High quality Bi2Te3 and Sb2Te3 topological insulators films were epitaxially grown on GaAs (111) substrate using solid source molecular beam epitaxy. Their growth and behavior on both vicinal and non-vicinal GaAs (111) substrates were investigated by reflection high-energy electron diffraction, atomic force microscopy, X-ray diffraction, and high resolution transmission electron microscopy. It is found that non-vicinal GaAs (111) substrate is better than a vicinal substrate to provide high quality Bi2Te3 and Sb2Te3 films. Hall and magnetoresistance measurements indicate that p type Sb2Te3 and n type Bi2Te3 topological insulator films can be directly grown on a GaAs (111) substrate, which may pave a way to fabricate topological insulator p-n junction on the same substrate, compatible with the fabrication process of present semiconductor optoelectronic devices.
I. INTRODUCTION
Q2Te3 (Q = Bi, Sb) is a typical V-VI narrow gap semiconductor, having a rhombohedral unit cell that can be considered as groupings of -Te-Q-Te-Q-Te- planes, referred to as quintuple layers (QLs). Within the QL unit, the chemical bond provides the binding force, while the weak van der Waals (vdW) force acts between adjacent QLs. Q2Te3 is a traditional thermoelectric (TE) material (e.g. ZT ≈ 1 for Bi2Te3 at 300 K),1 that has caused attention in the past decade due to its superior TE performance. Recently, Q2Te3-based research has increased because of the discovery of a new state of matter known as a three-dimensional (3D) “topological insulator” (TI). This material is insulating in bulk with a finite band gap but possesses a gapless surface state protected by time reversal symmetry (TRS),2 which has been confirmed by angle-resolved photoemission spectroscopy (ARPES)3–5 and transport measurements.6–10 The emergence of TIs has brought an increase in the study of exotic quantum physics, such as Majorana fermions and magnetic monopoles,11–13 which may pave the way for quantum computation applications. These investigations are also accelerating the development of spintronics due to the spin helical structure of the TI surface state.14 Additionally, a very recent prediction shows that TI may be a promising candidate for use in high performance photodetectors in the terahertz (THz) to infrared (IR) frequency range because of high absorbance.15
With so many potential applications, it has become important to understand how to fabricate these materials. For example, molecular beam epitaxy (MBE) is one way to grow high quality TIs due to the advantages of ultra-high vacuum (UHV) background pressure and precise control of growth parameters.5 In this work, we have systematically investigated the fabrication of high-quality Q2Te3 films by MBE on GaAs substrates. Up to now, Si, SrTiO3, sapphire, and graphene have already been used as substrates for the growth of Q2Te3 films by MBE.3,16–18 There have also been a few investigations on the MBE growth of Q2Te3 on a GaAs substrate. For example, Liu et al. demonstrated epitaxial growth of both Bi2Te3 and Bi2Se3 films on GaAs (100) directly by van der Waals epitaxy (vdWe).19 They reported that rotation domains formed in the TIs films due to the lattice symmetry mismatch between the hexagonal lattices of Bi2Te3 and the cubic symmetry of the GaAs (001) surface.19 The vdWe growth method has also been used for the growth of Bi2Se3 on a vicinal Si (111) surface.20 However, an amorphous intermediate layer formed in this approach which