (10−9 Torr) metal deposition process is suggested to ensure high interface quality at the metal/2D material junction, which yields a threefold increase in the electrical contact conductance.[10]Thermal management and heat transport at the metal/2D material interface have attracted less attention than electrical performance. Nevertheless, poor heat dissipation exacerbates the degradation of devices as their dimensionality is reduced. Interfacial heat transport is sensitive to the material struc-ture and chemical bonding; therefore, vacancies, interstitial impurities, and adsorbates on the MoS2 surface will affect the thermal conductance. To ensure uniformity and cleanliness of large-area semiconductor devices, a wet chemical cleaning process is commonly used. N-Methyl-2-pyrrolidone (NMP), which is used for cleaning and transferring 2D materials or layered materials via liquid-phase exfoliation, is a good hydro-philic solvent for enhancing the interfacial quality of metal/MoS2.[11–13] Lan and co-workers have reported that NMP cleaning can improve the electrical performance of monolayer MoS2 field-effect transistors.[14] However, whether NMP modi-fies the thermal properties of the metal/2D material interface is unclear.MoS2 is a promising thermoelectric material because of its unique density of states (DOS), which can enhance the Seebeck coefficient and result in a power factor as large as 8.5  mW m−1 K−2 for a bilayer MoS2 variant at room tempera-ture.[15] Thermoelectric properties can be improved through phonon engineering,[16] engineering of the strain effect,[17] and introducing various dopants for p- and n-type materials.[18,19] We previously proposed an interface design of inorganic composite thin films via machine learning and nanostructure optimiza-tion to achieve ultralow thermal conductivity for thermally insu-lating thin films or thermoelectric materials.[20,21] The machine learning techniques can accelerate the material exploration for development[22–24] and efficiency management of long-term control.[25,26] This strategy holds promise for reducing the total thermal conductivity along the cross-plane direction via nanoengineering for metal/MoS2 interfaces, which exhibit low thermal conductance. The metal can preserve the electrical conductance, and the metal/MoS2 interfaces can impede heat transfer to ensure a low thermal conductivity.Here, we demonstrate a tunable interfacial thermal and electrical conductance effect across a metal/MoS2 monolayer using the NMP cleaning method. We used density functional theory (DFT) calculations to show that the amounts of oxygen adsorbates at the Au/MoS2 interface affect the thermoelectric properties of the material: medium oxygen adsorbate con-tents (25–50%) correspond to materials with better thermal and electrical conductivities. Therefore, we used the NMP wet cleaning method to remove adsorbates at the Au/MoS2 inter-faces. We found that the cleaning efficiency of NMP is tempera-ture dependent and that, remarkably, the experimental results agree with the simulation results. The atomic binding, struc-ture, and tensile strain properties that affect the thermal and electrical conductance at the metal/MoS2 monolayer interface were characterized by Raman spectroscopy, X-ray photoelectron spectroscopy (XPS), and transmission electron microscopy (TEM). The tunability of the thermal resistance was verified using frequency-domain thermoreflectance (FDTR),[27] and the device electrical resistance at the Au/MoS2 interface was analyzed.2. Results and Discussion2.1. Simulation ResultsThe effect of different amounts of oxygen adsorbates or con-taminants on the thermal and electrical conductance at the Au/MoS2 interface were first analyzed via DFT calculations. The constructed unit cell of the Au/MoS2 has a thickness of ≈1.9  nm; the unit cells with various oxygen percentages are shown in Figure 1a. Because the simulation is based on a peri-odic structure, which behaves more like a Au/MoS2 composite, a direct comparison of the simulation and experimental values is not possible. However, the simulation results can be used to analyze the effect of various oxygen amounts on the thermal and electrical properties in the Au/MoS2 systems. In order to easily analyze the simulated results, here, the relaxation time (τ) of all structures is assumed to be constant. We compared the tendency of thermal conductivity/relaxation time (κ/τ) and electrical conductivity/relaxation time (σ/τ), and the units are W m−1 K−1 s−1 and Ω−1 m−1 s−1, respectively. The simulated κ/τ, σ/τ, and Seebeck coefficient versus temperature in the range from 100 to 500 K are shown in Figure 1b–d. The percentages of oxygen adsorbates (surface coverage) at the Au/MoS2 inter-faces are 0%, 25%, 50%, 75%, and 100%.The κ/τ of all of the samples increases with increasing temperature, and