⇦Chinese Physics Letters, 2018, Vol. 35, No. 8, Article code 086801⇨ Influence of Surface Structures on Quality of CdTe(100) Thin Films Grown on GaAs(100) Substrates * Yi Gu(顾义)1,2,6, Hui-Jun Zheng(郑慧君)3, Xi-Ren Chen(陈熙仁)4, Jia-Ming Li(李家明)2, Tian-Xiao Nie(聂天晓)5, Xu-Feng Kou(寇煦丰)1,2** Affiliations 1Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, Shanghai 200050 2School of Information Science and Technology, ShanghaiTech University, Shanghai 201210 3School of Physical Science and Technology, ShanghaiTech University, Shanghai 201210 4Shanghai Institute of Technical Physics, Chinese Academy of Science, Shanghai 200083 5Fert Beijing Institute, BDBC, and School of Electronic and Information Engineering, Beihang University, Beijing 100191 6University of Chinese Academy of Sciences, Beijing 100049 Received 19 March 2018, online 14 July 2018 ^*Supported by the National Key Research and Development Program of China under Grant Nos 2017YFB0405704 and 2017YFA0305400 and the Shanghai Sailing Program under Grant No 17YF1429200.
^**Corresponding author. Email: kouxf@shanghaitech.edu.cn Citation Text: Gu Y, Zheng H J, Chen X R, Li J M and Nie T X et al 2018 Chin. Phys. Lett. 35 086801 Abstract We report the epitaxial growth of single-crystalline CdTe(100) thin films on GaAs(100) substrates using molecular beam epitaxy. By controlling the substrate pre-heated temperature with adjustable Te flux, three different reconstructed surfaces are realized, and their influence on the subsequent CdTe growth is investigated. More importantly, we find that both the presence of a thin native oxide layer and the formation of Ga-As-Te bonds at the interface enable the growth along the (100) orientation and help to reduce the threading dislocations and other defects. Our results provide new opportunities for compound semiconductor heterogeneous growth via interfacial engineering. DOI:10.1088/0256-307X/35/8/086801 PACS:68.55.-a, 81.15.Hi, 78.55.Et, 78.55.-m © 2018 Chinese Physics Society Article Text The growth of high-quality CdTe thin films by molecular beam epitaxy (MBE) has been one of the main semiconductor research topics over the past four decades.[1,2] Due to its large optical absorption coefficient (i.e., $5\times10^{5}$ cm$^{-1}$) and direct band gap (i.e., 1.45 eV) which is matched to the optimal solar spectrum for photovoltaic energy conversion, CdTe has shown great promise in solar cell applications.[3-5] Meanwhile, significant interest has been focused on the HgTe/CdTe-based heterostructures because of their indispensable role in both infrared detector applications,[6,7] as well as emerging topological quantum physics.[8-10] To date, single-crystalline CdTe thin film growth has been achieved on different substrates such as Si,[11,12] Ge,[13,14] InSb,[15,16] GaAs[17-25], and sapphire.[26,27] Among them, GaAs, though with a 14.6% lattice constant mismatching to CdTe, turns out to be a suitable substrate of choice because not only are large-size GaAs wafers readily available but also detailed GaAs surface reconstruction analysis and a clean process are well established.[28-32] In this regard, previous research has demonstrated that both CdTe(111) and CdTe(100) can be epitaxially grown on the GaAs(100) substrates, depending on the CdTe-GaAs interfacial conditions.[19-21,24,25] In particular, a pristine fully oxide-desorbed GaAs surface would lead to the CdTe(111) thin film while a relatively Te-rich GaAs surface with low-temperature reconstruction structures could give rise to the (100) nucleation of CdTe, hence enabling the following epitaxial (100) growth.