A Surface by Molecular-Beam-Epitaxy Growth of Tellurium Monolayer

Research Article www.acsami.org

An Effective Approach to Improving Cadmium Telluride (111)A Surface by Molecular-Beam-Epitaxy Growth of Tellurium Monolayer Jie Ren,†,‡ Li Fu,*,† Guang Bian,§ Jie Su,† Hao Zhang,† Saavanth Velury,⊥ Ryu Yukawa,# Longxiang Zhang,‡ Tao Wang,† Gangqiang Zha,† Rongrong Guo,† Tom Miller,‡ M. Zahid Hasan,§ and Tai-Chang Chiang*,‡ †

State Key Laboratory of Solidification Processing, School of Materials Science and Engineering, Northwestern Polytechnical University, Xi’an 710072, China ‡ Department of Physics, University of Illinois at Urbana−Champaign, Urbana, Illinois 61801-3080, United States § Joseph Henry Laboratory and Department of Physics, Princeton University, Princeton, New Jersey 08544, United States ⊥ Department of Physics, University of California, Berkeley, California 94720, United States # Institute for Solid State Physics, University of Tokyo, Kashiwanoha, Chiba 277-8581, Japan ABSTRACT: The surface cleansing treatment of non-natural cleavage planes of semiconductors is usually performed in vacuum using ion sputtering and subsequent annealing. In this Research Article, we report on the evolution of surface atomic structure caused by different ways of surface treatment as monitored by in situ core-level photoemission measurements of Cd-4d and Te-4d atomic levels and reflection high-energy electron diffraction (RHEED). Sputtering of surface increases the density of the dangling bonds by 50%. This feature and the less than ideal ordering can be detrimental to device applications. An effective approach is employed to improve the quality of this surface. One monolayer (ML) of Te grown by the method of molecular beam epitaxy (MBE) on the target surface with heating at 300 °C effectively improves the surface quality as evidenced by the improved sharpness of RHEED pattern and a reduced diffuse background in the spectra measured by high-resolution ultraviolet photoemission spectroscopy (HRUPS). Calculations have been performed for various atomic geometries by employing first-principles geometry optimization. In conjunction with an analysis of the core level component intensities in terms the layer-attenuation model, we propose a “vacancy site” model of the modified 1 ML-Te/CdTe(111)A (2 × 2) surface. KEYWORDS: surface reconstruction, dangling bonds, adlayer, layer-attenuation model, atomic geometry, 1 ML-Te/CdTe(111)A (2 × 2) surface cleavage areas are usually small.11−15 Other noncleavage planes such as (100), (111), and (211) are particularly useful in wafer device fabrication, and they have to be cleaned using special vacuum cleaning methods among which the most popular one is ion sputtering followed by annealing. Surface reconstruction is often induced in the process and the surface chemical stoichiometry can change, leading to a rearrangement of the surface atoms to lower the surface energy.16−18 The atomic rearrangement on the polar CdTe (111) surfaces when compared to the “ideal” unreconstructed (1 × 1) structure shows the appearance of atom dimers (trimers) in order to reduce the dangling bond density.19−21 The standard treatment can also introduce surface atom vacancies and defects leading to extra dangling bonds with a detrimental effect on device performance.16,20−23

1. INTRODUCTION Cadmium telluride (CdTe) is one of the most widely used semiconductors in the zinc-blend II−VI semiconductor family. Compared to other semiconductors such as Si and GaAs, it offers several advantages including a wide band gap and high optical absorption coefficients in the visible range. Thus, it is an outstanding material that can be used in the fabrication of various optoelectronic devices.1−3 Moreover it is qualified as an ideal substrate for epitaxial growth of other II−VI compounds such as HgxCd1−xTe.4−6 For CdTe-based devices, widely used configurations include metal−semiconductor, semiconductor− semiconductor, insulator−semiconductor, and oxide−semiconductor heterostructures. The performance of these devices depends largely on the structural quality of the starting CdTe surfaces and the heterostructure preparation method.7−10 A clean, flat, well-ordered, and chemically stable substrate surface is necessary for precise fabrication of optoelectronic devices. The natural cleavage plane of CdTe is (110), and excellent surfaces can be obtained by direct cleavage in ultrahigh vacuum, but the © 2015 American Chemical Society

Received: October 16, 2015 Accepted: December 16, 2015 Published: December 16, 2015 726

DOI: 10.1021/acsami.5b09863 ACS Appl. Mater. Interfaces 2016, 8, 726−735

Research Article

ACS Applied Materials & Interfaces

The calculation is performed under the generalized gradient approximation (GGA) with the Perdew−Wang1991 (PW91) functional31 as implemented in the CASTEP code.32 The ionic cores are represented by ultrasoft pseudopotentials for Cd and Te atoms. The Cd 4d10, 5s2 electrons and Te 5s2, 5p4 electrons are explicitly treated as valence electrons. The plane-wave cut off energy is 380 eV and the Brillouin-zone integration was performed over 5 × 5 × 5 grid sizes using the Broyden−Fletcher−Goldfarb−Shenno (BFGS) minimization technique for zinc-blende structure optimization. This set of parameters assures total energy convergence of 5.0 × 10−6 eV/atom, the maximum force of 0.01 eV/ Å, the maximum stress of 0.02 GPa and maximum displacement of 5.0 × 10−4 Å.

