InAs/InAsSb Strained-Layer Superlattice

• InAs/InAsSb strained-layer superlattices are Ga-free type-II III–V heterostructures with alternating InAs and InAsSb layers that create tunable bandgaps and spatially separated electron and hole minibands.
• Strain balancing in these superlattices not only enables coherent growth on GaSb substrates but also serves as a critical band-engineering variable, influencing effective masses and optical transitions.
• Advanced atomistic modeling and Raman spectroscopy reveal that interface quality, optical anisotropy, and defect-engineered planar isolation are pivotal for optimizing mid- to far-infrared photodetector performance.

Searching arXiv for recent and foundational papers on InAs/InAsSb strained-layer superlattices and related device physics. InAs/InAs$_{1-x}$Sb$_x$ strained-layer superlattices are Ga-free, antimony-based type-II III–V heterostructures in which alternating InAs and InAsSb layers form a periodic potential, producing electron and hole minibands, a tunable effective bandgap, and wavelength-selective infrared absorption. In this system, the conduction-band minimum resides primarily on the InAs layers while the valence-band maximum resides on the InAsSb layers, so electrons and holes are spatially separated in the type-II sense; this separation is central to the long-carrier-lifetime and dark-current arguments that motivate the platform for mid-, long-, and, in modeled structures, far-infrared detection (Dehzangi, 2022, Hussain et al., 2022, Dehzangi et al., 17 Jul 2025). The term “strained-layer” denotes deliberate balancing of tensile and compressive constituents across each period so that thick periodic stacks can be grown coherently, most commonly on GaSb, without immediate relaxation. Within the recent literature, the platform is positioned as a manufacturable alternative to HgCdTe and to InAs/GaSb type-II superlattices, while also serving as a testbed for atomistic band-structure theory, polarization-dependent optics, Raman-based structural diagnosis, and planar focal-plane-array processing (Dehzangi, 2022, Liu et al., 2017).

1. Material system, heterointerface, and strain-balanced design

The InAs/InAs$_{1-x}$Sb$_x$ superlattice belongs to the “6.1 Å family” of III–V zincblende semiconductors and is grown epitaxially on GaSb substrates. Its defining electronic feature is type-II band alignment: electrons localize in InAs quantum wells, whereas holes localize in InAs$_{1-x}$Sb$_x$ quantum wells. The literature reviewed here treats this carrier separation as the core reason the system is considered advantageous for infrared photodetectors, because it suppresses radiative and Auger recombination, supports longer minority-carrier lifetimes, and permits broad bandgap tunability through joint control of Sb fraction $x$ and layer thicknesses (Dehzangi et al., 17 Jul 2025).

The structure is “strained-layer” because InAs and InAs$_{1-x}$Sb$_x$ are not closely lattice matched across the full alloy range, even though GaSb provides a practical substrate. A standard design condition given for superlattices is that the net average strain vanish, written as $\sum_i t_i \varepsilon_i = 0$, with $_x$0; a more explicit bilayer stress-neutrality form reported for coherently strained periods is

$_x$1

where $_x$2 is the biaxial modulus for (001) films. In the InAs/InAsSb periods discussed for GaSb growth, InAs is tensile in-plane on GaSb, while InAs$_x$3Sb$_x$4 is compressive for $_x$5 (Dehzangi, 2022, Liu et al., 2017).

This strain balance is not merely a mechanical constraint. It is also a band-engineering variable, because the miniband energies, effective masses, and optical transition energies depend on the in-plane lattice constant and on how elastic relaxation redistributes strain between the two constituents. A plausible implication is that the superlattice should be understood as a coupled electronic–elastic system rather than as a simple sequence of isolated quantum wells.

2. Band structure, minibands, and bandgap control

In the type-II InAs/InAsSb superlattice, the effective bandgap $_x$6 is set by the energy separation between the first electron miniband $_x$7 and the first hole miniband $_x$8, including quantum confinement and interlayer coupling. The cutoff wavelength scales inversely with this gap according to $_x$9. The immediate engineering parameters are the InAsSb composition $_{1-x}$0 and the layer thicknesses per period, but substrate-induced strain adds a second control axis by shifting band edges and miniband dispersions (Dehzangi, 2022).

