1. Introduction

Due to their great performance, infrared lasers have a wide range of uses today, including environmental gas detection, infrared imaging, space-free optical communication, laser ranging, and medical detection [1–4]. Mid-infrared laser applications are particularly prevalent. Quantum Cascade Lasers (QCLs) are unipolar lasers based on intersubband electron leaps in quantum wells. Unlike interband semiconductor lasers, which excite photons through electron/hole complexes between the conduction and valence bands, the wavelength of QCLs is determined by the energy level difference of the coupled quantum wells in the active region of the laser rather than by the material's intrinsic band gap [5–7]. The QCLs is a semiconductor laser that combines molecular beam epitaxy and energy band technology.

Since many researchers choose to concentrate more on the characteristics of QCLs lasers, such as high temperature, high power, high gain, and low threshold, the impact of energy band design and waveguide structure on QCLs performance has received more attention in scientific studies [8–10]. Potential well/barrier material development has received less attention than the more basic substrate epitaxy [11–13]. In this study, we thus build InP-based GaInAs/AlInAs ternary alloy potential well and barrier materials using molecular beam epitaxy (MBE) technology to examine the impact of growing temperature on quality of superlattice. The growth of superlattice structures to optimize the epitaxial process enables the intuitive analysis of the lattice features of the material, such as component and strain, while the optimized process conditions can be directly used for epitaxy in the active core region of the QCLs.

2. Material growth and methods

The experimental samples were grown on a Komponenten Octoplus 400 multi-chamber 3-inch wafer ultrahigh vacuum MBE system equipped with an arsenic (As) valve cracking cells and dual-filament thermal evaporation cells for indium (In) and gallium (Ga), a cold-lip filament infiltration cell for aluminum (Al). To grow a buffer epilayer with lattice-matched, the wafer was first restored at 500 °C for 10 min, and the As valve was opened when the substrate temperature exceeded 300 °C during the process. Then a 270 nm Ga_0.47In_0.53As buffer epilayer was grown on an n-type doped InP(100) substrate. The growth time of the buffer epilayer was 50 minutes, and its deposition rate was 0.9 Å/s. Next, the superlattice structure of strain-balanced is grown for 50 stages in the same way as in ref. [14]. The sample epilayer structure is shown in Fig. 1. The one-period superlattice structure consists of single-layer Ga_0.31In_0.69As and Al_0.64In_0.36As in order to easily calculate the composition and thickness of Ga_xIn_1-xAs and Al_yIn_1-yAs by HRXRD fitting [14,15].

 

Fig. 1. Schematic diagram of epitaxial growth structure.

Download Full Size | PDF

In this study, epitaxial wafers grown at various temperatures were created by changes in the superlattice's growth temperature. Buffer was grown first by keeping the temperature at 440°C, and then the superlattice structure was produced thereafter. The samples A, B, C, D, and E are epitaxial wafers that were grown at 440, 420, 400, 380, and 360 °C, respectively. The buffer epilayer was grown with a matched layer of Ga_0.47In_0.53As approximately 270 nm thick. Figure 2 illustrates the buffer layer's excellent morphology.

 

Fig. 2. Morphology of the buffer epilayer in AFM (The surface has a step shape, the scanning range size is 50 × 50 $\mu \mathrm{m}$, and the root mean square is 0.3 nm).

Download Full Size | PDF

The components and thicknesses of superlattice were calculated by fitting the peaks of HRXRD diffraction spectra. Before the fitting, the surface of the epitaxial wafers was first observed by atomic force microscopy (AFM) to measure the roughness of the morphology and to ensure that the surface was atomically monolayer. The purpose is to reduce the interfacial roughness scattering of carriers at low temperatures. Then, the strain of the lattice and the defects of the epilayer were characterized by the lattice diffraction peaks of the superlattice. Finally, the thicknesses and components of the potential well/barrier (Ga_xIn_1-xAs and Al_yIn_1-yAs) are calculated by fitting the diffraction peaks to the data. The optimal growth temperature is determined based on the RSM, lattice components and thicknesses. It also provides a reference for further research work by adjusting the V/III, Ga/In, and Al/In parameters to achieve the required lattice components and thicknesses for the target QCLs band structure.