the order of them is 25% O > 50% O > 75% O = 100% O >  0% O at temperatures greater than 300 K, as shown in Figure  1b. The residual oxygen adsorbates after the 50  °C NMP cleaning play an important role in modifying the interface region, resulting in better phonon transport. The σ/τ in all samples with different O percentages decreases with increasing temperature, as shown in Figure 1c. All of the energy bandgaps of the MoS2 monolayers with O adsorbates on their surface decrease compared with the energy bandgap of the as-deposited MoS2. In addition, the overlap of DOS between MoS2 and Au decreases with increasing oxygen percentage. The detailed electron DOS of the MoS2 monolayer and the Au/MoS2 interfaces can be found in Figure S2 (Supporting Information). The order of σ/τ at temperatures greater than 300 K is 25% O >  50% O >  100% O ≈ 75% O >  0% O. These results indicate that adequate oxygen would improve the thermal and electrical conductance but might also increase the interfacial vacancies or roughness in a real experimental procedure, both of which need to be carefully controlled.The simulated Seebeck coefficient does not exhibit strong temperature dependence at temperatures greater than 300 K in Figure  1d. The samples with adsorbate contents of 0% O and 50% O exhibit p-type characteristics, whereas those with adsorbate contents of 75% O and 100% O exhibit n-type charac-teristics; the samples with 25% O undergo a conversion from p- to n-type at temperatures greater than 300 K. We had per-formed some experiments related to the conversion from n- to p-type in MoS2 and other TMDs, and the conversion could be controlled by oxide layers,[28] the doping via implantation tech-nique,[29] and electron-beam irradiation.[30] The oxygen doping Adv. Mater. Interfaces 2020, 2000364www.advancedsciencenews.comwww.advmatinterfaces.de2000364  (3 of 9) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheimis one of commonly used methods to achieve the conversion. According to our experiences, we simulated the oxygen content at Au/MoS2 interfaces and predicted the Seebeck coefficient at different temperatures, as can be seen in Figure  1d. The See-beck coefficient of the p-type samples is slightly greater than that of the n-type samples. The order of Seebeck coefficients at temperatures less than 200 K is 25% O ≈ 50% O > 0% O > 75% O ≈ 100% O, whereas the 0% O samples exhibit the highest Seebeck coefficient at temperatures greater than 300 K. Thus, higher Seebeck coefficients were achieved with oxygen adsorb-ates at temperatures less than 200 K. These simulation results suggest that residual (adequate) oxygen adsorbates at the inter-face would provide better thermal and electrical conductivities. Next, we will analyze the experimentally determined thermal and electrical conductivity results.2.2. CharacterizationWe assessed the quality of our large-area monolayer MoS2 films using Raman spectroscopy. Figure 2a shows the Raman spectra of the MoS2 monolayer before (dotted line) and after (solid line) the NMP chemical treatment at different temperatures. To exclude the position effect, the Raman spectra were recorded at the same location on each sample before and after the NMP cleaning. The two characteristic peaks at ≈381 and ≈403 cm−1 correspond to the in-plane vibration (E2g1 ) of the Mo and S atoms and to the out-of-plane vibration (A1g) of the S atoms, as displayed in Figure  2a, respectively. The A1g peak of the sap-phire substrate, which was used to align the data, also appears at 414 cm−1 in the spectra of all of the samples. The frequency difference (∆ω) between the E2g1  and A1g modes of MoS2 is ≈21–22 cm−1, which is slightly greater than previously reported values of 18–20 cm−1 for MoS2 monolayers.[31–33] However, the characteristic peaks and ∆ω not only depend on the layer number but also on the applied laser wavelength, position, and synthesis method (e.g., chemical vapor deposition (CVD) or exfoliation).[32] The precise estimation of the MoS2 thickness was further confirmed by TEM images presented later in this section. In all samples shown in Figure  2a, the characteristic peaks do not exhibit a substantial change in frequency or inten-sity after the NMP cleaning at temperatures ranging from 25 to 110 °C, indicating that the MoS2 structure remained intact after the NMP cleaning.The Raman spectra of MoS2 coated with 5 nm thick Au are shown in Figure 2b. The Raman signal from the Au/MoS2 inter-faces is still visible in the spectra of the samples with a 5 nm Au layer but not in the spectra of the samples with 150 nm thick Au layer, which were used for the FDTR measurements. Com-pared with the spectrum of the as-deposited MoS2 without Au (green curve in Figure 2b), that of the Au-coated sample shows a split of the characteristic peak of the E2g1  mode into two peaks at 375.44 and 381.82 cm−1 (∆ω ≈ 6 cm−1), whereas the A1g mode shows no clear shift. A similar redshift (∆ω) has been reported for separated peaks at 6.6 and 6.38 cm−1 in the spectra of Au/Ti/MoS2 and Au/MoS2 interfaces.