⇦Chinese Physics Letters, 2018, Vol. 35, No. 2, Article code 026802 Surface Structure and Reconstructions of HgTe (111) Surfaces * Xin-Yi Yang(杨心怡)1, Guan-Yong Wang(王观勇)1, Chen-Xiao Zhao(赵晨晓)1, Zhen Zhu(朱朕)1, Lu Dong(董璐)1, Ai-Min Li(李爱民)1, Yang-Yang Lv(吕洋洋)2, Shu-Hua Yao(姚淑华)2, Yan-Bin Chen(陈延彬)3, Dan-Dan Guan(管丹丹)1,5, Yao-Yi Li(李耀义)1,5, Hao Zheng(郑浩)1,5, Dong Qian(钱冬)1,5, Canhua Liu(刘灿华)1,5, Yu-Lin Chen(陈宇林)4, Jin-Feng Jia(贾金锋)1,5** Affiliations 1Key Laboratory of Artificial Structures and Quantum Control (Ministry of Education), Department of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai 200240 2National Laboratory of Solid State Microstructures & Department of Materials Science and Engineering, Nanjing University, Nanjing 210093 3National Laboratory of Solid State Microstructures & Department of Physics, Nanjing University, Nanjing 210093 4Clarendon Laboratory, Department of Physics, University of Oxford, Oxford OX1 3PU, UK 5Collaborative Innovation Center of Advanced Microstructures, Nanjing 210093 Received 23 October 2017 *Supported by the National Key Research and Development Program of China under Grant Nos 2016YFA0301003 and 2016YFA0300403, and the National Natural Science Foundation of China under Grant Nos 11521404, 11634009, U1632102, 11504230, 11674222, 11574202, 11674226, 11574201 and U1632272.
**Corresponding author. Email: jfjia@sjtu.edu.cn Citation Text: Yang X Y, Wang G Y, Zhao C X, Zhu Z and Dong L et al 2018 Chin. Phys. Lett. 35 026802 Abstract HgTe (111) surface is comprehensively studied by scanning tunneling microscopy/spectroscopy (STS). In addition to the primitive $(1\times 1)$ hexagonal lattice, six reconstructed surface structures are observed: $(2\times 2)$, $2\times 1$, $4\times 1$, $3\times \sqrt{3}$, $2\sqrt{2}\times 2$ and $\sqrt{11}\times 2$. The $(2\times 2)$ reconstructed lattice maintains the primitive hexagonal symmetry, while the lattices of the other five reconstructions are rectangular. Moreover, the topographic features of the $3\times \sqrt{3}$ reconstruction are bias dependent, indicating that they have both topographic and electronic origins. The STSs obtained at different reconstructed surfaces show a universal dip feature with size $\sim $100 mV, which may be attributed to the surface distortion. Our results reveal the atomic structure and complex reconstructions of the cleaved HgTe (111) surfaces, which paves the way to understand the rich properties of HgTe crystal. DOI:10.1088/0256-307X/35/2/026802 PACS:68.35.B-, 68.35.bg, 68.37.Ef, 68.47.Fg © 2018 Chinese Physics Society Article Text Mercury telluride (HgTe) is one of the II–VI group semiconductors and it has drawn great attention for its intriguing properties and tremendous application potentials, especially for those related to the topological electronic structures.[1-10] For example, mercury telluride-cadmium telluride (HgTe-CdTe) quantum well was predicted and experimentally proved to be the first quantum spin-Hall system, or 2D topological insulator (TI), which possesses insulating bulk states and dissipationless conducting edge states.[11-19] Later, bulk HgTe was predicted to be a strong TI or Weyl semimetal depending on the applied uniaxial stress.[20-23] Moreover, HgTe-based compounds Hg$_{1-x}$Cd$_{x}$Te and HgTe/CdTe superlattice are widely studied for acquiring high-performance infrared emission and detection devices,[24,25] optical isolator and optical modulator.