Structural Properties and Energy Spectrum of Novel GaSb/AlP Self-Assembled Quantum Dots

In this work, the formation, structural properties, and energy spectrum of novel self-assembled GaSb/AlP quantum dots (SAQDs) were studied by experimental methods. The growth conditions for the SAQDs’ formation by molecular beam epitaxy on both matched GaP and artificial GaP/Si substrates were determined. An almost complete plastic relaxation of the elastic strain in SAQDs was reached. The strain relaxation in the SAQDs on the GaP/Si substrates does not lead to a reduction in the SAQDs luminescence efficiency, while the introduction of dislocations into SAQDs on the GaP substrates induced a strong quenching of SAQDs luminescence. Probably, this difference is caused by the introduction of Lomer 90°-dislocations without uncompensated atomic bonds in GaP/Si-based SAQDs, while threading 60°-dislocations are introduced into GaP-based SAQDs. It was shown that GaP/Si-based SAQDs have an energy spectrum of type II with an indirect bandgap and the ground electronic state belonging to the X-valley of the AlP conduction band. The hole localization energy in these SAQDs was estimated equal to 1.65–1.70 eV. This fact allows us to predict the charge storage time in the SAQDs to be as long as >>10 years, and it makes GaSb/AlP SAQDs promising objects for creating universal memory cells.


Introduction
The systems for long-term information storage with the possibility of fast access [1,2] are important for the development of computing technologies. The so-called universal memory cells combining the fast data access peculiar to the dynamic random-access memory (DRAM) and non-volatile long-term data storage will provide a significant increase in the performance and energy efficiency of memory elements that opens up prospects for a revolution in computer architecture. One of the promising methods in this research field is troscopy (PL) and supplemented by theoretical calculations. It was found that, regardless of the substrate type, Ga x Al 1−x Sb SAQDs are formed. The elastic strain in the SAQDs is almost completely relaxed, while the fraction of Al atoms in the SAQD composition does not exceed 10%. The strain relaxation in GaP/Si-based SAQDs does not lead to a decrease in the SAQDs' luminescence efficiency. Comparatively, the introduction of dislocations into GaP-based SAQDs leads to a strong decrease in the SAQDs' luminescence efficiency. Probably, this is caused by the appearance of Lomer 90 • -dislocations without uncompensated atomic bonds in GaP/Si-based SAQDs, while threading 60 • -dislocations are introduced into GaP-based SAQDs. The studies of SAQDs energy structure showed that they are characterized by an energy spectrum of type II, with the ground electronic state lying in the X valley of AlP and the ground hole state lying in Ga x Al 1−x Sb. The E loc value was estimated as 1.65-1.70 eV, so charge storage in SAQDs for a long time (>>10 years) is expected [37].

SAQDs Heterostructure Growth
The HSs were grown by MBE on matched GaP substrates with a (100) crystallographic orientation, as well as on artificial GaP/Si substrates of the same crystallographic orientation. A small, about 0.1%, mismatch between the AlP and GaP lattice constants [35], makes it possible to grow pseudomorphic AlP layers on GaP. The artificial GaP/Si substrates were used to clarify the question about a possible monolithic integration of the GaSb/AlP heterostructures with SAQDs and Si substrates. The monolithic integration of III-V HSs and Si substrates may allow combining the advantages provided by the features of III-V materials with the well-established and widely used Si technology [38][39][40][41]. The growth technique of artificial GaP/Si substrates suitable for the III-V low-dimensional HSs formation was reported in [42]. The density of threading dislocations in the near-surface layers of the artificial GaP/Si substrate was about 10 8 cm −2 .
All HSs were grown by the MBE technique in an improved UHV chamber of the MBEsetup Shtat-type (Ryazan, Russia). It was equipped with crucible sources of fluxes of Al and Ga atoms and Sb 4 molecules with aperture dampers, as well as two-zone valve sources of P 2 molecules [43]. The flux densities of P 2 and Sb 4 molecules, as well as of Al and Ga atoms, were determined from the values of the ion current of an ionization manometric transducer introduced, during the measurements, into direct fluxes to the substrate position [44]. This method makes it possible to determine the fluxes of atoms of groups III and V with an accuracy of ±2% and ±6%, respectively. The substrate temperature (T s ) was controlled by the thermocouple of the substrate heater, which was calibrated for each sample by the RHEED method in reference to the transition temperatures of surface superstructures on GaP(100) in the absence of a phosphorus flux. This technique was previously developed for the GaAs layer epitaxy [45] and was used here for the GaP growth after small corrections. The temperature determination accuracy was ±5 • C.
A series of HSs without GaSb/AlP SAQDs (HSs A and C) and with GaSb/AlP SAQDs (HSs of type B and type D) was grown on various substrates. The HSs are schematically present in Figure 1. The growth rate of GaP and AlP layers was one monolayer per second (ML/s). The GaP buffer layers with a thickness of 300 nm were grown at T s = 580 • C in the HSs grown on a matched GaP substrate. In HSs grown on an artificial GaP/Si substrate, the 300 nm thick GaP buffer layers were grown at T s = 600 • C. The buffer layer formation temperature was adjusted to optimize the quality of the growth surface morphology, which was controlled in situ by RHEED. After the buffer layer growth, the T s level was decreased to 420 • C for the HSs grown on a matched GaP substrate, and to 450 • C for the HSs grown on an artificial GaP/Si substrate. Then, the AlP layers were grown. The T s value for the AlP growth was tuned in accordance with the results of studies on the epitaxial growth of AlP layers [46,47]. The AlP layer thickness was 300 nm in all HSs. The SAQDs were located in the center of the AlP layer, as shown in Figure 1. The SAQDs' formation was performed at different T s values in the range of 360-480 • C. The material deposition rate during SAQD formation was 0.23 ML/s. To induce the SAQDs' formation, 1.6 MLs of GaSb was deposited. The SAQDs formation was in situ controlled by RHEED. After the deposition of the required amount of material, the growth was interrupted, and the growth surface was kept for 30 s without fluxes of atoms of both groups III and V. The AlP layers were covered by a 25 nm GaP layer grown at the T s value gradually increasing from 420 (450 • C) to 500 • C. In the HSs, where SAQDs were formed at 420 • C and 450 • C for HS of type B and type D, respectively, and 25 nm thick AlP layers were additionally grown, and unburied SAQDs were formed under conditions similar to the conditions used for the growth of buried SAQDs. This was implemented in order to study the SAQD array morphology using the AFM technique.
growth, the Ts level was decreased to 420 °C for the HSs grown on a matched GaP substrate, and to 450 °C for the HSs grown on an artificial GaP/Si substrate. Then, the AlP layers were grown. The Ts value for the AlP growth was tuned in accordance with the results of studies on the epitaxial growth of AlP layers [46,47]. The AlP layer thickness was 300 nm in all HSs. The SAQDs were located in the center of the AlP layer, as shown in Figure 1. The SAQDs' formation was performed at different Ts values in the range of 360-480 °C. The material deposition rate during SAQD formation was 0.23 ML/s. To induce the SAQDs' formation, 1.6 MLs of GaSb was deposited. The SAQDs formation was in situ controlled by RHEED. After the deposition of the required amount of material, the growth was interrupted, and the growth surface was kept for 30 s without fluxes of atoms of both groups III and V. The AlP layers were covered by a 25 nm GaP layer grown at the Ts value gradually increasing from 420 (450 °С) to 500 °С. In the HSs, where SAQDs were formed at 420 °С and 450 °С for HS of type B and type D, respectively, and 25 nm thick AlP layers were additionally grown, and unburied SAQDs were formed under conditions similar to the conditions used for the growth of buried SAQDs. This was implemented in order to study the SAQD array morphology using the AFM technique.