A first-principles study of an InAs/InAs$_{1-x}$1Sb$_{1-x}$2 superlattice with an 8.04 nm InAs segment and a 2.53 nm InAs$_{1-x}$3Sb$_{1-x}$4 segment, giving a 10.57 nm period along (001), treated the structure as fully strained at three in-plane lattice constants corresponding to InAs, GaSb, and AlSb. The calculated direct gap at $_{1-x}$5 decreases monotonically with increasing substrate lattice constant: $_{1-x}$6 eV for $_{1-x}$7 Å, $_{1-x}$8 eV for $_{1-x}$9 Å, and $_x$0 eV for $_x$1 Å, corresponding to wavelengths of approximately 10.69, 14.25, and 23.4 $_x$2m, respectively (Hussain et al., 2022). This establishes a direct strain route from LWIR into FIR response.

The same calculations found a direct $_x$3-point fundamental gap and a distinct multiband valence structure. Rather than a single heavy-hole band at the valence-band edge, the superlattice hosts two heavy-hole subbands at $_x$4, with the second heavier than the first in-plane and with both becoming nearly degenerate at $_x$5 because of zone folding. The lowest conduction band is predominantly In 5s in character and delocalized across the superlattice, whereas the highest valence band is predominantly Sb 5p on the In(As,Sb) side. In-plane electron masses remain small, while hole masses are strongly anisotropic; for the GaSb-lattice-constant case, the reported growth-axis values are $_x$6, $_x$7, and $_x$8 (Hussain et al., 2022). This combination of light in-plane carriers and very heavy cross-plane holes is directly relevant to detector design, because it implies strong lateral optical coupling together with suppressed transport along the growth direction.

A separate atomistic tight-binding study emphasizes the same band-engineering logic from a device-design perspective. It reports that increasing Sb fraction generally lowers the hole energy in the InAs$_x$9Sb$_{1-x}$0 layers and narrows the type-II gap, while thicker layers reduce inter-well coupling and also lower $_{1-x}$1. For a representative [(InAs)$_{1-x}$2–(InAs$_{1-x}$3Sb$_{1-x}$4)$_{1-x}$5] design grown along [001], the modeled superlattice bandgap is approximately 83 meV at 77 K, corresponding to $_{1-x}$6m, in excellent agreement with an experimentally reported 50% cutoff near 15 $_{1-x}$7m (Dehzangi et al., 17 Jul 2025).

3. Atomistic modeling, optical anisotropy, and interface non-idealities

Two complementary quantum-mechanical frameworks appear in the recent literature. One is a relativistic density-functional treatment using VASP with PAW potentials, a 300 eV plane-wave cutoff, spin–orbit coupling, and a modified Becke–Johnson exchange–correlation potential. For the InAs/InAs$_{1-x}$8Sb$_{1-x}$9 superlattice, bulk-tuned $_x$0 values were reported as non-transferable because they yielded metallic behavior; a single $_x$1 for all atoms was adopted to recover an $_x$2 eV consistent with experiment. Optical spectra were then calculated in the independent-particle RPA without local-field effects, with the authors explicitly noting that this overestimates dielectric constants and absorption and that Bethe–Salpeter corrections would red-shift $_x$3 peaks and reduce $_x$4 (Hussain et al., 2022).

The second framework is a modified $_x$5 empirical tight-binding method with spin–orbit interaction, a virtual-crystal approximation for ternary InAs$_x$6Sb$_x$7, and a bowing term applied only to the s on-site energy of the substituted sublattice. In this model, strain is incorporated geometrically through distorted nearest-neighbor vectors rather than through a continuum Bir–Pikus formalism. The average lattice constant is written as

$_x$8

with in-plane strain taken as $_x$9 and out-of-plane strain as $x$0. The same work introduces a simple Sb segregation model,

$x$1

using $x$2, $x$3, and $x$4, and states that segregation-induced bandgap shifts are smaller than uncertainties from tight-binding parameters or valence-band offsets but improve agreement across multiple structures (Dehzangi et al., 17 Jul 2025).

The optical response of the superlattice is strongly anisotropic. For the InAs/InAs$x$5Sb$x$6 structure, absorption is much larger for electric field polarized orthogonal to the growth axis than for polarization along the growth axis across all substrates studied. The reported static dielectric constants are $x$7 for the InAs-lattice-constant case, $x$8 for GaSb, and $x$9 for AlSb, with the anisotropy increasing as the in-plane lattice constant increases (Hussain et al., 2022). The dominant low-energy $_{1-x}$0 features arise from vertical transitions at $_{1-x}$1, and the entire low-frequency absorption spectrum is markedly higher for in-plane polarization. This suggests that detector architectures that preferentially couple fields with $_{1-x}$2 growth should realize a larger fraction of the available interband oscillator strength.