of course is not good for the coupling of spins between the substrate and the TI film, and could hinder the integration of TIs on present semiconductor optoelectronic devices. On the other hand, GaAs (111) should be a good substrate to realize coherent heteroepitaxy of Q2Te3 due to the lattice symmetry. For example, He et al. focused on the measurement of the magnetoresistance properties of Bi2Te3 thin films grown on semi-insulating GaAs (111) with ZnSe as a buffer layer21 while Richardella et al. demonstrated coherent heteroepitaxy of Bi2Se3 on GaAs (111)B.22
In this paper, we investigate the growth and behavior of Q2Te3 on both vicinal and non-vicinal GaAs (111) (V-GaAs (111) and Nv-GaAs(111)) substrates by reflection high-energy electron diffraction (RHEED), atomic force microscopy (AFM), X-ray diffraction (XRD) and high resolution transmission electron microscopy (HRTEM). Together with Hall and magnetoresistance (MR) measurements, our observations indicate that p type Sb2Te3 and n type Bi2Te3 topological insulator films can be directly grown on a GaAs (111) substrate, enhancing the opportunity to fabricate topological insulator p-n junctions.23
II. EXPERIMENTAL SECTION
A Riber 32P solid source MBE system was used to grow Q2Te3 films on epi-ready GaAs (111) substrates. The growth chamber is equipped with a RHEED (Staib) and a transmission optical system (kSA BandiT) monitoring the substrate band edge to give accurate growth temperature down to 150 °C. High purity 6N Bi, Sb and Te sources were evaporated to provide the beam fluxes for the film growth with a ratio of about 1:12. The morphology of as-grown films were quenched by turning off the manipulator heater right after the growth, and a Veeco ambient AFM was used to characterize the sample surface. High resolution XRD (HRXRD) was performed on a PANalytical X’Pert MRD system equipped with a parabolic mirror and PIXcelTM detector. The interface was characterized on FEI Titan 80-300 TEM. The Hall bar structure for transport measurement was fabricated using standard photo-lithography method as detailed in Figure 1S.24
The Q2Te3 films were grown on GaAs (111)A and GaAs (111)B substrates. As a result, two kinds of different RHEED evolutions were observed during growth due to the difference between A and B planes of GaAs (111) as show in Figure 1S and Figure 2S.24 As a comparison, both V-GaAs (111) and Nv-GaAs(111) substrates were used as substrates for growth of Q2Te3 films. This is motivated by the fact that a vicinal Si (111) substrate was observed to improve the quality of Bi2Se3 epitaxial films grown by vdWe.20 After deoxidizing GaAs substrates at 610 °C, a two steps growth method, i.e. a high temperature GaAs buffer layer growth at 590 °C and a low temperature Q2Te3 film growth at 250 °C, was performed. More than ten kinds of different samples including those shown in TABLE 1S24 were investigated and here we will show the data from the six kinds of samples as shown in Table I. Sample A (200 nm Bi2Te3/V-GaAs (111)A-3°) and sample B (400 nm Sb2Te3/V-GaAs (111)A-3°) were a 200 nm thick Bi2Te3 film and a 400 nm thick Sb2Te3 film grown on V-GaAs (111)A substrate (3° miscut) with a 330 nm thick GaAs buffer layer. Samples C and D were 30 nm thick Bi2Te3 films grown on V-GaAs (111)A and Nv-GaAs (111)A with the same thick GaAs buffer layer (3 nm). Samples E (200 nm thick Bi2Te3) and F (400 nm thick Sb2Te3) were grown on V-GaAs (111)B substrate with a 330 nm thick GaAs buffer layer.