а) b)
Nanomaterials 2023, 13 The SAQDs' formation process was in situ controlled by the RHEED technique using homemade equipment (ISP SB RAS, Novosibirsk, Russia). The unburied SAQDs array structure was ex situ studied by AFM using a Solver P47 microscope (NT-MDT, Moscow, Russia) operating in the semi-contact mode. The steady-state PL was excited by the radiation from a GaN laser diode (ISP SB RAS, Novosibirsk, Russia) with a photon energy of 3.06 eV (405 nm) and a power density (P ex ) varied in the range 0.37-25 W/cm 2 . The PL radiation was analyzed using an SDL-1 double grating monochromator (LOMO, Saint Petersburg, Russia) and recorded using a nitrogen-cooled Ge photodiode (Edinburgh Instruments, Edinburgh, Great Britain) in the synchronous detection mode. The samples were placed in a helium-filled cryostat Utreks-R (Kharkov, Ukraine), which maintains the temperature in the range of 5-300 K.

RHEED
The RHEED pattern of two-dimensional growth was observed during the GaSb deposition until the nominal amount of deposited material reached 1.6 ML. The 2D RHEED pattern was dramatically changed into a typical 3D diffraction pattern when the nominal amount of the deposited material reached 1.6 ML. This transformation indicates the formation of nanosized 3D islands [48,49]. The 3D RHEED patterns corresponded to the SAQDs formation in the HS of type B grown at T s = 420 • C and the HS of type D grown at T s = 450 • C are shown in Figure 2a,b, respectively. The character of the RHEED patterns was the same for all HSs of type B and type D independently of T s variations in the range of 360-480 • C during the SAQDs formation. A detailed analysis of the RHEED pattern for HS D was performed. Coupled with SAQD 3D Bragg reflections, we observed 2D reflections associated with the electron diffraction on the AlP surface superstructure. The boundaries of these reflections are marked in Figure 3 by vertical black dotted lines. As can be seen in the pattern, the horizontal distances between the zero-order reflection and first-order 3D Bragg reflections and between the zero-order reflection and first-order 2D surface reflections differ by 10 ± 1%. The distance between 2D surface reflections is inversely proportional to the AlP lattice constant, and, therefore, the lattice constant of the SAQD material is 10 ± 1% larger than the AlP lattice constant. Taking into account the mismatch between the lattice constants of GaSb and AlP (10.5% [35]), this also indicates an almost complete strain relaxation into the SAQDs. In addition, these data made it possible to estimate the composition of SAQD material. In general, due to different mixing processes, SAQDs can be formed from the GaxAl1−xSbyP1−y quaternary alloy. However, the RHEED results show that the fraction of P atoms in the SAQDs composition is insignificant, since, at a significant element mixing over group V, a proportional change in the alloy lattice constant would occur [35]. Unfortunately, this logic cannot be applied to the mixing of group III materials, since the lattice constants in AlSb/GaSb and AlP/GaP pairs are practically the same [35]. Therefore, we can state that SAQDs consist of an almost unstrained ternary alloy GaxAl1−xSb. An analysis of these RHEED patterns allows us to obtain the in situ information on the strain in the formed SAQDs. It is known that the horizontal distance between diffraction reflections under the conditions of 3D Bragg diffraction is L~1/d 011 , while the vertical distance is H~1/d 100 , where d 011 and d 100 are the corresponding interplanar distances in the island material [49]. The distances are governed by the lattice constants of the island material in the corresponding directions: d 011~a|| /2 0.5 and d 100~a⊥ . Here, a || is the lattice constant in the HS plane, and a ⊥ is the lattice constant in the growth direction. In the strained islands, lattice constant values in different directions are not equal to each other, since the compression of the island lattice in the HS plane leads to its expansion in the growth direction [50]. The analysis of the RHEED patterns showed that the L/H ratio for the HS of type B is 1.434, and for the HS of type D, it is 1.466, which differs from 2 0.5 by less than 5% (1.414 for the case of complete strain relaxation). This allows stating that the strain in the obtained SAQDs is almost completely relaxed.
The absence of visible dynamics in the 3D RHEED pattern during the SAQDs' formation shows that the strain relaxation occurred almost immediately after the SAQDs' formation and the stage of strained islands is not stable. The variation in T S in the range of 360-480 • C during the SAQD formation did not lead to noticeable changes in the SAQD formation process. Regardless of T S , almost completely relaxed SAQDs are formed. The HS of type B with SAQDs grown at 420 • C and the HS of type D with SAQDs grown at 450 • C will be discussed further and marked like HS B and D, respectively.
A detailed analysis of the RHEED pattern for HS D was performed. Coupled with SAQD 3D Bragg reflections, we observed 2D reflections associated with the electron diffraction on the AlP surface superstructure. The boundaries of these reflections are marked in Figure 3 by vertical black dotted lines. As can be seen in the pattern, the horizontal distances between the zero-order reflection and first-order 3D Bragg reflections and between the zero-order reflection and first-order 2D surface reflections differ by 10 ± 1%. The distance between 2D surface reflections is inversely proportional to the AlP lattice constant, and, therefore, the lattice constant of the SAQD material is 10 ± 1% larger than the AlP lattice constant. Taking into account the mismatch between the lattice constants of GaSb and AlP (10.5% [35]), this also indicates an almost complete strain relaxation into the SAQDs. In addition, these data made it possible to estimate the composition of SAQD material. In general, due to different mixing processes, SAQDs can be formed from the Ga x Al 1−x Sb y P 1−y quaternary alloy. However, the RHEED results show that the fraction of P atoms in the SAQDs composition is insignificant, since, at a significant element mixing over group V, a proportional change in the alloy lattice constant would occur [35]. Unfortunately, this logic cannot be applied to the mixing of group III materials, since the lattice constants in AlSb/GaSb and AlP/GaP pairs are practically the same [35]. Therefore, we can state that SAQDs consist of an almost unstrained ternary alloy Ga x Al 1−x Sb. Figure 2. RHEED patterns obtained after the SAQDs formation in the HS of (a) typ = 420 °С and in the HS of (b) type D grown at Ts = 450 °С. Thin red dotted lines mark H.
A detailed analysis of the RHEED pattern for HS D was performed. SAQD 3D Bragg reflections, we observed 2D reflections associated with th fraction on the AlP surface superstructure. The boundaries of these reflectio in Figure 3 by vertical black dotted lines. As can be seen in the pattern, distances between the zero-order reflection and first-order 3D Bragg refle tween the zero-order reflection and first-order 2D surface reflections diff The distance between 2D surface reflections is inversely proportional to constant, and, therefore, the lattice constant of the SAQD material is 10 ± 1 the AlP lattice constant. Taking into account the mismatch between the la of GaSb and AlP (10.5% [35]), this also indicates an almost complete strain the SAQDs. In addition, these data made it possible to estimate the compos material. In general, due to different mixing processes, SAQDs can be for GaxAl1−xSbyP1−y quaternary alloy. However, the RHEED results show that th atoms in the SAQDs composition is insignificant, since, at a significant el over group V, a proportional change in the alloy lattice constant would occ tunately, this logic cannot be applied to the mixing of group III materials, s constants in AlSb/GaSb and AlP/GaP pairs are practically the same [35]. The state that SAQDs consist of an almost unstrained ternary alloy GaxAl1−xSb. Thus, as a result of RHEED in situ studies of the GaSb/AlP SAQD growth parameters were determined for the SAQD array formation, and were obtained concerning the strain and chemical composition of the SAQ

Atomic Force Microscopy
The surface of HSs with unburied SAQDs was studied by the AFM te scanning was carried out immediately after removing the HSs from the gr Thus, as a result of RHEED in situ studies of the GaSb/AlP SAQD formation, the growth parameters were determined for the SAQD array formation, and also the data were obtained concerning the strain and chemical composition of the SAQDs.