The polar CdTe (111)A surface after surface treatment by ion sputtering and subsequent annealing shows a (2 × 2) reconstruction.16,22−24 The density of dangling bonds on the surface was shown to increase significantly in previous studies.23,25 As a consequence, absorbability for adatoms of the surface is enhanced, and various undesired features such as disordering, impurities and defects on the surface are introduced. Minimizing surface dangling bonds and increasing the surface order are of great importance, and our work reported herein is directed at this issue. Specifically, we discovered a novel method for modifying a typical cation-vacancy reconstruction structure, the CdTe (111)A (2 × 2) surface with Cd vacancies, and thus obtaining a better ordered surface. Here, the CdTe (111)A (2 × 2) surface with Cd vacancies is prepared by several cycles of Ar+ sputtering and annealing in an ultrahigh vacuum chamber. We investigated the evolution of Cd atoms on the outlayer and the change in surface structure during the surface treatment process by using core-level photoemission spectroscopy to measure the Cd 4d and Te 4d levels and RHEED to assess the surface reconstruction and degree of order. Deposition of one ML Te on the surface of CdTe by molecular beam epitaxy (MBE) was employed to improve the surface atomic structure. To verify the surface quality, we analyzed the band-structure mapping results deduced from HRUPS. It is known that these techniques are extremely sensitive to the surface condition of solids.26−28 By examining the changes in the HRUPS spectra and the sharpness of RHEED patterns, we conclude that the so-treated 1 ML-Te/CdTe (111)A (2 × 2) surface exhibits a better ordered surface structure than the surface before Te treatment, and thus demonstrating an effective approach for improving the surface quality of CdTe (111)A (2 × 2). Furthermore, we have performed theoretical simulations on the atomic geometry of the unmodified and modified structures and studied thoroughly the effects of Te adlayer on the relaxed (2 × 2) reconstruction surface and the atomic arrangement of 1 ML-Te/CdTe (111)A (2 × 2) surface.

3. RESULTS AND DISCUSSION The Laue diffraction pattern of the CdTe (111) wafer sample, shown in Figure 1A, reveals the in-plane orientation. The

Figure 1. (A) Laue diffraction pattern of the CdTe(111) crystal. (B) Crystal structure and the (111) plane in CdTe. (C) Side view of CdTe lattice structure along [11] direction. (D) Schematic illustration of configuration of ARPES measurements using synchrotron radiation (SR).

high-symmetry directions ΓKM′ and ΓM are identified. The cubic zinc blende structure of CdTe crystal and its (111)A surface, marked by blue shaded planes, are shown in Figure 1B. The ideal (111)A surface is terminated with a Cd layer and breaks the tetrahedral bonding configuration. Therefore, each Cd atom at the surface is with a single dangling bond. The two-dimensional atomic structure of the ideal (111) surface is illustrated in Figure 1C. Figure 1D presents the schematic configuration of ARPES measurements using synchrotron radiation (SR), showing the SR incident direction, the normal emission direction, N, and a non-normal emission direction, NN. 3.1. Evolution of the (2 × 2) Reconstructed Surface. The evolution of the atomic arrangement of the surface can be characterized by four stages: the initial state before any surface treatment, the nonreconstructed state, the initially reconstructed state and the fully reconstructed state. The four stages are labeled i−iv in Figure 2, where core-level photoemission spectra of synchrotron radiation excited Cd 4d and Te 4d and RHEED patterns are presented. In core-level photoemission measurements, the incident photon energy used for exciting the Cd-4d and Te-4d levels are 51 and 81 eV, respectively. The corresponding electron escape depth λ at these energies is about 6−6.5 Å.33−35 The polar angles of the emitted photoelectrons were set to be centered either at 0° and 60°

2. EXPERIMENTAL SECTION 2.1. Surface Preparation and Measurements. The undoped CdTe crystal was grown by the modified Bridgman method in our laboratory.29 CZT samples were cut into the wafers of 5 × 10 × 1.5 mm3 with the surface parallel to (111) planes. Then CdTe samples were first mechanically polished using 0.5 μm size magnesia suspension and then chemically polished with a 2% Br-MeOH solution for 5 min. Afterward, they were rinsed in deionized water and finally dried with nitrogen.30 The Laue diffraction was used to evaluate the quality of the samples used before each experiment. The experiments of HRUPS and RHEED were performed at the Synchrotron Radiation Center (SRC) of the University of Wisconsin−Madison of the USA. Briefly, the systems work under UHV conditions (better than 1 × 10−10 mbar). The beamline covers the energy range from 10 to 250 eV, and the energy resolution is about 15 meV. The analysis chamber is equipped with a VG Scienta electron analyzer with a 2D detector for exciting the Cd-4d and Te-4d levels and HRUPS measurements, an ion gun for Ar+ sputtering, a retractable optics for RHEED. The MBE chamber comprises a quartz crystal microbalance (QCM) for monitoring the deposition rates. The sample surface was cleaned in situ by sputtering for 10 min with 500 eV and 1 × 10−5 Torr Ar ions, followed by annealing for 3−5 min at a temperature of 300 °C. 2.2. Atomic Geometries Simulation. First-principles geometry optimization method based on density-functional theory is adopted to optimize the atomic geometry of CdTe. Plane waves and ultrasoft pseudopotentials were employed in the numerical evaluation. 727


Research Article


Figure 2. Core-level photoemission spectra of Cd-4d and Te-4d orbitals and RHEED patterns at different stages, (i) before cleaning with contamination on surface, (ii) formation of nonreconstruction surface with bulk-like 1 × 1 periodicity, (iii) initial formative stage of 2 × 2 reconstruction surface, (iv) stable formative stage of 2 × 2 reconstruction surface.

Table 1. Line-Shape Parameters from Fittting Te-4d and Cd-4d Core-Level Data for the Surfaces in Stages i−iv Studied stage i θ = 0° Cd spin−orbit splitting [eV] branching ratio surface shift [eV] Gaussian width [eV] B S Lorentzian width S/B intensity ratio Cd/Te intensity ratio

1.01

Te

stage ii θ = 60°

Cd

θ = 0°

stage iii θ = 60°

θ = 0°

stage iv θ = 60°

θ = 0°

θ = 60°

Te

Cd

Te

Cd

Te

Cd

Te

Cd

Te

Cd

Te

Cd

Te

1.36 0.58

1.31 0.61

0.71 0.48

1.46 0.67

0.71 0.4

1.47 0.53

0.71 0.3 0.61

1.47 0.55

0.71 0.54 0.60

1.47 0.52

0.71 0.39 0.61

1.47 0.54

0.71 0.47 0.61

1.46 0.56

0.58

0.57

0.53

0.54

0.58

0.55

0.5

0.26

0.28

0.26

0.45 0.45 0.24 0.29

0.52

0.28

0.4 0.52 0.24 0.23 0.89

0.51

0.27

0.6 0.39 0.27 0.78

0.49

0.27

0.41 0.44 0.27 0.49 0.93

0.99

0.24

0.24

0.2

0.2

contributions from the bulk (B) and the surface (S), the line shapes were analyzed by fitting the data with a Voigt line shape (convolution of a Gaussian and a Lorentzian) for each spin− orbit-split component. The fitting parameters are the