[21] In this Letter, we report the epitaxial growth of high-quality CdTe(100) thin films on the GaAs(100) substrate using a solid source MBE system. By precisely controlling the substrate pre-anneal conditions, we manage to create three different GaAs surface structures, namely the original ($1\times1$) surface with thin GaO$_{x}$ residual oxide, the Ga-terminated ($8\times2$) surface, and the Te-stabilized ($6\times1$) surface. More importantly, we demonstrate that such surface reconstruction conditions result in different 3D nucleations of CdTe during the initial growth steps, and in turn are reflected by the discrepancies of the low-temperature photoluminescence spectra. Our results highlight the importance of the surface treatment on the CdTe epitaxial growth, hence unveiling new opportunities for compound semiconductor heterogeneous growth via interfacial engineering. High-quality single-crystalline CdTe(100) films were grown in an ultra-high vacuum DCA P700 system with the base pressure of $7\times10^{-11}$ torr. Epi-ready semi-insulating ($\rho>10^{6}$ $\Omega\cdot$cm) GaAs (100) substrates were In-soldered on molybdenum sample holders and pre-degassed in the preheating chamber for 1 h before being loaded into the growth chamber. To quantitatively investigate the influence of surface structures on the quality of as-grown samples, we adopted three different strategies during the pre-anneal process. Specifically: the GaAs substrate of sample 1 was annealed to 540$^{\circ}\!$C when the streaky 2D RHEED pattern began to appear; the substrate of sample 2 was annealed till 580$^{\circ}\!$C to completely remove the native oxide; and the substrate of sample 3 was annealed up to 580$^{\circ}\!$C under the Te flux-rich condition with pressure background about $6\times10^{-7}$ torr. As a result, three different GaAs surface structures were formed, which in turn resulted in different surface energies for the following CdTe thin film growth.

Fig. 1. Real-time RHEED patterns after pre-anneal (top panels) and after CdTe thin film growth (middle panels), and the GaAs surface structures (bottom panels) under three different pre-anneal conditions. (a) Sample 1, annealed at 540$^{\circ}\!$C to keep the original (1$\times$1) pattern. (b) Sample 2, annealed at 580$^{\circ}\!$C to obtain the Ga-stabilized (1$\times$1) pattern. (c) Sample 3, annealed at 580$^{\circ}\!$C with a Te flux-rich environment. The RHEED patterns along both the [$\bar{1}1$] and [031] azimuths are plotted to reveal the Ga-As-Te bonded (6$\times$1) surface. For all the three samples, the $d$-space shrink by $-$14.6% after the sample growth, indicating the CdTe(100) growth on the GaAs(100) substrates.

During the following sample growth, all substrates were maintained at 320$^{\circ}\!$C, and a high purity CdTe (99.9999%) source material was evaporated from a dual filament effusion cell to produce a stable beam flux with the growth rate around 2 Å/s. Epitaxial growth was monitored by an in-situ high energy electron diffraction (RHEED) technique, and the as-grown surface configuration was traced by recording the real-time RHEED patterns with a KSA400 system built by K-space Associates, Inc. After growth, all CdTe samples were investigated by an atomic force microscope (AFM), high-resolution transmission electron microscopy (HRTEM), and 77 K photoluminescence (PL). Relevant results are discussed in the following. Figure 1 summarizes the evolutions of the RHEED patterns under three different pre-anneal conditions. All epi-ready GaAs substrates were annealed without the presence of the arsenic flux, and the 2D original RHEED pattern became visible when the substrate temperature $T_{\rm sub}$ was above 500$^{\circ}\!