Atomic Force Microscopy
The surface of HSs with unburied SAQDs was studied by the AFM technique. AFM scanning was carried out immediately after removing the HSs from the growth chamber in order to minimize the effects of possible AlP interaction with oxygen and water in the atmosphere. The AFM images of the HSs B and D surface, as well as surface height profiles for various sections, are presented in Figure 4. As can be seen, an SAQD array is actually formed in both HSs. The SAQD array parameters (SAQD density (N QD ), SAQD diameter (D QD ), and SAQD height (h QD )) are presented in Table 1.
in order to minimize the effects of possible AlP interaction with oxygen and water in the atmosphere. The AFM images of the HSs B and D surface, as well as surface height profiles for various sections, are presented in Figure 4. As can be seen, an SAQD array is actually formed in both HSs. The SAQD array parameters (SAQD density (NQD), SAQD diameter (DQD), and SAQD height (hQD)) are presented in Table 1.   In addition, as it is shown in Figure 4, the AlP surface is characterized by a smooth relief with "waves". The lateral extension of these "waves" is D AlP and the vertical magnitude is h AlP . The values of these parameters are also presented in Table 1.
In our opinion, an increase in the h AlP value for the smooth AlP relief in the GaP/Sibased HS, in comparison with the GaP-based HS, can be caused by (1) a larger RMS of the initial GaP growth surface in the case of artificial GaP/Si substrate; and (2) the influence of threading dislocations, which are present in the layers grown on an artificial GaP/Si substrate. An increase in the SAQD array N QD is observed in parallel with an increase in the vertical magnitude of the smooth AlP relief h AlP , while the lateral extension of the AlP relief is approximately the same. We assume that this effect can be attributed to an increase in the atomic step density on the AlP surface and, thus, to an increase in the number of places where the SAQD formation is preferable. Note, that an approximately twofold increase in h AlP (and steps density also) results in a proportional increase in N QD, and this is in favor of our suggestion. The same assumption can be used to explain the decrease in the D QD [51]. Moreover, an effect of the threading dislocation, which is present in GaP/Si-based HSs, on the SAQD formation process, cannot be excluded. It is possible that the dislocation outcrops play the role of a source of mobile adatoms because of the crystal structure distortion in the dislocation core. It can have an effect on lateral adatom diffusion during the SAQDs' formation. Unfortunately, our experimental data are not enough for a correct estimation of the contribution of each discussed mechanism. Therefore, this question is out of the scope of this study.
It is necessary to note that the SAQD arrays are characterized by a sufficiently high density (several units of 10 10 cm −2 ), and a high charge plane density may be expected when the SAQD states are filled with holes. This is useful from the point of view of the future application of these HSs as floating gates in memory cells since it will increase the efficiency of the SAQD layer effect on the underlying FET channel conductivity.
Thus, the ex situ AFM studies of an array of unburied GaSb/AlP SAQDs grown on various substrates made it possible to obtain information on the AlP surface morphology used for the SAQD array formation, as well as on the geometric parameters of the SAQD arrays.