[26,27] However, so far the investigations of HgTe surface structure by scanning tunneling microscopy (STM) are rarely reported,[9,28-30] hindered by the difficulty to obtain a clean and atomically flat surface. Surface reconstruction refers to the rearrangement of surface atoms caused by imbalanced force acting on atoms, atomic defects, external strain or adsorbed foreign atoms. Customarily, the reconstructed surface lattices possessing the same symmetry with primitive lattice are denoted by parentheses, e.g., $(n\times m)$, and those having different symmetries are denoted with no parentheses, indicating the change of basis vectors in both direction and magnitude, e.g., $n\times m$.[31-35] In addition, the rearranged surface structure will inevitably affect the distribution of electronic states on the surface and will modify the band structure of the surface states,[21,36] which in turn lead to modification of properties of the materials. Therefore, the study of surface structure of HgTe is very important for further understanding the material and applications. STM is one of the most powerful and popular technics used in surface science, especially in the determination of surface structure, reconstruction and electronic structure of surface states. The direct observation in real space with sub-atomic resolutions by STM provides detailed information about the location of every atom on the surface, which will be of great help to investigate the possible origin of rearrangement of surface atoms. In this work, the HgTe (111) surface is studied by STM. A well-defined (111) surface of HgTe with perfect $(1\times 1)$ lattice is clearly observed. Moreover, six types of reconstructions of HgTe (111) surfaces, $(2\times 2)$, $2\times 1$, $4\times 1$, $3\times \sqrt{3}$, $2\sqrt{2}\times 2$ and $\sqrt{11}\times 2$, are acquired in different areas. Among these reconstructions, only $(2\times 2)$ reconstruction keeps the same hexagonal symmetry as the unconstructed primitive $(1\times 1)$ lattice. A simple structure model is proposed to elucidate its formation mechanism. The $3\times \sqrt{3}$ reconstruction has a rectangular symmetry and shows distinct topography at different scanning biases, which is attributed to the inhomogeneous distribution of electronic states at reconstructed surface. Other reconstructions are also systematically studied. Finally, we analyze the universal dip feature shown in all STSs, and regard it as the result of surface distortion. The samples were grown by the vapor transport method. Polycrystalline HgTe and high purity iodine (5 mg$\cdot$cm$^{-3}$) were ground and loaded into a quartz ampoule sealed under vacuum pressure $\sim$4$\times$10$^{-6}$ Torr. Subsequently, the quartz ampoule was placed in a double zone furnace with a temperature profile of 400–500$^{\circ}\!$C. Millimeter-sized polyhedral [111] oriented single crystals with metallic luster were successfully obtained after growth of a period over 10 d. All experiments were performed with a commercial ultrahigh vacuum (UHV) MBE-STM combined system (Unisoku) with a base pressure of $1\times10^{-10}$ Torr. The (111) surface of HgTe was obtained by cleavage in UHV at room temperature. Then the cleaved samples were transferred to a cooling stage kept at 77 K for in situ STM measurement. All STM topographic images were taken at a tunneling current of 0.1 nA with tungsten tip.
cpl-35-2-026802-fig1.png
Fig. 1. Crystal structures of HgTe and topographical images of HgTe (111) surface. (a) Crystal structure of HgTe. (b) Top view of HgTe (111) surface. (c) Side view of HgTe (111) surface. (d) Typical topography of cleaved HgTe (111) surface, $V_{\rm s}=2.0$ V. (e) Atomic resolved image of HgTe (111) surface, $V_{\rm s}=1.63$ V, and the inset is the FFT image. (f) Line profile along the black dashed line in (d).
