3.a. In Situ Reflectance Measurement Comments
A semiconductor laser operating at a wavelength of 880 nm was used to monitor and control the growth steps and conditions during the growth. In situ reflectance measurement is an important measurement technique that gives valuable parameters such as growth rate, surface states, layer thickness during the growth [24].
We show in fig .1 the in-situ reflectance and growth temperature versus growth time for all samples. The In concentration of the InxAl1−xAs layer is indicated as xln. The temperature was increased to 555°C and the InP buffer layer was grown at this temperature. There was no oscillation observed during the buffer growth because of the same refractive index of substrate and epilayer.
Then, the growth temperature was reduced to 540°C and InxAl1−xAs layers with different In concentration were grown. As can be seen from the figure that oscillations were observed while growing InxAl1−xAs due to the refractive index difference between epilayer and InP buffer. The growth rate and the total film thickness were determined by using the interference patterns. The equations used to determine the film thickness and growth rate are given below:
$$\varDelta d=\frac{\lambda }{2n} and r=\frac{\varDelta d}{T}$$
1
Where Δd is an oscillation thickness, λ is the wavelength of laser light, n is the refractive index of the layer, r is the growth rate, and T is the time for a complete oscillation.
High growth rate causes short periods of reflectance oscillations, while low growth rate causes long periods. The growth rate of the S1, S2, and S3 are 0.394, 0.422, and 0.444 nm/s, respectively. The phase and amplitude of the reflectance oscillations depend on the wavelength of the incident light, the optical constants of the materials, and the thickness of the growing layer. In addition, surface roughness is another important parameter that affects the reflectance curve [25].
During the three sample growth, the wavelength of the excitation light was heft constant, and the growth rate was really identical. However, a change in In content induce a change in the refraction index of the InxAl1−xAs layer.
The thickness of the layers and the wavelength of the incident light are the same, but with the increase of the In fraction, the refractive index of the InxAl1−xAs layer will increase, i.e. approach the refractive index of InP. As the difference between the refractive index of the buffer layer and InxAl1−xAs layer decreases, the amplitude of the oscillations will also decrease. Therefore, the surface situation cannot be interpreted from the amplitudes of the oscillations, but the fact that the reflectance intensity did not decrease for all of the samples is an indication of the quality of the surface.
3.b. HR-XRD results
Table 1
HR-XRD results of the InxAl1−xAs/InP QCL structures with different In alloy composition.
Sample | Indium Contents (%) | Lattice (Å) | Lattice mismatch (%) | FWHM (degree) | Dislocation density (109 cm− 2) |
---|
S1 | 37.1 | 5.8078 | -1.01% | 0.093 | 0.173 |
S2 | 33.8 | 5.7946 | -1.26% | 0.110 | 0.243 |
S3 | 29.8 | 5.7787 | -1.53% | 0.142 | 0.408 |
HRXRD measurements were employed to determine the indium alloy composition (x) of the grown structures. Figure 2 exhibits the recorded 2θ/ω X-ray diffraction spectra of the three InxAl1-xAs/InP structures. It is evident from Fig. 2 that the high-intensity peaks originate from the InP substrate, while the remaining peaks emanate from the InxAl1-xAs layers. Due to the increasing Indium alloy composition within the layer, the peaks associated with the InxAl1-xAs layers are further separated from the InP substrate peaks. The peak separation angle can be utilized to ascertain the In composition within the alloy. Hence, the xIn values were determined as 37.1%, 33.8%, and 29.8% for S1, S2, and S3 structures, respectively, by employing the formula provided in Reference [26] and adapting it to the InxAl1-xAs/InP structure.