Photoluminescence
The steady-state PL spectra of all considered HSs were measured under the nonresonant excitation of nonequilibrium charge carriers in an AlP matrix at the liquid nitrogen temperature (77 K) and P ex = 25 W/cm 2 . The measured results are shown in Figure 5. As it is presented in Figure 5a, the spectrum of HS A contains two PL bands with the maxima at energies 1.01 and 0.83 eV. We attribute these bands to the charge carrier recombination through deep centers in the AlP and/or GaP layers. The spectrum of HS B also contains both PL bands with the maxima at energies 1.01 and 0.83 eV, as can be seen from the deconvolution of the PL spectrum of HS B into the sum of two Gaussian components ( Figure 5a). Unfortunately, no additional PL bands that could be associated with the charge carrier recombination in SAQDs were detected. It is also necessary to note that the integrated PL intensity of HS B is noticeably (four times) lower than the integrated PL intensity of HS A. At the same time, a comparison of the PL spectra of HSs C and D presented in Figure 5b made it possible to reveal the PL band which is, probably, associated with the charge carrier recombination in SAQDs. This band with a maximum at 0.85 eV is As it is presented in Figure 5a, the spectrum of HS A contains two PL bands with the maxima at energies 1.01 and 0.83 eV. We attribute these bands to the charge carrier recombination through deep centers in the AlP and/or GaP layers. The spectrum of HS B also contains both PL bands with the maxima at energies 1.01 and 0.83 eV, as can be seen from the deconvolution of the PL spectrum of HS B into the sum of two Gaussian components (Figure 5a). Unfortunately, no additional PL bands that could be associated with the charge carrier recombination in SAQDs were detected. It is also necessary to note that the integrated PL intensity of HS B is noticeably (four times) lower than the integrated PL intensity of HS A. At the same time, a comparison of the PL spectra of HSs C and D presented in Figure 5b made it possible to reveal the PL band which is, probably, associated with the charge carrier recombination in SAQDs. This band with a maximum at 0.85 eV is marked with a vertical arrow in Figure 5b. Detailed studies of the luminescence properties of HSs B and D were carried out. The low-temperature (5 K) steady-state PL spectra of the HSs were measured, and the dependences of the PL spectra on temperature and P ex were analyzed. The obtained results are presented in Figures 6-8. bination through deep centers in the AlP and/or GaP layers. The spectrum of contains both PL bands with the maxima at energies 1.01 and 0.83 eV, as can be the deconvolution of the PL spectrum of HS B into the sum of two Gaussian co (Figure 5a). Unfortunately, no additional PL bands that could be associated charge carrier recombination in SAQDs were detected. It is also necessary to no integrated PL intensity of HS B is noticeably (four times) lower than the inte intensity of HS A. At the same time, a comparison of the PL spectra of HSs C a sented in Figure 5b made it possible to reveal the PL band which is, probably, with the charge carrier recombination in SAQDs. This band with a maximum a marked with a vertical arrow in Figure 5b. Detailed studies of the luminescence of HSs B and D were carried out. The low-temperature (5 K) steady-state PL spe HSs were measured, and the dependences of the PL spectra on temperature an analyzed. The obtained results are presented in Figures 6-8.  As it is presented in Figure 6, the spectrum of HS C without SAQD consists of bands whose shape is well described by Gaussian functions with the maxima at the energies of 0.88, 0.96, and 1.09 eV, with the full width at half maximum (FWHM) values of 230, 120, and 110 meV, respectively. We attribute these PL bands to the charge carrier recombination through deep levels in the AlP and/or GaP layers. A detailed elucidation of the nature of these deep levels is beyond the scope of this work. The spectrum of SAQD HS D contains PL bands, whose shape is also described by the Gaussian function, and their spectral position, the FWHM and amplitude values are close to the corresponding bands in the spectrum of the HS C without SAQDs. The bands are characterized by the maxima at energies of 0.88, 0.95, and 1.08 eV, with FWHM values of 230, 112, and 120 meV, respectively. The change in the PL band intensity compared to the spectrum of the HS without SAQDs is no more than 30%. We also attribute these PL bands to the recombination through deep levels in the bulk material. Furthermore, the spectrum contains an additional PL band with a maximum at 0.86 eV and an FWHM value of 72 meV, which is marked in Figure 6 with a thick blue line. We attribute this additional band to the charge carrier recombination in the SAQDs. As can be seen, the PL data obtained at 5 K are in good agreement with the PL data obtained at 77 K ( Figure 5). The PL spectra of the SAQD HS were