[Figure 1D] to give relatively bulk-sensitive and surfacesensitive core-level spectra, respectively. The upper and lower spectra in each panel of Figure 2A and B correspond to emission angle 0° and 60°, respectively. To quantify the 728


Research Article


more intense, as evident in Figure 2ivC,D. More cleaning cycles beyond this stage yielded no further changes in the core-level spectra and RHEED patterns. This proves that the surface in stage vi is the targeted clean (111)A surface. The surface might still have defects, but it has reached a saturation state based on the sputtering and annealing procedure. 3.2. Atomic Arrangement Model. First-principles geometry optimization was performed to simulate the atomic geometry. Because of approximately one-quarter of surface Cd atoms are missing as observed in the experiments, we assumed that the (2 × 2) reconstruction involves Cd deficiency. As the ideal Cd-terminated polar A face is electrostatically unstable, it is likely to reconstruct, and a likely consequence is a reduction of the Cd surface concentration to minimize the surface dipole. Accordingly, a slab with one-quarter of Cd atoms missing in the outermost layer is used to simulate the surface structure. Top and side views of the relaxed (2 × 2) reconstruction are shown in Figure 3A and B respectively, where blue and gold

magnitude of spin−orbit splitting, the branch ratio between the two spin−orbit components, Gaussian and Lorentzian widths, the surface shift, and the intensities of the bulk and surface contributions. The stoichiometric ratio (Cd/Te ratio) at the surface is also an important factor for determining the surface atomic arrangement. For each stage, the results from fitting the Cd-4d and Te-4d orbitals are plotted in Figure 2A,B and listed in Table 1. The raw data and fitted curves are shown as shortdashed blue and black solid curves, respectively, for the spectra. Also shown are the surface and bulk components from the fit. The RHEED patterns shown in Figure 2C and D correspond to electron diffraction along the [11̅0] and [112̅] directions, separately. In stage i, besides the core-level of at 10.7 eV, there are two peaks, one at 7.1 eV and the other at 4.1 eV that can be observed in the spectrum taken with 51 eV photons, shown in Figure 2iA. The peak at 7.1 eV is most likely the O-2p corelevel, which is from the oxide contamination on the surface. The other peak at 4.1 eV corresponds to the valence band of CdTe. Because of the presence of contamination on the surface of CdTe, the spin−orbit−split (SOS) components corresponding to Cd-4d3/2 and Cd-4d5/2 cannot be resolved (see Cd-4d line shape in Figure 2iA,B). The line shape of Te-4d3/2 and Te-4d5/2 SOS components is skewed with an elevated tail on the higher binding-energy side, shown in Figure 2iB. A few faint RHEED spots are observed in Figure 2iC,D. The weakness of the spots and a high background are consistent with the existence of contaminants on the surface. The surface was then cleaned twice using the sputtering-annealing treatment cycle. The surface showed afterward additional features in the new stage, ii. The Cd-4d and Te-4d SOS components are resolved clearly. While the oxide peak has disappeared, the main peaks are broad and have long tails (Figure 2iiA,B). The corresponding RHEED patterns show fuzzy spots and there is no evidence for reconstruction (Figure 2iiC,D). Apparently a 1 × 1 periodic structure emerged after removing the oxide contamination layer. The Cd/Te ratio is estimated to be ∼1 in both stages i and ii, meaning roughly an equal number of Cd and Te atoms existing in the surface layers. The above cleaning cycle was repeated four more times. The resulting stage iii shows much sharper spectra and diffraction patterns. In the Cd-4d core-level spectrum, as shown in Figure 2iiiA, a shifted component appeared on the higher binding-energy side of the bulk component. The emergent peak can be attributed to the surface component and is related to the reconstruction. The surface and bulk components obtained from the fitting are labeled S and B, respectively. The surface-to-bulk intensity (S/B) ratio of the surface-sensitive spectrum (off-normal emission at 60°) is higher than that of the bulk-sensitive spectrum (normal emission at 0°), as shown in Table 1. This analysis confirms that the shifted component is indeed surface related. By contrast, the Te-4d core-level spectrum (see Figure 2iiiB) does not reveal any surface-shifted component. The Cd/Te ratio changes to 0.93, implying that a reduced concentration of Cd near the surface. These changes are accompanied by surface reconstruction, as revealed by the RHEED patterns shown in Figure 2iiiC and D. In addition to the main spots from the 1 × 1 structure, faint half-order spots appear, some of which are indicated by red dashed circles. In stage iv, obtained after more surface treatment cycles, the Cd/Te ratio decreases to 0.89 (Table 1), close to the theoretical value of 0.875 which is calculated from a model with one-quarter of surface Cd atoms removed (see below). The half-order spots become sharp and

Figure 3. Calculated geometric structure of the relaxed (2 × 2) reconstruction: (A) top view and (B) side view.

spheres represent Cd and Te atoms, respectively. A Cd vacancy is indicated by the yellow dashed cycle. The remaining threequarters of the surface Cd atoms move toward the adjacent Te layer by 0.896 Å and at the same time, attract the same number of Te atoms in the nearest layer to come closer together. These Te atoms relax outward by 0.066 Å. These relaxations give rise to distorted hexagonal rings consisting of the nearest-neighbor Cd and Te atoms. One of such rings is indicated by the white shaded area in the Figure. The 3-fold coordinated Te atoms are laterally relaxed toward the Cd vacancy along the Cd−Te bond direction, which is marked by yellow arrows. The result is a (2 × 2) reconstruction (indicated by the yellow dashed quadrangle). To examine the validity of the simulated surface structure, the component ratio ICd/ITe was theoretically analyzed based on a layer attenuation model.36,37 In this simulation, the model can be described as follows. If I Cd (0) is the photoelectron intensity produced by one Cd atom at a depth d below the surface of a solid dCd‑n at a particular electron kinetic energy E, the photoelectron intensity ICd‑n emitted from the surface with NCd‑n Cd atoms is given by ⎡ −d ⎤ ICd‐n = NCd‐nICd(0) exp⎢ Cd‐n ⎥ ⎣ λ cos θ ⎦