[34,35] The tensile strain induced by the lattice mismatch of Au and MoS2 causes the redshift of the E2g1  mode, whereas the interfacial bonding variation induced by Au deposition is negligible. Nevertheless, the peak inten-sity of the out-of-plane vibration (A1g) slightly decreases when the sample temperature during the NMP cleaning process is increased, as shown in Figure 2b, implying that the interfacial bonding of Au/MoS2 might be weakened by NMP cleaning.MoS2 samples were analyzed by XPS after being transferred from sapphire to Au substrates to prevent interference from the oxygen atoms of the sapphire substrate. The XPS results in Figure 1.  a) Unit cells of Au/MoS2 with different oxygen atom contents of 0%, 25%, 50%, 75%, and 100% (surface coverage). The simulated structure is composed of periodic unit cells separated by a 15 Å vacuum spacing to ensure no interactions between the Au/MoS2 composites. The simulated b) κ/τ, c) σ/τ, and d) Seebeck coefficient of a Au/MoS2 monolayer (at different percentages of oxygen adsorbates) versus NMP temperatures. For comparison with the experimental results, the simulated samples with 75–100% O correspond to the as-deposited MoS2, those with 25–50% O cor-respond to the 50 °C NMP-cleaned MoS2, and those with 0% O correspond to the 110 °C NMP-cleaned MoS2. The sample with intermediate oxygen adsorbates (25–50%) shows optimized thermal and electrical conductance.Adv. Mater. Interfaces 2020, 2000364www.advancedsciencenews.comwww.advmatinterfaces.de2000364  (4 of 9) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, WeinheimFigure 2c reveal that the carbon and oxygen contents decrease by 35.5% and 44.65%, respectively, after the 110 °C NMP cleaning. At the same time, the Mo3d, S2s, Mo3p, and S2p binding energies show no obvious variations (details are available in Figure S3 in the Supporting Information). These results imply that the NMP cleaning can effectively remove the undesirable adsorbates or contaminants from the surface while not affecting the physical properties of the MoS2. The effect of the NMP residue or the adsorbates on the transfer the samples from the sapphire sub-strate to the Au substrate can be excluded; details are available elsewhere.[14]The lattice structure of the Au/MoS2/sapphire interfaces was characterized by TEM, as shown in Figure  2e–g. A broad and mixed interfacial region at the Au/as-deposited MoS2 interface with a thickness of 1.43–1.79 nm of MoS2 is shown in Figure 2e. The interface of the Au/110 °C NMP-cleaned MoS2 in Figure 2g is smooth and clear and the thickness of MoS2 is 0.71–1.07 nm, which is exactly the monolayer thickness of MoS2. The Au/50 °C NMP-cleaned MoS2 in Figure 2f shows intermediate thickness (0.83–1.4  nm) and interfacial cleanness compared with the interfaces in Figure 2e,g. This comparison shows that the NMP cleaning can eliminate the adsorbates from the MoS2 surface, consistent with the XPS results, and that the efficiency of cleaning increases with increasing temperature. In addi-tion, the adsorbates or contaminants on the as-deposited MoS2 surface form a mixed region at the interface of Au and MoS2. The extra oxygen (or carbon) will likely form new bonds such as OS and AuO at the Au/MoS2 interface, corresponding to the intensity change of A1g in Figure 2b. The change in the chemical characteristics at the Au/MoS2 interfaces will affect the thermal and electrical transport across the interfaces.2.3. Thermal PropertiesThe total thermal conductance of Au/MoS2/sapphire is plotted in Figure 3a as a function of the NMP cleaning temperature. The thermal conductance shown in Figure  3a was evaluated on the basis of the reciprocal of the total thermal resistance in Figure S1 (Supporting Information). The Au surface was heated by a 405  nm pump laser, and the thermoreflectance was detected by a 635 nm probe laser, as shown in Figure 3b. The thermal conductance at different NMP temperatures are listed in Table 1. The oxygen percentage of the as-deposited interface is assumed as 100% and that of other interfaces decreases as NMP temperatures increase. According to the Figure 2.  