In order to explore the sample’s surface morphology, atomic force microscope (AFM) topography images were acquired, the results of these ones are illustrated in Fig. 2. In some cases, the AFM images present some structures as artifacts due to the tip adherence and contamination. The vertical scale is presented in nanometres, it can be observed in the case of the 3D images that the maximum height is around 20 nm in the analysed area which was 2.0 µm×2.0 µm. As can be seen in the surface images presented corresponding to the grown layers of the ZnTe compound, these are highly uniform in their surface morphology. The measured roughness was 32.37 nm in terms of the average roughness and 39.69 nm as the root mean square.
As a result of HR-XRD measurement, the peak corresponding to the (004) diffraction plane of ZnTe was identified and investigated. In Fig. 3, the thin peak about 33° corresponds to the (004) diffraction plane from the GaAs substrate, which was used as a reference for the ZnTe layer peak position adjustment and its comparison. The wide peaks at the left of it correspond to the (004) diffraction planes from the epitaxial layers, their intensities were amplified by 200 times due to the weaknesses of these peaks as a consequence of the thin thickness involved on the diffraction. The full width at half maximum (FWHM) of the diffraction peaks was very wide (600 arcsec to 800 arcsec) indicating a highly distorted lattice due mainly to mosaicity. Additionally, in this figure as is expected it can be observed that the intensities of the peaks increase with the thickness of the layers. Moreover, as the layer thickness increases, the peak position shifts away from the peak corresponding to the GaAs substrate towards the position that must have the correspondence to the ZnTe in bulk. This effect could be attributed to the relaxation of the layers and the increase of dislocation defects in the interface, due to the fact that for highly dislocated layers, the peak position is determined by the mean distortion of the lattice. For the samples whose diffractograms are showed in Fig. 3, the calculated thicknesses are ranged between 135 nm to 270 nm. In the case of the ZnTe layers grown on GaSb, it was not possible to resolve clearly the ZnTe diffraction peak from the one corresponding to the GaSb substrate. This is a consequence of the smaller lattice mismatch between the ZnTe layers and the GaSb substrate, since the lattice constants of the two materials are very close, aGaSb=6.0959 Å and aZnTe=6.1034 Å. In addition, the interface defects and the nanometric thickness of the ZnTe layers are related to the width of the diffraction peak,. The described above is shown in Fig. 4, which presents an X-ray diffraction (XRD) curve of a typical 125 nm thick ZnTe layer that was obtained by performing 410 growth cycles on the GaSb substrate. In Fig. 4, as a result of the deconvolution treatment, the peak corresponding to the ZnTe layer near the dominant peak can be observed due to the (004) diffraction plane of the GaSb substrate. Compared with the GaSb substrate peak, the great peak width of the layer and its weakness are evident.
Figure 3. Diffractograms of ZnTe on GaAs where peak position shift is related with the increase of interface dislocation.
Back scattering configuration was used to obtain the Raman spectra of the (100) GaSb and GaAs substrates that are shown in Fig. 5. GaSb Raman spectrum presents a dominant band at 237 cm−1 that is associated to longitudinal optical (LO) phonon frequency of GaSb, and the other is related to metallic tellurium observed at 274 cm−1, which was found experimentally. The weak shoulder on the low-frequency side of the GaSb-LO band at 229 cm−1 is the GaSb-transverse optical (TO) mode, in principle forbidden in this experimental geometry. Its weak intensity in the Raman spectrum indicates that the crystalline quality of the Te-doped GaSb substrate is very good. Similarly, the GaAs substrate only shows a Raman peak at 293 cm−1 that is associated with the GaAs-LO mode. Raman scattering measurement was performed to the substrates in order to have a reference for the clear identification of the peaks of the grown films. For assessing the quality of the ZnTe grown layers, Raman spectroscopy measurement was performed. The obtained results are shown in Fig. 6. It is noteworthy that the first- and second-order Raman dispersion was obtained of the grown samples. The TO-ZnTe mode was forbidden in the experimental measurement configuration for the (001) substrate orientation, which becomes active by the breakdown of the selection rules in the backscattering configuration. This breakdown is attributed to the generation of structural defects as misfit dislocation in the layer-substrate interface and other crystallinity problems originated from compositional fluctuations and by elastically scattering from ionized doping impurities. Fig. 6 shows the Raman spectra of the samples grown on the GaSb and GaAs substrates. Fig. 6 (a) illustrates the deconvolution of two Raman spectra, which present the first- and second-order Raman scattering of ZnTe, corresponding to the frequencies that are shown in Table 1. In these ones, there are two dominant peaks, associated with the tellurium related bands around 123 cm−1 and 141 cm−1-. The discussion of Raman scattering will be divided into two parts, corresponding to the Raman of the first and second orders.
