Quintuple-layer epitaxy of thin films of topological insulator Bi2Se3

APPLIED PHYSICS LETTERS 95, 053114 共2009兲

Quintuple-layer epitaxy of thin films of topological insulator Bi2Se3 Guanhua Zhang, Huajun Qin, Jing Teng, Jiandong Guo, Qinlin Guo, Xi Dai, Zhong Fang, and Kehui Wua兲 Institute of Physics, Chinese Academy of Sciences, Beijing 100080, People’s Republic of China

共Received 29 June 2009; accepted 17 July 2009; published online 6 August 2009兲 Atomically smooth, single crystalline Bi2Se3 thin films were prepared on Si共111兲 by molecular beam epitaxy. Scanning tunneling microscopy, low-energy electron diffraction, x-ray photoelectron emission spectroscopy, and Raman spectroscopy were used to characterize the stoichiometry and crystallinity of the film. The films grow in a self-organized quintuple layer by quintuple-layer mode, and atomically smooth films can be obtained, with controllable thickness down to one quintuple layer 共⬃1 nm兲. © 2009 American Institute of Physics. 关DOI: 10.1063/1.3200237兴 Bi2Se3 is a narrow gap semiconductor with rhombohe5 ¯ m兲 space group. 共R3 dral crystal structure belonging to the D3d Traditionally, it is an important material for thermoelectric applications.1,2 Very recently, it has regenerated great interest by being predicted to be a three-dimensional topological insulator 共TI兲, a new state of quantum matter.3–10 The calculation by Zhang et al.5 indicated a large bulk energy gap and a topological surface state described by a single Dirac cone at the ⌫ point. Independent angle-resolved photoelectron emission spectroscopy 共ARPES兲 study on Bi2Se3 indeed revealed such a surface Dirac cone.6 These results strongly suggested that Bi2Se3 is a fascinating TI: the simple surface state is ideal for realizing magnetoelectric effects,7 and the large bulk gap may favor high temperature spintronics applications. To date, all ARPES measurements on Bi2Q3 共Q = Se, Te兲 were performed on cleaved single crystal surfaces, while for transport measurements or device applications, thin films are desired. In literature, various methods to prepare Bi2Q3 thin films have been reported, such as electrodeposition,11,12 chemical bath deposition,13 solvo thermalization,14 successive ionic layer adsorption and reaction,15 thermal evaporation,16 reactive evaporation,17 metal-organic chemical vapor deposition,18 and compound evaporation,19,20 etc. However, these methods normally produce polycrystalline films aiming for thermoelectric applications, which are far from the current TI studies. So, growing high quality, single crystalline Bi2Se3 thin films is today an urgent task for the TI society. In this letter, we report on the growth of atomically smooth, high-quality, single crystalline Bi2Se3 films on Si共111兲 by molecular beam epitaxy 共MBE兲 under ultrahigh vacuum 共UHV兲 condition. We found an interesting selforganization phenomenon, in that the film thickness is always an integer number of quintuple layers along the c axis. These quintuple atomic layers are the fundamental building blocks of the bulk crystallographic unit cell, which contains three such layers stacked along c.21 Atomically smooth films can be obtained even with only one quintuple layer. The films were characterized by scanning tunneling microscopy 共STM兲, low-energy electron diffraction 共LEED兲, x-ray phoa兲

Author to whom correspondence should be addressed. Electronic mail: [email protected].