a) Raman spectra of the MoS2 monolayer of the as-deposited, 25, 50, 80, and 110 °C NMP-cleaned devices. The Raman spectra were recorded at the same position before and after NMP cleaning; these spectra are represented by dotted and solid lines, respectively. The yellow and green atoms of the vibration modes (E2g1  and A1g) represent the S and Mo atoms, respectively. The characteristic peaks do not change after the NMP cleaning process at various temperatures. b) The Raman spectra of an MoS2 monolayer with 5 nm Au top layer. The as-deposited MoS2 and the 50 and 110 °C NMP-cleaned MoS2 are represented by blue, red, and black curves, respectively. The spectrum of the as-deposited MoS2 without the Au top layer is indicated by a green curve and is provided for reference. The observed redshift of the E2g1  mode is due to the tensile strain induced by the lattice mis-match between Au and MoS2. The decrease in intensity of the A1g mode is attributed to the lower interfacial bonding at the Au/MoS2 interface. The XPS spectra for the c) C 1s and d) O 1s core peaks of the MoS2 sample transferred onto a Au substrate with and without NMP cleaning. The carbon and oxygen contents decrease by 35.5% and 44.65% after NMP cleaning. TEM images of the Au/MoS2 monolayer/sapphire substrate, where the MoS2 layer is e) as-deposited, f) 50 °C NMP-cleaned, and g) 110 °C NMP-cleaned. The interfacial region decreases after NMP cleaning, as revealed by a com-parison of the (e) as-deposited sample with a thicker interfacial region and the (g) 110 °C NMP-cleaned sample with the sharpest Au/MoS2 interface.Adv. Mater. Interfaces 2020, 2000364www.advancedsciencenews.comwww.advmatinterfaces.de2000364  (5 of 9) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheimpositive correlation between cleaning efficiency and NMP temperature discussed above (Figure  2c–g), the simulated samples in Figure  1a with 75–100% O can be represented as the as-deposited MoS2, those with 25–50% O as the 45–50 °C NMP-cleaned MoS2, and those with 0% O as the 110 °C NMP-cleaned MoS2.The total thermal conductance of the sample with the as-deposited MoS2 was 19.89 MW m−2 K−1. The relatively low thermal conductance of metal/MoS2 interfaces indicates that such interfaces present a bottleneck in heat dissipation.[36] When the NMP cleaning temperature was increased to 45  °C, the thermal conductance increased by 3.39 times, to 87.49 MW m−2 K−1. This increase in thermal conductance is attributed to the removal of undesirable adsorbates on the MoS2 surface by the NMP cleaning process. However, when the cleaning temperature was increased to greater than 50 °C, the thermal conductance became lower than that of the as-deposited sample, decreasing by 74.37% to 5.12 MW m−2 K−1 at an Figure 3.  a) Total thermal conductance of the Au/MoS2/sapphire interface versus NMP cleaning temperature from 25 to 110 °C. The thermal conduct-ance increased by 339.87% after 45 °C NMP cleaning and conversely decreased by 74.37% after 110 °C NMP cleaning. The as-deposited sample is marked in green, whereas the green dashed line is used for comparing the thermal conductance with that of the as-deposited sample. b) Schematic of the thermal conductance measurement of Au/MoS2 monolayer on the sapphire substrate (see the details of the 1D heat conduction equation in the Experimental Section (Experimental Methods)). c) I–V curves of the as-deposited MoS2 and MoS2 that was subjected to 50 and 110 °C NMP cleaning. The 50 °C NMP cleaning dramatically improves (decreases) the contact resistance. The bottom illustration shows the measured samples with patterned Au pads corresponding to the photo in (e). d) I–V curves of the as-deposited MoS2 and the 50 °C NMP-cleaned MoS2. Linear I–V curves are observed after the NMP cleaning. e) The patterned Au pad of the 50 °C NMP-cleaned sample (the scale bar represents 1 mm).Table 1.  The thermal conductance of the Au/MoS2 monolayer under different NMP cleaning temperatures. The as-deposited interface is assumed to have the most oxygen adsorbates at the interface, and the oxygen percentage decreases with increasing the NMP temperatures corresponding to the TEM images in Figure 2e–g.O percentage [%] at the interface100 0NMP temperature [°C] As-deposited 25 45 50 60 70 80 100 105 110Thermal conductance [MW m−2 K−1]19.89 36.76 87.49 62.70 16.65 14.39 12.68 11.11 10.46 5.12Adv. Mater. Interfaces 2020, 2000364www.advancedsciencenews.comwww.advmatinterfaces.de2000364  (6 of 9) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, WeinheimNMP cleaning temperature of 110  °C. As the TEM images in Figure 2g show, a smooth and clean Au/MoS2 interface can be achieved through NMP cleaning at 110 °C; however, the phonon density of states (PDOS) of Au and MoS2 have a large mis-match above 5 THz that remarkably hinders the heat transport and leads to low interfacial thermal conductance.[37]The high thermal conductance of the 50  °C NMP-cleaned sample, whose TEM image is presented in Figure  2f, implies that adsorbates such as oxygen may increase the overlap region of the PDOS between Au and MoS2. Zhan et al. have reported that the PDOS is modified in the interface regions, resulting in a better match than in the bulk region.