Photoluminescence and minority carrier lifetime of quinary GaInAsSbBi grown on GaSb by molecular beam epitaxy Free

The need for manufacturable high performance mid- to long-wave infrared technologies is growing due to their applications in thermal imaging,¹ gas sensing,² and growing potential in noninvasive medical detection/diagnostics.³ The telecommunication industry has taken advantage of the high manufacturability of near- to short-wave infrared III–V semiconductor materials to support the ever increasing performance requirements of our world's telecommunications networks. Given the pervasiveness of this industry and its ever-expanding markets, there would be great value in leveraging that same industrial base for mid-wave infrared optoelectronic applications. In fact, this is the advantage held by III–V mid-wave infrared superlattice materials, such as strain-balanced InAs/InAsSb and InGaAs/InAsSb, which are highly manufacturable due to their high yield and use of large-area commercially available GaSb and GaAs substrates, and can be produced with technologically relevant minority carrier lifetimes (1–10 μs).^4,5

While the bandgap engineering flexibility of superlattice material systems has furthered our capabilities in the mid-wave infrared,⁶ superlattices gain this tunability at the expense of electron and hole wavefunction coupling and localization effects due to superlattice thickness variations,^7,8 which leads to lower absorption coefficients⁹ and to very low vertical mobility,¹⁰ respectively. Given these challenges, a more ideal mid-wave infrared solution would be a bulk III–V alloy with sufficient tunability in its bandgap and band edge alignments to form effective heterostructures with other alloy systems at the lattice constant of a large-area commercially available substrate. One potential design space that could enable this technology is an alloy of Bi in a III-AsSb material. Bi is the largest, non-radioactive and low-toxicity group-V element that dramatically reduces the bandgap of InAs at a rate of ∼55 meV/% Bi.¹¹ Incorporating into InAsSb(Bi), the quaternary material takes advantage of that large bandgap tunability while maintaining a lattice-match to the nearby GaSb substrate which provides a multitude of neighboring alloys to form heterostructures.^12–15 All that remains then is achieving acceptable minority carrier lifetimes, which can be realized with the inclusion of Ga into the quinary alloy GaInAsSbBi.

In order to evaluate GaInAsSbBi as a candidate for mid-wave infrared detection, a sufficient quantity of Bi should be incorporated to achieve a 5 μm wavelength cutoff (∼2% Bi mole fraction for a 120 K operating point, InAsSb has a 3 μm cutoff for comparison).^14,15 Fundamental optoelectronic quality metrics, such as the alloy's minority carrier lifetime τ_mc, need to be investigated. The minority carrier lifetime is a strong function of the growth conditions utilized during the material's growth, with particularly strong dependence on the growth temperature.¹⁶ Low growth temperatures below 350 °C have traditionally been utilized to facilitate more efficient Bi incorporation in InAsBi and InAsSbBi;^14,15 however, photoluminescence has yet to be observed in thick bulk layers grown under these conditions. In contrast, InAsSbBi grown at higher temperatures (between 360 and 380 °C) exhibits strong photoluminescence and minority carrier lifetimes comparable to equivalent InAsSb samples grown at the same temperatures (∼100s of ns). However, Bi incorporation is severely reduced at these temperatures (∼0.2–0.5% mole fraction incorporated to date),¹² and Bi incorporation efficiency will be further reduced at temperatures >400 °C where the longest lifetimes can likely be achieved.⁹ This leaves InAsSbBi in a compromised position where it is difficult to simultaneously achieve sufficient Bi incorporation to reach a 5 μm mid-wave infrared cutoff while maintaining a technologically relevant minority carrier lifetime (>1 μs).