3. Results and discussion

3.1. Roughness of surface

Scanning of the sample surface is carried out by the NCM module of the AFM, this procedure does not damage the sample is nondestructive, the AFM test results of samples A∼E are shown in Fig. 3, and the size of the scanning map is 50 × 50 $\mu \mathrm{m}$. According to the scale in the figure, sample A's maximum peak height of the surface morphology is approximately 1.5 nm, sample B's highest point is approximately 1.2 nm, sample C's highest point is approximately 4.5 nm, and samples D and E's highest points are respectively 3 nm and 6.5 nm. Therefore, for the maximum peak height of the surface (Sp), sample B has a better surface morphology. Regarding the arithmetic mean height of the surface (Sa). With the exception of sample D, which had an arithmetic mean height of 0.4 nm, every other sample surface's arithmetic mean height (Sa) was 0.3 nm, making the mean height essentially constant. The root-mean-square roughness of the surface morphology (Sq), from samples A to E are essentially 0.4 nm. The difference between samples A and D, Sq of 0.3 and 0.5 nm, respectively, is 2 Å smaller than the size of one atom.

 

Fig. 3. AFM test of sample surface (A∼E correspond to sample A∼E respectively, B-3d is a three-dimensional scan of sample B).

Download Full Size | PDF

Judging from the Sq and Sp of the sample, it can be said that sample B's surface morphology has a good surface morphology due to its narrow fluctuation range and minimal roughness change of the surface peak height. Figure B-3d shows the three-dimensional scan of sample B. It is evident that the surface morphology is largely flat and devoid of visible protrusions and depressions. Sample B is devoid of the same white “bright spot” as sample E, according to observations of the sample surface. This “bright spot” has a peak height of roughly 6 nm. As a result, sample B is thought to have good surface morphology.

3.2. Analysis of superlattices

In this study, the lattice quality of the epitaxial wafer was analyzed by a high-resolution X-ray diffraction test, and the diffraction pattern of the sample is shown in Fig. 4. SLS-0 and SLS-1 represent the zero-level peak and the first-level peak of the superlattices, respectively, and the substrate's diffraction peak is denoted as Sub. The zero-level peak of the superlattice coincides with the position of the substrate peak, which shows that the growth of the superlattices is coherent growth with the substrate. As the growth temperature lowers, the distance between the first-order peak of the superlattice layer and the substrate peak gradually increases and shifts to the left. Compared to the first-order peak of sample B, the first-order peak of sample E has increased by about 200 arcseconds. In addition, the diffraction peak of sample E also shows the phenomenon of “split peaks”, which may be caused by changes in the composition or thickness of the superlattice. The distance between the Buffer and Sub. peaks of samples C, D, and E gradually increased, which is directly related to the decreasing of the Ga_xIn_1-xAs comp. X in the Buffer layer.

 

Fig. 4. HRXRD diffraction spectrum of the sample. (The left inset shows the local diffraction peak of Sub/SLS-0, and the right inset shows the local diffraction peak of SLS-1).

Download Full Size | PDF

The components and thickness of the samples were calculated by AMASS, as shown in Fig. 5. First, the model was established based on the growth structure of the epitaxial wafer. Then, the test's crystal orientation, the number of stages, and buffer and monolayer potential well/base thickness ranges were set. At last, the precise components and thicknesses of each sample are calculated by optimization iterations. Each peak in the sample matches the measured diffraction peaks precisely during the calculation. The estimated component and thickness are more accurate with greater matching overlap. The component and thickness results for each of the samples A, B, C, D, and E were obtained by calculation, as shown in Fig. 6 - right.

 

Fig. 5. Fitting calculations for samples B and E.

Download Full Size | PDF

 

Fig. 6. Buffer composition x for samples A∼E (left), Composition and thickness of potential well/base (right).

