Growth of II-VI/III-V heterovalent quantum structures Free

The range of material properties achievable with monocrystalline semiconductor structures grown on commercially available III-V substrates can be broadened substantially by integration of III-V materials with lattice-matched II-VI counterparts. Properties such as carrier mobilities, band offsets, and refractive index contrast in these heterovalent structures can reach values unobtainable with conventional isovalent III-V heterostructures alone. However, the effects of interface bonding and significantly different growth conditions for the III-V and II-VI semiconductors could render these benefits irrelevant without employing an acceptable growth method to reduce structural defects in the epilayers.^1,2 In addition, the risk of cross contamination in dedicated III-V growth chambers has acted as a proverbial bottleneck for achieving optoelectronic devices which utilize heterovalent heterostructures. Although the growth of II-VI layers on III-V semiconductor substrates has been studied for decades,^3–5 few successful demonstrations of III-V growth on II-VI layers have been demonstrated.^6–9 

Heterovalent II-VI/III-V structures suitable for devices such as transistors,¹⁰ solar cells,¹¹ and photodetectors¹² have been proposed and demonstrated. The key for producing high quality heterovalent structures lies in the growth conditions of the interface between the III-V and II-VI materials.¹³ It has been reported that group-VI atoms such as Te and Se displace group-V atoms on the III-V surface and are energetically favored to form compounds with group-III atoms,^14,15 resulting in discontinuities of the crystal structure that degrade the subsequent epilayer.

In this study, nearly lattice-matched ZnSe/GaAs and ZnTe/GaSb quantum wells have been grown by molecular beam epitaxy (MBE) using a newly upgraded single-chamber system, and the material quality has been compared with a thicker ZnSe/GaAs double heterostructure grown using a dual-chamber MBE system. To maintain good material quality during the III-V quantum well growth, migration-enhanced epitaxy (MEE) and a substrate temperature ramp technique were utilized to minimize the defects caused by the formation of 3D islands near the interface. Abrupt interfaces between the II-VI and III-V layers were confirmed using transmission electron microscopy (TEM) and x-ray diffraction (XRD), and photoluminescence (PL) was observed from the heterovalent structures.

The quantum well samples were grown in a VG-V80 MBE system equipped with Ga, In, As, Sb, Cd, Zn, Te, and Se effusion cells in the growth chamber, which allows for more control of the growth conditions at the heterovalent interface. The cross contamination between the III-V and II-VI materials in the single-chamber system was minimized by reducing the idle temperature of the group-III cells during the II-VI layer growth, and by limiting the substrate temperature after deposition of the II-VI layers. No increase in the partial pressure of Zn, Se, or Te was observed on the growth chamber mass spectrometer during the III-V layer growth. The double heterostructure sample was grown in a dual-chamber VG-V80 MBE system with separate III-V and II-VI growth chambers to reduce cross contamination. The sample was transferred between the two growth chambers under ultrahigh vacuum conditions to minimize any surface adsorption that could contaminate the interfaces. A temperature ramp was used for the GaAs layer of the double heterostructure to achieve suitable growth conditions without damaging the ZnSe layers. Flux ratios were measured using a retractable ion gauge to relate the beam equivalent pressure to the cell temperatures and valve positions. The V/III flux ratios were calibrated using the change in the surface reconstruction observed by reflection-high-energy electron diffraction (RHEED) to locate the 1:1 stoichiometry point at the substrate temperature used for the buffer layer growth. The VI/II ratios were determined using RHEED oscillations with a constant Zn cell temperature to find the point at which the growth rate becomes Zn-limited. The substrate temperatures were measured using an optical pyrometer, and extrapolation of thermocouple/pyrometer results was used for temperatures below 300 °C.

The PL spectra were recorded with a SPEX 1404 double-grating spectrometer with a photomultiplier tube detector for the ZnSe/GaAs samples, and a Nicolet Magna-IR 760 Fourier Transform Infrared Spectrometer for the ZnTe/GaSb samples. The influence of absorption in the II-VI layers on the emission of the quantum wells was measured by using different laser wavelengths to toggle absorption in the barriers. Cross-sections of the samples were prepared for TEM imaging by standard mechanical polishing and argon-ion-milling at 2.0–2.5 keV with the sample held at liquid-nitrogen temperature to minimize ion-induced damage artifacts.¹⁶ The TEM observations were carried out with a Philips-FEI CM200 FEG TEM and a JEOL 2010F FEG TEM, both operated at 200 keV.