0003-6951/2009/95共5兲/053114/3/$25.00

toelectron emission spectroscopy 共XPS兲, and Raman spectroscopy. Experiments were performed in an UHV system 共base pressure 5 ⫻ 10−11mbar兲 equipped with STM, LEED, and XPS. The Si共111兲 substrates were cut from a phosphorusdoped 共n-type兲 wafer with a resistivity of 2 ⍀ cm. Clean Si共111兲-7 ⫻ 7 surfaces were obtained by repeatedly flashing the sample to about 1400 K. High purity Bi 共99.997%兲 and Se共99.999%兲 were evaporated from Knudsen cells and the fluxes were calibrated by a quartz microbalance flux monitor. The typical growth rate of Bi2Se3 films was about 0.5 nm/ min in our experiments. At first, we used the Si共111兲-7 ⫻ 7 surface as substrate. Bi and Se were codeposited to the 7 ⫻ 7 surface at room temperature, followed by annealing to about 390 K for 5 min. A typical STM image of the as-prepared films is shown in Fig. 1共a兲. The film appears smooth and continuous in larger areas, and ordered atomic structure can be observed within a single crystalline domain 共not shown here兲. However, the LEED pattern 关Fig. 1共c兲兴 shows strong background with circularly elongated faint spots, indicating the coexistence of islands, or crystallites, that are rotated randomly around the c axis, similar to the case of Bi2Te3 crystallites on sapphire共0001兲.22 The LEED result is consistent with the STM observation of many grain boundaries and additional islands on the film surface 关Fig. 1共a兲兴.

FIG. 1. 共Color online兲 STM images of Bi2Se3 films 共a兲 directly deposited on the Si-7 ⫻ 7 and 共b兲 deposited on the ␤- 冑 3-Bi substrate. 共c兲 and 共d兲 are LEED patterns corresponding to 共a兲 and 共b兲, respectively.

95, 053114-1

© 2009 American Institute of Physics

Downloaded 12 Oct 2009 to 159.226.37.69. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp

053114-2

Zhang et al.

Appl. Phys. Lett. 95, 053114 共2009兲

FIG. 3. 共Color online兲 STM images of Bi2Se3 films, with thicknesses from one quintuple layer to 15 quintuple layers. The films are atomically smooth in large area. The steps in the images correspond to the original Si steps. Some additional small islands and holes may exist on the surface due to a slight excess or insufficient amount of Bi2Se3 deposited. These islands and holes all exhibit an identical height 共depth兲 of 9.5 Å.

FIG. 2. 共Color online兲共a兲 The STM image of a Bi2Se3 film with coverage slightly higher than four quintuple layers. The inset is an atomically resolved image. 共b兲 The height profile along AA⬘ marked in 共a兲, showing that the islands are of 9.5 Å in height. Note that 3.1 Å corresponds to the height of Si共111兲 steps 共c兲 The crystal structure of Bi2Se3, the dashed panel shows a unit of Bi–Se quintuple layers.

To further improve the film crystallinity, we tried to modify the interface to favor a better epitaxial growth. First, the 7 ⫻ 7 reconstruction is removed by depositing 1 ML Bi 关here 1 ML refers to the atomic density of the Si共111兲 plane兴 and annealing to 750 K to form a ␤- 冑 3-Bi surface.23 In this surface, the Si共111兲 substrate recovers to a bulk-terminated 1 ⫻ 1 structure underneath the 1ML Bi adatoms. Such an interface is atomically much sharper compared with the Si共111兲-7 ⫻ 7. Bi and Se, with a coverage ratio of 2:3, were codeposited to the ␤- 冑 3-Bi surface at room temperature, followed by annealing to about 390 K for 5 min. The thusprepared film is atomically flat, with a very sharp LEED pattern, as shown in Figs. 1共b兲 and 1共d兲. From the LEED pattern and the atomically resolved STM image 关inset in Fig. 2共a兲兴, the lattice constant a of the films was determined to be 4.13⫾ 0.10 Å, quite consistent with that of a bulk Bi2Se3 structure. More interestingly, in STM images one often sees triangular islands on the terraces, as illustrated in Fig. 2共a兲. The triangular shape reflects the hexagonal structure of bulk Bi2Se3. It is important that all these islands exhibit identical height of 9.5 Å, as indicated by the height profile in Fig. 2共c兲. In addition, on films with holes 共not shown here兲, the depth of all holes is also 9.5 Å. It is known that the structure of Bi2Se3 consists of a periodic stacking of Bi–Se quintuple layers along the c axis 关Fig. 2共b兲兴, and the unit cell spans three quintuple layers with the lattice constants a ⬃ 4.14 Å and c ⬃ 28.64 Å.24,25 The observed island height and hole depth coincide with 1/3 c, i.e.,