Figure 6. Raman spectra of the samples grown on GaAs and GaSb substrates: (a) ZnTe Raman spectra of the samples with different thicknesses and substrates and (b) deconvolution of the measured Raman spectra into individual components.
Order Raman Mode assignation GaAs GaSb First Te2O-related 126 128 Te-related 144 146 LO-ZnTe 208 208 Second 2TA(L)-ZnTe 96 96 2TA(X)-ZnTe 105 106 (W1+W2)-ZnTe 159 159 (TO(X)+TA(X))-ZnTe 226 228 (LO(X)+TA(X))-ZnTe 238 238 2LA(X)-ZnTe 273 275
Table 1. Frequencies of active Raman lines detected for the ZnTe samples grown on the two different substrates
ZnTe has a crystal structure of the body-centered cubic (BCC) zinc blende type with two atoms per unit cell. Thus, it would expect three degenerate optical modes of vibrations at the centre of the Brillouin zone (q=0); however, the binding is partially ionic and in this case Poulet has shown that the macroscopic electric field associated with the longitudinal mode of vibration increases its frequency above that of the transverse mode of vibration. The first-order Raman spectrum consists of two bands: The highest in frequency is associated to the LO branch and the lowest is the doubly degenerate TO branch. Thus, the peak observed at 208 cm–1 corresponds to the ZnTe LO phonon-. The Raman spectra exhibit peaks related to the substrate characteristics. For the ZnTe Raman spectra of the layers grown on the GaSb substrate, three dominant peaks between 90 cm–1 and 150 cm–1 can be observed, which are associated with Te. As in the case of the samples grown on the GaAs substrate, the peak at 208 cm–1 corresponds to LO-ZnTe,. In addition, as can be observed in Fig. 6 (b), the weakness of the peak corresponding to LO-ZnTe is slightly displacement in comparison with the reported for the ZnTe bulk, possibly due to the layers stress and the nanometric nature of these layers, as has been reported by other authors.
In the second-order scattering process, the energy of the scattered photons differs from the energy of the incident photons by the energy of two vibrational quanta. The two phonons involved can be from the same branch (overtones) or from different branches (sums or differences). The wave vector of the phonon involved is usually orders-of-magnitude larger than that of the photons and conservation of crystal momentum. Therefore it is required that the wave vectors of the two phonons are equal and opposite in sign. Thus, the two phonons involved, although they may be from different branches of the spectrum, must come from the same point in the zone. This of course still allows two phonon scattering from any point in the zone and the scattered spectrum will consist of a continuous background with peaks corresponding to critical points in the two-phonon density of states.
The critical points in the zinc blende phonon spectrum are at the points Γ(0, 0, 0), X(1, 0, 0), L(1/2, 1/2, 1/2), and W(l, 1/2, 0) in the Brillouin zone. Birman has worked out the selection rules for the second-order Raman scattering from these critical points in zinc blende structures. Fig. 6 observes the features of the second-order spectrum and Table 1 contains a list of the frequencies corresponding to these features. Column 2 of Table 1 gives the mode assignments of these features. The features below 160 cm–1 could be due to the combinations of transverse acoustic phonons (sums and overtones) or to differences such as LA(X)-TA(X).
In these experiments, it was observed that the width of all the features below 160 cm–1 decreased at the same rate with increasing temperatures, and in addition was strongly polarized. On the basis of these observations the low-energy features were assigned to overtones and sums of transverse acoustic phonons.