The addition of Ga in GaInAsSbBi may present a solution to this impasse in InAsSbBi. GaAsBi with ∼1% mole fraction of Bi has been grown at 400 °C, and higher Bi incorporation rates have been demonstrated at similar temperatures.^17,18 This suggests that Bi forms a stronger bond to Ga than to In at ∼400 °C growth temperatures, possibly due to Ga's smaller atomic radius, which lends itself to an increase in atomic orbital overlap. Thus, co-alloying Bi with Ga could result in a lower alloy formation energy and better dilute Bi incorporation efficiency as a result.¹⁹ Here, we demonstrate a quinary GaInAsSbBi alloy grown at 400 °C by molecular beam epitaxy, which achieves the same cutoff wavelength due to Sb and Bi incorporation as a quaternary InAsSbBi alloy grown under similar flux conditions at 360 °C. The quinary's substantially longer minority carrier lifetime is consistent with the higher growth temperature. The structural and optical properties are examined by Nomarski interference contrast microscopy, x-ray diffraction, Rutherford backscattering spectroscopy, steady-state photoluminescence, and time-resolved photoluminescence. A recombination rate analysis is performed on the temperature dependent minority carrier lifetime to determine the intrinsic doping density of the quinary GaInAsSbBi and quaternary GaInAsSb reference.

The samples are grown on (100)-oriented n-type GaSb substrates using a VG-V80H molecular beam epitaxy system with valved group-V sources. The growth temperatures are measured using a Fluke Endurance series emissivity-corrected pyrometer, model E2ML sensing at 1.6 μm with a minimum temperature limit of 250 °C. The emissivity setting is 0.655, calculated using the refractive index of GaSb. The quaternary and quinary GaInAsSb(Bi) samples are sandwiched between a 500 nm thick buffer and a 30 nm thick cap of lattice matched InAsSb. The 30 nm thickness of the cap layer is chosen so that it is sufficiently thick to provide confinement to photogenerated carriers, yet thin enough to produce a spectral range, which is uniquely characteristic of Bi in the Rutherford backscattering yield spectrum. The quaternary InAsSbBi sample from Ref. 12 was sandwiched between a 400 nm thick buffer and a 100 nm cap of lattice matched InAsSb.

Before a quinary is grown with desired constituent mole fractions, group-III growth rates and V/III flux ratios are calibrated,²⁰ followed by three alloy composition calibration growths. The first calibration is a lattice-matched InAsSb layer grown at 440 °C with an In-limited growth rate of 1 μm/h, an As/In flux ratio of ∼1.4, and an Sb/In flux ratio of ∼0.11 to achieve lattice-match on the GaSb substrate and produce a high quality InAsSb reference. These growth conditions are repeated for the second composition calibration, with the growth temperature reduced to 400 °C and As/In flux ratio reduced to near unity, resulting in increased Sb incorporation and intentionally compressive InAsSb. Next, Ga is introduced in the third composition calibration with a target 0.027 μm/h Ga growth rate and 0.973 μm/h In growth rate to compensate the compressive strain, producing lattice-matched GaInAsSb. The total group-III growth rate target is 1 μm/h so that the Sb/III and As/III flux conditions are maintained. Finally, the lattice-matched quaternary GaInAsSb is grown again, but with the addition of Bi to grow quinary GaInAsSbBi. Table I summarizes the growth conditions and the (004) layer-substrate peak separation (in arcseconds) of each layer measured by x-ray diffraction.