[38] Hahn et  al. pro-posed a thermal conductance analysis of Si/Ge interfaces with mixed regions by molecular dynamics simulations; their results showed that, compared with the sharp interface, the 0.5 nm mixed region exhibited increased thermal conductance, whereas a larger mixed region (1 or 2  nm) exhibited lowered thermal conductance.[39] Accordingly, there would be a trade-off to enhancing the thermal conductance across the Au/MoS2 interfaces: the broad mixed region (Figure 2e) and sharp inter-face (Figure  2g) both adversely affect thermal transport, and the intermediate interface (Figure  2f) has optimized thermal conductance. That is, the thermal conductivity of the Au/50 °C NMP-cleaned MoS2 is greater than that of the Au/110 °C NMP-cleaned MoS2, whereas the Au/as-deposited MoS2 exhibits the lowest thermal conductivity, in good agreement with the simu-lation results in Figure 1b.2.4. Electrical CharacterizationFigure  3c shows the current–voltage (I–V) curves of the as-deposited, 50 °C NMP-cleaned, and 110 °C NMP-cleaned sam-ples. Multiple devices with the same channel length between two electrodes for each of the three cases were used to check  the consistency and reproducibility of the results. A representative  device is shown in Figure 3e. Note that the I–V curves change substantially from nonlinear to linear after the NMP cleaning, as shown in Figure  3d, indicating that ohmic-like behavior appears only after NMP cleaning. This result was reproducibly measured in different devices, demonstrating high consistency. Furthermore, the total electrical resistance was dramatically  reduced by ≈2 orders of magnitude (from 3910 to 50 MΩ) by the 50  °C NMP cleaning, which is attributed to the improve-ment of contact resistance at the Au/MoS2 interface. Higher contents of adsorbates such as O and C on the as-deposited sample may cause impurity scattering, adversely affecting the electron transport. However, the residual adsorbates after 50 °C NMP cleaning (Figure 2f) enhance the electron transport at the Au/MoS2 interface compared with that at the sharp interface of the 110  °C NMP-cleaned sample (Figure  2g). Several authors have proposed improving contact through surface treatment or surface engineering via interfacial chemical conditioning. Houssa et al. reported a large discrepancy in contact resistance of the MoS2-based metal/semiconductor heterojunctions by considering the interaction of 1T-MoS2 with various chemical species (hydrogen, oxygen). The atoms or molecules of these species lower the work function of the 1T-MoS2 during the local transformation, leading to low contact resistances.[40] Leong et al. reported that the presence of zigzag edges on nickel–gra-phene electrodes enhances electrical contact via a tunneling effect and lower work function.[41] Bhattacharjee et  al. found that applying a sulfur-based (ammonium sulfide) treatment to MoS2 could reduce its contact resistance and variability with high-work-function metals.[42] The similar chemical enhance-ment at interfaces of metal/graphene has been applied as well to increase the performance for sensitive surface detection[43] and surface-enhanced Raman spectroscopy.[44] The charge transport can be tuned by the degree of interaction between the metal and TMD layer via interlayers, doping strategy, or defect engineering.[4,45,46] The interlayer between the material and metal electrode (buffer layer) which plays as a role for car-rier transport has also been used for thin film solar cells,[47] and the bandgap alignment and the carrier concentration at donor or acceptor level are essential for the interfacial conductive characteristics.Neither the broad interfacial region of the as-deposited sample nor the sharp interface of the 110  °C NMP-cleaned sample were optimized interfaces for electrical contacts from our electrical measurement, although the 110 °C NMP cleaning process provided MoS2 with ohmic-like contact (linear I–V characteristics). That is, the Au/50 °C NMP-cleaned MoS2 inter-face exhibited better electrical conductivity than the Au/110 °C NMP-cleaned MoS2 and the Au/as-deposited MoS2, which is consistent with the simulated electrical conductivity results in Figure  1c. The intermediate interfaces of the 45–50  °C NMP-cleaned samples, which exhibit high thermal conductance and low electrical contact resistance, have strong potential for use in high-performance transistors on MoS2 or other TMD materials. On the other hand, the 110 °C NMP-cleaned Au/MoS2 interface, which exhibits low thermal conductance and good electrical contact, can be a good candidate for thermoelectric composites via interface nanoengineering.3. ConclusionWe proposed an effective and simple NMP cleaning method for tuning the thermal and electrical conductance at the Au/MoS2 interface. The thermal conductance can be increased by 3.39 times by increasing the wet cleaning temperature by ≈45–50 °C; on the other hand, it