Thermal annealing effects on the physical properties of GaAsBi/GaAs/GaAs:Si structure

GaAsBi is a new material called a highly mismatched alloy that has drawn attention regarding its special physical properties. The alloying of the GaAs matrix by Bi atom gives rise to a huge restructuring of the band structure. A rapid shrinkage in the bandgap energy and a splitting of the spin–orbit interaction band are noted. But the synthesis of this material requires unusual growth conditions in order to avoid the appearance of droplets on the surface and a native defects due to the non-stoichiometry. Consequently, an improvement of the physical properties is required to be used in device applications. In this perspective, we report an investigation of the effect of thermal annealing on the GaAsBi/GaAs/GaAs:Si structure. Photoreflectance, Spectroscopic ellipsometry are used to study the optical characteristics of this structure. High Resolution X-Ray Diffraction and Atomic Force Microscopy are employed as structural techniques for investigation.


Introduction
In the previous years, the bismide semiconductor alloys have attracted major attention due to their possible applications in electronic and optoelectronic devices [1-3]. The main advantage of using III-V-Bi semiconductors in the development of the devices is their low sensitivity to temperature fluctuations. In fact, using these materials in devices can lead to a reduced temperature dependence which is attractive for temperature insensitive lasers, optical amplifiers widely applied in telecommunications [4,5]. Another vital property of these materials is the opportunity to decrease the band gap of the III-V host matrix by adding a small amount of Bi. The GaAsBi alloy is one of these families that have unusual properties compared with classical alloys such as InGaAs and AlGaAs. Indeed, it is reported that the band gap shrinks by 38-83 meV per percent of Bi [6,7]. This band gap reduction depends on the growth technique. It was stipulated that Bi incorporation rearranges the band structure of GaAs which gives rise to a large bowing parameter and to a valence band splitting. This opens up attractive new possibilities for efficient photonic devices, such as near-and mid-infrared lasers without losses due to Auger recombination. Recently, there have been a number of attempts to improve the performance of some proposed devices basing on the engineering of quantum nanostructure [8,9]. However, several complexities associated with producing high quality films 1 3 have been announced by many research groups [7,10,11]. Experimentally, GaAsBi has been developed as a bulk or quantum structure using metalorganic vapor phase epitaxy (MOVPE) or molecular beam epitaxy (MBE). In fact, any deviation from GaAsBi optimal growth conditions can lead to the appearance of rough surfaces and the inevitable emergence of defect centers. Specifically, an increase in the Bi content or the thickness of the layer can multiply the density of Bi droplets on the surface. It is also reported that as-grown layers of GaAsBi have a low efficiency of luminescence due to the localized states that affect the carrier recombination processes [12][13][14][15][16]. Generally, a low temperature is required for the successful synthesis of these materials, which generate different types of defects. In spite of the significance of the GaAsBi semiconductor, the physical properties of this material are partially mastered so far. Supplementary efforts are required to achieve a synthesized material with a large range of Bi composition and acceptable physical properties. Several procedures, such as the step-by-step epitaxy process [15,16] and thermal annealing [17] were previously used to improve the properties of this alloy. In some essays [18,19], it was noted that thermal annealing can have an effect on the surface morphology and improve its optical properties. Nevertheless, the effect of thermal annealing on the GaAsBi alloy is not completely achieved and depends on numerous factors. Consequently, further studies are still needed in order to comprehend the effect of the post-growth thermal annealing on the physical properties of this material.
In this paper, we investigate the effect of the thermal annealing on a GaAsBi/GaAs/GaAs:Si structure by using photoreflectance, ellipsometry and high Resolution X-Ray Diffraction (HRXRD). The experimental data are analyzed based on different theoretical physical concepts.

Sample preparation and characterization techniques
The sample consists of a layer of relatively thin GaAsBi deposited on an intermediate GaAs layer called a buffer layer with a thickness of about 100 nm on a Si-doped GaAs (001) oriented substrate. The growth is made by the MOVPE technique in a horizontal reactor using trimethylbimuth (TMBi), trimethygallium (TMGa), and arsine (AsH 3 ) as sources. A photo of the essential parts of the used system is given in Fig. 1. The precursor bubblers of TMGa and TMBi are kept in thermo-baths at a regular temperature of 22 °C and 0 °C, respectively. High purity H 2 as a carrier gas has been applied in every part of the growth system. The growth temperature is instantaneously measured by a thermocouple inserted into the graphite susceptor inside the reactor. First, the GaAs substrate is cleaned by thermal desorption in a mixture flow of H 2 and AsH 3 at a temperature of 700 °C for 10 min. After that, the temperature is decreased and stabilized at 600 °C. This growth temperature is optimal to grow the GaAs buffer layer using V/III ratio of 30. Then the active layer of GaAsBi is epitaxied with V/III ratio of 9.5, a TMBi flow of 0.2 µmol/min, and a growth temperature of 420 °C [10,20]. In all cases the total flow rate in the reactor is kept at 3 l/min. Our setup ensures data acquisition and devices command via a PC. The sample as grown and annealed is characterized by PR, spectroscopic ellipsometry, and HRXRD. The annealing is made in the MOVPE reactor in the presence of AsH 3 (6.25 cc/min) and 3 l/min H 2 flow at a temperature of 600 °C.
The photoreflectance technique was achieved with a 5 mW argon laser beam as a modulated source. The polychromatic source is a tungsten halogen lamp with 250 mW. Two monochromators and a lens system are arranged to detect the signal by a silicon detector. Spectroscopic ellipsometry (VASE) is used in the energy range of 0.7-3.35 eV at an incidence angle of 60 • . AFM surface topography is carried out in the tapping mode by a digital D3100 Nanoscope instrument. HRXRD spectra are recorded by a D8 discover diffractometer characterized by a wavelength λ CuKα1 = 1.54056 Å in ω/2θ scan.