measured at the temperature of 5 K and Pex varying in the range of 0.375-25 W/cm 2 . The measured results are shown in Figure 8. As can be seen in Figure 8a, the integral SAQD PL intensity is nearly proportional to Pex. According to the results of [52], this indicates that the internal quantum yield of radiative recombination of nonequilibrium charge carriers in the SAQDs is close to 100%, and almost all electron-hole pairs captured into the SAQDs recombine radiatively. In addition, a spectral shift in the SAQD PL band is observed with the Pex increasing. The energy shift dependence on Pex is shown in Figure 8b. As can be seen from the curves, the SAQDs PL band energy shift is proportional to Pex 1/3 . In accordance with the results reported in [53][54][55], this indicates that the SAQDs have a band alignment of type II, which implies the spatial separation of electrons and holes localized in the SAQDs. The PL band shift is caused by the energy band distortion, and, as a consequence, the shift in the quantum confinement levels with a change in the number of charge carriers localized in the SAQDs. The PL spectra of SAQD HS D were measured at different temperatures in the range of 5-190 K. The measurements were carried out at P ex = 25 W/cm 2 . The results are shown in Figure 7. As the temperature increases, the SAQD PL quenches. The temperature dependence of the integrated SAQD PL intensity is shown in Figure 7a. The experimental dependence is well described by the Arrhenius function with the PL quenching activation energy E a = 15 ± 2 meV. As is shown further by the energy spectrum calculation, the value of E a is due to the fact that the SAQD PL quenching occurs due to the nonequilibrium electron ejection from the SAQD into the AlP matrix. A shift in the spectral position of the PL bands is also observed. In Figure 7b are the dependences of the spectral shift magnitude of the PL bands with the maxima at 0.95 eV, marked with black dots and number 1, and with a maximum at energy of 0.86 eV, marked with blue dots and the number 2.
As can be seen from the curves, the SAQD PL band at 0.86 eV shows a shift that is well described by the temperature dependence of the AlP bandgap. However, the PL band at 0.95 eV is almost not shifted with temperature. The bands with the maxima at 0.88 eV and 1.08 eV behave in a similar way. The observed character of the temperature dependences of the integrated PL intensity and the spectral shift in PL bands, additionally, indicates that the PL band with the maximum at 0.86 eV is associated with the charge carrier recombination in SAQDs, while the remaining spectral bands are caused by the recombination on deep levels in AlP and/or GaP. Thus, the PL experiments made it possible to unambiguously identify the PL band associated with the recombination of nonequilibrium charge carriers in the SAQDs grown on artificial GaP/Si substrates. We list here the main experimental factors, which allow us to state that the founded PL band is associated with the SAQDs: 1. This PL band exists in the spectrum of HS D with SAQDs and is absent in the spectrum of HS C without SAQDs. 2. This PL band shifts towards lower energies with increasing temperature (Figure 7b). This behavior is opposite to deep-level PL bands (0.8-1 eV), which have a stable energy position. Moreover, the SAQD PL band shift is in excellent relation to the temperature dependence of Eg of AlP. It indicates that PL energy is governed by the energy level position in the SAQD. The PL spectra of the SAQD HS were measured at the temperature of 5 K and P ex varying in the range of 0.375-25 W/cm 2 . The measured results are shown in Figure 8. As can be seen in Figure 8a, the integral SAQD PL intensity is nearly proportional to P ex . According to the results of [52], this indicates that the internal quantum yield of radiative recombination of nonequilibrium charge carriers in the SAQDs is close to 100%, and almost all electron-hole pairs captured into the SAQDs recombine radiatively. In addition, a spectral shift in the SAQD PL band is observed with the P ex increasing. The energy shift dependence on P ex is shown in Figure 8b. As can be seen from the curves, the SAQDs PL band energy shift is proportional to P ex 1/3 . In accordance with the results reported in [53][54][55], this indicates that the SAQDs have a band alignment of type II, which implies the spatial separation of electrons and holes localized in the SAQDs. The PL band shift is caused by the energy band distortion, and, as a consequence, the shift in the quantum confinement levels with a change in the number of charge carriers localized in the SAQDs.
Thus, the PL experiments made it possible to unambiguously identify the PL band associated with the recombination of nonequilibrium charge carriers in the SAQDs grown on artificial GaP/Si substrates. We list here the main experimental factors, which allow us to state that the founded PL band is associated with the SAQDs:

1.
This PL band exists in the spectrum of HS D with SAQDs and is absent in the spectrum of HS C without SAQDs.

2.
This PL band shifts towards lower energies with increasing temperature (Figure 7b). This behavior is opposite to deep-level PL bands (0.8-1 eV), which have a stable energy position. Moreover, the SAQD PL band shift is in excellent relation to the temperature dependence of E g of AlP. It indicates that PL energy is governed by the energy level position in the SAQD.

3.
SAQD PL band shits~P ex 1/3 when excitation density P ex is changed. This behavior is peculiar to low-dimensional systems with band alignment of type II [53][54][55] and not peculiar to deep-level PL.

Summarizing Experimental Results
At the end of the section, we would like to list the most important experimental results, in our opinion. The GaSb/AlP HSs were studied by RHEED, AFM, and steady-state PL spectroscopy. As a result, it was shown: 1.
The deposition of 1.6 ML GaSb on the AlP surface at T S = 360-480 • C, where AlP was grown on matched GaP, as well as on artificial GaP/Si substrates, leads to the SAQD array formation.

2.
An almost complete strain relaxation is observed in the SAQDs regardless of the substrate type employed for the SAQD growth.

3.
SAQDs grown on an artificial GaP/Si substrate consist of almost unstrained Ga x Al 1−x Sb.

4.
The geometric parameters of the SAQD arrays were determined, as well as their correlations with the morphology of the AlP surface on which the SAQDs were formed (Table 1): an increase in the AlP surface roughness leads to an increase in the SAQDs density and a decrease in their sizes.

5.
No PL bands that could be associated with the recombination of nonequilibrium charge carriers in SAQDs grown on matched GaP substrates were found. 6.
The SAQD formation in HSs grown on matched GaP substrates leads to a significant (fourfold) decrease in the integrated intensity of PL bands associated with the nonequilibrium charge carrier recombination through deep levels in AlP and/or GaP.

7.
A low-temperature PL band with a maximum at about 0.86 eV was found, and the band is associated with the nonequilibrium charge carrier recombination in GaSb/AlP SAQDs grown on artificial GaP/Si substrates. 8.
The thermal quenching of the SAQD PL occurs at E a = 15 ± 2 meV. 9.
SAQDs grown on an artificial GaP/Si substrate are characterized by a high internal luminescence quantum yield close to 100%. 10. A spectral shift in the PL band of SAQDs grown on an artificial GaP/Si substrate was observed. This shift is~P ex 1/3 , and it indicates that SAQDs have a band alignment of type II.

Discussion
In this section, we discuss the structural properties of the considered GaSb/AlP SAQDs, as well as the SAQD energy spectrum.