(1)

where θ the polar angle (indicated in Figure 1D) and λ the mean free path with assumed value of 6−6.5 Å for both Cd and Te, which are denoted by yellow dashed lines in Figure 3b. Considering the λ value, it is reasonable to assume that the emission electrons of Cd-4d are predominantly from the contributions of the first and second layers, which are marked by Cd-1 and Cd-2 in Figure 3B. Their escape depths are dCd‑1 729


Research Article

ACS Applied Materials & Interfaces and d Cd‑2, respectively. dCd‑1 is the covalent radius of Cd whose value is 1.48 Å. dCd‑2 is the sum of covalent radius of Cd and the distance between Cd(1) and Cd(2), which can be obtained from the simulated geometry structure. Its value is 4.46 Å in this study. To calculate the S/B ratio of Cd atoms from the simulated results, it is necessary to separate the bulk and surface components from the top two layers. It is reasonable to assume that the second layer contributes only to the bulk component, while the top first layer consists of contributions to both the surface and the bulk components.35,38 Three types of Cd atoms exist in the top layer as shown in Figure 3A, and they are indicated by A, B, and C, respectively. Three possible cases are summarized in Table 2. They are classified as follows: 1/3 of

have lower atomic mass and higher vapor pressure, therefore they are more easily knocked off by Ar+ ions. Figure 4A depicts the process of Cd escaping from the outermost layer in stages ii and iii. Figure 4B shows the stable CdTe-(111)A (2 × 2) structure with Cd-vacancy after enough cleaning cycles. In Figure 4C we compare our calculated result with a recent scanning tunneling microscopy (STM) measurement of CdTe (111)A surface.25 The Figure in front is the calculated structure with atoms denoted by blue and red balls, and the image at back is the result of high resolution STM mapping. They are in good accordance with the CdTe (111)A (2 × 2) model with Cd-vacancy. Since one-quarter of surface Cd atoms are missing, the density of surface dangling bonds increases by 50%. In the STM image, the Cd vacancies would correspond to the black spots. A surface supercell is indicated by the yellow dashed lines. Clearly, our proposed model agrees well with the high resolution STM image. ARPES spectra along ΓKM′, or the [11̅0] direction, taken from the CdTe (111)A (2 × 2) surface with 19 and 25 eV photons are shown in Figure 5A and B. The corresponding

Table 2. Calculated S/B Ratio of Cd Atoms for Three Possible Absorption Models of CdTe (111)A-(2 × 2) Surface Cd contribution to S S/B ratio (θ = 0°, λ = 6 Å) S/B ratio (θ = 60°, λ = 6 Å) S/B ratio (θ = 0°, λ = 6.5 Å) S/B ratio (θ = 60°, λ = 6.5 Å)

1/3 of NCd‑1 to S

2/3 of NCd‑1 to S

full of NCd‑1 to S

0.222 0.280

0.569 0.779

1.219 1.154

0.217

0.556

1.909

0.272

0.746

1.783

NCd‑n atoms in the top layer contribute to the S component and the rest to the B component, 2/3 of NCd‑n atoms contribute to the S component, so all of the NCd‑n atoms contribute to the S component. The calculated results of the (S/B) ratio for the surface-sensitive (polar angle 60°) spectrum and bulk-sensitive (polar angle 0°) spectrum are listed in Table 2. Comparing S/B ratios in Table 1 and Table 2, the experimental results are in good agreement with the model calculation with 1/3 surface atoms in the top layer contributing to the S component. Our model of the deficiency of Cd atoms on the CdTe (111)A surface is schematically shown in Figure 4A and B. The surface loses Cd atoms during the Ar+ ion sputtering and annealing process. Ar+ ions are more likely to break the bonds of the atoms with longer bond length indicated by A than the shorter bonds denoted by B (see Figure 3A). Thus, in the repeated sputtering process, it is the Cd atoms that get preferentially removed. Also, compared to Te atoms, Cd atoms Figure 5. APRES mapping taken along ΓKM′ measured with (A) 19 eV photons and (B) 25 eV photons. (C and D) Corresponding second derivative maps of panels A and B, respectively.

second derivatives along the energy axis of the same data are displayed in Figure 5C and D for an enhanced view of the band features. There are two sets of bands sitting close to the Fermi levels labeled as Γ7 and Γ8 in terms of the representations of the point group at Γ (marked with the red arrows), as depicted in Figure 5B. Most of the images of the bands taken at 19 and 25 eV are blurry. The band Γ′8+ cannot be seen in the region indicated by the white dashed cycle. These features indicate the nonideal condition of the CdTe (111)A (2 × 2) surface (see below). 3.3. Role of Te Adlayer in Surface Improvement. As discussed above, the sputtering-annealing surface treatment increases dangling bonds and yields distorted hexagonal rings in

Figure 4. (A) Schematic illustrations of Cd escape from outermost layer during the surface cleaning process. (B) Atomic structure of CdTe (111)A-(2 × 2) surface with Cd-vacancy. (C) Calculated results (front) and high resolution STM image (back) of CdTe (111)A(2 × 2) structure. Reproduced with permission from ref 25. Copyright 2011, AIP Publishing LLC. 730


Research Article


Figure 6. Core-level photoemission spectra from CdTe (111)A-(2 × 2) surfaces with the deposition of 1 ML Te. (A and B) Core level mapping of Cd-4d and Te-4d orbitals, respectively, (E) before heating and (F) and after heating. (C and D) RHEED patterns of the same surface as that in panels A and B with horizontal diffraction along [11̅0] and [112̅], respectively, (G and H) RHEED patterns of the same surface as that in panels E and F.

