Enhanced thermoelectric power in two-dimensional transition metal dichalcogenide monolayers

Transcription

1 Enhanced thermoelectric power in two-dimensional transition metal dichalcogenide monolayers Item Type Article Authors Pu, Jiang; Kanahashi, Kaito; Cuong, Nguyen Thanh; Chen, Chang- Hsiao; Li, Lain-Jong; Okada, Susumu; Ohta, Hiromichi; Takenobu, Taishi Citation Enhanced thermoelectric power in two-dimensional transition metal dichalcogenide monolayers 2016, 94 (1) Physical Review B Eprint version Publisher's Version/PDF DOI /PhysRevB Publisher American Physical Society (APS) Journal Physical Review B Rights Archived with thanks to Physical Review B Download date 08/12/ :32:18 Link to Item

2 Supplemental Material Enhanced Thermoelectric Power in Two-Dimensional Transition Metal Dichalcogenide Monolayers Jiang Pu, 1* Kaito Kanahashi, 1 Nguyen Thanh Cuong, 2 Chang-Hsiao Chen, 3 Lain-Jong Li, 4 Susumu Okada, 5 Hiromichi Ohta, 6 and Taishi Takenobu 1,7* 1 Department of Advanced Science and Engineering, Waseda University, Tokyo , Japan 2 International Center for Young Scientist (ICYS) and International Center for Materials Nanoarchitectonics (MANA), National Institute for Materials Science (NIMS), Ibaraki , Japan 3 Department of Automatic Control Engineering, Feng Chia University, Taichung 40724, Taiwan 4 Physical Science and Engineering Division, King Abdullah University of Science and Technology (KAUST), Thuwal , Kingdom of Saudi Arabia 5 Graduate School of Pure and Applied Science, University of Tsukuba, Ibaraki Japan 6 Research Institute for Electronic Science, Hokkaido University, Sapporo , Japan 7 Kagami Memorial Laboratory for Material Science and Technology, Waseda University, Tokyo , Japan * hokoh.apple@fuji.waseda.jp, takenobu@waseda.jp

3 S1. Spectroscopic characterizations of CVD-grown MoS 2 monolayers. Highly crystalline and uniform MoS 2 monolayer films were grown on sapphire substrates (1 cm 2 ) using the CVD method established by Lee et al. and characterized by spectroscopic methods [1]. In brief, the sapphire substrates were placed in the center of the tube furnace on a quartz board, 0.3 g MoO 3 in a Al 2 O 3 crucible was placed 3 cm away from substrates, and S powder in a quartz tube was placed 8 cm away from the furnace open end at the upstream in a quartz tube. The furnace was first heated to 150 C at 10 C /min with 80 sccm Ar at 10 Torr and then annealed for 20 min. After that, the temperature was increased to 650 C at a rate of 25 C /min and then maintained for 10 min. The S powder was heated via a heating belt at 160 C when the furnace reached 400 C. After the growth, the furnace was slowly cooled to room temperature. Figures S1(a) and S1(b) reveal the Raman and Photoluminescence (PL) spectra, respectively, for two devices, based on which we fabricated EDLTs and performed thermoelectric measurements. We compared the observed Raman peaks with those of the reported mechanically exfoliated monolayer MoS 2. The two characteristic Raman peaks can be assigned to the E 1 2g mode and the A 1g mode. Importantly, the energy difference between the two characteristic peaks is approximately 19.7 cm 1, which is in good agreement with the results obtained for the exfoliated monolayer MoS 2 [2]. In addition, the strong PL shown in Fig. S1(b) is attributed to the direct intraband recombination of the photogenerated excitons [3,4]. These results clearly indicate that the synthesized films for each device are monolayers. Note that the sharp peak at approximately 1.78 ev shown in Fig. S1(b) originates from the Raman scattering of the sapphire substrate. FIG. S1. (a) Raman and (b) PL spectrum of CVD-grown MoS 2 monolayer films.

4 S2. EDLT fabrication and electrical characterization First, for the source and drain electrodes, Au contacts with Ni adhesion layers (80 nm / 2 nm) were thermally deposited onto the WSe 2 and MoS 2 film surfaces. Next, the ion gels, which are a mixture of an ABA-type triblock copolymer, poly(styrene-block-methyl methacrylate-block-styrene) (PS-PMMA-PS; M PS = 4.3 kg/mol, M PMMA = 12.5 kg/mol, M w = 21.1 kg/mol), and an ionic liquid, 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([EMIM][TFSI]) in an ethyl propionate solution, were drop cast onto the film surface and the source/drain electrodes; this was followed by annealing them in an N 2 atmosphere at 70 C for 12 h. The weight ratio of the copolymer, ionic liquid, and solvent was maintained at 0.7:9.3:20. The transistor channel region was then covered with a thin Pt foil (50 μm thick) to form a top-gate electrode. Finally, a thin gold wire was inserted into the gel films between the channel and the top-gate electrode as the quasi-reference-electrode. The transport properties of the fabricated EDLTs were characterized using a semiconductor parameter analyzer (Agilent Technologies, Inc. E5270) in a dark shield probe station inside an N 2 -filled glove box. Impedance measurements were performed to measure the capacitance of the EDLTs using a frequency response analyzer (a Solartron 1252A frequency response analyzer with a Solartron 1296 dielectric interface controlled by ZPlot and ZView software), and the frequency range was set to Hz. In particular, for the DC voltage dependence of capacitance measurement, the frequency was set to 10 Hz with a fixed AC voltage amplitude of 5 mv, and the DC voltage was applied. All electrical measurements were performed at room temperature.