In the case of CdTe samples, these were grown on the (100) oriented GaAs and GaSb substrates as in the ZnTe case, although the range of growth temperatures for the CdTe deposition was slightly different. Temperatures higher than 400 °C were necessary for the CdTe growth. As in the ZnTe case, the exposure time under the vapour sources was selected as 3 s. Figs. 7 and 8 show the XRD curves corresponding to the (004) CdTe diffraction plane for the samples obtained with different growth temperatures and different substrates. All the presented samples have the same number of grown cycles which is directly related with the layer thickness. The sample labelled as CdTe29 was grown at 405 °C, which is close to the minimum temperature required for the layer growth. As shown Fig. 7, the (004) plane diffraction peak presents a shift toward angles corresponding to lower lattice constants, which is indicative of stress accumulation. In comparison, the sample grown at 415 °C shows a shift corresponding to the increase of the lattice constant, which suggests that the stress produced by the lattice constants difference between the layer and substrate has been relaxed due to the dislocation and defects generation mechanism, favoured by the increased used growth temperature. In addition, the defects propagated from the substrate and the interface layer-substrate.
Fig. 9 shows the Raman spectra for CdTe nanolayers excited by 532 nm wavelength light. Two Raman spectra were fitted with Lorentzian lines (black circles) that are shown in Fig. 9 (b). The peaks at 95 cm–1 to 98 cm–1, 120 cm–1 to 125 cm–1, and 140 cm–1 to 143 cm–1 have been identified from trigonal Te. The 120 cm–1 to 125 cm–1 peak is due to a phonon with A1 symmetry, and the 95 cm–1 to 98 cm–1 and 140 cm–1 to 143 cm–1 peaks are originated from phonons with an E symmetry. However, the 140 cm–1 to 143 cm–1 peak could also be assigned to the TO phonon in CdTe, because it is very close to the reported one-. According to the Raman studies of CdTe,, the peak located at 157 cm–1 to 160 cm–1 can be assigned to the LO phonon of CdTe, in spite of a difference from 171 cm–1 and 167 cm–1 reported by Islam et al. and Amirtharaj et al., respectively. To the best of our knowledge, there have been no reports of the assignment for the broad band located at 265 cm–1 to 270 cm–1, which probably originates from the combination bands and one tentatively assigns them to the overtones of E and A1 modes in Te. The 295 cm–1 peak could be assigned at LO-GaAs. The peak at 321 cm–1 is associated with the phononic replica of the LO mode of the CdTe. The results of the measurement are summarized in Table 2. A high concentration of tellurium exists in the as-prepared CdTe nanolayers, as shown by the Raman study. Similar phenomena were reported in the CdTe film synthesized through the liquid-phase epitaxy method as well as the surface of single crystal CdTe in bulk form. One possible explanation could be that the free energy of Te is smaller than that of CdTe and thus Te crystals can be easily formed during the synthesis of CdTe. However, further study is still needed to confirm this one. One implication of it is that the surface property of the nanolayers may not be as expected. Isolated elemental domains (Te) are formed or precipitated on the surface of nanolayers; it may affect their applications in optoelectronic devices since its elemental properties are different from its compound properties.