Table 3. Line-Shape Parameters from Fitting of Te-4d and Cd-4d Core-Level Data for 1 ML-Te/CdTe (111)A-(2 × 2)-Cd-Vacancy Surface before and after Heating before heating θ = 0° spin−orbit splitting [eV] branching ratio surface shift [eV] Gaussian width [eV] B S Lorentzian width S/B intensity ratio Cd/Te intensity ratio

after heating θ = 60°

θ = 0°

θ = 60°

Cd

Te

Cd

Te

Cd

Te

Cd

Te

0.71 0.4 0.61

1.46 0.54

0.71 0.46 0.61

1.46 0.56

0.71 0.41 0.61

1.46 0.55

0.71 0.48 0.61

1.46 0.58

0.4 0.47 0.25 0.16 0.69

0.51

0.45 0.42 0.25 0.17

0.52

0.42 0.47 0.22 0.24 0.78

0.47

0.47 0.41 0.22 0.29

0.48

0.22

0.22

0.21

0.21

measurements as shown later. The fitting parameters and calculated results such as S/B in Cd core-level spectra and Cd/ Te ratios are listed in Table 3. Compared to the CdTe (111)A (2 × 2) surface prepared by just sputtering and annealing, both the S/B and Cd/Te ratios decreased remarkably for the surface with 1 ML-Te adlayer. These changes can be attributed to the additional Te gained from the Te deposition, which screens the contributions from the surface Cd, but cannot remove it completely. After heating, because of the desorption of the excess Te adatoms,40 the S/B and Cd/Te ratios increased. RHEED patterns from the surface with 1 ML-Te adlayer before and after heating are shown in Figure 6C-D and 6G-H, respectively, where 6C and 6G are along [11̅0] and 6D and 6H along [2̅11]. The RHEED spots from the annealed surface are much sharper than before. Thus, annealing improved the surface order significantly. The RHEED pattern is also much better than the unmodified surface (Figure 2ivC,D). To further confirm this, ARPES band mapping using different photon energies was performed on the 1 ML-Te/CdTe (111)A (2 × 2) surface after annealing. Comparing it with the mapping on the

the surface atomic structure. The vacancy-induced reconstruction after the surface treatment is likely not a perfectly ordered structure. The nonideal geometry and undesired electronic structure are indicated by the blurry features in the RHEED patterns and ARPES spectra. To saturate the remaining dangling bonds and thus improving the surface order, an appropriate amount of passivating material can be grown on the target surface, which has been proven to be very effective in previous works.39 According to a previously published results,40 excessive Te atoms can be easily desorbed by heating. Therefore, Te atoms were used as adatoms in this study. 1 ML-Te adlayer was grown by MBE on the target surface, followed by heating at 300 °C for 5 min to remove any excessive Te atoms. Figure 6A and B show the Cd-4d and Te-4d core-level line shapes of the surface with 1 ML-Te adlayer deposited, respectively. Figure 6E and F present the corresponding line shapes after annealing. In these results, the surface component S is located at the same position as the one taken from the unmodified surface (Figure 2viA), implying that they have the same surface nature. This is also verified by the ARPES 731


Research Article


Figure 7. ARPES mapping taken along ΓKM′ measured for CdTe (111) (2 × 2) surfaces with 1 ML-Te adlayer with 19 eV photons: (A) before heating treatment and (B) after heating treatment. (C and D) The corresponding second derivative maps, (E) the corresponding EDC taken at the Γ̅ point, (F and G) with 25 eV photons before heating treatment and after heating treatment, respectively, (H and I) the corresponding second derivative maps, and (J) the corresponding EDC taken at the Γ̅ point.

bridge site and the on-top site.41 On the (111) surface, the calculated difference between adsorptions on the fcc and the hcp hollow sites has shown to be negligible.42,43 Therefore, we calculate the atomic geometries corresponding to only three high symmetry sites: the fcc hollow, the bridge, and the on-top sites, and the results are shown in Figure 8. Figure 8A,B show the top and side views of the relaxed hollow-site reconstruction, respectively. Some adatoms on the CdTe surface are off from the default bulk lattice sites by a distance more than 1.7 Å due to the significant relaxation (measured from Figure 8A). The behavior of the relaxation is fairly complicated. The Te atoms in the atomic layer right below the surface indicated by yellow arrows in Figure 8A were laterally relaxed toward the Cd vacancies. Three quarters of Te atoms form trimers in the outermost layer (see Figure 8A) while the rest are located in the center of hexagonal rings formed by nearest-neighbor Cd and Te atoms. This leads to the relaxation of Cd atoms in the hexagonal rings toward the Cd vacancies while the adjacent Te atoms are pushed away from the Cd vacancies. As a result, this gives rise to huge distortions in the hexagonal rings, denoted by the white shaded area. The top and side views of the relaxed configuration of the bridge site are shown in Figure 8C and D. Besides the Te atoms in the atomic layer right below the surface, indicated by yellow arrows in Figure 8C, half of the Te atoms in the adlayer, marked by white arrows, are also relaxed toward the Cd vacancies. The rest of the Te atoms in the adlayer are on the bridge sites. Though this arrangement yields a small distortion in hexagonal rings, denoted by the white shaded area, the adatoms are still significantly off from the ideal bulk lattice sites. Top and side views of the relaxed on-top site reconstruction are shown in Figure 8E, F respectively. According to previous studies,44 the adsorption at on-top sites of the unreconstructed CdTe(111) surface is unstable. This conclusion is further confirmed by comparing the Cd/Te ratio between the experimental and calculated results. The Cd/Te ratio is calculated by the layer