Smooth, droplet-free surfaces are observed for all samples with Normarski imaging. The inset in Fig. 1 shows the Rutherford backscattering spectroscopy yield signal (red curve), and the backscattering model fit of 0.13% ± 0.02% Bi mole fraction in the GaInAsSbBi sample (black curve) leading to a predicted 7.15 meV redshift in bandgap. To maximize RBS sensitivity to Bi, a 5 MeV Fe³⁺ beam is employed. An annular partially depleted silicon surface barrier detector is placed at 180 degrees and approximately 5 μC of charge is accumulated for both the reference and quinary samples at room temperature. Figure 1 shows the (004) x-ray diffraction pattern of the GaInAsSb sample, which exhibits −22 ˝ compressive strain (blue curve), while the GaInAsSbBi (red curve) is slightly more compressive at −54 ˝ due to the incorporation of 0.13% Bi. The tetragonal distortion inferred from the measured strain is used to determine the As and Sb mole fractions of the samples, given the calibrated group-III fluxes and Bi mole fraction from Rutherford backscattering, resulting in alloy compositions of Ga_0.029In_0.971As_0.882Sb_0.118 in the quaternary and Ga_0.029In_0.971As_0.883Sb_0.116Bi_0.001 in the quinary. Despite the small degree of strain observed, the 1 μm thick samples are still well within the critical thickness and exhibit high structural and interface quality as evidenced by the Pendellösung fringes in both the quaternary Ga_0.029In_0.971As_0.882Sb_0.118 and quinary Ga_0.029In_0.971As_0.883Sb_0.116Bi_0.001. In contrast, the InAs_0.911Sb_0.081Bi_0.008 sample grown at 360 °C in Ref. 9 is still closely lattice-matched with slightly greater compressive strain at −81 ˝ (green curve); however, the Pendellösung fringes are characteristic of only the cap layer of the sample.

A comparison of the Bi mole fractions in the Ga-free quaternary InAs_0.911Sb_0.081Bi_0.008 grown at 360 °C and quinary Ga_0.029In_0.971As_0.883Sb_0.116Bi_0.001 grown at 400 °C provides only marginal evidence for the enhanced Bi incorporation efficiency in the presence of Ga. Detailed analysis of the Bi incorporation in InAsSbBi as a function of growth temperature indicates that the Bi sticking coefficient decreases with the increasing growth temperature with a characteristic slope of 20.56 °C.¹¹ Thus, the 40 °C increase in growth temperature should result in a factor of 0.14 reduction in Bi incorporation coefficient, resulting in an expected 0.11% Bi in an equivalent InAsSbBi alloy grown at 400 °C. Rutherford backscattering analysis shows that the Bi mole fraction is 0.13% ± 0.02% in the quinary Ga_0.029In_0.971As_0.883Sb_0.116Bi_0.001 (see Fig. 1 inset), indicating that if Ga enhances the Bi incorporation efficiency, then the 2.9% Ga flux used in this sample was too low to conclusively observe the effect. This finding suggests that future growths with higher Ga concentrations should be investigated.

Steady-state photoluminescence is measured from the GaInAsSb(Bi) samples using a Bruker 80 V Fourier transform infrared spectrometer and a 785 nm wavelength pump laser. Double modulation is utilized to increase the signal to noise ratio, with the laser modulated at 50 kHz.²¹ The steady-state 120 K photoluminescence spectra of the GaInAsSb(Bi) samples are shown in Fig. 2 along with the quaternary InAs_0.911Sb_0.081Bi_0.008 sample for comparison. The red curve in Fig. 2 corresponds to the Ga_0.029In_0.971As_0.883Sb_0.116Bi_0.001 sample grown with the same growth conditions as the quaternary Ga_0.029In_0.971As_0.882Sb_0.118 alloy (blue curve), but with Bi introduced during growth. The 6 meV red shift in the rising edge of the photoluminescence is consistent with the measured 0.13% Bi mole fraction measured by Rutherford backscattering (Fig 1 inset), and results in a cutoff wavelength identical to InAs_0.911Sb_0.081Bi_0.008 due to the latter's lower Sb content. However, the photoluminescence signal of the quinary Ga_0.029In_0.971As_0.883Sb_0.116Bi_0.001 is brighter (∼9% higher integrated photoluminescence intensity) than that of the quaternary InAs_0.911Sb_0.081Bi_0.008 grown at 360 °C [0.12 μs (Ref. 12)], affirming the benefit of utilizing a higher growth temperature.