. | Q2Te3 . | Buffer Layer . | . | . | . |
---|---|---|---|---|---|
. | & . | & . | . | . | . |
Sample. . | Thickness . | Thickness . | Substratea . | Misalignment Angle . | Conduction Type . |
A | Bi2Te3 | GaAs (111)A | N type | ||
& | & | V-GaAs (111)|${\rm A} \to (\overline 2 11)$| | 3° | n | |
200 nm | 330 nm | ||||
B | Sb2Te3 | GaAs (111)A | N type | ||
& | & | V-GaAs (111)|${\rm A} \to (\overline 2 11)$| | 3° | p | |
400 nm | 330 nm | ||||
C | Bi2Te3 | GaAs (111)A | N type | ||
& | & | V-GaAs (111)|${\rm A} \to (\overline 2 11)$| | 3° | – | |
30 nm | 3 nm | ||||
D | Bi2Te3 | GaAs (111)A | Semi-insulating | ||
& | & | Nv-GaAs (111)A | ±0.1° | n | |
30 nm | 3 nm | ||||
E | Bi2Te3 | GaAs (111)B | N type | ||
& | & | V-GaAs (111)|${\rm B} \to (\overline 2 11)$| | 3° | n | |
200 nm | 330 nm | ||||
F | Sb2Te3 | GaAs (111)B | N type | ||
& | & | V-GaAs (111)|${\rm B} \to (\overline 2 11)$| | 3° | p | |
400 nm | 330 nm |
. | Q2Te3 . | Buffer Layer . | . | . | . |
---|---|---|---|---|---|
. | & . | & . | . | . | . |
Sample. . | Thickness . | Thickness . | Substratea . | Misalignment Angle . | Conduction Type . |
A | Bi2Te3 | GaAs (111)A | N type | ||
& | & | V-GaAs (111)|${\rm A} \to (\overline 2 11)$| | 3° | n | |
200 nm | 330 nm | ||||
B | Sb2Te3 | GaAs (111)A | N type | ||
& | & | V-GaAs (111)|${\rm A} \to (\overline 2 11)$| | 3° | p | |
400 nm | 330 nm | ||||
C | Bi2Te3 | GaAs (111)A | N type | ||
& | & | V-GaAs (111)|${\rm A} \to (\overline 2 11)$| | 3° | – | |
30 nm | 3 nm | ||||
D | Bi2Te3 | GaAs (111)A | Semi-insulating | ||
& | & | Nv-GaAs (111)A | ±0.1° | n | |
30 nm | 3 nm | ||||
E | Bi2Te3 | GaAs (111)B | N type | ||
& | & | V-GaAs (111)|${\rm B} \to (\overline 2 11)$| | 3° | n | |
200 nm | 330 nm | ||||
F | Sb2Te3 | GaAs (111)B | N type | ||
& | & | V-GaAs (111)|${\rm B} \to (\overline 2 11)$| | 3° | p | |
400 nm | 330 nm |
V-GaAs (111) substrate was made by AXT Inc., whereas, Nv-GaAs(111) was made by Wafer Technology Ltd.
III. RESULT AND DISCUSSION
A. Surface morphology studies with atomic force microscopy
Figure 1 shows a set of AFM images, which reveal the different growth modes of Q2Te3 films observed on both V-GaAs (111) and Nv-GaAs(111) substrates. Fig. 1(a) is the AFM image of sample A (200 nm Bi2Te3/V-GaAs (111)A-3°). The big terraces and steps run nearly parallel. In most areas, about 1 μm wide terraces and 35 nm high steps can be found. The root-mean-square (RMS) roughness is about 8 nm for 5 × 5 μm2 area. Furthermore, small steps with height of about 1 nm were observed on the flat terrace as seen in Fig. 1(b) and 1(c) (500 × 500 nm2), consistent with a single QL thickness for each step. This surface morphology covers the entire sample surface based on more AFM measurements at the different positions on the surface of sample A. Sample B (400 nm Sb2Te3/V-GaAs (111)A-3°) shows the similar surface morphology as shown in Fig. 1(d). Obviously, this is a typical step-bunched surface,25 which forms with the big steps and terraces through a step-flow growth mode. Such surface morphology is very different compared to reported result of spiral growth.20,26 In order to avoid concerns over variation in the thickness of GaAs buffer layer or Q2Te3 film, both samples C (30 nm Bi2Te3/V-GaAs (111)A-3°) and D (30 nm Bi2Te3/Nv-GaAs (111)A) were grown on the same molybdenum block simultaneously using the same thick GaAs buffer layer (3 nm) and Q2Te3 (30 nm) films. As shown in Fig. 1(e), sample C still shows an initial step-bunched surface. However, spiral growth was observed for sample D as shown in Fig. 1(f)–1(h). For the as-grown Q2Te3 film, AFM investigations indicate that a V-GaAs (111) substrate is conducive to the formation of a step bunched surface through the tendentious step-flow growth induced by the high density of steps providing the high energy barrier to grow over the steps in its surface.27 Nv-GaAs (111) substrates sustain a smooth surface during spiral growth or partial step-flow growth due to the lower step density associated with non-vicinal substrate surface, confirmed by the more AFM investigation results as shown in Figure. 4S.24