the thickness of a quintuple layer. We thus conclude that the film grows by a stacking of Bi–Se quintuple layers. This is a very important self-organization behavior that makes it easy to achieve a stoichiometric film, even with not well controlled Bi and Se coverage ratio. In fact, the film can be grown nicely with a little bit excess of the Se coverage. The excess Se atoms are spontaneously desorbed during annealing, as has been proven by our XPS measurements before and after annealing. Furthermore, by precisely controlling the Bi/Se coverage, we have obtained atomically smooth films, with tunable thickness quintuple layer by quintuple layer. The smallest thickness of the films reaches to one quintuple layer, namely ⬃1 nm, as shown in Fig. 3. To prove the chemical stoichiometry of our films, we measured in situ the XPS spectra of the films. Figure 4 shows the spectra recorded for a 4 nm Bi2Se3 film. In Fig. 4共a兲 the two peaks correspond to the Se 3d5/2 and 3d3/2, with binding energies of 53.5 and 54.1 eV, respectively, and in Fig. 4共b兲 the Bi 4f 5/2 and 4f 7/2 peaks at 158.1 and 163.4 eV are shown. The binding energies of the Se 3d peaks show a redshift of about 2.2eV with respect to those of pure bulk Se,26 while the binding energies of Bi 4f peaks show a blueshift of about 1.1eV. The opposite shifts are caused by the Se–Bi bond and the charge transfer from Bi to Se. In addition, from the XPS spectra one can also estimate the surface composition by ␳Se / ␳Bi = 共ISe / SSe兲 / 共IBi / SBi兲, where ␳ is the atomic density, I the integrated peak intensity, and S the atomic sensitivity factor. Considering that the film is very thin, surface atomic sensitivity factors were used here: SSe = 0.44 and SBi = 2.9.27 We got ␳Se / ␳Bi ⬃ 1.5, indicating that the films are normally stoichiometric. Note that small deviations from stoichiometry that impact the carrier type and concentration are not observable by this method. Figure 5 shows a typical Raman spectrum taken from a 40 nm film. There are two characteristic peaks, at 131.5 and 171.5 cm−1, which are, respectively, consistent with the two 1 , reported for the Bi2Se3 single vibrational modes, E1g and Ag1 28 2 mode, whose frequency is crystal. Note that the Ag1 97 cm−1 is out of the measurable range 共⬎100 cm−1兲 of our Raman spectroscopy.


053114-3

Appl. Phys. Lett. 95, 053114 共2009兲

Zhang et al.

work is helpful to further investigations on the physical properties, such as the two-dimensional transport properties and thickness dependent behaviors, of TI. We thank W.R. Fang for the help in Raman measurements. This work was supported by the National Natural Science Foundation 共Grant No. 10874210兲 and MOST 共Grant No. 2007CB936800兲 of China. 1

FIG. 4. 共Color online兲 XPS spectra of a 4 nm Bi2Se3 film: 共a兲 the Se 3d peaks and 共b兲 the Bi 4f peaks.

In conclusion, we obtained Bi2Se3 single crystalline films on Si共111兲 by MBE. The key to improve the crystal quality is to remove the 7 ⫻ 7 reconstruction by using the ␤- 冑 3-Bi surface as the substrate. Several techniques have been used to prove that the as-grown films are stoichiometric and single crystalline. The film thickness is controllable in a wide range 共⬎1 nm兲, with a quintuple layer as a unit. This

FIG. 5. 共Color online兲 Raman spectrum of a 40 nm film. The two characteristic peaks are the E1g and A1g1 modes of Bi2Se3 single crystal. The inset shows these two vibrational modes.