unmodified surface allows us to identify the improvement of the surface quality. For a clear comparison, we present the ARPES mapping on the unmodified surface again together with the result from the modified surface in Figure 7. The ARPES spectra taken with 19 eV photons along ΓKM′ from the CdTe (111)A (2 × 2) surface and the surface with 1 ML-Te after heating are shown in Figure 7A and B respectively. The corresponding second derivatives are displayed in Figure 7C and D. Results using 25 eV photons are presented in Figure 7F−I. Two sets of valence bands are marked with the red arrows and a surface state, with its binding energy in the bulk gap, is denoted with S in Figure 7I. The positions of the bands do not change after Te deposition but apparently the bands of the modified surface after heating are much sharper, compared to those from the unmodified surface. Specifically, the band Γ8′+ of the modified surface can be resolved well, while it cannot be seen in the data for the unmodified surface. All other bands of the modified surface after heating show clear dispersions as a function of the crystal momentum k. On the other hand, most of these bands appear fainter or fuzzier in the data for the unmodified surface. To highlight these bands, Figure 7E and J show the photoemission energy distribution curves (EDC) at the Γ̅ point with photon energies of 19 and 25 eV. EDCs form the unmodified surface and the modified surface are plotted with golden and black solid curves, respectively. Comparing the EDC curves, the results of the modified surface show more details of the emission from bulk and surface states than the results of the unmodified surface. Taken the above facts together, this suggests that the treatment procedure of Te deposition and heating on the CdTe (111)A-(2 × 2) surface with Cd vacancies is an effective method to improve the surface quality. 3.4. Modeling Te Adsorption on CdTe (111)A-(2 × 2) Surface. Four different high-symmetry adsorption sites were considered for Te-adatom on the CdTe (111)A-(2 × 2) surface, namely the fcc hollow site, the hcp hollow site, the 732


Research Article


Figure 9. (A) top view and (B) side view of calculated geometric structure of the relaxed vacancy-site (2 × 2) reconstruction.

right below the surface toward the Cd vacancies, shown in Figure 3A, 8A, and 8C were offset by the Te atoms introduced to the vacancy sites. Instead the simulation shows a slight outward relaxation of those Te atoms right below the surface, indicated by yellow arrows in Figure 9A. These relaxation features are accompanied by the negligible distortion of the hexagonal rings, denoted by the white shaded area. We also calculated Cd/Te ratios of the vacancy-site model and the result is summarized in Table 4. The results demonstrate clearly that the vacancy site model matches the experimental data very well. A schematic of the vacancy-site absorption model is shown in Figure 10.

Figure 8. Calculated geometric structures of the relaxed (2 × 2) reconstruction. (A) Top view and (B) side view for the hollow site model. (C) Top view and (D) side view for the bridge-site model. (E) top view and (F) side view for the on-top site model.

attenuation formula which depends on the photoelectron intensity produced by one Cd (Te) atom ICd(0) (ITe(0)), the number of atoms of each layer contributing to photoemission, NCd‑n (NTe‑n), and the escaped depth, dCd‑n (dTe‑n). Although ICd(0) (ITe(0)) cannot be measured directly, we can estimate the value of ICd(0)/ITe(0) from formula 1 with the measured Cd/Te ratio of CdTe (111)A-(2 × 2) structure (Table 1). NCd‑n (NTe‑n) and dCd‑n (dTe‑n) were obtained directly from the geometry optimization results. The results are summarized in Table 4. The calculated Cd/Te ratios of absorption on the fcc

Figure 10. Schematic illustrations of formation of 1 ML-Te/CdTe (111)A-(2 × 2)-Cd-vacancy at different stages: (A) deposition of 1 ML Te atoms onto the surface, (B) the configuration of a full Te adlayer, (C) the removal of the excessive nonon-site Te atoms from the outer layer, and (D) the model of modified on-site 1 ML-Te/ CdTe (111)-(2 × 2) surface.

4. CONCLUSION In summary, we systematically studied the effects of surface treatment on CdTe (111)A surface, and showed the formation of (2 × 2)-Cd vacancy surface reconstruction. We further propose an effective approach to improve the surface quality by depositing and heating 1 ML Te adlayer and establish the vacancy-site absorption model to explain the mechanism for surface improvement. The methodology explored in this work is likely to be broadly applicable and can be used as a standard tool for smoothening and improving covalent semiconductor surfaces, and therefore, it is of great practical value to the semiconductor industry.

Table 4. Calculated Results of S/B Ratio for the VacancySite Model of 1MLTe/CdTe (111)A-(2 × 2)-Cd-Vacancy Surface Cd/Te ratio

fcc hollow sites

bridge site

top site

vacancy site

0.24

0.31

0.21

0.76

hollow, bridge, and top sites are much lower than the experimental result. These results provide additional support to our proposal. For the on-top model, the surface free energy decreases noticeably as one-quarter of the Te atoms are placed at the location of Cd vacancies. Considering minimization of surface free energy, it is reasonable to assume that threequarters of Te atoms in the on-top site model are removed during heating, resulting in a new absorption model which we call a vacancy-site model. The relaxed top and side views of the vacancy-site model are shown in Figure 9A and B, respectively. The distortions induced by the adatoms are small; less than 1 Å. The prominent relaxation of Te atoms in the atomic layer

■

AUTHOR INFORMATION

Corresponding Authors

*Phone: 00-82-029-88494083. Fax: 00-82-029-88494083. E-mail: [email protected]. *Phone: 00-1-217-3332593. Fax: 00-1-217-2442278. E-mail: [email protected]. 733