The temperature dependent minority carrier lifetime of the quaternary and quinary alloys is measured using time-resolved photoluminescence. The GaInAsSb(Bi) samples are pumped with a 1535 nm (0.81 eV) pulsed laser with samples mounted in a liquid nitrogen cooled cryostat. The laser pulses are 3.5 ns long, and the excitation is varied to inject 10¹¹–10¹² photons/cm² per pulse into the GaInAsSb(Bi) active region using a motorized half-waveplate compensator and polarizing beam splitter combination. The photoluminescence signal is collected and collimated with a 2 in. diameter f/2 90° off-axis parabolic mirror and then focused with a second off-axis parabolic mirror, transmitted through a 2.4 μm cutoff long-pass filter, and measured by a 6 μm cutoff VIGO Systems PVI-4TE detector. A Teledyne Lecroy HD 4096 oscilloscope averages 100 000 time-resolved photoluminescence decays to acquire one photoluminescence decay signal per excitation condition per temperature from 77 to 300 K. An optical schematic of the system can be found in Ref. 22.

The excitation conditions are selected to establish low-injection conditions in the samples. Supposing all the photoexcited electron–hole pairs distribute across the 1 μm active region absorber, then the lowest excitation of 10¹¹ photons/cm² per pulse reaching the absorber results in an initial carrier density of 10¹⁵ electron–hole pairs/cm³. This is just higher than the mid-high 10¹⁴ cm⁻³ background carrier density determined by the recombination rate analysis in the materials indicating low-injection conditions are quickly established after the initial excitation. As the excitation level is increased, non-single exponential decay is observed at short time scales, indicating high injection conditions and a corresponding transient reduction of the lifetime. However, even in the high injection case, after the short transient of non-single exponential decay, the system returns to low-injection behavior with a characteristic slope consistent with low-injection pumping.

The minority carrier lifetime is determined as a function of temperature by fitting the characteristic slope of low-injection regime photoluminescence signal with a single exponential decay. Figure 3 shows the time-resolved photoluminescence decays for the quaternary Ga_0.029In_0.971As_0.882Sb_0.118 and quinary Ga_0.029In_0.971As_0.883Sb_0.116Bi_0.001 samples under low excitation conditions at 117 K. The black curves are the photoluminescence decay data, while the subset of gray points show the data used to fit the single exponential slope in the low excitation range. The red and blue dashed lines are the best fit exponential slopes for the quinary and quaternary, respectively. It can be seen by the photoluminescence signals and slope fits that the quinary Ga_0.029In_0.971As_0.883Sb_0.116Bi_0.001 has a longer minority carrier lifetime (0.34 μs) than the quaternary (0.09 μs).

The recombination rate analysis is performed on the GaInAsSb(Bi) samples by fitting the three recombination rate mechanisms to the temperature dependent minority carrier lifetime determined by

In Eq. (1), the minority carrier lifetime $τmc$ is modeled as a function of the radiative lifetime $τrad$ scaled by the photon recycling factor $ϕ$⁠, the Shockley–Read–Hall (SRH) lifetime $τSRH$⁠, and the Auger lifetime $τAuger$⁠. Further description of the recombination rate analysis can be found in Ref. 4 and references therein.

Figure 4 shows the temperature dependent minority carrier lifetime of the samples grown in this study alongside the solid curves showing the resultant recombination rate fit in Eq. (1). The temperature-dependent minority carrier lifetime of the quaternary Ga_0.029In_0.971As_0.882Sb_0.118 (circles and blue curve) is lower than the quinary Ga_0.029In_0.971As_0.883Sb_0.116Bi_0.001 (squares and red curve). This could be due to the intrinsic defects introduced in the InAsSb ternary by adding Ga into the system.²³ Once Bi is introduced in the quinary growth, however, the minority carrier lifetime increases by >3× in the SRH-limited regime, possibly a result of the surfactant behavior of Bi,^13,24 an additional benefit of Bi being introduced during growth.