B. Structural studies with X-ray diffraction and transmission electron microscopy
To determine the crystal quality and epitaxial orientation of as-grown Q2Te3 films, XRD was performed. Figures 2(a) and 2(b) are the 2θ-ω scans within 0-80° range for samples E (200 nm Bi2Te3/V-GaAs (111)B-3°) and F (400 nm Sb2Te3/V-GaAs (111)B-3°). Only Q2Te3 (003) and GaAs (111) families of reflections were observed, indicating the high c-axis orientation of Q2Te3 films along GaAs [111] direction, i.e. Q2Te3 (001)//GaAs (111). Compared with the peak position of GaAs substrates, the QL thickness (equals to 1/3 of lattice constant c) was given by dQL = 1.011 nm for Bi2Te3 and dQL = 1.007 nm for Sb2Te3, respectively, consistent with the RHEED observation results that the Q2Te3 lattice is fully relaxed. The full-width-half-maximum (FWHM) is 0.09° and 0.014° for Bi2Te3 (006) and Sb2Te3 (00,15), respectively, as shown in Fig. 2(c) and 2(d), indicating high-quality growth. To distinguish the subtle difference of as-grown Q2Te3 films, reciprocal space mapping (RSM) was performed on samples F (400 nm Sb2Te3/V-GaAs (111)B-3°), C (30 nm Bi2Te3/V-GaAs (111)A-3°) and D (30 nm Bi2Te3/Nv-GaAs (111)A). The RSM results of symmetric planes in Fig. 2(e)–2(g) distinctly demonstrated the tilted angle of 0.055°, 0.046° and 0.002° between Q2Te3 (00,18) and GaAs (222), for samples F, C and D, respectively, indicating higher c-axis epitaxial orientation than vdWe on a vicinal substrate.21 For these three samples, Sample D grown on non-vicinal substrate has a better c-axis orientation than samples C and F. The RSM result of non-symmetric planes of GaAs (440) and Bi2Te3 |$(\overline 2 2,27)$| for sample D in Fig. 2(h) presents the in-plane lattice mismatch between Bi2Te3 film and GaAs (111) substrate is around 9.3%, which is consistent with the RHEED observation of small critical thickness. Besides, the peak of sample F grown on vicinal substrate with thick GaAs buffer layer is much broader than that of samples C and D, indicating worse crystalline quality. The substrate miscut dependence of the tilted orientation angles for these three samples proves that as-grown Q2Te3 films are sensitive to the surface lattice orientation of the substrates and V-GaAs (111) substrates can reduce their crystalline quality through the unfavorable step bunching process.
To further clarify the effect of V-GaAs (111) substrates on the crystal quality of as-grown Q2Te3 film, we investigated the interface between Q2Te3 films and GaAs buffer layer for these three samples by HRTEM. Figure 3(a) gives the HRTEM images of sample F. It is obvious that the interface is disordered and a lot of defects were formed in both the GaAs buffer layer and Sb2Te3 film. Furthermore, the fast Fourier transform (FFT) analysis near the interface confirmed the RSM result that there is a small tilted angle of about 0.055° between the Sb2Te3 film and GaAs (111) substrate. For sample C, a similar result is observed in Fig. 3(b), i.e. disordered interface and tilted orientation, except for that fewer defects were formed due to the thinner GaAs buffer layer. In contrast, both a sharp interface and nearly parallel orientation were found for sample D, as shown in Fig. 3(c) and 3(d). Based on these observations, we can conclude that an Nv-GaAs (111) substrate is better to use than a vicinal substrate in order to fabricate high quality Q2Te3 films on GaAs (111) substrate.