decreases by 74.37% when the cleaning temperature is increased to 110 °C. The total electrical resistance can be reduced by 2 orders of magnitude when the NMP cleaning is performed at 50 °C. We found that the ability of the NMP cleaning procedure to remove undesirable adsorb-ates from the MoS2 surface while preserving the physical prop-erties of the material is temperature dependent. The Au/MoS2 interface changed from a broad and mixed interfacial region to a cleaner and thinner region when the cleaning was conducted at 50 °C and changed gradually to a sharp interface as the NMP cleaning temperature was increased at 110 °C. The intermediate interfacial region at the Au/MoS2 interface (45–50  °C NMP-cleaned) exhibited excellent interfacial thermal conductance and the lowest electrical contact resistance among the investi-gated samples. The experimental results are in good agreement with the simulated results, indicating that adequate oxygen adsorbates at the Au/MoS2 interface enhance the thermal and Adv. Mater. Interfaces 2020, 2000364www.advancedsciencenews.comwww.advmatinterfaces.de2000364  (7 of 9) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheimelectrical conductance. We have shown that the interfacial thermal and electrical conductance of Au/MoS2 can be tuned by changing the interfacial chemical properties via NMP wet cleaning, which is promising for both enhancing the heat dis-sipation of functional ultrascaled electronics and for interface design of thermal insulating thin films.4. Experimental SectionExperimental Methods: The continuous monolayer MoS2 films were grown on sapphire substrates via CVD in a hot-wall furnace by sulfuring MoO3 powders at 750 °C. The sulfur stream was generated by heating sulfur powder at 190 °C under an Ar carrier gas. The sapphire substrate was placed downstream at the center of the furnace. The sample was immersed into NMP solution for 20 min at various temperatures from 25 to 110  °C. All samples were rinsed with isopropyl alcohol (IPA) to remove any residual chemical species from their surface, and they were subsequently dried under blown nitrogen gas. The samples were placed in a vacuum chamber for Au deposition followed by NMP cleaning within 30 min via sputtering (CFS-4EP-LL, 5 × 10−4 Pa). Samples with 5 and 150 nm thick Au layers were used for the Raman spectroscopy and thermal property measurements, respectively. Meanwhile, a metal mask with a lateral size of 200 µm × 2 mm was used to deposit 100 nm Au pads using an electron beam evaporator (EIKO, 5 × 10−5 Pa) for electrical measurement purposes. The detailed flow of the NMP wet cleaning process for MoS2 samples is shown in Figure 4.For the XPS measurements, the MoS2 monolayer was transferred onto a Au substrate from the sapphire substrate to avoid interference from the oxygen signals of the oxide substrate. Raman spectra were collected with a micro-Raman spectrometer (HORIBA-JOBIN-YVON, model T64000) with low-frequency (<200 cm−1) capability; the spectrometer was equipped with a single-mode laser with a wavelength of 514  nm. The thermal resistances of the Au/MoS2 monolayer/sapphire structures were measured under vacuum (<0.02  Pa) at room temperature using the FDTR method. The Au film layer functioned as a transducer heated by a pump laser with a certain angular frequency. The pump laser was a multimode diode laser with a wavelength of 405 nm (360  mW) modulated by a lock-in amplifier (Stanford SR8300). The temperature response at the surface of the Au film was measured by a probe laser with a wavelength of 635 nm (5 mW). Measurements were carried out at three positions on each sample at four frequencies from 1 to 8 kHz for greater accuracy and stability. Schematics of the fabricated structure and the measurement technique are shown in Figure 3b. The temperature at the Au surface, T(0), was obtained using the 1D heat conduction equation[38,48–51](0)21 1042 201 120 02 2001 12 211TqdeCR RCCd CCdiωλλλ λλλ λ= + + + − + −π−  (1)where q is the heat flux, d is the film thickness, ω is the frequency, λ is the thermal conductivity, C is the volumetric heat capacity, and the subscripts 0, 1, and 2 refer to the Au, MoS2 monolayer, and the sapphire substrate, respectively; R01 and R12 are the interfacial thermal resistance of the Au/MoS2 and the MoS2/sapphire, respectively, as shown in Figure  3b. The fourth and fifth terms in Equation (1) represented the thermal resistance of the Au and MoS2 monolayer, respectively. The total thermal resistance could be evaluated by the sum of the second to  fifth terms in Equation (1), which was the intercept of the linear plot of (0)0Tqd  versus 1ω  obtained from the measurement. In this work, the total thermal conductance in various samples was calculated by the reciprocal of the total thermal resistance. The measured thermal resistance with experimental standard deviation is provided in Figure S1 (Supporting Information).Simulation Procedure: All DFT calculations were performed using the Vienna Ab-initio Simulation Package.