The 0.34 μs lifetime observed in the quinary grown at 400 °C is consistent with the trends in Ref. 9, where the lifetime in the InAsSbBi alloys grown in the same molecular beam epitaxy system is consistently just under the expectation for lattice-matched InAs_0.91Sb_0.09 (0.39 μs at 400 °C). While the conclusions that can be drawn from comparisons of material quality metrics like minority carrier lifetime in samples grown at different times are typically limited, careful periodic benchmarking of the lifetime of material produced by this molecular beam epitaxy system enables the comparisons here. The minority carrier lifetime of a mid-wave infrared InAs/InAsSb superlattice benchmark structure has been carefully tracked over time and is comparable at a Shockley–Read–Hall-limited lifetime of 2 μs during the growth campaigns which produced the GaInAsSb(Bi) samples examined here as well as the Ga-free InAs_0.911Sb_0.081Bi_0.008 sample from Ref. 9. This indicates that the state of the molecular beam epitaxy system and quality of the mid-wave infrared InAsSb-based material produced in these time frames are comparable.

Table II provides the best-fit parameters to the temperature dependent minority carrier lifetime. The majority carrier concentration determined by a recombination rate analysis of the time-resolved photoluminescence becomes increasingly less sensitive as the majority carrier concentration decreases (particularly in undoped samples); however, this approach is sufficient for the purposes of this report.^4,6 The defect levels of both samples seem to be shallow toward the conduction band. The defect cross section-concentration product is consistent with the observed minority carrier lifetime increase in the quinary GaInAsSbBi sample, further highlighting that Bi may have aided in growing a higher quality material with fewer defects. The Bloch overlap parameter |F₁F₂| typically has large influence on the minority carrier lifetime at higher temperatures where the Auger recombination intrinsically dominates. However, the fact that the minority carrier lifetime of the quaternary GaInAsSb is heavily dominated by the SRH recombination could suggest that the model is less sensitive to the Bloch overlap parameter in these samples.

More growth and minority carrier lifetime studies are required to further understand the benefits of adding Ga and understand its relationship with Bi incorporation efficiency. Evaluation of the photoluminescence spectra and time-resolved photoluminescence decays show that Bi extends the cutoff of GaInAsSb to longer wavelengths and results in improved minority carrier lifetime in quinary GaInAsSbBi. InAsSbBi can achieve the same cutoff wavelength but requires higher Bi mole fractions achieved with lower temperature growth conditions that degrade its minority carrier lifetime. The findings presented here provide evidence that the inclusion of Ga in quinary GaInAsSbBi may be the enabling factor to achieve long minority carrier lifetime mid-wave infrared material; however, whether or not Ga significantly modifies the Bi sticking coefficient remains to be determined.

In conclusion, quinary GaInAsSbBi is grown at 400 °C, evidenced by Rutherford backscattering. The quinary achieves a similar cutoff wavelength as a quaternary InAsSbBi grown under similar flux conditions at 360 °C as seen by steady-state photoluminescence. An increase in growth temperature improves the minority carrier lifetime up to 0.34 μs at 120 K but significantly decreases the Bi incorporation efficiency for InAsSbBi. Incorporating Ga enables growth of GaInAsSbBi at higher temperatures than InAsSbBi, where longer minority carrier lifetimes and similar cutoffs are attainable, opening a new design space to explore for mid-wave infrared sensing applications.

The authors acknowledge financial support through research sponsored by the Air Force Research Laboratory, Contract No. 9453-19-2-0015. This work was performed, in part, at the Center for Integrated Nanotechnologies, an Office of Science User Facility operated for the U.S. Department of Energy (DOE) Office of Science. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy's National Nuclear Security Administration under Contract No. DE-NA0003525. The views expressed in the article do not necessarily represent the views of the U.S. DOE or the United States Government. The authors would like to thank Professor Michael P. Short for his support and assistance. Approved for public release: distribution is unlimited. AFMC PA No. 2021-3891.