$$\frac{\varDelta a}{a}= -\frac{\varDelta \theta }{\text{tan}\left({}_{B}\right)}$$
2
Where \(\varDelta \theta\) is the measured angular spacing between the epitaxial layer and substrate diffraction peaks and \({}_{B}\) is the Bragg angle for the substrate. All samples exhibited InxAl1−xAs peaks positioned to the right of the InP peak, signifying that the film was under tensile strain, leading to an expansion of the out-of-plane lattice parameter. Additionally, the FWHM values of XRD peaks for the S1, S2, and S3 structures were determined to be 0.093°, 0.110°, and 0.172°, respectively. The QCL191 structure, with the lowest FWHM value (0.093°), is the closest to the lattice-matched InxAl1−xAs/InP structure. Utilizing the FWHM values, it is possible to approximate the dislocation density (Ndis) in such epitaxial layers according to the following formula [27–28]:
$${\text{N}}_{\text{d}\text{i}\text{s}}=2 \frac{{\left(\text{F}\text{W}\text{H}\text{M}\right)}^{2}}{{9\text{a}}_{0}^{2}}$$
3
Where FWHM is expressed in radians and a0 represents the lattice constant of the epitaxial layer, determined according to Vegard's Law. Table I summarizes the crystal structural parameters, including the lattice parameter (a), dislocation density (Ndis), and strain (e), for each sample. Notably, the Ndis and e values of S1 and S2 are lower than those of S3. A mere 7.3% reduction in the Indium content results in a significant increase in dislocation density, approximately 135%. Additionally, thickness fringes are clearly visible in Fig. 2 for these two samples, indicating high crystalline quality and sharp interfaces between the InxAl1−xAs epilayer and the InP substrate. Based on these results, it can be concluded that the In0.371Al0.629As epitaxial layer of sample S1, with the smallest FWHM, exhibits the minimum dislocation density.
3.c. Raman results
Figure 3 presents typical Raman scattering spectra at 300 K in InxAl1−xAs of various xIn. The disorder-activated longitudinal acoustic (DALA) phonon of InxAl1−xAs with the middle composition range typically appears in the broad Raman band from 90 to 150 cm-1 [10, 29]. This observation is attributed to the interactions between atomic clusters that occur during the growth of the epitaxial layer, leading to a disordered crystal arrangement at the interface. An increase in the degree of disorder (i.e., the alloy phase separating into indium-rich regions and aluminum-rich regions) contributes to an increase in dislocation density [10, 29].
To accurately determine the peak frequencies of the various Raman modes, the Raman spectra were deconvoluted using a Lorentzian shape function. Figure 3(b) depicts the fitting curves corresponding to the Raman data of sample S1. The five identified peaks, which are located at 190, 250, 310, 355, and 269 cm-1, correspond to LO-InAsP, LO-InAs, TO-InP, LO-InP, and LO-AlAs, respectively. The same spectral features were observed in samples S2 and S3. Notably, as the Indium content increases, the lattice parameter of the InxAl1−xAs alloy shifts between that of InAs and AlAs. This phenomenon is reflected in Fig. 3(a), where the LO-InAs, LO-AlAs, and LO-InAsP modes exhibit a slight shift towards lower wavenumbers with increasing In content. This shift is attributed to the tensile strain induced by the lattice mismatch between InP and InxAl1−xAs [29–30], and it aligns well with the HR-XRD results.According to previous studies [27, 31], the relationship between the shift of InP-like LO frequency and the residual stress \(\left(\mathcal{R}\right)\) in the InxAl1−xAs epitaxial layer is given by:
$$\mathcal{R}=\frac{{2{{\omega }}_{0}^{LO}\varDelta {\omega }}^{LO}}{\left({\text{S}}_{11}+{2\text{S}}_{12}\right)\left(\text{p}+2\text{q}\right)-\left({\text{S}}_{11}-{\text{S}}_{12}\right)\left(\text{p}-\text{q}\right)}$$
4
$${\varDelta {\omega }}^{\text{L}\text{O}}={{\omega }}^{\text{L}\text{O}}-{{\omega }}_{0}^{\text{L}\text{O}}$$
5
$${{\omega }}_{0}^{\text{L}\text{O}}=7.096{x}^{2}-78.5x+404.1$$
6
where, p and q are the optical phonon deformation constants, S11 and S12 are the elastic compliance constants, ωLO is the measured AlAs-like LO frequency in epitaxial layer, \({\omega }_{0}\) is the AlAs-like LO frequency in the ideal strain-free bulk InxAl1-xAs alloy as a function of composition x. All parameters and values of Eq. 4 are summarized in Table 2.
Table 2
The Raman results of the InxAl1−xAs epitaxial layers with different In content.