5 S3. The comparison of the conductivity, the Seebeck coefficient, and the power factor in MoS 2 The comparison of the electrical conductivity (σ), the Seebeck coefficient (S), and the power factor between two MoS 2 devices is displayed in Fig. S2. The measured S shows good reproducibility in each device because the S is originated from band structure (Fig. S2(b)). This surely indicates the reliable thermoelectric measurements in our system. However, the absolute value of obtained power factor is different in each device (Fig. S2(c)). We consider that this variation is caused by transport properties, especially for the σ change, depending on each film quality. Because the transport properties are easily affected by adverse effects, such as grain boundaries and/or defects originated from chalcogen atoms, the σ shows variation in each film (Fig. S2(a)). Therefore, although the S is almost consistent, the σ change is the main reason to induce the increase/decrease of the power factor. FIG. S2. The comparison of transport and thermoelectric properties of two MoS 2 EDLTs.

6 S4. Spectroscopic characterizations of CVD-grown WSe 2 monolayers. Highly crystalline and uniform WSe 2 monolayer films were also grown on sapphire substrates (1 cm 2 ) using the CVD system developed by Huang et al. and characterized by spectroscopic methods [5]. As the first step, 0.3 g of WO 3 powders were placed in a ceramic boat located in the heating zone of a 1 tube furnace. Se powders were then placed in a separate ceramic boat on the upstream of the furnace and maintained at 270 C during the reaction. The sapphire substrates for the WSe 2 growth were placed on the downstream, where the Se and WO 3 vapors were carried to the targeting sapphire substrates by an Ar/H 2 flow (Ar = 60 sccm, H 2 = 6 sccm, chamber pressure = 5 Torr). Next, the center heating zone was heated to 925 C at a rate of 25 C /min. The temperature of the sapphire substrates was set at 750 C when the center heating zone reached 925 C. After that, the heating zone was maintained for 15 min, and the furnace was finally allowed to cool to room temperature. Figures S3(a) and S3(b) show the Raman and PL spectra, respectively, for two devices. As shown in Fig. S3(a), two characteristic Raman peaks, i.e., the E ' (or E 1 2g) mode at 250 cm 1 and the A ' 1 (or A 1g ) mode at 260 cm 1, are identified. Other higher energy Raman peaks observed at 360 cm 1 and 375 cm 1, as shown in the inset of Fig. S3(a), are assigned to the 2E 1g and A 1g +LA modes, respectively. It should be noted that the Raman peak around 310 cm 1, which has been predicted to be related to a rigid layer shear mode, did not exist. These observed Raman bands agree with the Raman spectrum of the monolayer WSe 2 produced by mechanical exfoliation [6]. Moreover, the strong PL of Fig. S3(b) originates from the direct bandgap of the monolayer WSe 2 ; these results also indicate that the synthesized films are monolayer films [7]. FIG. S3. (a) Raman and (b) PL spectrum of CVD-grown WSe 2 monolayer films.

7 S5. The comparison of the conductivity, the Seebeck coefficient, and the power factor in WSe 2 To further evaluate the reproducibility of thermoelectric performance, we additionally fabricated three WSe 2 devices to confirm their properties. As shown in the distribution of Fig. S4, the S shows excellent reproducibility in all devices. However, because the σ is depended on each film quality, the obtained maximum power factor is ranging from 200 to 300 μw m 1 K 2, which indicates the reproducible thermoelectric performance in our devices. FIG. S4. The comparison of transport and thermoelectric properties of fabricated WSe 2 EDLTs.

8 S6. The computational details of energy band calculation. All calculations were performed in the framework of DFT [8,9] with OPENMX code [10] using a linear combination of pseudo-atomic orbitals as a basis set. The exchange-correlation potential was treated with a generalized gradient approximation (GGA) [11]. Norm-conserving pseudopotentials [12] with partial core correction [13] were adopted to describe the electron-ion interaction. Pseudo-atomic orbitals (PAOs) were generated by a confinement scheme [14] with a cutoff radius of 7.0 (a.u.) for all atomic species. We used the PAOs specified by Mo-s3p2d2f1, W-s3p2d2f1, S-s3p3d2f1, and Se-s3p3d2f1, where Mo, W, S, and Se are atomic symbols, and s3, for example, indicates the employment of three orbitals for the s component. Real-space grid techniques were used with an energy cutoff of 300 Ry in numerical integration [15]. The spin-orbit coupling was included in all calculations via a relativistic j-dependent pseudopotential scheme in the noncollinear DFT formalism [16]. Here, we used a slab model for monolayer MoS 2 and WSe 2 with a 1 1 unit cell in the lateral direction and a 12-Å vacuum region. The Brillouin-zone integration was performed using a uniform k-point grid for the neutral system, and a uniform k-point grid for the electron/hole doping system. The DOS was calculated using the tetrahedron scheme with a k-point grid [17].