Research Article

ACS Applied Materials & Interfaces Notes

(15) Ren, J.; Bian, G.; Fu, L.; Liu, C.; Wang, T.; Zha, G. Q.; Jie, W. Q.; Neupane, M.; Miller, T.; Hasan, M. Z.; Chiang, T. C. Electronic structure of the quantum spin Hall parent compound CdTe and related topological issues. Phys. Rev. B: Condens. Matter Mater. Phys. 2014, 90, 205211−205217. (16) Wu, Y. S.; Becker, C. R.; Waag, A.; Kraus, M. M.; BicknellTassius, R. N.; Landwehr, G. Correlation of the Cd-to-Te ratio on CdTe surfaces with the surface structure. Phys. Rev. B: Condens. Matter Mater. Phys. 1991, 44, 8904−8911. (17) Ren, J.; Fu, L.; Bian, G.; Wong, M. H.; Wang, T.; Zha, G. Q.; Jie, W. Q.; Miller, T.; Hasan, M. Z.; Chiang, T. C. Spectroscopic studies of CdTe(111) bulk and surface electronic structure. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 91, 235303−235308. (18) Ebina, A.; Takahashi, T. Studies of clean and adatom treated surfaces of II−VI compounds. J. Cryst. Growth 1982, 59, 51−64. (19) Benson, J. D.; Wagner, B. K.; Torabi, A.; Summers, C. J. Surface stoichiometry and reaction kinetics of molecular beam epitaxially grown (001) CdTe surfaces. Appl. Phys. Lett. 1986, 49, 1034−1036. (20) Tatarento, S.; Bassani, F.; Klein, J. C.; Saminadayar, K.; Cibert, J.; Etgens, V. H. Surface reconstructions of (001) CdTe and their role in the dynamics of evaporation and molecular-beam epitaxy growth. J. Vac. Sci. Technol., A 1994, 12, 140−148. (21) Tatarenko, S.; Daudin, B.; Brun, D.; Etgens, V. H.; Veron, M. B. Cd and Te desorption from (001), (111)B, and (110) CdTe surfaces. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50, 18479−18488. (22) Duszak, R.; Tatarenko, S.; Cibert, J.; Magnéa, N.; Mariette, H.; Saminadayar, K. Study of CdTe(1̅1̅1̅) surface reconstructions by RHEED and XPS. Surf. Sci. 1991, 251, 511−515. (23) Janowitz, C.; Manzke, R.; Skibowski, M.; Orlowski, B. A. Surface band structure of CdTe(111)-2 × 2 by angle-resolved photoemission. Surf. Sci. 1991, 247, 100−105. (24) Wu, Y. S.; Becker, C. R.; Waag, A.; von Schierstedt, K.; BicknellTassius, R. N.; Landwehr, G. Surface sublimation of zinc blende CdTe. Appl. Phys. Lett. 1993, 62, 1510−1512. (25) Egan, C. K.; Jiang, Q. Z.; Brinkman, A. W. Morphology and reconstructions of polar CdTe(111)A, B surfaces by scanning tunneling microscopy. J. Vac. Sci. Technol., A 2011, 29, 11021−11029. (26) Tersoff, J.; Kevan, S. D. Nonlifetime effects in photoemission linewidths. Phys. Rev. B: Condens. Matter Mater. Phys. 1983, 28, 4267− 4271. (27) Theilmann, F.; Matzdorf, R.; Meister, G.; Goldmann, A. Influence of surface structural disorder on linewidths in angle-resolved photoemission spectra. Phys. Rev. B: Condens. Matter Mater. Phys. 1997, 56, 3632−3635. (28) Neave, J. H.; Joyce, B. A.; et al. Dynamics of film growth of GaAs by MBE from Rheed observations. Appl. Phys. A: Solids Surf. 1983, 31, 1−8. (29) Li, G. Q.; Jie, W. Q.; Wang, T.; Yang, G. Impurities in CdZnTe crystal grown by vertical Bridgman method. Nucl. Instrum. Methods Phys. Res., Sect. A 2004, 534, 511−517. The CdTe single crystals are supplied by Imdetek Ltd. (30) Bai, X. X.; Jie, W. Q.; Zha, G. Q.; Zhang, W. H.; Li, P. S.; Hua, H.; Fu, L. XPS and SRUPS study of oxygen adsorption on Cd0.9Zn0.1Te (1 1 1)A surface. Appl. Surf. Sci. 2009, 255, 7966−7969. (31) Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M. R.; Singh, D. J.; Fiolhais, C. Atoms, molecules, solids, and surfaces: Applications of the generalized gradient approximation for exchange and correlation. Phys. Rev. B: Condens. Matter Mater. Phys. 1992, 46, 6671−6687. (32) Segall, M. D.; Lindan, P. J. D.; Probert, M. J.; Pickard, C. J.; Hasnip, P. J.; Clark, S. J.; Payne, M. C. First-principles simulation: ideas, illustrations and the CASTEP code. J. Phys.: Condens. Matter 2002, 14, 2717−2744. (33) Liu Y. Angle-Resolved Photoemission Studies of TwoDimensional Electron Systems. Ph.D. Thesis, University of Illinois at Urbana-Champaign, Urbana, IL, U.S.A., 2010. (34) Brudle, C. R. The application of electron spectroscopy to surface studies. J. Vac. Sci. Technol. 1974, 11, 212−221.

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors thank I. Matsuda for his valuable support with ARPES experiment and M. Bissen and M. Severson for assistance with beamline operation at the Synchrotron Radiation Center, which was supported by the University of Wisconsin-Madison. TM and the beamline operations were partially supported by NSF Grant No. DMR 13-05583. This work was supported by National Natural Science Foundation of China under Grant No. 51172185 (L.F.), the Project of Key areas of innovation team in Shaanxi Province under contract No. 2014KCT-12, Introducing Talents of Discipline to Universities No. B08040 and the U.S. Department of Energy (DOE), Office of Science (OS), Office of Basic Energy Sciences, under Grant No. DE-FG02-07ER46383 (TCC) and Grant No. DE-FG-02-05ER46200 (MZH). M.N. is partially supported by the US National Science Foundation Grant No. NSF-DMR-1006492.