Sample | xIn (%) | \({\varvec{\omega }}_{0}^{\mathbf{L}\mathbf{O}}\left({\varvec{c}\varvec{m}}^{-1}\right)\) | ωLO \(\left({\varvec{c}\varvec{m}}^{-1}\right)\) | \({\varDelta \varvec{\omega }}^{\mathbf{L}\mathbf{O}} \left({\varvec{c}\varvec{m}}^{-1}\right)\) | \(\mathcal{R}\) \(\left(\varvec{G}\varvec{P}\varvec{a}\right)\) |
---|
S1 | 37.1 | 375.95 | 366.33 | -9.62 | 1.12 |
S2 | 33.8 | 378.37 | 368.05 | -10.32 | 1.21 |
S3 | 29.8 | 383.33 | 370.91 | -12.42 | 1.48 |
Table 2 compares the residual strain of the InxAl1−xAs epitaxial layer with varying Indium content. With decreasing Indium content, the residual strain in epitaxial layers increases, and sample S1 exhibits the lowest residual strain value, indicating the highest crystalline quality. This observation corroborates the findings of our previous HR-XRD measurements.
3.d. Photoluminescence spectroscopy:
Figure 4 compares the normalized PL intensity of InxAl1−xAs samples with varying In contents, measured at 10 K with a laser intensity of 0.1 W/cm2. A noticeable blue-shift (approximately 20 meV) in the InxAl1−xAs emission is observed as the In content decreases. The broadening of the PL peak (63 meV) with decreasing In content indicates a deterioration in optical quality, attributed to alloy phase separation (clustering effect) and stacking faults in the InxAl1−xAs layers [10, 11]. These results suggest that clustering has a minimal impact on sample S1, and higher In content leads to improved optical quality.
Two additional transitions appear at energies lower than those described in the previous section for bulk processes. The first transition at 1.37 (± 2) eV is associated with the normal interface or (e–A) transition in InP material [32]. The last transition, peaking at 1.17–1.24 eV, likely originates from a thin InAsP layer at the InP/ InxAl1−xAs interface, resulting in a higher wavelength (lower energy) luminescence band [10, 32]. It can also be inferred that S3 is more affected by interface roughness fluctuations and compositional disorder.
We conducted a detailed examination of the three samples through excitation-dependent PL measurements at 10 K in the 1.1–1.3 eV range. Figures 5a-c present the PL spectra of the three samples with decreasing excitation power. For clarity, each spectrum is normalized based on the maximum PL intensity and shifted upward. The integrated intensity, FWHM, and PL peak energy of the three samples are extracted and depicted in Figs. 5d-f as functions of excitation intensity.
As the power intensity increases from 1 to 46 µW, the peak energy of all samples exhibits a blue shift (see Fig. 5f), indicating a type-II band alignment and the formation of triangular wells due to band bending at the interface. This power dependence can be associated with localized states arising from alloy fluctuations and state filling effects, phenomena observed in systems displaying similar power dependencies [33–35]. Upon close inspection, the power dependence fit demonstrates a logarithmic relationship, consistent with systems exhibiting the spatial separation of charges expected from alloy fluctuations at this graded interface. As excitation increases from 1 to 46 µW, the spectrum gradually becomes symmetric, eventually displaying a perfect Gaussian peak. Simultaneously, the FWHM narrows from 61 to 50 meV for S3, from 21 to 27 meV for S2, and from 24 to 27 meV for S1, as shown in Fig. 5d. Subsequently, the PL spectra become asymmetric again, with a shoulder on the higher-energy side. These features can be well explained by the filling effect of the localized exciton at low excitation and a free exciton energy state at high excitation intensity [11].
Figure 5d also illustrates that the integrated PL intensity of the three samples increases with rising excitation power intensity. We observed that the exponent "n" approaches unity, indicating no saturation at higher excitation power. This suggests that this PL transition is not attributed to any impurity or defects but is an intrinsic recombination (band-to-band). Previous studies have demonstrated that the three transitions are indirect with a type II transition [10].
Figures 6a-c present the results of temperature-dependent photoluminescence (PL) spectra for the three samples, revealing distinct behaviors for each. Firstly, in Fig. 6e, the FWHM initially narrows as the temperature increases from 10 to 50 K for samples S1 and S2, and from 10 to 80 K for sample S3. This behavior is attributed to localized excitons initially occupying a broad distribution of inhomogeneous fluctuations, which are then thermally excited into the narrower distribution of the free exciton state. Subsequently, the FWHM continuously widens with further temperature increases. The kinetic energy of the free exciton increases with temperature, leading to a larger FWHM.