HgTe has a zinc-blende structure with lattice constant of 0.646 nm, as depicted in Fig. 1(a). Each Hg (Te) atom bonds with 4 Te (Hg) atoms and all bonds are energetically identical. Apparently, HgTe is hard to cleave since it is not a van der Waals layer material. Figure 1(b) shows the top view of (111)-orientated surface of HgTe, which shows a honeycomb structure consisting of two sub-lattices. Thus bulk HgTe can be seen as stacking of such honeycomb layers without considering Hg-Te bonds along [111] direction, which are easy to break in the cleavage process (see Fig. 1(c)). The thickness of one honeycomb layer is 370 pm, as marked in Fig. 1(c). A typical morphology of cleaved HgTe (111) surface is shown in Fig. 1(d), where flat terraces and sharp steps can be observed clearly. Figure 1(e) exhibits a clearly atomic resolved image obtained on a terrace. The lattice constants, deduced from fast Fourier transform (FFT) of the lattice (shown in the inset), is 0.46 nm, which is well matched with the theoretical value of 0.456 nm (see Fig. 1(b)). Moreover, only steps with heights of 370 pm and 740 pm have been found in the line profile along the black dashed line in Fig. 1(d), corresponding to one and two honeycomb layers, respectively. The STM results confirmed that our cleaved surface is HgTe (111) surface indeed.
cpl-35-2-026802-fig2.png
Fig. 2. Several reconstructions observed on HgTe (111) surfaces: (a) $({2\times 2})$ reconstruction, $V_{\rm s}=2.0$ V. (b) $2\times 1$ reconstruction, $V_{\rm s}=1.32$ V. White circles represent the primitive $(1\times 1)$ lattice and blue circles represent the lattice of reconstruction. (c) $4\times 1$ reconstruction, $V_{\rm s}=1.22$ V. White circles represent the primitive $(1\times 1)$ lattice and blue circles represent the lattice of reconstruction. (d) $3\times \sqrt{3}$ reconstruction, $V_{\rm s} =1.96$ V. (e) $2\sqrt{2}\times 2$ reconstruction, $V_{\rm s}=1.0$ V. (f) $\sqrt{11}\times 2$ reconstruction, $V_{\rm s}=1.74$ V. The unit cells of different reconstructions are marked out by white dashed hexagon and rectangles, respectively.
Further studies on HgTe (111) surface show complex reconstructions in addition to the primitive $(1\times 1)$ lattice structure. The $(2\times 2)$, $2\times 1$, $4\times 1$, $3\times \sqrt{3}$, $2\sqrt{2}\times 2$ and $\sqrt{11}\times 2$ reconstructions were observed by STM (as shown in Figs. 2(a)–2(f)). Only the $(2\times 2)$ reconstruction has the same hexagonal symmetry as the primitive lattice (see Fig. 2(a)). Other reconstructed lattices show rectangular symmetry. The $2\times 1$ reconstruction shows a strip characteristic along $b'$ direction and the strip contains two columns of atoms (shown in Fig. 2(b)). The rearranged atoms are marked out by blue circles. The atomic distance within the strip is 0.46 nm, which is the same as the primitive lattice (white circles), while the atomic distance between two strips is enlarged. The same method is used to analyze $4\times 1$ reconstruction, which shows alternating bright and dark strips, but with different widths. The bright strip contains three atom columns, while the dark strip contains only one column. The atomic distances within and between strips are the same, but larger than 0.46 nm. The $3\times \sqrt{3}$ reconstruction is formed by highlighted atoms, and this type of reconstruction shows a bias-dependent character which will be discussed in the following. More complex reconstructions, e.g., $2\sqrt{2}\times 2$ and $\sqrt{11}\times 2$, are also found on HgTe (111) surface. Strictly speaking, the unit cell of $\sqrt{11}\times 2$ reconstruction is not a rectangular, but a parallelogram, due to a slight angular deviation from 90$^{\circ}$ between $a'$ and $b'$. There are mainly two reasons for the reconstructions observed here: (1) the large stress induced by the mechanical cleavage; (2) desorption of Hg atoms due to its high vapor pressure.
cpl-35-2-026802-fig3.png
Fig. 3. The $({2\times 2})$ reconstruction. (a) Topographical image of $({2\times 2})$ reconstruction, $V_{\rm s} =2.0$ V, and the inset is atomically resolved STM image of $(1\times 1)$ surface of HgTe (111). (b) A possible model of $({2\times 2})$ reconstruction.