The contribution of the buffer layer and the substrate
In order to localize the effect of the buffer layer deposited on Si-doped GaAs substrate, we present in Fig. 2 the response of photoreflectance measured on this structure. The photoreflectance response is classified into two regimes: weak or strong electric field, according to the electro-optical energy ratio. In the case of stacking layers, the existence of these two field regimes is possible. This is related to several parameters such as the thickness of the deposited layer, the excitation beam's penetration length, and the electric field's intensity at the interface. This regime is called the mixed regime.
Surprisingly, we note the absence of FKO structure, which indicates the weak magnitude of the electric field. Contrarily to the simple case of GaAs/undoped GaAs, the determination of the structure parameters relative to GaAs/Si-doped GaAs is determined by using the TDFF It is interesting to remark that no other structures have been observed. This fact eliminates any response in the GaAsBi structure far from the energy gap. Fig. 1 Photo of MOVPE setup a pneumatic valves for gas control and distribution in tubing system b the reactor shape and heated susceptor

Study of physical parameters of GaAsBi/GaAs/ Si-doped GaAs as grown structure
In Fig. 3, we report the PR measurement on as grown GaAsBi structure as a function of the temperature. It should be noted that measurements were taken using 10-25 K as a step of temperature, but for more visibility of the figure, we have ignored some temperatures. We can easily remark that the change in the temperature has an intensive effect on the shape of PR signal. Indeed, the decrease of the temperature appears a negative peak around 1.48 eV. According to Fig. 3, this peak could be attributed to the contributions of the GaAs buffer layer and the Si-doped substrate. In fact, for temperatures up to 200 K the GaAsBi contribution is dominant. Such a situation is among the most complex and requires the use of a mixed model. This regime is characterized by the existence of two behaviors of the field in the same heterostructure. This overlap makes it very difficult to apply the two previous methods. For this reason, several authors have developed theoretical models to simulate the PR [21] and electroreflectance (ER) [22] responses. The development of this model is based on the optical properties of the studied materials, such as the refractive index, the extinction coefficient, and the complex dielectric function.
In the vicinity of the critical points, the contribution of the perturbation in the dielectric function Δε (= Δε 1 + iΔε 2 ) is given by: where H(z) is the Aspnes electro-optical function which depends on complex Airy functions , E g is the energy of the conduction band critical point (gap energy), Γ 0 is the broadening parameter expressed in unit of energy and ℏθ is the electro-optical energy related to the electric field F by (ℏ ) 3 = e 2 ℏ 2 F 2 2μ , µ is the inter-band effective mass. The knowledge of ε and ∆ε for each layer permits to calculate the reflectance R with and without variation of the electric field (R on and R off respectively). The photoreflectance response (∆R/R) can be related to the dielectric function by the expression of Séraphin [23]: where α and β are the Séraphin parameters.
The combination of all previous formulas gives the photoreflectance as: In Fig. 4, for example, we report the full range of measurement in ambient temperature of a grown structure with data fit by using this model. A good agreement between experimental and fit results is observed.
(1) When the temperature increases, the gap energy of such semiconductor material decreases due to two main phenomena: • The thermal expansion of the crystal lattice causes an increase in the interatomic distance, so a decrease in the gap is noted. • The electron-phonon interaction is a thermally activated phenomenon. This interaction is dominant at high temperatures, from about 100 K, and reduces the bandgap.
Two formulas are usually used to analyze these effects. The dominantly used one is that of Varshni law [24] which is given by: where E g (0) is the bandgap energy value at T = 0 K. , and are tow adjustable parameters. The experimental results are fitted by this formula, and the result is plotted in Fig. 5. For comparison, we report the result relative to GaAs material. All fit parameters are summarized in Table 1. It is clear that the incorporation of bismuth in the GaAs host matrix causes a significant decrease in bandgap energy. Indeed, at ambient temperature a reduction of about 48 meV/% is observed. This value is in good agreement with the values determined for samples grown by the MOVPE technique [10]. But, a higher rate of about 88 meV/% is reported for samples grown by MBE [24]. The difference in these rates has been explained by several factors, including those related to the growth method, such as the intrinsic mechanism of bismuth incorporation in matrix. For temperatures above 200 K, however, a linear dependence of bandgap energy with temperature is observed. This decrease is characterized by a slope of about 0.25 meV/K which is 75% less than that determined for GaAs. Such a result has vital interesting for implication of the GaAsBi in devices with less sensitivity to temperature fluctuations.

