As lower cost infrared imaging goals drive research in alternatives to state-of-the-art HgCdTe, the unparalleled bandgap engineering flexibility afforded by the heaviest group‑V element Bi offers a unique III-V analog to HgCdTe in InAsSbBi. By varying the mole fraction of constituent lattice-matched ternaries InAs_0.91Sb_0.09 (4 µm wavelength at 120 K) and InAs_0.93Bi_0.07 (12 µm wavelength), quaternary InAsSbBi spans the technologically relevant mid- to long-wave infrared spectrum and can be grown lattice-matched on large area GaSb substrates. Moreover, InAsSbBi’s compositional likeness to conventional bulk InAsSb and the InAs/InAsSb superlattice put it in a unique position to take advantage of recent technological innovation discovered and implemented in these related infrared systems. Molecular beam epitaxy grown InAsSbBi alloys are examined using temperature-dependent steady-state and time-resolved photoluminescence spectroscopy, X-ray diffraction, reflection high-energy electron diffraction (RHEED), and Nomarski imaging. RHEED patterns show that the InAsSbBi layer grows with a droplet-free (2×3) surface reconstruction at 380 °C. The surface is observed to remain specular and droplet free during growth, and Nomarski imaging further verifies that a smooth surface morphology is obtained. The InAsSbBi layers are 1 µm thick sandwiched between lattice-matched InAsSb layers providing carrier confinement. Comparison of the tetragonal distortion measured by X-ray diffraction and the bandgap energy measured by steady-state photoluminescence provides an evaluation of the Sb and Bi content of each sample. The target InAsSbBi mole fractions yielding 5 µm wavelength emission are 6.0% Sb and 2.2% Bi, and detailed examination of the photoluminescence properties and molecular beam epitaxy growth conditions show the growth progression towards this goal. Bandgap characteristics and recombination dynamics evaluated from the photoluminescence experiments are compared to equivalent 5 µm wavelength InAs/InAsSb superlattices and 4 µm wavelength lattice-matched InAsSb bulk samples.

GaAs_1-xBi_x alloys are challenging to grow by molecular beam epitaxy due to the surfactant-like nature and low solubility of bismuth (Bi) in this system. The key to good Bi incorporation in GaAsBi is to provide enough Bi flux to the surface to stabilize a surfactant layer, but not so much as to induce Ga-Bi droplet formation. The point at which Ga-Bi droplets form, or the Bi saturation point, limits the maximum Bi fraction obtainable with the growth conditions used. Samples grown with Ga-Bi droplets have reduced Bi composition and significant lateral and vertical phase separation. Strain-stabilization, by growing on partially relaxed InGaAs buffer layers, can be used to overcome Bi saturation in GaAsBi_0.07-0.09 films. We propose that reducing the compressive strain decreases the total energy of the system, allowing more Bi to incorporate and Ga-Bi droplet formation to be avoided. We have explored this trend for various GaAsBi epilayer thicknesses, as shown in Figure 1. By growing on In_0.105GaAs buffer layers, droplet-free films >100 nm can be achieved for Bi fractions x<0.09. This is an improvement over samples grown with high compressive strain on GaAs, where films of >100 nm are limited to compositions of x<0.07.

III-V semiconductors alloyed with dilute amounts of bismuth have been shown to have dramatically reduced bandgaps, making them well suited for optoelectronic applications in the mid- and far-infrared. Given that GaSb has a bandgap already in the near-IR range (Eg = 0.726 eV), GaSb_1-xBi_x has great potential for pushing out to these long wavelength regimes. Unfortunately, bismuth tends to surface segregate and form droplets at sufficiently high Bi flux rather than incorporate into the growing film. The highest Bi content GaSb_1-xBi_x reported to date had x=0.14, and only x=0.11 has been achieved without droplets forming on the film surface [1]. We have recently observed that droplets are also likely to form upon reaching some critical thickness for a given Bi content in GaAsBi films, and these droplets can lead to significant lateral and vertical phase separation in the GaAsBi system.

We have found the solubility of Bi in GaSb to be different than the solubility of Bi in GaAs. The same Bi flux that leads to x=0.0315 in GaAs_1-xBi_x (grown at 250^oC) leads to negligible Bi incorporation in GaSb(Bi) (grown at 285^oC). A comparison of the GaSb(Bi) and GaAsBi surfaces is shown in Fig. 1. This work seeks to determine the effect of increasing bismuth flux and film thickness on surface morphology and film homogeneity in GaSbBi films as compared to GaAsBi. Samples were grown on GaSb substrates in a Veeco GENxplor MBE system using a valved Sb cracker and solid source effusion cells for Ga and Bi, with growth monitored in-situ via RHEED. Bismuth content was determined via HRXRD and confirmed for select samples with Rutherford backscatter spectrometry. The presence/absence of droplets was determined via optical microscopy and SEM imaging, while cross-sectional film homogeneity and phase separation was examined via TEM

[1] O. Delorme, L. Cerutti, E. Tournie, J.-B. Rodriguez. J. Cryst. Growth, 447 (2017)

Long wavelength IR III-V based devices have long been of interest for potential applications such as chemical sensing and large format IR imaging. Within the III-V family, extending to the longest wavelength requires the use of InAs_1-xSb_x (x ≤ 0.6) which offers the lowest bandgap energy (E_g) ranging from 0.05-0.35 eV [1]. However, the lack of conventional substrate at the desired lattice constant has restricted progress on the growth and study of this material system. To overcome this limitation, we employ a metamorphic step-graded InAs_1-xSb_x buffer on GaSb, enabling the study of low-E_g InAs_1-xSb_x as a function of growth conditions. Using this method, we previously presented the effect of substrate temperature (T_sub) and V/III on Sb incorporation of the lowest-E_g cap layer [2]. Here, we investigate the effect of V/III on Sb incorporation as a function of x and use x-ray reciprocal space mapping (RSM) to examine the effect of growth conditions on strain and dislocation dynamics.

We grew several InAs_1-xSb_x step-graded structures in which the Sb/(As+Sb) flux ratio was varied from 0.05 to 0.50 in 0.05 increments, under various T_sub and V/III, and identified the Sb composition in each layer using RSM along [110] with (004) and (115) reflections. This allows comparison of Sb-content as a function of Sb/(As+Sb) for various V/III, given in Fig. 1. These results suggest that V/III has little effect on Sb incorporation, in direct conflict with our previous photoluminescence (PL) results [2]. To understand the discrepancy between PL and RSM, we measured (004) RSM of the same three samples with the x-ray beam incident along [1-10], revealing extremely different strain relaxation compared to the [110] case. Asymmetric strain relaxation has been observed in other III-V graded buffer systems and has been explained by different dislocation formation energies and glide velocities along each direction resulting from the core structure of the dislocation being terminated with either a group-III or a group-V element [3]. To develop an understanding of this mechanism within InAs_1-xSb_x graded buffers, we will employ the x-ray analysis of Ayers [4] to quantify the threading dislocation density (TDD) in these films along both [110] and [1-10], thus enabling a comparison of TDD as a function of growth conditions, thickness, and propagation direction. AFM and TEM will further support these results. Taken together, we will develop a picture of dislocation dynamics during the growth of metamorphic InAs_1-xSb_x.