We propose a possible process to elucidate the formation of $(2\times 2)$ reconstruction as shown in Fig. 3(b). The top-most Hg atoms are represented by large grey balls and the lower Te atoms are represented by small orange balls. Considering the relatively high vapor pressure of Hg atoms and comparing the reconstructed $(2\times 2)$ lattice with the primitive lattice, we find that the missing of Hg atoms marked by dashed circles would lead to the doubling of basis vectors and would give a $(2\times 2)$ lattice. When losing the dashed Hg atoms, the primitive unit cell (grey solid parallelogram) turns into the grey solid parallelogram, leading to the double periodicity. The similar phenomenon was also observed on HgTe (001) surface.[29]
cpl-35-2-026802-fig4.png
Fig. 4. Bias-dependent topographical images of $3\times \sqrt 3$ reconstruction. (a) Atomic resolved image obtained at 1.96 V, and the white circles shows the primitive $(1\times 1)$ lattice. (b) Atomic resolved image obtained at 1.86 V, the black circles show the position of highlighted atoms in (a), A and B represent two different trimers respectively. (c) and (d) The FFT images of (a) and (b), and the primitive and reconstructed unit cells are represented by large and small dashed white circles, respectively.
Figures 4(a) and 4(b) show the bias-dependent STM images of $3\times \sqrt{3}$ reconstruction. The two different images are acquired under 1.96 V and 1.86 V at the same area. The periodicity of the unit cell is unchanged, which is marked by white rectangle in the figures. The FFTs shown in Figs. 4(c) and (d) also confirm the same unit cell. The atomic arrangement shown in Fig. 4(a) is quite similar to the primitive $(1\times 1)$ lattice, but has a highlighted atom per unit cell. This highlighted atom gives rise to the reconstructed $3\times \sqrt{3}$ periodicity. The atomic arrangement shown in Fig. 4(b) shows a quite different trimer character. Two types of trimers can be observed in the image: the brighter one (A) and the darker one (B). Rows formed by two kinds of trimers are alternately arranged. To obtain a clearer comparison, the position of highlighted atoms shown in Fig. 4(a) is marked out by black circles in Fig. 4(b). We find that they correspond to the top left atoms of type A trimer. This bias-dependent feature is attributed to the variation of electronic state detected by STM tip at different biases and reflecting the inhomogeneous distribution electronic states at the reconstructed $3\times \sqrt{3}$ HgTe (111) surface. Typical $dI/dV$ curve taken by STM on HgTe $(1\times 1)$ surface is shown in Fig. 5(a). The density of states decreases rapidly in the vicinity of the Fermi level and shows a dip feature with size $\sim$100 mV. Above the Fermi level, it shows a shoulder located at $\sim$50 mV. A similar dip feature is also observed at the HgTe (110) surface, which is attributed to the surface relaxation.[9] Thus it is reasonable for us to ascribe the dip feature to surface distortion. The dip is also observed on $2\times 1$, $4\times 1$, $3\times \sqrt{3}$, $2\sqrt{2}\times 2$ reconstructed surfaces (see Figs. 5(b)–5(e)). It should be noted that the $dI/dV$ curves are inhomogeneous at the same reconstructed surface, which is probably caused by the inhomogeneity of distortion and stress at different positions. The nonzero LDOS inside the dip may be caused by the topological surface states.[9]
cpl-35-2-026802-fig5.png
Fig. 5. The $dI/dV$ (STS) curves acquired on HgTe (111) surface under 77 K ($f=991$ Hz, $V_{\rm s}=0.1$ V, and $I=100$ pA). (a) The $dI/dV$ curve measured on HgTe $(1\times 1)$ surfaces. (b)–(e) The $dI/dV$ spectra measured on HgTe (111) $2\times 1$, $4\times 1$, $3\times \sqrt 3$, $2\sqrt{2}\times 2$ reconstruction surfaces, respectively.