Download Full Size | PDF

The change in the comp.X of Ga_xIn_1-xAs buffer may be of more interest than the thickness of the material. Because of growth in a rich-As environment, Ga_xIn_1-xAs can be thought of as a combination of GaAs and InAs, so the variation of the component is more influenced by Ga and In. For the Ga_xIn_1-xAs comp.X of the buffer layer, there is no significant change between sample A and sample B. Sample E shows a significant decrease in comp.X compared to sample B, as shown in Fig. 6 - left. The growth of the buffer layer did not change factors such as the growth temperature; however, it led to a decrease in comp.X, which may be related to the cell molecular beam flow of the Ga. With the use of the thermal evaporation source cell during growth, the Ga beam current decreases accordingly, while In is more likely to evaporate to form InAs, so the influence of the Ga cells beam current leads to a decrease in the Buffer component, which also explains the decrease in the angle of the Buffer diffraction peaks for sample E in Fig. 4. Therefore, when growing many slices or single slices with a multi-stage thickness of epitaxial wafers, the small change in the molecular beam current of the cell has a significant impact on the material growth, which influences the growth quality of the epitaxial wafers. In order to maintain the stability of the growth beam current, the variation of the molecular beam current can be adjusted by stage compensation, while the beam current data of each cell is tested before the material growth.

For the growth of superlattice Al_0.64In_0.36As layers, the single-stage superlattice structure consists of a layer of GaInAs and a layer of AlInAs. Compared to GaInAs, which is the potential well layer in the superlattice, the component of the potential barrier layer AlInAs is less affected by the change in growth temperature, and the thickness will change with the change in comp. X. component's average value for all samples is roughly 0.635, which is just slightly less than the lattice-matched component of 0.64. The superlattice barrier layer AlInAs shows an average thickness of roughly 3.38 nm, with samples C and E having significantly different thicknesses of 0.29 nm. The difference is slightly smaller than the individual atomic sizes. The reason for the change in thickness of the barrier layer is related to the change in comp. X. The lattice constant of AlAs is smaller than that of InAs, and a decrease in comp. X indicates a decrease in AlAs, so the thickness of the barrier layer increases. Temperature has less of an impact on the formation of Al_0.64In_0.36As because the barrier layer comp. X is comparatively constant.

For the growth of superlattice Ga_0.31In_0.69As layers, the components of Ga are more variable than those of Al by temperature. The desorption of GaAs at high temperatures may be the cause of the decrease in comp. X of sample A in comparison to sample B [16]. In addition, desorption of InAs also exists, but since In participates in both the potential well GaInAs and the barrier AlInAs when superlattices are growing and a rich-As environment is present, an excess of InAs is produced, which lowers the Ga comp. X. The discrepancy between the thicknesses of samples A and B can also be explained by the fact that InAs has a higher lattice constant than GaAs and that as InAs content rises, the GaInAs potential well layer thickens. The component X of samples D and E decreases compared to sample B because the growth of GaInAs at low temperatures produces defects such as lattice inversions and vacancies, resulting in a decrease in Ga comp. X [17,18].

The deposition rates of superlattice (potential well GaInAs and barrier AlInAs) are shown in Fig. 7. The average deposition rate of the barrier AlInAs layer is about 1.52 Å/s, and the population variance coefficient is 0.002. The average deposition rate of the GaInAs layer is about 0.90 Å/s, and the population variance is 0.0001. Compared with the potential barrier layer, the deposition rate of the potential well layer has less fluctuation. The rate of deposition in the barrier layer is 1.7 times higher than in the potential well layer. The barrier layer has a higher deposition rate, which may be related to the composition of the material. Since the targeted component of Ga was at 0.31 and the component of Al was at 0.64, a higher deposition rate for the barrier layer of AlInAs is required. The deposition rate of GaInAs is relatively stable, which may be related to the physical properties of Ga. Ga has a stronger metallicity than Al, and it is easy to provide a significant volume of Ga beam, which leads to a more constant rate throughout the growth phase. The sources of group III fluxes are responsible for layer composition and growth rate [19].

 

Fig. 7. Deposition rates of GaInAs and AlInAs.

Download Full Size | PDF

4. Conclusions

In summary, this study explores the effects of temperature on the growth of a quantum cascade laser's core barrier/well layer. Based on the analysis using high-resolution X-ray diffraction spectroscopy and AFM experiments, it has been found that the superlattice barrier layer AlInAs has a little tiny temperature effect, and its components are more stable, with the thickness changing with small changes in the components. Temperature has a significant impact on the potential well layer GaInAs, with overly high or low temperatures leading to a decrease in the comp.X of Ga_xIn_1-xAs. Different factors that contribute to the component's reduction are not the same; high temperature growth causes GaAs desorption, while low temperature growth produces lattice defects. The optimal growth conditions for determining the temperature for growing the complete QCLs potential barrier/well layer structure are finally indicated in this work.