$C, indicating the desorption of the native oxide layer. For sample 1, the non-reconstructed GaAs(100) surface (top panel of Fig. 1(a)) is maintained through the low-temperature anneal process (i.e., the substrate heater was turned off immediately after clear Kikuchi lines were observed on the fluorescent screen at $T_{\rm sub} =540^{\circ}\!$C). On the contrary, sample 2 was annealed to 580$^{\circ}\!$C to ensure the complete removal of the GaO$_{x}$ residual layer. Under such circumstances, the pristine GaAs surface is found to reconstruct to the Ga-stabilized ($8\times2$) structure, as confirmed by the RHEED pattern along the [0$\bar{1}1$] axis in the top panel of Fig. 1(b).[33] It is noted that further increasing the substrate temperature would result in the re-appearance of the ($1\times1$) structure with an excessive diffraction background, probably owing to the noncongruent evaporation of Ga atoms at higher temperature. Based on our previous experience dealing with the Te-passivated GaAs samples,[34,35] sample 3 was given with a modified strategy, which is to pre-anneal the GaAs(100) substrate under the Te flux-rich environment (i.e., the Te shutter was open once the substrate temperature was above 400$^{\circ}\!$C). Accordingly, the robust Ga-As-Te bonding is expected to be formed at the GaAs surface.[20] Experimentally, when sample 3 was heated at 580$^{\circ}\!$C, unique 1/2 order streaks are distinguishable along the [031] axis, which is 63.4$^{\circ}\!$ from the primary [0$\bar{1}1$] azimuth, and such patterns are well-maintained when the sample was later cooled down to the growth temperature at $T_{\rm sub} =320^{\circ}\!$C. Along with the RHEED patterns taken at other critical angles (i.e., there is no 1/2-order pattern along the symmetrical [0$\bar{3}1$] direction), we thus may verify that the resultant GaAs surface after the Te-assisted pre-annealing is reconstructed to be the ($6\times1$) morphology, as illustrated in the bottom panel of Fig. 1(c). During the succeeding CdTe thin film growth, all the three samples exhibited similar growth behavior. Particularly, spotty RHEED patterns always occurred at the initial stage, implying the 3D (100) CdTe nucleation process. Then after stabilizing for 10 min, these 3D dots gradually elongated into streaks, and the 2D growth mode persisted for the entire subsequent growth. As highlighted by the middle panels of Fig. 1, the RHEED patterns of samples 1–3 all show sharp 2D streaky lines with bright zero-order specular spots, indicative of the single-crystalline features of the samples. More importantly, since the RHEED patterns directly reflect the in-plane atom morphology in the reciprocal $k$-space, we can thus inspect the as-grown surface configuration using the $d$-spacing evolution between the two first-order diffraction lines. From Fig. 1, it can be calculated that the magnitudes of the $d$-spacing shrink by 14.6% in all three samples after growth. Consequently, it is concluded that high-quality CdTe(100) thin films are grown on the GaAs(100) substrates with the original ($1\times1$), the Ga-stabilized ($8\times2$), and the Te-terminated ($6\times1$) surface structures.

Fig. 2. Surface morphology of the CdTe(100) thin films grown under different pre-anneal conditions. (a) Sample-1 non-reconstructed GaAs(100), (b) sample-2 (8$\times$2)-Ga surface, and (c) sample-3 (6$\times$1)-Te surface. The sizes of the AFM images are 5 μm$\times$5 μm.

Fig. 3. X-ray diffraction patterns of CdTe(100) thin films. The step angle 2$\theta$ is varied from 20$^{\circ}\!$ to 70$^{\circ}\!$. Only (100)-related CdTe reflections are observed in the XRD rocking curves for samples 1–3. Data are shifted vertically for convenient comparison.