In summary, atomic-resolved structures of HgTe (111) surface have been observed for the first time by STM. In addition to the $(1\times 1)$ surface, six typical reconstructed surfaces, $(2\times 2)$, $2\times 1$, $4\times 1$, $3\times \sqrt{3}$, $2\sqrt{2}\times 2$ and $\sqrt{11}\times 2$, are also observed and studied in detail by STM. The stresses induced by cleavage and desorption of HgTe atoms are considered as the main reasons for the complex reconstructions. We propose a possible explanation of the formation of $(2\times 2)$ reconstruction based on the fact that it is easy for Hg atoms at HgTe surface to escape. Moreover, a bias-dependent $3\times \sqrt{3}$ reconstruction is observed and is attributed to the change of electronic states. Further studies are needed to reveal the origins of various reconstructions. Finally, the STSs show a universal dip feature that is caused by surface distortion existing on both primitive $(1\times 1)$ and reconstruction surfaces. Our results are helpful to understand the complex surface structures of HgTe and to investigate other II–VI group semiconductors.[29,30,37,38] Moreover, the reconstructions may provide a way to tune band structures, including the surface states, which would be beneficial to detect the topological properties of HgTe and to improve the performance of HgTe-based infrared devices. References Electron-Elastic-Wave Interaction in a Two-Dimensional Topological InsulatorTopological insulators: The next generationColloquium : Topological insulatorsThe birth of topological insulatorsQuantum Hall Effect from the Topological Surface States of Strained Bulk HgTeTopological surface states of strained Mercury-Telluride probed by ARPESElectronic structure of the quantum spin Hall parent compound CdTe and related topological issuesTunable spin helical Dirac quasiparticles on the surface of three-dimensional HgTeStrained HgTe plates grown on SrTiO 3 investigated by micro-Raman mappingQuantum Spin Hall Effect in GrapheneTopological insulators and superconductorsQuantum Spin Hall Effect and Topological Phase Transition in HgTe Quantum WellsQuantum Spin Hall Insulator State in HgTe Quantum WellsNonlocal Transport in the Quantum Spin Hall StateThe Quantum Spin Hall Effect: Theory and ExperimentFinite size effects on the quantum spin Hall state in HgTe quantum wells under two different types of boundary conditionsThe robustness of the quantum spin Hall effect to the thickness fluctuation in HgTe quantum wellsWave-vector filtering effect of the electric-magnetic barrier in HgTe quantum wellsTopological Insulators in Three DimensionsHelical edge and surface states in HgTe quantum wells and bulk insulatorsAb initio study of topological surface states of strained HgTeSymmetry-protected ideal Weyl semimetal in HgTe-class materialsElectronic structures of HgTe and CdTe surfaces and HgTe/CdTe interfacesThe CdTe/HgTe superlattice: Proposal for a new infrared materialCarrier recombination lifetime characterization of molecular beam epitaxially grown HgCdTeMagneto-Optical Investigations of a Novel Superlattice: HgTe-CdTeInduced Superconductivity in the Three-Dimensional Topological Insulator HgTeTermination, surface structure and morphology of the molecular beam epitaxially grown HgTe(001) surfaceMechanisms of molecular beam epitaxial growth of (001) HgTeMapping the Diffusion Potential of a Reconstructed