Order Raman Mode assignation GaAs GaSb First Te-related 98 95 Te2O-related 125 120 TO-CdTe 143 140 LO-CdTe 157 160 UI 189 – LO-GaAs 295 – 2LO-CdTe – 321 Second (E+A1)-CdTe 270 264
Table 2. Frequency of active Raman lines detected for the CdTe samples grown on the two different substrates
After successfully obtaining the ZnTe and CdTe layers, it was proceeded to grow the layers of Cd1-xZnxTe ternary alloy, which was on the (100) oriented GaAs substrate. The used temperature range was between 400 °C and 425 °C. In order to grow the ternary alloy layers, a precursor was used as a solid source prepared by weighing the corresponding amounts of zinc and cadmium and mixing them. The procedure for preparing the precursor source was accomplished by combining Zn and Cd in different amounts which then were melted into a liquid forming at high temperatures. The liquid was then cooled rapidly for its solidification to form a new solid. The resulting solid was used as the cation precursor source. Its vapour pressure was controlled by controlling the temperature. The samples studied in this work and presented in Fig. 8 used the combinations of Zn and Cd weights presented in Table 3, in which, the column labelled by “x in the grown layer” represents the final molar composition of the solid Cd1-xZnxTe layers. The molar composition of these layers were evaluated by using the Vergard’s law a(x)=6.103x+6.482(1–x), so, the Zn molar fraction x was determined by the observed D-values obtained from the positions of the respective (004) diffraction peaks. The results of the measurement are summarized in Table 3, from which it can be observed that with the increase of the molar concentration x of the cation solid source, the molar concentration in the final grown solid layer was increased. The obtained concentrations using these three different solutions were: For the sample labelled as T1 was 92.5%, for the sample T5 was 90.9%, and for the sample T7 was 89.7%. From these results, it can be concluded that different molar concentrations of Zn in the layers of the ternary alloy can be obtained by preparing the solid precursor source with different Zn and Cd combinations. In the presented samples, the Zn molar composition was varied from 83.2% to 43.4%, but the variation of the molar composition of the grown layer was from 92.5% to 89.7%, respectively. FWHM of these peaks is above 3600 arcsecs, which can be considered large and attributed to mosaicity, combination with the nanometric dimensions of layers, and the generation of misfit defects.
Sample Solid sources composition (%) x in the grown layer (%) Zn Cd T1 83.2 16.8 92.5 T5 58.0 42.0 90.0 T7 43.4 56.6 89.7
Table 3. Cation source composition used for growing the ternary Cd1-xZnxTe
Characterization of Cd1-xZnxTe (0≤x≤1) Nanolayers Grown by Atomic Layer Deposition on GaSb and GaAs (001) Oriented Substrates
- Received Date: 2017-09-27
- Rev Recd Date: 2018-04-26
- Publish Date: 2019-06-01
- III-V substrates /
- atomic layer deposition (ALD) /
- defect generation mechanism /
- ternary alloy Cd1-xZnxTe /
- Zn and Cd mixture
Abstract: ZnTe, CdTe, and the ternary alloy CdZnTe are important semiconductor materials used widely for the detection of an important range of electromagnetic radiation as gamma ray and X-ray. Although, recently these materials have acquired renewed importance due to the new explored nanolayer properties of modern devices. In addition, as shown in this work they can be grown using uncomplicated synthesis techniques based on the deposition in vapour phase of the elemental precursors. This work presents the results obtained from the deposition of nanolayers of these materials using the precursor vapour on GaAs and GaSb (001) substrates. This growth technique, extensively known as atomic layer deposition (ALD), allows the layers growth with nanometric dimension. The main results presented in this work are the used growth parameters and the results of the structural characterization of the layers by the means of Raman spectroscopy measurements. Raman scattering shows the peak corresponding to longitudinal optical (LO)-ZnTe, which is weak and slightly redshift in comparison with that reported for the ZnTe bulk at 210 cm–1. For the case of the CdTe nanolayer, Raman spectra presented the LO-CdTe peak, which is indicative of the successful growth of the layer. Its weak and slightly redshift in comparison with that reported for the CdTe bulk can be related with the nanometric characteristic of this layer. The performed high-resolution X-ray diffraction (HR-XRD) measurement allows to study some important characteristics such as the crystallinity of the grown layer. In addition, the HR-XRD measurement suggests that the crystalline quality has dependence on the growth temperature.
|Citation:||Joel Díaz-Reyes, Roberto Saúl Castillo-Ojeda, José Eladio Flores-Mena. Characterization of Cd1-xZnxTe (0≤x≤1) Nanolayers Grown by Atomic Layer Deposition on GaSb and GaAs (001) Oriented Substrates[J]. Journal of Electronic Science and Technology, 2019, 17(2): 97-108. doi: 10.11989/JEST.1674-862X.70929112|