After sample growth, AFM was performed to compare the surface morphology of the CdTd thin films. Figure 2 shows the AFM images of samples 1–3 within the 5 μm$\,\times5\,μ$m region. It can be observed that all the samples have similar surface roughness with the measured rms values ranging from 1.8 nm (sample 3) to 2.5 nm (samples 1 and 2). Given that the thicknesses of these samples are around 600 nm, we hence may claim that all the three initial surface structures would enable the high-quality CdTe(100) thin film growth with a flat surface. Meanwhile, the crystalline qualities of the as-grown CdTe samples were characterized by x-ray diffraction (XRD). A typical $\theta$–$2\theta$ angular scan with step angles varying from 20$^{\circ}\!$ to 70$^{\circ}\!$ was successively performed, and the results are shown in Fig. 3. Sharp (100)-related CdTe reflections in addition to the GaAs(100) substrate peaks are clearly observed in the XRD rocking curves for samples 1–3, hence validating the highly ordered CdTe(100) epitaxial growth on the GaAs(100) substrate. It is also noted that the full widths at half maximum (FWHM) from the CdTe(400) peaks are found to be around 800–1300 arcsec, similar to previous reported data.[36,37]

Fig. 4. HRTEM images and SAED patterns of the CdTe(100) thin films grown on the GaAs(100) substrates. [(a), (d), (g)] HRTEM cross-sectional images of samples 1–3 all show the epitaxial single-crystalline CdTe(100) with zinc-blende structure, yet their interface situations are different. [(b), (e), (h)] Zoom-in HRTEM to highlight the high-quality of the MBE-grown CdTe(100) samples. [(c), (f), (i)] SAED patterns reveal two sets of diffraction spots with the same Miller indices but different radii, which confirms the epitaxial CdTe(100)-GaAs(100) configurations in all three samples.

To further investigate the detailed structural characteristics of the epitaxial films, HRTEM measurements were applied and the results are presented in Fig. 4. From the zoom-in TEM images shown in Figs. 4(b), 4(e), and 4(h), it is seen that highly ordered CdTe(100) epi-layers with zinc-blende configuration are formed on top of the GaAs(100) substrates. Meanwhile, the selected area electron diffraction (SAED) patterns of samples 1–3 all display two distinct sets of diffraction spots, which follow the same Miller indices yet their corresponding radii differ by 14%, again manifesting the epitaxial growth of CdTe(100) thin film on the GaAs(100) substrate with few twin and dislocation defects. Strikingly, the HRTEM images of the CdTe-GaAs interfaces show dramatic differences among the three thin films: sample 1 has a well-defined and relatively sharp interface, while samples 2 and 3 show the signature of mixed GaAs-to-CdTe transition structures. Considering the specific pre-anneal treatments received by each sample, we suggest that for the sample 1 case, the presence of the residual oxide layer after low-temperature pre-anneal helps to confine the CdTe-GaAs plane and in turn triggers the CdTe(100) growth. In contrast, since both samples 2 and 3 endure higher pre-anneal temperature (i.e., to completely remove the native GaO$_{x}$ oxide), the GaAs surface would form a relatively Te-rich structure due to the high sticking coefficient of Te and its relatively low Ga-Te bonding energy. As a result, the initial 3D CdTe nucleation on such Te-terminated surface would give rise to the formation of the CdTe-GaAs transition layer at the interface. Nevertheless, we need to point out that unlike previous reports of the CdTe(100) growth,[18,19,21] few interfacial threading dislocations are observed from the high angle annular dark field (HAADF) image in all our samples. This observation suggests that the pre-anneal methods applied in this work may be an effective approach to obtain low-defect CdTe thin films.

Fig. 5. Low-temperature photoluminescence spectra of samples 1–3. All the samples display pronounced peaks at 1.41 eV (defect-associated recombination) and 1.58 eV (bound exciton recombination). Inset: zoom-in PL spectra around 1.58 eV.