.
The second is that representing the Bose-Einstein formulation [27]. In this context, gap energy is expressed as: where E g (0) = E B − a B , θ B characterizes the average phonon temperature. a B is a coefficient relative to the electron-phonon interaction strength. The details of the origin of this expression are widely reported in the literature [26].   Figure 6 represents the experimental data of bandgap energy with temperature fitted by using this formulation. This analysis permits determining the empirical parameters, whose values are illustrated in Table 2.
It can be seen that this value changes when incorporating the bismuth in the GaAs host matrix. The explanation of this variation can be associated with the interaction between phonon-electron interactions.
In general, the temperature shift of the bandgap contains two contributions: that associated with thermal expansion and that related to electron-phonon coupling effects [27].

Study of physical parameters of GaAsBi/GaAs/ Si-doped GaAs structure after annealing
Similar procedures for PR measurements have been used for the same structure annealed at 600 °C in mixture gas (H 2 + AsH 3 ) inside an MOVPE reactor. In Fig. 7, we report the results of PR measurements as a function of the temperature. A qualitative observation of curves shows comparable shapes to those measured for as-grown structure. A small shift in PR peaks is observed. In order to investigate the effect of the annealing, we have applied the same subroutine to extract the physical parameters of the annealed structure.
As an example, the results of calculations and experimental data are reported in Fig. 8. The spectrum corresponds to the data measured at 300 K. The details of the curve show the presence of harmonic oscillations Bandgap energy (eV)

Temperature (K)
As grown Fig. 6 Bandgap energy data of as grown GaAsBi structure dependence with temperature fitted by Bose-Einstein formula  Fig. 9 combined with shoulders indicating the mixture regime. A good agreement between experimental data and theoretical fit is observed. We have applied the Varshni's model to adjust the experimental data and we report the fit parameters in Table 1.
The Varshni and Bose-Einstein models have been applied to fit the experimental data of the bandgap energy variation as a function of the temperature, as illustrated in Figs. 9 and 10. The fit parameters are reported in Table 2.
Basing on the previous analysis, we can conclude about the effect of annealing on optical properties of GaAsBi: • The annealing at 600 °C, increases the composition of the material. This fact is justified by the increase in bandgap energy. • The annealing does not enhance the non-sensitivity of the bandgap energy to the temperature. To conclude on this hypothesis, we show in Fig. 11 the translate curves of bandgap energies relative to the two investigated structures and compare them to GaAs. We can easily remark that the as-grown structure is less sensitive to the temperature variation. An annealing of the structure limits this behavior.

Study of spin-orbit interaction
Figures 12, a and b present the PR spectra recorded in the vicinity of the spin-orbit transition as a function of temperature in the range 20-300 K for the two studied samples. It should be noted that when the temperature increases, the PR peaks shift towards low energies but the shape of the spectra remains almost unchanged. This indicates the low sensitivity of this alloy to temperature variation for the studied transition. The transition energy (E g + Δ 0 ) is deduced by adjusting the experimental data by the TDFF function as plotted in Fig. 12. The values of (E g + Δ 0 ) issuing from this analysis are plotted in Fig. 13. It is observed that the energy E g + Δ 0 for both compositions decreases when the temperature increases. The experimental data of E g + Δ 0 as a function of temperature are fitted by Varshni's formula. The fit parameters are reported in Table 3.

3
For comparison, we have inserted Varshni's plot of GaAs data. For the temperature range 200-300 K, the figure shows that when the composition increases, the relative variation of the energy Eg + Δ0 decreases. In this range, the variation can be approximated by linear behavior. The analysis of the results proves that the slopes are 0.27 meV/K, 0.18 meV/K and 0.12 meV/K for GaAs, as grown structure, after annealed structure, respectively. These interesting results added to the atypical properties of the GaAsBi alloy make it of great interest for infrared and spinotronic applications.