Structural Properties of SAQDs
The main feature of the SAQD structure is an almost complete absence of strain, as indicated by the RHEED data. Strains caused by a mismatch between the lattice constants of the deposited material and the matrix material are one of the main reasons for the SAQDs' formation. The reorganization of the surface structure and the formation of 3D islands (SAQDs) leads to a decrease in the total energy of the system due to a decrease in the elastic energy in SAQDs, despite an increase in the surface energy [51,56,57]. However, an increase in the SAQDs' size leads to an increase in the elastic energy stored in them. When the elastic energy reaches a threshold value, the plastic relaxation of strains in SAQDs occurs by the introduction of dislocations [58][59][60][61][62][63]. It is known that the coalescence of two complementary 60 • dislocations (U-half-loops) leads to the formation of a complex containing a Lomer 90 • dislocation lying in the (001) plane and threading arms lying in planes of the (111) type [64][65][66][67]. Sliding of threading arms in planes of type (111) leads to their going beyond the SAQD, and, thus, only a 90 • dislocation remains inside and in the vicinity of the SAQD. This dislocation does not cross the SAQD volume. The schematic image of the dislocation complex in the SAQD when threading arms exist in the SAQD volume is presented in Figure 9. As shown in [68], the core of the Lomer dislocation segment does not contain uncompensated atomic bonds, unlike threading arms. Therefore, the presence of 90 • dislocations does not lead to the formation of deep centers and, consequently, to an increase in the probability of nonradiative recombination in SAQDs. We have already observed a similar effect in GaSb/GaP [69] and GaAs/GaP [70,71] heterosystems. At the same time, if no nucleation and effective coalescence of complementary 60 • dislocations occurred during the formation of the SAQD and its plastic relaxation, then 60 • dislocations remain in the SAQD. This leads to a sharp quenching of PL due to an increase in the rate of nonradiative recombination. It is necessary to note that in both cases the threading arms are growing from the SAQD into the volume of the AlP matrix and affect the concentration of photoexcited nonequilibrium charge carriers by increasing the rate of nonradiative recombination. However, the absence of nonradiative centers directly in SAQDs, where Lomer dislocations are formed, makes it possible to observe the PL of such SAQDs. Our PL data show that the PL of GaSb/AlP SAQDs grown on an artificial GaP/Si substrate is characterized by a high internal PL quantum yield (about 100%), and it indicates that the introduction of dislocations had no noticeable effect on the efficiency of radiative recombination of nonequilibrium charge carriers. erials 2023, 13, x FOR PEER REVIEW image of the dislocation complex in the SAQD when thread volume is presented in Figure 9. As shown in [68], the core o ment does not contain uncompensated atomic bonds, unlike the presence of 90° dislocations does not lead to the formation quently, to an increase in the probability of nonradiative rec have already observed a similar effect in GaSb/GaP [69] and G tems. At the same time, if no nucleation and effective coalesc dislocations occurred during the formation of the SAQD and 60° dislocations remain in the SAQD. This leads to a sharp increase in the rate of nonradiative recombination. It is necessa the threading arms are growing from the SAQD into the vol affect the concentration of photoexcited nonequilibrium charg rate of nonradiative recombination. However, the absence of n in SAQDs, where Lomer dislocations are formed, makes it p such SAQDs. Our PL data show that the PL of GaSb/AlP SA GaP/Si substrate is characterized by a high internal PL quant it indicates that the introduction of dislocations had no notice of radiative recombination of nonequilibrium charge carriers. Thus, it can be suggested that the strain relaxation occur Lomer dislocations located at the SAQD/matrix heterointe Thus, it can be suggested that the strain relaxation occurs due to the introduction of Lomer dislocations located at the SAQD/matrix heterointerface in the GaSb/AlP HSs grown on artificial GaP/Si substrates. At the same time, the absence of an SAQD PL band for GaP-based HSs, as well as a decrease in the PL associated with the recombination at deep levels in AlP and/or GaP as a result of the SAQDs formation, suggests that the strain relaxation in GaP-based SAQDs is provided by the introduction of 60 • -dislocations penetrating the SAQD volume. Today, available experimental data are not enough to unambiguously reveal the reason for such a drastic change in the strain relaxation mode; however, we assume that this may be caused by the presence of threading dislocations in the volume of GaP and AlP layers in HSs grown on artificial GaP/Si substrates. Indeed, the presence of threading arm dislocations in GaP and AlP layers grown on a GaP/Si substrate can facilitate efficient coalescence of 60 • dislocations formed during SAQD formation. This assumption is supported by the results obtained in [67], where it was shown that the presence of an initial density of threading dislocations in a relaxing layer has a significant effect on the mechanism of introducing dislocations and leads to an increase in the probability of the formation of the complementary pairs of 60 • dislocations.