Finally, low-temperature PL measurements were taken on samples 1–3 to further evaluate their photoelectric properties. All the samples were photo-excited at 532 nm laser with a power of 60 mW. A Fourier Transform-Infrared Spectroscopy (FTIR) spectrometer and a silicon detector were used to resolve and detect the PL signals across the [1.0 eV, 1.8 eV] range, respectively. From Fig. 5, it can be clearly seen that all the samples show pronounced PL peaks at 1.41 eV and 1.58 eV (we have checked the bare GaAs control sample to exclude the possibility that such PL peaks may come from the substrate). Compared with the intrinsic bound exciton recombination peak at 1.58 eV, which almost remains the same shape and intensity for all the samples (inset of Fig. 5), the defect-associated recombination peak at 1.41 eV exhibits strong sample-dependent behavior among samples 1–3. Combined with RHEED, XRD, and HRTEM results discussed in Figs. 1–4, it is implied that the PL intensity differences at 1.41 eV may relate to the CdTe-GaAs interface discrepancies, yet more quantitative experiments (including further optical characterizations) need to be carried out to understand the interface-related phenomena as well as to further justify the thin film quality. In conclusion, we have demonstrated high quality epitaxial growth of CdTe(100) on the GaAs(100) substrate. Different GaAs surface reconstruction configurations are achieved by changing the substrate pre-anneal conditions, and both the presence of the oxide residual layer and the Te-rich surface enable initial 3D nucleation of CdTe followed by the subsequent epitaxial growth along the (100) orientation. HRTEM experiments confirm the highly ordered single-crystalline structures of samples and show that the CdTe(100)-GaAs(100) interface strongly depends on the surface structures after pre-annealing. Our results suggest that the interfacial engineering plays a significant role in determining both the epitaxial orientation and the photoelectric properties of the as-grown CdTe sample. References van der Waals epitaxy of CdTe thin film on grapheneMolecular beam epitaxial growth and optical properties of the CdTe thin films on highly mismatched SrTiO 3 substratesTheoretical Considerations Governing the Choice of the Optimum Semiconductor for Photovoltaic Solar Energy ConversionGrain boundaries in CdTe thin film solar cells: a reviewReview on first-principles study of defect properties of CdTe as a solar cell absorberGrowth of cubic and hexagonal CdTe thin films by pulsed laser depositionEffect of substrate temperature on the structure and optical properties of CdTe thin filmQuantum Spin Hall Effect and Topological Phase Transition in HgTe Quantum WellsQuantum Spin Hall Insulator State in HgTe Quantum WellsThe quantum spin Hall effect and topological insulatorsCurrent status of direct growth of CdTe and HgCdTe on silicon by molecular-beam epitaxyMolecular beam epitaxial growth of CdTe and HgCdTe on Si (100)Preparation of CdTe thin films on Ge substrates by molecular beam epitaxyMolecular beam epitaxial growth of high structural perfection, heteroepitaxial CdTe films on InSb (001)Microstructural studies of CdTe and InSb films grown by molecular beam epitaxyLow defect density CdTe(111)‐GaAs(001) heterostructures by molecular beam epitaxyGrowth of (100)CdTe films of high structural perfection on (100)GaAs substrates by molecular beam epitaxyCdTe-GaAs(100) interface: MBE growth, rheed and XPS characterizationInfluence of Ga‐As‐Te interfacial phases on the orientation of epitaxial CdTe on GaAsEpitaxial growth of CdTe on GaAs by molecular beam epitaxyPhotoluminescence