Spectroscopic ellipsometry characterization
To determine the optical properties of the two structures, we have used a theoretical approach to fit the experimental data using a simple model composed of the substrate material and a thin oxide overlayer. This methodology is generally adopted as a correction to take account of the spontaneous oxidation of GaAs. A thin layer of GaO3 or GaO2 will be formed when GaAs is permanently exposed to air. The results for the two structures are illustrated in Fig. 14.
The calculated fit permits determining the real ( 1 ) and imaginary ( 2 ) parts of the dielectric function. The analysis of the change of 1 and 2 (not shown here) gives attractive  Fig. 13 Variation of the spin-orbit E g + Δ 0 coupling for as grown and annealed GaAsBi structures compared to GaAs layer For the harmonic oscillator approach, the dielectric function is given by [28] The amplitude A j , the energy E j and broadening Γ j are the jth CP parameters. n the number of conventional CP.
The second derivative of dielectric function is written as   Fig. 15. The parameters allowing the best fits are illustrated in Table 4. These results agree with those determined by PR measurements.

High resolution XRD characterization
In order to determine the structural properties of the asgrown structure, the HRXRD spectrum has been recorded on (004) reflection, as presented in Fig. 16. The plot is dominated by two peaks. The right peak is associated with (004) GaAs reflection, and the second one is relative to GaAsBi response. The angular separation between the two peaks denoted is directly related to the lattice parameter (a GaAsBi ) in a perpendicular direction through Brag's law: where d hkl is the inter-reticular distance. In our case, B is the Bragg angle relative to GaAsBi layer given by Assuming GaAs = 33.026 • (or lattice parameter of GaAs a GaAs = 5.6533 Å), we can deduce the lattice parameter (a GaAsBi ) of GaAsBi layer. Vegard law is used to estimate the Bi composition.
Our calculation estimates the Bi content to 3.7%. We can also observe some harmonic oscillations in the base of the peaks. These oscillations are named Pendellösung oscillations, a synonym for the good interface between the GaAsBi layer and the substrate. The period of oscillations permits us to determine the thickness (d) of GaAsBi layer using following formula: where λ is the X-ray wavelength, θ is the Bragg angle for recorded reflection. Δω av is the average period of Pendellösung oscillations. For such reflection, the average period is calculated from Où ∆ω i is the value of the ith period calculated for the n observed periods.
Our calculation gives a thickness of 43 nm. The surface morphology of the structure is scanned by using Atomic Force Microscope (AFM). As shown in the inset of Fig. 16, the surface shows spherical islands. As reported in the literature [20] these islands are Bi droplets. The density and the size of these defects depend on growth conditions and the thickness of the active layer. The appearance of these droplets represents a major (14) x = a GaAsBi − a GaAs a GaBi − a GaAs handicap for using this material in device applications. It will be beneficial to find a solution to this problem.
The annealing of the structure shows an evaporation of the droplets, as indicated in the inset of Fig. 17. Some holes in the surface are observed after annealing the structure up to 600 °C. It is noted (not reported here) that droplets evaporate when temperatures reach 580 °C. This fact can represent a solution for cleaning the GaAsBi surface layer from the Bi droplets. Now, more effort is focused to reduce the density and the size of appeared holes after annealing.
On the other hand, HRXRD is applied to characterize the structure after annealing, and the spectrum is reported in Fig. 17. Two differences are remarked when comparing this spectrum to that of Fig. 16: -The angular separation between the two peaks () increased, indicating an increase in the Bi composition. Indeed, our calculation of the new lattice parameter gives a 4.8% Bi content. This fact is probably due to a new redistribution of Bi atoms inside the host matrix. -The period of Pendellösung oscillations remains constant but is more damped. This attenuation of oscillations is probably related to a slight increase in roughness due to the evaporation of droplets and the appearance of the holes. Our calculation indicates a similar thickness for the layer (46 nm).

Conclusion
Thin layer GaAsBi grown on buffer layer and Si-doped substrate by MOVPE has been studied as grown and after annealed at 600 °C. Optical and structural characterizations of this structure by PR, HRXRD, and AFM techniques have been carried out. It should be noted that the incorporation of Bi in the GaAs matrix results in a significant reduction of Bangap energy and spin-orbit interaction. Such reduction is mainly due to the intrinsic properties of Bi atom. Also, the bandgap energy of GaAsBi is less sensitive to temperature variation compared to GaAs. This fact is confirmed by using different models to describe this behavior. On the other hand, the thermal annealing of GaAsBi increases the Bi composition, leading to more shrinking in bandgap energy and spin-orbit coupling energy. In spite of the increase in composition due to thermal annealing, the sensitivity of bandgap energy to temperature remains the same. A slight change is observed for spin-orbit coupling energy.
Author contributions BOA: Ellipsometric measurements, HHHA: Ellipsometric analysis of the data results, IZ: Photoreflectance measurements and data results, HF: Growth of the samples and HRXRD measurements, AR: Supervisor of the work and writing of the paper.
Funding The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.
Data availability Data will be made available on reasonable request.

Declarations
Conflict of interest On behalf of all authors, there is no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.