Energy Spectrum
In order to determine the energy spectrum of the SAQDs, in particular the E loc value, the SAQD energy spectrum was calculated and the results were compared with the experimental PL data. Following the results of [29,33], we use a truncated tetrahedral pyramid for modeling a SAQD shape. According to the AFM data, the pyramid base length-toheight ratio is equal to 15:1. Note that obtaining precise information about the SAQD shape from the AFM data is complicated by the convolution effect during scanning the surface features of sizes comparable with the probe tip radius. Thus, the SAQD is modeled as a truncated pyramid consisting of a Ga x Al 1−x Sb alloy of uniform composition. Since available experimental data did not allow the obtaining of information about the Ga (Al) content in the SAQDs, different alloy compositions were considered. The Ga x Al 1−x Sb alloy parameters were estimated from the known parameters of GaSb and AlSb within the quadratic approximation [35]: where P GaSb and P AlSb are the values of the corresponding parameters for GaSb and AlSb, and C GaAlSb is the bowing parameter. Calculations of the energy level positions for electrons and holes were implemented in the framework of the single band approximation, and the exciton effect was not taken into account. The variations in the SAQD sizes were also accounted for in the calculations. A detailed discussion of the approaches used in the calculations can be found in our previous work [36] related to the calculations of the InGaSb/AlP SAQDs energy spectrum. The values of the AlP, GaSb, and AlSb material parameters, such as the band gap at the Г, X, and L points of the Brüllien zone; spin-orbit splitting value in the valence band; valence band offset (VBO); and charge carrier effective masses, as well as the corresponding bowing parameters, were reported in [35,72]. The material parameters used in the calculations are presented in Table 2. The calculations were performed using the Nextnano++ program package [73]. This program package is commonly used for III-V SAQD energy spectrum calculations [28,31]. The results are shown in Figure 10.
The calculations show that, independently from the SAQD composition and sizes, the ground electronic state lies in the X valley of the AlP conduction band, and the ground hole state lies in the heavy hole subband of Ga x Al 1−x Sb. This is in excellent agreement with our experimental PL data, which indicate a type-II band alignment for the SAQDs. In addition, the observed E a of the SAQD PL temperature quenching is 15 ± 2 meV, which is in good agreement with the weak localization of electrons in the vicinity of the SAQD of type II. x FOR PEER REVIEW 16 of 20 addition, the observed Ea of the SAQD PL temperature quenching is 15 ± 2 meV, which is in good agreement with the weak localization of electrons in the vicinity of the SAQD of type II. The calculated dependences of the SAQD optical transition energy are shown in Figure 10a. An increase in the SAQD sizes leads to a decrease in the optical transition energy, which is explained by the quantum confinement effect. An increase in the content of Al atoms in the SAQD composition leads to an increase in the optical transition energy, which is caused by an increase in the GaxAl1−xSb alloy bandgap. The comparison of the calculated optical transition energy values with the experimental PL data (shaded area in Figure 10a) is performed taking into account the PL bandwidth. It reveals the range of admissible values of SAQD sizes and composition. As can be seen from the curves, the SAQD cannot contain more than 10% Al in the composition, and the SAQD height cannot be less than 4 nm. At the same time, our AFM data show that the vertical SAQD sizes do not exceed 4 nm. Possible reasons for this discrepancy are now discussed. The underestimation of the SAQD height in the AFM measurements can be caused by the following reasons: (1) the effect of convolution in the AFM measurements of surface morphology when the sizes of features are comparable with the probe tip radius [74]; (2) SAQDs' degradation during the HS cooling with unburied SAQDs in the residual atmosphere of the growth chamber; the time of which noticeably exceeds the growth pause time for buried SAQDs; and (3) the formation of an oxide layer on the surface during the HS exposure to the atmosphere. As seen in Figure 10b, the performed calculations show that the Eloc value in GaxAl1−xSb/AlP is 1.65-1.70 eV, depending on the composition and size of SAQDs. According to [37], the hole storage time can be estimated by:  The calculated dependences of the SAQD optical transition energy are shown in Figure 10a. An increase in the SAQD sizes leads to a decrease in the optical transition energy, which is explained by the quantum confinement effect. An increase in the content of Al atoms in the SAQD composition leads to an increase in the optical transition energy, which is caused by an increase in the Ga x Al 1−x Sb alloy bandgap. The comparison of the calculated optical transition energy values with the experimental PL data (shaded area in Figure 10a) is performed taking into account the PL bandwidth. It reveals the range of admissible values of SAQD sizes and composition. As can be seen from the curves, the SAQD cannot contain more than 10% Al in the composition, and the SAQD height cannot be less than 4 nm. At the same time, our AFM data show that the vertical SAQD sizes do not exceed 4 nm. Possible reasons for this discrepancy are now discussed. The underestimation of the SAQD height in the AFM measurements can be caused by the following reasons: (1) the effect of convolution in the AFM measurements of surface morphology when the sizes of features are comparable with the probe tip radius [74]; (2) SAQDs' degradation during the HS cooling with unburied SAQDs in the residual atmosphere of the growth chamber; the time of which noticeably exceeds the growth pause time for buried SAQDs; and (3) the formation of an oxide layer on the surface during the HS exposure to the atmosphere.
As seen in Figure 10b, the performed calculations show that the E loc value in Ga x Al 1−x Sb/ AlP is 1.65-1.70 eV, depending on the composition and size of SAQDs. According to [37], the hole storage time can be estimated by: where T is the temperature, σ inf is the capture cross-section at a high temperature and γ is the coefficient independent of temperature. Localization energies of 1.65-1.70 eV are high enough for the hole storage times of >>10 years, at typical σ inf values of 10 −12 -10 −9 cm 2 for GaSb/GaP [32], InGaAs/GaP [30], and InGaSb/GaP [33] SAQDs, according to the available experimental results. It makes the Ga x Al 1−x Sb/AlP SAQDs a promising object for the fabrication of universal memory cells.

Conclusions
The formation processes, structural properties, and energy spectrum of novel GaSb/ AlP HSs with SAQDs grown by MBE on matched GaP and artificial GaP/Si substrates were studied. Artificial substrates were characterized by a threading dislocation density of about 10 8 cm −2 . The growth conditions were found to form the SAQD arrays on the AlP layers grown on different substrates. It was found that the strains in the SAQDs relax almost completely independently from the substrate type. The strain relaxation in GaP/Si-based SAQDs does not lead to a decrease in the SAQD luminescence efficiency, while the dislocations present in GaP-based SAQDs lead to a strong decrease in the SAQD luminescence efficiency. Probably, this is caused by the introduction of Lomer 90 • -dislocations without uncompensated atomic bonds in GaP/Si-based SAQDs, while threading 60 • -dislocations are introduced into GaP-based SAQDs. It was shown that SAQDs grown on an artificial GaP/Si substrate consist of an almost unstrained Ga x Al 1−x Sb alloy with a fraction of Al atoms not exceeding 10%. The comparison of experimental PL data and calculation results made it possible to reveal that SAQDs have a band alignment of type II with an indirect bandgap and the ground electronic state belonging to the X-valley of the AlP conduction band and the ground hole state lying in Ga x Al 1−x Sb. The estimated E loc value was in the range of 1.65-1.70 eV. This allows us to predict the charge storage time in SAQDs at room temperature to be as long as >> 10 years, and that makes these SAQDs a promising object for the fabrication of universal memory cells.