[52–55] The projector-augmented-wave method[56,57] was used in conjunction with the generalized gradient approximation and Perdew–Burke–Ernzerhof[58] exchange-correlation functional. The Kohn–Sham orbitals were expanded in a plane-wave basis set with a kinetic energy cutoff of 500 eV. The convergence threshold was set to 10−5 eV for the total electronic energy in the self-consistent loop. The atomic positions were relaxed using the quasi-Newton algorithm until the x-, y-, and z-components of the unconstrained atomic force were smaller than 1 × 10−2 eV Å−1.The calculated lattice constants for the bulk Au and MoS2 unit cells were 4.15 and 3.18 Å, respectively, in good agreement with the experimental values.[59,60] In the Au/MoS2 composites, the Au(111) surface was adopted, which was modeled by a six-layer slab within a (2 × 2) lateral supercell, where the lowest three layers were fixed; the MoS2 sheet was also modeled in a (2 × 2) lateral supercell. The lattice mismatch of the Au/MoS2 composites was controlled to be smaller than 3%. For the summation in the Brillouin zone, the Monkhorst–Pack mesh k-point[61] was set to (8 × 8 × 8), (8 × 8 × 1), and (4 × 4 × 1) for the Au(111) surface, MoS2 unit cell, and Au/MoS2 composites, respectively. The composites were separated with a vacuum spacing of 15 Å to ensure no interactions occurred between them. In addition, oxygen adsorption (surface coverage) of 25–100% was carried out on the MoS2 sheet to represent various oxygen concentrations in the experiment. Each oxygen Figure 4.  The workflow of the NMP cleaning process on MoS2 samples. The thickness and pattern of the deposited Au layers were varied depending on the measurement technique for which the sample was prepared (Raman, FDTR, or electrical measurement).Adv. Mater. Interfaces 2020, 2000364www.advancedsciencenews.comwww.advmatinterfaces.de2000364  (8 of 9) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheimatom was adsorbed onto each top site of the sulfur atoms, resulting in the most stable structures, as shown in Figure  1a. The thermoelectric properties were calculated using the BoltzTraP code,[62] which adopted the constant relaxation time and rigid band approximations via the semiclassical Boltzmann transport theory.Supporting InformationSupporting Information is available from the Wiley Online Library or from the author.AcknowledgementsThis work was supported by the “Materials Research by Information Integration” Initiative (MI2I) project of the Support Program for Starting Up Innovation Hub from Japan Science and Technology Agency (JST). The National Center of High-Performance Computing (NCHC) contributed to this project by allowing access to their computer facilities and donating computer time.Conflict of InterestThe authors declare no conflict of interest.Keywordselectrical conductance, interfacial thermal resistances, molybdenum disulfide, monolayers, wet cleaningReceived: February 28, 2020Revised: April 11, 2020Published online: [1]  J.  Pu, Y. Yomogida, K. K.  Liu, L. J.  Li, Y.  Iwasa, T. Takenobu, Nano Lett. 2012, 12, 4013.[2]  K. Wang, L. Li, W. Xue, S. Zhou, Y. Lan, H. Zhang, Z. Sui, Int. J. Elec-trochem. Sci. 2017, 12, 8306.[3]  M. Tan, Y. Hao, X. Ren, J. Electron. Mater. 2014, 43, 3098.[4]  Q.  Cheng, J.  Pang, D.  Sun, J.  Wang, S.  Zhang, F.  Liu, Y.  Chen, R. Yang, N. Liang, X. Lu, Y.  Ji, J. Wang, C. Zhang, Y. Sang, H. Liu, W. Zhou, InfoMat 2020, 2, 656.[5]  K. F. Mak, C. Lee, J. Hone, J. Shan, T. F. Heinz, Phys. Rev. Lett. 2010, 105, 136805.[6]  L.  Yang, K.  Majumdar, H.  Liu, Y.  Du, H.  Wu, M.  Hatzistergos, P. Y. Hung, R. Tieckelmann, W. Tsai, C. Hobbs, P. D. Ye, Nano Lett. 2014, 14, 6275.[7]  R. Kappera, D. Voiry, S. E. Yalcin, B. Branch, G. Gupta, A. D. Mohite, M. Chhowalla, Nat. Mater. 2014, 13, 1128.[8]  K. A. Duerloo, Y. Li, E. J. Reed, Nat. Commun. 2014, 5, 4214.[9]  D. Jena, K. Banerjee, G. H. Xing, Nat. Mater. 2014, 13, 1076.[10]  C. D. English, G. Shine, V. E. Dorgan, K. C. Saraswat, E. Pop, Nano Lett. 2016, 16, 3824.[11]  V. Kaushik, S. H. Wu, H. Jang, J. Kang, K. Kim, J. W. Suk, Nanoma-terials 2018, 8, 587.[12]  X.  Cui, C. Z.  Zhang, R.  Hao, Y. L.  Hou, Nanoscale 2011, 3,  2118.[13]  Y. Xu, H. Cao, Y. Xue, B. Li, W. Cai, Nanomaterials 2018, 8, 942.[14]  P.-C.  Chen, C.-P.  Lin, C.-J.  Hong, C.-H.  Yang, Y.-Y.  Lin, M.-Y.  Li, L.-J. Li, T.-Y. Yu, C.-J. Su, K.-S. Li, Y.-L. Zhong, T.-H. Hou, Y.-W. Lan, Nano Res. 2019, 12, 303.[15]  K.  Hippalgaonkar, Y.  Wang, Y.  Ye, D. Y.  Qiu, H.  Zhu, Y.  Wang, J. Moore, S. G. Louie, X. Zhang, Phys. Rev. B 2017, 95, 115407.[16]  Z. Jin, Q. Liao, H. Fang, Z. Liu, W. Liu, Z. Ding, T. Luo, N. Yang, Sci. Rep. 2015, 5, 18342.