D. M. Rowe, in CRC Handbook of Thermoelectrics, edited by D. M. Rowe 共CRC, Boca Raton, 1995兲, p. 90. 2 H. J. Goldsmid, Thermoelectric Refrigeration 共Pion Ltd., London, 1986兲. 3 L. Fu and C. L. Kane, Phys. Rev. B 76, 045302 共2007兲. 4 Y. L. Chen, J. G. Analytis, J. H. Chu, Z. K. Liu, S. K. Mo, X. L. Qi, H. J. Zhang, D. H. Lu, X. Dai, Z. Fang, S. C. Zhang, I. R. Fisher, Z. Hussain, and Z. X. Shen, Science 325, 178 共2009兲. 5 H. J. Zhang, C. X. Liu, X. L. Qi, X. Dai, Z. Fang, and S. C. Zhang, Nat. Phys. 5, 438 共2009兲. 6 Y. Xia, D. Qian, D. Hsieh, L. Wray, A. Pal, H. Lin, A. Bansil, D. Grauer, Y. S. Hor, R. J. Cava, and M. Z. Hasan, Nat. Phys. 5, 398 共2009兲. 7 Q. Liu, C.-X. Liu, C. Xu, X.-L. Q, and S.-C. Zhang, Phys. Rev. Lett. 102, 156603 共2009兲. 8 Y. S. Hor, A. Richardella, P. Roushan, Y. Xia, J. G. Checkelsky, A. Yazdani, M. Z. Hasan, N. P. Ong, and R. J. Cava, Phys. Rev. B 79, 195208 共2009兲. 9 J. Moore, Nat. Phys. 5, 378 共2009兲. 10 L. Fu and C. L. Kane, Phys. Rev. Lett. 100, 096407 共2008兲. 11 A. P. Torane, C. D. Lokhande, P. S. Patil, and C. H. Bhosale, Mater. Chem. Phys. 55, 51 共1998兲. 12 A. P. Torane and C. H. Bhosale, Mater. Res. Bull. 36, 1915 共2001兲. 13 R. K. Nkum, A. A. Adimado, and H. Totoe, Mater. Sci. Eng., B 55, 102 共1998兲. 14 W. Z. Wang, Y. Geng, Y. Qian, Y. Xie, and X. M. Liu, Mater. Res. Bull. 34, 131 共1999兲. 15 B. R. Sankapal, R. S. Mane, and C. D. Lokhande, Mater. Chem. Phys. 63, 230 共2000兲. 16 D. Nataraj, K. Prabakar, S. K. Narayandass, and D. Mangalaraj, Cryst. Res. Technol. 35, 1087 共2000兲. 17 K. J. John, B. Pradeep, and E. Mathai, Solid State Commun. 85, 879 共1993兲. 18 J. Waters, D. Crouch, P. O’Brien, and J. H. Park, J. Mater. Sci.: Mater. Electron. 14, 599 共2003兲. 19 S. Augustine, S. Ampili, J. K. Kang, and E. Mathai, Mater. Res. Bull. 40, 1314 共2005兲. 20 S. Augustine, J. Ravi, S. Ampili, T. M. A. Rasheed, K. P. R. Nair, T. Endo, and E. Mathai, J. Phys. D 36, 994 共2003兲. 21 S. Nakajima, J. Phys. Chem. Solids 24, 479 共1963兲. 22 Y. Iwata, H. Kobayashi, S. Kikuchi, E. Hatta, and K. Mukasa, J. Cryst. Growth 203, 125 共1999兲. 23 K. J. Wan, T. Guo, W. K. Ford, and J. C. Hermanson, Phys. Rev. B 44, 3471 共1991兲. 24 H. Lind, S. Lidin, and U. Häussermann, Phys. Rev. B 72, 184101 共2005兲. 25 H. Okamoto, J. Phase Equilib. 15, 195 共1994兲. 26 J. F. Moudler, W. F. Stickle, P. E. Sobol, and K. D. Bomben, Handbook of X-Ray Photoelectron Spectroscopy 共Perkin-Elmer, Eden Prairie, 1992兲. 27 C. D. Wagner, J. Electron Spectrosc. Relat. Phenom. 32, 99 共1983兲. 28 W. Richter, H. Köhler, and C. R. Becker, Phys. Status Solidi B 84, 619 共1977兲.