in CdTe grown on GaAs substrates by molecular beam epitaxyHigh resolution electron microscope study of epitaxial CdTe‐GaAs interfacesInterface structure in heteroepitaxial CdTe on GaAs(100)Control of orientation of CdTe grown on clean GaAs and the reconstruction of the precursor surfacesHigh quality epitaxial films of CdTe on sapphireProperties and applications of CdTe/sapphire epilayers grown by molecular beam epitaxySurface reconstructions of GaAs(100) observed by scanning tunneling microscopyAtomic structure of GaAs(100)‐(2×1) and (2×4) reconstructed surfacesA surface analytical study of GaAs(100) cleaning proceduresFirst-principles study of the atomic reconstructions and energies of Ga- and As-stabilized GaAs(100) surfacesX‐ray photoelectron spectroscopic study of the oxide removal mechanism of GaAs (100) molecular beam epitaxial substrates in i n s i t u heatingBonding direction and surface‐structure orientation on GaAs (001)Interplay between Different Magnetisms in Cr-Doped Topological InsulatorsReview of 3D topological insulator thin-film growth by molecular beam epitaxy and potential applicationsInfluence of a ZnTe buffer layer on the structural quality of CdTe epilayers grown on (100)GaAs by metalorganic vapor phase epitaxyDetermining the [001] crystal orientation of CdTe layers grown on (001) GaAs

[1]	Mohanty D et al 2016 Appl. Phys. Lett. 109 143109
[2]	Tang K et al 2016 J. Alloys Compd. 685 370
[3]	Loferski J J 1956 J. Appl. Phys. 27 777
[4]	Major J D 2016 Semicond. Sci. Technol. 31 093001
[5]	Yang J H et al 2016 Semicond. Sci. Technol. 31 083002
[6]	Pandey S et al 2005 Thin Solid Films 473 54
[7]	Sathyamoorthy R et al 2003 Sol. Energy Mater. Sol. Cells 76 339
[8]	Bernevig B A et al 2006 Science 314 1757
[9]	Konig M et al 2007 Science 318 766
[10]	Qi X L and Zhang S C 2010 Phys. Today 63 33
[11]	Sporken R et al 1992 J. Vac. Sci. Technol. B 10 1405
[12]	Sporken R et al 1989 Appl. Phys. Lett. 55 1879
[13]	Matsumura N et al 1985 J. Cryst. Growth 71 361
[14]	Zanatta J et al 1998 J. Cryst. Growth 184 1297
[15]	Farrow R et al 1981 Appl. Phys. Lett. 39 954
[16]	Wood S et al 1984 J. Appl. Phys. 55 4225
[17]	Ballingall J et al 1985 Appl. Phys. Lett. 47 599
[18]	Bicknell R et al 1984 Appl. Phys. Lett. 44 313
[19]	Faurie J et al 1986 Surf. Sci. 168 473
[20]	Feldman R et al 1986 Appl. Phys. Lett. 48 248
[21]	Kolodziejski L et al 1986 J. Vac. Sci. Technol. A 4 2150
[22]	Leopold D et al 1986 Appl. Phys. Lett. 49 1473
[23]	Otsuka N et al 1985 Appl. Phys. Lett. 46 860
[24]	Ponce F, Anderson G and Ballingall J 1986 Surf. Sci. 168 564
[25]	Srinivasa R, Panish M and Temkin H 1987 Appl. Phys. Lett. 50 1441
[26]	Cole H, Woodbury H and Schetzina J 1984 J. Appl. Phys. 55 3166
[27]	Myers T, Giles T N, Yanka R, Bicknell R, Cook J J, Schetzina J, Jost S, Cole H and Woodbury H 1985 J. Vac. Sci. Technol. A 3 71
[28]	Biegelsen D, Bringans R, Northrup J and Swartz L E 1990 Phys. Rev. B 41 5701
[29]	Chadi D 1987 J. Vac. Sci. Technol. A 5 834
[30]	Lu Z, Lagarde C, Sacher E, Currie J and Yelon A 1989 J. Vac. Sci. Technol. A 7 646
[31]	Qian G X, Martin R M and Chadi D 1988 Phys. Rev. B 38 7649
[32]	Vasquez R, Lewis B and Grunthaner F 1983 Appl. Phys. Lett. 42 293
[33]	Cho A 1976 J. Appl. Phys. 47 2841
[34]	Kou X F, Lang M R, Fan Y B, Jiang Y, Nie T X, Zhang J M, Jiang W J, Wang Y, Yao Y G, He L and Wang K L 2013 ACS Nano 7 9205
[35]	He L, Kou X F and Wang K L 2013 Phys. Status Solidi-R 7 50
[36]	Leo G, Longo M, Lovergine N, Mancini A, Vasanelli L, Drigo A, Romanato F, Peluso T and Tapfer L 1996 J. Vac. Sci. Technol. B 14 1739
[37]	Shtrikman H, Oron M, Raizman A and Cinader G 1988 J. Electron. Mater. 17 105