The ZnSe/GaAs double heterostructure sample was grown in a dual-chamber MBE system on a 2 in. n-type GaAs (001) substrate, as illustrated by the schematic in Fig. 1(a). The native substrate oxide was removed in the III-V chamber under As overpressure at a temperature of 610 °C until a 2 × 4 surface reconstruction was observed with RHEED. A 500 nm GaAs buffer layer was then grown at 580 °C at a growth rate of 10 nm/min under an As/Ga flux ratio of 1.5:1. After growth of the GaAs buffer layer, the substrate was ramped down under an As overpressure and then transferred to the II-VI chamber. Once in the II-VI chamber, the substrate temperature was increased to 280 °C before being exposed to a Zn overpressure to avoid Se–Ga bonding at the interface. The ZnSe deposition began with the application of Se flux at a Se/Zn flux ratio slightly higher than unity. The growth rate for ZnSe was calibrated for 10 nm/min using RHEED and XRD. The thickness of the II-VI barrier layers was 200 nm. The RHEED reconstruction transitioned immediately to a streaky 2 × 1 when the ZnSe growth was initiated, indicating a smooth interface formation. After the ZnSe barrier layer, the sample was moved back to the III-V chamber for the GaAs growth on ZnSe. The GaAs layer was initiated by opening the Ga shutter at a substrate temperature of 280 °C after a 3 min As soak at 250 °C. The substrate temperature was then ramped up to 500 °C at a rate of 15 °C/min as the GaAs was grown at a rate of 10 nm/min. Thus, the growth temperature reached 500 °C after approximately 150 nm. Indeed, the RHEED reconstruction started off spotty at the beginning of the GaAs layer growth at 280 °C, but became streaky at a thickness of around 150 nm, as shown in Figs. 1(b) and 1(c), indicating a smooth surface. The final ZnSe barrier was grown using a similar method as before to obtain an As–Zn bonding configuration at the interface.

In addition to the double heterostructure, ZnSe/GaAs quantum well samples were grown using a single-chamber MBE system with both II-VI and III-V growth occurring in the same chamber. The thickness of the GaAs well was too small to use a temperature ramp to obtain adequate GaAs layers starting from an initial growth temperature of 280 °C. Instead, MEE was utilized in tandem with a substrate temperature ramp that slowed the growth rate and simultaneously increased the Ga adatom diffusion length on the surface at low substrate temperatures. The MEE procedure is illustrated in Fig. 2. The layer structure of the ZnSe/GaAs quantum well samples was 20 nm of GaAs confined by 60 nm of ZnSe. The presence of As adatoms on the ZnSe surface was enhanced by utilizing a low substrate temperature As soak before the GaAs layer,¹⁷ whereby the substrate temperature was dropped to 75 °C for a 3 min As soak before initiation of the GaAs MEE growth by closing the As shutter and opening the Ga shutter for 1 monolayer at a time in between As deposition times of 10 s. The RHEED reconstruction showed a streaky 2 × 4 surface reconstruction at a substrate temperature of 500 °C after approximately 15 nm of GaAs growth.

ZnTe/GaSb quantum well structures were grown on ¼ 2 in. n-type GaSb substrates. The layer structure mirrored the ZnSe/GaAs sample, with 20 nm GaSb confined by 60 nm ZnTe. The substrate oxide was removed at 540 °C under Sb overpressure using sporadic Ga flashing to retain a smooth GaSb surface. A 500 nm GaSb buffer layer was then grown at a substrate temperature of 500 °C at a growth rate of 10 nm/min and V/III ratio of 1.5:1. The transition from GaSb to ZnTe growth was done in much the same way as for the ZnSe/GaAs samples. The GaSb growth was terminated by closing the Ga shutter and ramping the Ga cell temperature down to 700 °C. The substrate temperature was then ramped down to 300 °C under an Sb overpressure before closing the Sb shutter and opening the Zn shutter for 30 s. The substrate temperature was ramped up to 325 °C once the Zn flux was initiated, and Te flux was applied to begin the ZnTe deposition 30 s after the Zn shutter was opened while the temperature was still being ramped. The ZnTe growth rate was 10 nm/min, and a streaky Te-rich 2 × 1 surface reconstruction was observed by RHEED immediately after the growth was initiated. To mitigate the Te-rich growth conditions, a Zn overpressure was applied after the ZnTe growth to achieve a Zn-terminated c2 × 2 surface reconstruction. An Sb soak of 30 s on the ZnTe surface at a temperature of 325 °C was utilized to promote Sb–Zn bonding. Unlike the ZnSe/GaAs quantum well samples, the initial low-temperature group-V elemental soak procedure was not used for the ZnTe/GaSb quantum wells since the Sb soak at a substrate temperature of 325 °C was sufficient to achieve Sb–Zn bonding at the interface. During the GaSb well growth, the RHEED reconstruction shifted from a spotty pattern during the first few monolayers to a 3 × 1 GaSb surface after deposition of approximately 4 nm of GaSb. A constant substrate temperature of 325 °C was used throughout the growth of the GaSb layer on ZnTe. The second ZnTe layer was grown at the same temperature, and an immediate transition of the RHEED pattern to a streaky 2 × 1 was observed at the start of the layer growth.

XRD measurements of the heterostructures show strong peaks from the III-V and II-VI layers, and the Pendellösung fringes surrounding the peaks indicate abrupt interfaces. The double heterostructure sample did not utilize a low-temperature As deposition, but very good crystal quality was retained throughout the structure as evidenced by the narrow GaAs and ZnSe peaks along with the prominent fringes, as shown in Fig. 3(a). The quantum well samples, on the other hand, required a low-temperature As soak to achieve an adequate structural quality in the GaAs well layer. Pendellösung fringes and a prominent ZnTe diffraction peak observed in the ZnTe/GaSb quantum well sample suggest the 30 s Sb soak adequately covers the ZnTe surface to protect from Ga–Te bonding at the start of the GaSb layer growth [see Fig. 3(b)].