C. Transport measurements of Q2Te3 thin films
In order to demonstrate the topological insulator behavior of as-grown Q2Te3 films, transport measurements were also performed. For our samples, all the Sb2Te3 films show p type conducting behavior, whereas, all the Bi2Te3 films show n type conducting behavior. Figure 4 shows the typical transport measurement results of Q2Te3 films on GaAs (111)B, and the one for Q2Te3 films on GaAs (111)A was shown in Figure 5S.24 Based on the linear behavior of Hall resistance vs. magnetic field (Rxy-H) below 30 kOe, the carrier density and mobility can be estimated as 7.81 × 1018 cm−3 and 959 cm2/Vs, 9.13 × 1018 cm−3 and 781 cm2/Vs for samples E (200 nm Bi2Te3/V-GaAs (111)B-3°) and F (400 nm Sb2Te3/V-GaAs (111)B-3°), respectively, which are typical parameters in TI films.3,28 Nonlinear Hall behavior was observed over 30 kOe for sample E, as shown in Fig. 4(a), which is likely an indication of more than one transport channel (surface and bulk) in the sample.29 Magnetoresistance (MR) measurements were performed for three different directional fields up to 80 kOe. In the perpendicular field (H), the R(H)/R(0) increases from 1 to 1.73 and 1.56 for samples E and F at 1.8 K, respectively, over the range of 0 to 80 kOe; linear MR was observed above 20 kOe, which may be attributed to the quantum linear MR (QLMR) of the surface states,30 as shown in Fig. 4(c) and 4(d). For the parallel field situation (Fig. 4(e) and 4(f)), R(H)/R(0) increases to 1.15 ∼ 1.30 at 80 kOe, which are smaller than that in perpendicular field. In a close-up view as shown in Fig. 4(g), a clear magnetoconductance peak (Δσ) was observed for sample E in small perpendicular field regime, which can be attributed to the weak anti-localization effect (WAL).8,21,31–34 According to the HLN theory35
where ψ(x) is the digamma function and lϕ is the phase coherence length. α is a coefficient reflecting the strength of the spin-orbital coupling and magnetic scattering. For Bi2Te3 film, Δσ at 1.8 K in the perpendicular field fits the HLN equation quite well in the low magnetic field and yields α = −0.2 and lϕ = 709 nm. Meanwhile, the temperature dependence of normalized resistance (R/Rmin) for sample F has an upturn below 4 K with field 0 kOe and 20 kOe, as shown in Figure 4(h), which is reminiscent of the electron-electron interaction in TI films.8
IV. SUMMARY
In summary, we have successfully demonstrated the MBE growth of Q2Te3 topological insulator films directly on GaAs (111) substrates. Similar to the results for Bi2Se3 grown on graphene/SiC(0001) reported by Liu et al.,26 spiral growth mode was observed for Q2Te3 films on Nv-GaAs (111)A substrate too. However, an step-bunched surface was observed for Q2Te3 films on V-GaAs (111)A substrates. Both XRD and HRTEM results indicate that the Nv-GaAs (111) substrate is better than a vicinal substrate to provide high quality Q2Te3 films. The tilted angle between Q2Te3 film and GaAs (111) substrate may be related to the misfit dislocation with a nonzero net out-of-plane Burgers vector component since Nagai's steps model is not applicable here due to the large lattice mismatch,36,37 and more interface investigations need to be done to establish the tilting mechanism in this kind of high-misfit heteroepitaxial systems. The observation of linear MR and WAL provides proofs for the possible TI behavior of as-grown Q2Te3 films.10,30 Hall effect measurements show that unintentional doping happens to as-grown Q2Te3 films: Sb2Te3 films always show p type conducting behavior, whereas, Bi2Te3 films always show n type conducting behavior, creating the potential to fabricate topological p-n junction on the same GaAs (111) substrate, which may offer a new platform to realize exciton condense in the interface. Further, our study paves a way to integrate TI-based devices with present semiconductor optoelectronic devices, generating brand-new multifunction devices.
ACKNOWLEDGMENTS
The authors gratefully acknowledge the NSF financial support through Grant No. DMR-0520550, the National Basic Research Program of China (Grant No. 2013CB934600), the National Natural Science Foundation of China (Grant No. 11174007 & No. 11222434), the Penn State MRSEC under NSF grant DMR-0820404. We are grateful to Moses H. W. Chan and Meenakshi Singh for the help in the transport measurement.