Secondly, as depicted in Fig. 6f, the PL peak energy exhibits an "S-shape" behavior (red/blue/red-shift) with increasing temperature, particularly pronounced at low indium concentration. The S-shape behavior at low temperature is attributed to the strong localization of excitons at an energy level within the material band gap. However, the maximum redshift is 10 meV for S3 and only 5 meV for the other two samples. At around 60 K, the excitons thermalize and relax to the local minima (alloy disorder), resulting in the first red shift. As the temperature rises from 60 K to 100 K, localized carriers gain enough thermal energy to transfer to higher energy levels in the band tails until reaching the maximum of the band continuum, causing a blue shift in the PL peak energy. Finally, when the lattice temperature surpasses 100 K, the PL peaks of all samples shift to lower energies, with S1 and S2 redshifting faster than S3. In this phase, carriers are thermally activated and prevented from localization, indicating free carrier recombination. Similar anomalous behavior of PL peak energy with temperature increase has been observed in other QW InGaAs/InAlAs systems at lower temperatures, attributed to the presence of band tails in the density of states (DOS) due to potential fluctuations caused by the interface in the quantum well.
We used the Arrhenius relation to fit the experimental data of the integrated PL intensity to understand the mechanism of thermal quenching of carriers in this system [37–38]:
$$\varvec{I}\left(\varvec{T}\right)=\frac{{\varvec{I}}_{0}}{1+\sum _{\varvec{i}}{\varvec{C}}_{\varvec{i}}\varvec{e}\varvec{x}\varvec{p}\varvec{e}\varvec{x}\varvec{p}(-\frac{{\varvec{E}}_{\varvec{a}\varvec{i}}}{{\varvec{K}}_{\varvec{B}}\varvec{T}}) }$$
7
Where T is the measured temperature, I0 is a variable parameter intensity, I(T) is the integrated PL emission intensity, Ci are constants related to the densities of non-radiative recombination centers, Eai are the activation energies corresponding to the non-radiative recombination centers and KB is Boltzmann’s constant.
The Arrhenius fits, depicted in Fig. 6d along with the experimental data, reveal a noteworthy trend in activation energies. As illustrated, there is a clear correlation between the decrease in indium content and the rise in thermal activation energy [11]. This observation can be rationalized by the pronounced defect trapping effect in S3, demanding a higher activation energy for the release of charge carriers from their localized states. The distinct thermal quenching mechanisms exhibited by these three QWs are evident in the variability of their activation energies.
3.e. Time-resolved PL (TRPL) measurement
TRPL measurements were conducted to gain insights into the impact of indium concentration on carrier dynamics in the InxAl1−xAs QW and the inverted interface transition (type II transition). Figure 7 displays the PL lifetime curves for both the fundamental transition energy of the QW and the type II transition energy. To extract the PL lifetime from these curves, background subtraction from the detected signal and a single exponential model were employed.\(I\left(t\right)={I}_{0}\text{*}\text{e}\text{x}\text{p}(-\frac{t}{\tau })\) [3] where \({I}_{0}\) and τ are amplitude and decay time-constant, respectively. The lifetimes in the InxAl1−xAs QW (direct transition) are determined to be 0.37 ns, 0.53 ns, and 0.71 ns, while in the type II transition (indirect transition), they are measured to be 1.01 ns, 2.94 ns, and 6.47 ns for the S3, S2, and S1 samples, respectively. It is evident that in both transitions, the carrier lifetime increases with the rise in indium concentrations, indicating a significant difference in the material quality of these three samples. Two factors contribute to this increase in lifetime with higher indium concentration: the localization of excitons and the interface defect density. Exciton localization is known to increase radiation lifetime [39], while defect density at the interface decreases the radiation lifetime [3]. In this case, the effect of interface defect density appears to be predominant over the effect of exciton localization on the carrier lifetime. Notably, the sample S1, with the best crystalline quality, exhibits a high PL lifetime of 0.71 ns for the InxAl1−xAs epilayer and 6.47 ns for the type II transition. This sample, characterized by a high indium content close to the lattice mismatch between the InxAl1−xAs layer and the InP substrate (xIn=0.52), demonstrates superior optical quality, lower peak FWHM, higher crystalline quality, better interface quality, fewer hetero-interface scattering effects, reduced composition fluctuation, and fewer defects in the QCLs active region.