9 S7. Theoretical thermopower of monolayer WSe 2 and MoS 2 for the rigid band model. We calculated the theoretical thermopower, S, of the monolayer WSe 2 and MoS 2 by solving the semi-classic Boltzmann transport equation within the constant scattering time approximation on the basis of the energy band structures obtained by first-principles (DFT) calculations, as implemented in the BoltzTrap code [18]. The motion of an electron is treated semi-classically, in which the group velocity of an electron in a particular band can be derived from band energy by the equation: 1 ε i, υα ( i, k) = h k α k where i,k is the ith energy band at the point k, and k α is the th component of wave-vector k. From the group velocity, the S is calculated by following equations: σ σ S 2 e N ( ε ) = τυ i, k) υ ( i, k) δ ( ε i, k α ( ε ( T, μ) σ ( ε ) β i, k 1 f ( T, ε, μ) = Ω dε ε 1 f ( T, ε, μ) = Ω dε et σ ( T, μ) ε ( T, μ) σ ( ε )( ε μ) ) where α and β are tensor indices, Ω, μ, ƒ, e, and N are the volume of the unit cell, the chemical potential, the Fermi-Dirac distribution, the electron charge and the number of k points sampling, respectively. The effect of doping concentration (or the position of V V VB(CB) ) on the S is estimated based on the rigid band approximation. For comparison, we also calculated the theoretical S from the Mott equation (3) using the quantum capacitance C q calculated by the fixed-band approximation. Figure S5 shows the calculated and experimental S for the Boltzmann transport theory and Mott relation derived from the ESM-DFT and the rigid band models of the (a) p-type WSe 2 and (b) n-type MoS 2. The tendency of the calculated S is comparable with that of the experimental datasets; however, the large difference of the absolute value of S in the energy range near the band edges, especially for the n-type MoS 2, obviously indicates the significance of the charging effect for evaluating the thermoelectric properties in monolayer TMDCs.

10 FIG. S5. Theoretical and experimental thermopower for (a) p-type WSe 2 and (b) n-type MoS 2. Reference (associated with Supplemental Material) [1] Y.-H. Lee, X.-Q. Zhang, W. Zhang, M.-T. Chang, C.-T. Lin, K.-D. Chang, Y.-C. Yu, J. T.-W. Wang, C.-S. Chang, L.-J. Li, T.-W. Lin, Adv. Mater. 24, 2320 (2012). [2] C. Lee, H. Yan, L. E. Brus, T. F. Heinz, J. Hone, S. Ryu, ACS Nano. 4, 2695 (2010). [3] K. F. Mak, C. Lee, J. Hone, J. Shan, T. F. Heinz, Phys. Rev. Lett. 105, (2010). [4] A. Splendiani, L. Sun, Y. Zhang, T. Li, J. Kim, C.-Y. Chim, G. Galli, F. Wang, Nano Lett. 10, 1271 (2010). [5] J.-K. Huang, J. Pu, C.-L. Hsu, M.-H. Chiu, Z.-Y. Juang, Y.-H. Chang, W.-H. Chang, Y. Iwasa, T. Takenobu, L.-J. Li, ACS Nano 8, 923 (2014). [6] W. Zhao, Z. Ghorannevis, K. K. Amara, J. R. Pang, M. Toh, X. Zhang, C. Kloc, P. H. Tan, G. Eda, Nanoscale 5, 9677 (2013). [7] W. Zhao, Z. Ghorannevis, L. Chu, M. Toh, C. Kloc, P.-H. Tan, G. Eda, ACS Nano 7, 791 (2013). [8] W. Kohn, L. J. Sham, Phys. Rev. 140, A1133 (1965). [9] P. Hohenberg, W. Kohn, Phys. Rev. 136, B864 (1964). [10] T. Ozaki, Phys. Rev. B. 67, (2003). [11] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77, 3865 (1996). [12] I. Morrison, D. M. Bylander, L. Kleinman, Phys. Rev. B. 47, 6728 (1993). [13] S. G. Louie, S. Froyen, M. L. Cohen, Phys. Rev. B. 26, 1738 (1982). [14] T. Ozaki, H. Kino, Phys. Rev. B. 69, (2004). [15] T. Ozaki, H. Kino, Phys. Rev. B. 72, (2005). [16] G. Theurich, N. A. Hill, Phys. Rev. B. 64, (2011). [17] P. E. Blochl, O. Jepsen, O. K. Andersen, Phys. Rev. B. 49, (1994).

11 [18] G. K. H. Madsen, D. J. Singh, Comput. Phys. Commun. 175, 67 (2006).