■

REFERENCES

(1) Limousin, O.; Blondel, C.; Crétolle, J.; Dzitko, H.; Laurent, P.; Lebrun, F.; Leray, J. P.; Arques, M.; Baffert, N.; Mathy, F.; Noca, A.; Trystram, P.; Villard, P.; Baron, P.; Delagnes, E.; Rouger, M. The ISGRI CdTe gamma-ray camera: first steps. Nucl. Instrum. Methods Phys. Res., Sect. A 2000, 442, 244−249. (2) Gloeckler, M.; Sankin, I.; Zhao, Z. CdTe Solar Cells at the Threshold to 20% Efficiency. IEEE J. PHOTOVOLT. 2013, 3, 1389− 1393. (3) Yamaguchi, K.; Nakayama, N.; Matsumoto, H.; Ikegami, S. J. CdS−CdTe Solar Cell Prepared by Vapor Phase Epitaxy. J. Appl. Phys. 1977, 16, 1203−1209. (4) Ferekides, C. S.; Marinskiy, D.; Viswanathan, V.; Tetali, B.; Palekis, V.; Selvaraj, P.; Morel, D. L. High efficiency CSS CdTe solar cells. Thin Solid Films 2000, 361, 520−526. (5) Tanaka, A.; Masa, Y.; Seto, S.; Kawasaki, T. Zinc and selenium co-doped CdTe substrates lattice matched to HgCdTe. J. Cryst. Growth 1989, 94, 166−170. (6) Rogalski, A. HgCdTe infrared detector material: history, status and outlook. Rep. Prog. Phys. 2005, 68, 2267−2336. (7) Kim, K. H.; Cho, S. H.; Suh, J. H.; Won, J. H.; Hong, J. K.; Kim, S. U. Schottky-type polycrystalline CdZnTe X-ray detectors. Curr. Appl. Phys. 2009, 9, 306−310. (8) Nelson, A. J.; Conway, A. M.; Reinhardt, C. E.; Ferreira, J. L.; Nikolic, R. J.; Paye, S. A. X-Ray Photoemission Analysis of Passivated Cd(1‑x)ZnxTe Surfaces for Improved Radiation Detectors. Mater. Lett. 2009, 63, 180−181. (9) Burger, A.; Chen, H.; Tong, J.; Shi, D.; George, M. A.; Chen, K. T.; Collins, W. E.; James, R. B.; Stahle, C. M.; Bartlett, L. M. Investigation of Electncal Contacts for GdZnTe Nuclear Radiation Detectors. IEEE Trans. Nucl. Sci. 1997, 44, 934−938. (10) Li, Q.; Jie, W. Q.; Fu, L.; Wang, X. Q.; Zhang, X. G. Metal− CdZnTe contact and its annealing behaviors. Appl. Surf. Sci. 2006, 253, 1190−1193. (11) Mescher, M. J.; Schlessinger, T. E.; Toney, J. E.; Brunett, B. A.; James, R. B. Development of dry processing techniques for CdZnTe surface passivation. J. Electron. Mater. 1999, 28, 700−704. (12) Owens, A.; Peacock, A.; Bavdaz, M. Progress in Compound Semiconductors. Proc. SPIE 2002, 4851, 1059−1070. (13) Scheiber, C.; Giakos, G. C. Medical applications of CdTe and CdZnTe detectors. Nucl. Instrum. Methods Phys. Res., Sect. A 2001, 458, 12−25. (14) Schlesinger, T. E.; Toney, J. E.; Yoon, H.; Lee, E. Y.; Brunett, B. A.; Franks, L.; James, R. B. Cadmium zinc telluride and its use as a nuclear radiation detector material. Mater. Sci. Eng., R 2001, 32, 103− 189. 734


Research Article

ACS Applied Materials & Interfaces (35) John, P.; Miller, T.; Hsieh, T. C.; Shapiro, A. P.; Wachs, A. L.; Chiang, T. C. Photoemission studies of CdTe(100) and the AgCdTe(100) interface: Surface structure, growth behavior, Schottky barrier, and surface photovoltage. Phys. Rev. B: Condens. Matter Mater. Phys. 1986, 34, 6704−6712. (36) van der Veen, J. F.; Himpsel, F. J.; Eastman, D. E. StructureDependent 4f-Core-Level Binding Energies for Surface Atoms on Ir(111), Ir(100)-(5 × 1), and Metastable Ir(100)-(1 × 1). Phys. Rev. Lett. 1980, 44, 189−192. (37) John, P.; Miller, T.; Chiang, T. C. InSb(100) reconstructions probed with core-level photoemission. Phys. Rev. B: Condens. Matter Mater. Phys. 1989, 39, 1730−1737. (38) Maeda, F.; Watanabe, Y.; Scimeca, T.; Oshima, M. Surface structure of Se-treated GaAs(001) from angle-resolved analysis of core-level photoelectron spectra. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 48, 4956−4959. (39) Cho, S.; DiVenere, A.; Wong, G. K.; Ketterson, J. B.; Meyer, J. R.; Hong, J. L. Growth-mode modification of Bi on CdTe.111.A using Te monolayer deposition. Phys. Rev. B: Condens. Matter Mater. Phys. 1998, 58, 2324−2328. (40) Pridachin, D. N.; Yakushev, M. V.; Sidorov, Y. G.; Shvets, V. A. A study of tellurium adsorption processes on silicon by ellipsometry, RHEED and AES methods. Appl. Surf. Sci. 1999, 142, 485−489. (41) Carlsson, J. M.; Hellsing, B. First-principles investigation of the quantum-well system Na on Cu(111). Phys. Rev. B: Condens. Matter Mater. Phys. 2000, 61, 13973−13982. (42) Koper, M. T. M.; van Santen, R. A.; Wasileski, S. A.; Weaver, M. J. Field-dependent chemisorption of carbon monoxide and nitric oxide on platinum-group (111) surfaces: Quantum chemical calculations compared with infrared spectroscopy at electrochemical and vacuumbased interfaces. J. Chem. Phys. 2000, 113, 4392−4407. (43) Loffreda, D.; Simon, D.; Sautet, P. Dependence of stretching frequency on surface coverage and adsorbate−adsorbate interactions: a density-functional theory approach of CO on Pd (111). Surf. Sci. 1999, 425, 68−80. (44) Li, Y. P.; Zhang, H. F.; Fang, R. C.; Yang, F. Y. A Study on The Structural Stability of CdTe(111). Acta Phys. Sin.Ch. Ed. 1995, 44, 788−794.

735


A Surface by Molecular-Beam-Epitaxy Growth of Tellurium Monolayer

Recommend Documents