Extensive TEM observations were made to complement the XRD measurements. Examination of the ZnSe/GaAs quantum wells revealed {111}-type stacking faults in the initial ZnSe layer but these did not propagate into the GaAs layer, which was virtually defect-free. A bright-field image showing a typical area of the quantum well is shown in Fig. 4(a), and Fig. 4(b) is a high-resolution lattice-fringe image showing an enlarged view of the GaAs well. The excellent crystallinity of both ZnSe/GaAs and ZnTe/GaSb material systems is clearly evident, and the interfaces are well defined. There was no visible evidence for interfacial compounds or dislocations at the ZnSe/GaAs interface.

The PL spectra from the ZnSe/GaAs double heterostructure sample was measured using a 532 nm pump laser, with approximately 90% of the laser light absorbed in the GaAs layer between the ZnSe barriers. Two emission peaks, the expected 820 nm (1.51 eV) GaAs emission and another stronger, broader peak around 845 nm (1.47 eV), were observed in the double heterostructure grown in the dual-chamber system, as shown in Fig. 5. The below-band gap peak at 1.47 eV in the double heterostructure sample has been attributed to defect-related transitions in GaAs,¹⁸ and is likely related to the low substrate temperature during the initial growth of the GaAs layer. Measurements for the ZnSe/GaAs quantum well were made using a 405 and 532 nm pump laser to toggle absorption in the ZnSe barrier layers. Only a portion of the 532 nm light is absorbed by the GaAs well, whereas the 405 nm light is absorbed primarily in the ZnSe barriers and the GaAs well. The ZnSe/GaAs quantum well has a much broader emission spectrum but lacks the strong below-band gap peak that was observed in the double heterostructure sample. The FWHM linewidth of the quantum well peak was calculated to be approximately 53 meV. The broad peak linewidth can be primarily attributed to the nonuniform layer formation during the low-temperature GaAs quantum well growth process. No higher energy emission peaks were seen from the ZnSe barriers using the 405 nm pump laser, suggesting carriers are recombining nonradiatively in the defective ZnSe barrier layers.

The PL spectrum from the ZnTe/GaSb quantum well samples was measured using a 405 and 785 nm pump laser. Most of the 405 nm light is absorbed in the ZnTe barriers, whereas the ZnTe layers are transparent to the 785 nm light and therefore a significant portion of carriers are excited in the GaSb buffer. Strong below-band gap emission was observed from the quantum well samples, see Fig. 6. Additionally, a small peak corresponding to the 20 nm GaSb quantum well can be seen at 0.78 eV. PL emission was achieved at temperatures up to 75 K, but the emission from the quantum well was only seen at temperatures of 30 K and lower. The relative intensity of the quantum well emission peak compared to the substrate and defect peaks is higher for the 785 nm excitation wavelength compared to the 405 nm wavelength, likely due to defect sites in the ZnTe barriers. The linewidth of the quantum well peak was found to be 20 meV, significantly smaller than the ZnSe/GaAs sample grown in the same chamber. The effective Sb termination on the ZnTe surface compared to As on ZnSe plays an important role in the improved heterovalent interfaces, as the growth of the III-V layer can be initiated at a higher temperature without loss of group-V atoms at the interface.

Heterovalent II-VI/III-V heterostructures with well-defined interfaces were grown in single-chamber and dual-chamber MBE systems. Photoluminescence peaks from both the GaAs and GaSb quantum wells were observed, but the samples also showed significant below-band gap luminescence which can possibly be attributed to defects in the III-V material due to the low substrate temperature. The PL emission spectra from the ZnSe/GaAs double heterostructure is also dominated by below-band gap emission, suggesting that the low growth temperature at heterovalent interfaces remains a primary issue regardless of whether the growth of the II-VI and III-V materials is done in a single chamber or kept separate. Evidence of more abrupt interfaces in the ZnTe/GaSb quantum wells compared to the ZnSe/GaAs quantum wells grown in the same chamber is attributed to the higher initial III-V growth temperature for GaSb. Using short wavelength pump light to excite the II-VI barriers reduces the PL emission from the III-V quantum wells, likely because of the nonradiative recombination at the surface and in the stacking faults observed in the II-VI layers. XRD and TEM results showed abrupt interfaces between the III-V and II-VI layers, but TEM images also show stacking faults propagating in the II-VI layers. Quantification of the interdiffusion between these materials is still ongoing.

The authors gratefully acknowledge the use of the high-resolution x-ray diffraction system in the LeRoy Eyring Center for Solid State Science at Arizona State University, and the use of facilities in the John M. Crowley Center for High Resolution Electron Microscopy. This work was partially supported by AFOSR Grant No. FA 9550-15-1-0196.