[17]  J. X. Xiang, R. N. Ali, Y. F. Yang, Z. Zheng, B. Xiang, X. D. Cui, Phys. E 2019, 109, 248.[18]  S. Kong, T. M. Wu, M. Yuan, Z. W. Huang, Q. L. Meng, Q. K. Jiang, W. Zhuang, P. Jiang, X. H. Bao, J. Mater. Chem. A 2017, 5, 2004.[19]  S. Kong, T. Wu, W. Zhuang, P. Jiang, X. Bao, J. Phys. Chem. B 2018, 122, 713.[20]  D. Wu, L. Xie, X. Chao, Z. Yang, J. He, ACS Appl. Energy Mater. 2019, 2, 1616.[21]  Y.-J. Wu, M. Sasaki, M. Goto, L. Fang, Y. Xu, ACS Appl. Nano Mater. 2018, 1, 3355.[22]  Y.-J. Wu, L. Fang, Y. Xu, npj Comput. Mater. 2019, 5, 56.[23]  T. Zhang, L. Fang, Y. Xu, Sci. Rep. 2017, 7, 7109.[24]  Y.-J. Wu, T. Zhan, Z. Hou, L. Fang, Y. Xu, Sci. Data 2020, 7, 36.[25]  Y.  Zhou, Y.  Huang, J.  Pang, K.  Wang, J. Power Sources 2019, 440, 227149.[26]  Y. Zhou, Y. Wang, K. Wang, L. Kang, F. Peng, L. Wang, J. Pang, Appl. Energy 2020, 260, 114169.[27]  R. Kato, Y. Xu, M. Goto, Jpn. J. Appl. Phys. 2011, 50, 106602.[28]  C.-Y. Lin, K. B. Simbulan, C.-J. Hong, K.-S. Li, Y.-L. Zhong, Y.-K. Su, Y.-W. Lan, Nanoscale Horiz. 2020, 5, 163.[29]  Y.-W. Lan, P.-C. Chen, Y.-Y. Lin, M.-Y. Li, L.-J. Li, Y.-L. Tu, F.-L. Yang, M.-C. Chen, K.-S. Li, Nanoscale Horiz. 2019, 4, 683.[30]  C.-P.  Lin, P.-C.  Chen, J.-H.  Huang, C.-T.  Lin, D.  Wang, W.-T.  Lin, C.-C.  Cheng, C.-J.  Su, Y.-W.  Lan, T.-H.  Hou, ACS Appl. Electron. Mater. 2019, 1, 684.[31]  D.  Dumcenco, D.  Ovchinnikov, K.  Marinov, P.  Lazic, M.  Gibertini, N.  Marzari, O.  Lopez Sanchez, Y. C.  Kung, D.  Krasnozhon, M. W.  Chen, S.  Bertolazzi, P.  Gillet, A.  Fontcuberta i Morral, A. Radenovic, A. Kis, ACS Nano 2015, 9, 4611.[32]  B. M.  Foley, S. C.  Hernandez, J. C.  Duda, J. T.  Robinson, S. G. Walton, P. E. Hopkins, Nano Lett. 2015, 15, 4876.[33]  Y.  Liu, H.  Nan, X.  Wu, W.  Pan, W.  Wang, J.  Bai, W.  Zhao, L.  Sun, X. Wang, Z. Ni, ACS Nano 2013, 7, 4202.[34]  H.  Yuan, G.  Cheng, L.  You, H.  Li, H.  Zhu, W.  Li, J. J.  Kopanski, Y. S.  Obeng, A. R.  Hight Walker, D. J.  Gundlach, C. A.  Richter, D. E. Ioannou, Q. Li, ACS Appl. Mater. Interfaces 2015, 7, 1180.[35]  C. Gong, C. Huang, J. Miller, L. Cheng, Y. Hao, D. Cobden, J. Kim, R. S.  Ruoff, R. M.  Wallace, K.  Cho, X.  Xu, Y. J.  Chabal, ACS Nano 2013, 7, 11350.[36]  E.  Yalon, O. B.  Aslan, K. K. H.  Smithe, C. J.  McClellan, S. V.  Suryavanshi, F.  Xiong, A.  Sood, C. M.  Neumann, X.  Xu, K. E. Goodson, T. F. Heinz, E. Pop, ACS Appl. Mater. Interfaces 2017, 9, 43013.[37]  X. Liu, G. Zhang, Y.-W. Zhang, Nano Res. 2016, 9, 2372.[38]  T. Zhan, S. Minamoto, Y. Xu, Y. Tanaka, Y. Kagawa, AIP Adv. 2015, 5, 047102.[39]  K. R.  Hahn, M.  Puligheddu, L.  Colombo, Phys. Rev. B 2015, 91, 195313.[40]  M.  Houssa, K.  Iordanidou, A.  Dabral, A.  Lu, G.  Pourtois, V. Afanasiev, A. Stesmans, ACS Appl. Nano Mater. 2019, 2, 760.[41]  W. S. Leong, X. Luo, Y. Li, K. H. Khoo, S. Y. Quek, J. T. Thong, ACS Nano 2015, 9, 869.[42]  S. Bhattacharjee, K. L. Ganapathi, D. N. Nath, N. Bhat, IEEE Trans. Electron Devices 2016, 63, 2556.[43]  Y.  Yin, J.  Pang, J.  Wang, X.  Lu, Q.  Hao, E.  Saei Ghareh Naz,  X. Zhou, L. Ma, O. G. Schmidt, ACS Appl. Mater. Interfaces 2019, 11, 15891.Adv. Mater. Interfaces 2020, 2000364www.advancedsciencenews.comwww.advmatinterfaces.de2000364  (9 of 9) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim[44]  Y. Han, H. Wang, L. Qiang, Y. Gao, Q. Li, J. Pang, H. Liu, L. Han, Y. Wu, Y. Zhang, J. Mater. Sci. 2020, 55, 591.[45]  D.  Somvanshi, S.  Kallatt, C.  Venkatesh, S.  Nair, G.  Gupta, J. K.  Anthony, D.  Karmakar, K.  Majumdar, Phys. Rev. B 2017, 96, 205423.[46]  H. Qiao, Z. Huang, X. Ren, S. Liu, Y. Zhang, X. Qi, H. Zhang, Adv. Opt. Mater. 2020, 8, 1900765.[47]  Y.  Cao, X.  Zhu, H.  Chen, X.  Zhang, J.  Zhouc, Z.  Hu, J.  Pang, Sol. Energy Mater. Sol. Cells 2019, 200, 109945.[48]  Y. B. Xu, H. T. Wang, Y. Tanaka, M. Shimono, M. Yamazaki, Mater. Trans. 2007, 48, 148.[49]  T.  Zhan, Y.  Xu, M.  Goto, Y.  Tanaka, R.  Kato, M.  Sasaki, RSC Adv. 2015, 5, 49703.[50]  Y. Xu, R. Kato, M. Goto, J. Appl. Phys. 2010, 108, 104317.[51]  Y.  Xu, M.  Goto, R.  Kato, Y.  Tanaka, Y.  Kagawa, J. Appl. Phys. 2012, 111, 084320.[52]  G. Kresse, J. Hafner, Phys. Rev. B 1993, 47, 558.[53]  G. F. Kresse, J. Furthmüller, Comput. Mater. Sci. 1996, 6, 15.[54]  G. F. Kresse, J. Furthmüller, Phys. Rev. B 1996, 54, 11169.[55]  G. Kresse, J. Hafner, Phys. Rev. B 1994, 49, 14251.[56]  P. E. Blöchl, Phys. Rev. B 1996, 54, 11169.[57]  G. Kresse, D. Joubert, Phys. Rev. B 1999, 59, 1758.[58]  J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77, 3865.[59]  E.  Scalise, M.  Houssa, G.  Pourtois, V.  Afanas’ev, A.  Stesmans, Nano Res. 2012, 5, 43.[60]  E. R. Jette, F. Foote, J. Chem. Phys. 1935, 3, 605.[61]  H. J. Monkhorst, J. D. Pack, Phys. Rev. B 1976, 13, 5188.[62]  G. K. H. Madsen, D. J. Singh, Comput. Phys. Commun. 2006, 175, 67.Adv. Mater. Interfaces 2020, 2000364