Au(111) Surface at Nanometer Scale with 2D Molecular GasStructural and Electronic Properties of Sulfur-Passivated InAs(001) (2×6) SurfaceElemental structure in Si(110)-“16×2” revealed by scanning tunneling microscopyAtomic arrangements of 16×2 and (17,15,1) 2×1 structures on a Si(110) surfaceEngineering Topological Surface States: HgS, HgSe, and HgTeEC-STM Studies of Te and CdTe Atomic Layer Formation from a Basic Te SolutionDomain wall formation at the c ( 2 × 2 ) − ( 2 × 1 ) phase transition of the CdTe(001) surface
[1] Wu X G 2016 Chin. Phys. Lett. 33 027303
[2] Moore J 2009 Nat. Phys. 5 378
[3] Hasan M Z and Kane C L 2010 Rev. Mod. Phys. 82 3045
[4] Moore J E 2010 Nature 464 194
[5] Brune C, Liu C X, Novik E G, Hankiewicz E M, Buhmann H et al 2011 Phys. Rev. Lett. 106 126803
[6] Crauste O, Ohtsubo Y, Ballet P, Delplace P A L and Carpentier D et al 2013 arXiv:1307.2008[cond-mat.mes-hall]
[7]Oostinga J B, Maier L, Schüffelgen P, Knott D and Ames C et al 2013 Phys. Rev. X 3 021007
[8] Ren J, Bian G, Fu L, Liu C, Wang T et al 2014 Phys. Rev. B 90 205211
[9] Liu C, Bian G, Chang T R, Wang K, Xu S Y et al 2015 Phys. Rev. B 92 115436
[10] Lv M, Wang R, Wei L, Yu G, Lin T et al 2016 J. Appl. Phys. 120 115304
[11] Kane C L and Mele E J 2005 Phys. Rev. Lett. 95 226801
[12] Qi X L and Zhang S C 2011 Rev. Mod. Phys. 83 1057
[13] Bernevig B A, Hughes T L and Zhang S C 2006 Science 314 1757
[14] König M, Wiedmann S, Brüne C, Roth A, Buhmann H et al 2007 Science 318 766
[15] Roth A, Brüne C, Buhmann H, Molenkamp L W, Maciejko J et al 2009 Science 325 294
[16] König M, Buhmann H, W Molenkamp L, Hughes T, Liu C X et al 2008 J. Phys. Soc. Jpn. 77 031007
[17] Cheng Z, Chen R and Zhou B 2015 Chin. Phys. B 24 067304
[18] Guo H M, Zhang X L and Feng S P 2012 Chin. Phys. B 21 117301
[19] Zou Y L and Song J T 2013 Chin. Phys. B 22 037304
[20] Fu L, Kane C L and Mele E J 2007 Phys. Rev. Lett. 98 106803
[21] Dai X, Hughes T L, Qi X L, Fang Z and Zhang S C 2008 Phys. Rev. B 77 125319
[22] Wu S C, Yan B and Felser C 2014 Europhys. Lett. 107 57006
[23] Ruan J, Jian S K, Yao H, Zhang H, Zhang S C and Xing D 2016 Nat. Commun. 7 11136
[24] Schick J T, Bose S M and Chen A B 1989 Phys. Rev. B 40 7825
[25] Schulman J and McGill T 1979 Appl. Phys. Lett. 34 663
[26] Chang Y, Grein C H, Zhao J, Becker C R, Flatte M E et al 2008 Appl. Phys. Lett. 93 192111
[27] Guldner Y, Bastard G, Vieren J, Voos M, Faurie J and Million A 1983 Phys. Rev. Lett. 51 907
[28] Maier L, Oostinga J B, Knott D, Brune C, Virtanen P et al 2012 Phys. Rev. Lett. 109 186806
[29] Oehling S, Ehinger M, Gerhard T, Becker C R, Landwehr G et al 1998 Appl. Phys. Lett. 73 3205
[30] Oehling S, Ehinger M, Spahn W, Waag A, Becker C R and Landwehr G 1996 J. Appl. Phys. 79 748
[31]Oura K, Lifshits V G, Saranin A A, Zotov A V and Katayama M 2003 Surface Science (Berlin: Springer) chap 8 p 171
[32] Yan S C, Xie N, Gong H Q, Sun Q, Guo Y, Shan X Y and Lu X H 2012 Chin. Phys. Lett. 29 046803
[33] Li D F, Guo Z C, Li B L, Dong H N and Xiao H Y 2011 Chin. Phys. Lett. 28 086802
[34] Toshu A, Masamichi Y, Izumi O and Kazuyuki U 2000 Phys. Rev. B 61 3006
[35] Yamamoto Y 1994 Phys. Rev. B 50 8534
[36] Virot F, Hayn R, Richter M and van den Brink J 2013 Phys. Rev. Lett. 111 146803
[37] Lay M D and Stickney J L 2004 J. Electrochem. Soc. 151 C431
[38] Neureiter H, Tatarenko S, Spranger S and Sokolowski M 2000 Phys. Rev. B 62 2542