4. Device architectures and planar pixel isolation

For MWIR photodetection, the literature reviewed here focuses on Ga-free InAs/InAs$_{1-x}$3Sb$_{1-x}$4 absorbers embedded in unipolar-barrier designs, specifically nBn and pBn heterostructures. In the devices discussed, both detector types use the same MWIR absorber and the same barrier; only the top-contact polarity differs, with an n-type top contact for nBn and a p-type top contact for pBn. The barrier is a superlattice of AlAs$_{1-x}$5Sb$_{1-x}$6/InAs$_{1-x}$7Sb$_{1-x}$8 engineered to form a deep electron quantum well in the conduction band, thereby blocking majority-electron flow while allowing minority holes to be collected without a depletion junction at the surface (Dehzangi, 2022).

A central processing development is the use of ion implantation as a planar pixel-isolation method rather than for junction formation. Conventional SLS focal-plane arrays isolate pixels by deep mesa etching through absorber and barrier layers, followed by sidewall passivation. As pitch shrinks, that approach suffers from higher perimeter-to-area ratio, increased surface leakage, and more difficult passivation of narrow, deep mesas. The planar alternative uses implantation-induced defects to convert the material between diodes into a highly resistive state, effectively “drawing” isolation lanes into the epitaxial stack without deep etches. In the reported process, top metal is deposited first, an 800 nm SiO$_{1-x}$9 hard mask is then deposited over the full wafer, isolation lanes are opened lithographically, implantation is performed at 7° tilt with no cooling, a 15 s anneal at $_x$0 follows, and vias are finally opened to the top metal. Zn was used for pBn devices and Si for nBn devices; the tested energies were 100, 190, and 380 keV for pBn, and 100, 190, and 300 keV for nBn, each with doses of $_x$1, $_x$2, and $_x$3 (Dehzangi, 2022).

The optimized conditions identified empirically were 300 keV and $_x$4 for pBn, with an estimated ion-concentration peak depth of approximately 900 nm and straggle of approximately 100 nm, and 380 keV and $_x$5 for nBn, with an estimated depth of approximately 1000 nm and straggle of approximately 115 nm. Lower energies can leave only partial isolation if ions do not penetrate past the barrier into the absorber, so the electrically active area may exceed the nominal diode aperture; the reported consequence is artificially high optical signals, including QE $_x$6 anomalies (Dehzangi, 2022). A common misconception is therefore directly addressed in the literature: implantation here is not used to form the junction but to create resistive isolation lanes through damage-induced compensation and deep-level formation.

Electrical and optical performance under optimized conditions approached, but did not fully match, mesa-isolated controls from the same MBE wafer. For 200 $_x$7m circular diodes, the planar nBn device at $_x$8 mV showed $_x$9 and $\sum_i t_i \varepsilon_i = 0$0, while the planar pBn device at $\sum_i t_i \varepsilon_i = 0$1 mV showed $\sum_i t_i \varepsilon_i = 0$2 and $\sum_i t_i \varepsilon_i = 0$3. At $\sum_i t_i \varepsilon_i = 0$4 K, planar nBn dark current was approximately $\sum_i t_i \varepsilon_i = 0$5–$\sum_i t_i \varepsilon_i = 0$6 mesa and planar pBn approximately $\sum_i t_i \varepsilon_i = 0$7–$\sum_i t_i \varepsilon_i = 0$8 mesa; at 150 K, the planar devices exceeded mesa dark current by more than one order of magnitude. The optical penalty was smaller: the planar nBn device had a 50% cutoff $\sum_i t_i \varepsilon_i = 0$9m and peak responsivity $_x$00 A/W at $_x$01m at 77 K, rising to $_x$02 A/W with $_x$03m at 150 K, while the planar pBn device had $_x$04m and $_x$05 A/W at $_x$06m at 77 K, rising to $_x$07 A/W at 150 K. Saturated QE at the responsivity peak remained close to mesa values for nBn and somewhat lower for pBn, and QE was reported to remain consistent across diode diameters and temperatures under optimized implantation conditions (Dehzangi, 2022).

5. Raman spectroscopy, phonon modes, and structural diagnostics

Polarized Raman spectroscopy provides a non-destructive probe of strain, confinement, and vertical modulation in InAs/InAs$_x$08Sb$_x$09 superlattices on GaSb. A room-temperature study of coherently strained MBE and MOCVD structures concluded that the characteristic Raman response is governed by an InAs-like mode that behaves as a confined or quasi-confined mode at low Sb fraction and evolves into an extended superlattice mode as $_x$10 increases, together with a robust “forbidden” LO-like feature in cleaved-edge geometries that signals modulation-induced mode mixing (Liu et al., 2017).

The measurements were made in both (001) backscattering from the growth surface and (110) backscattering from the cleaved edge. Bulk reference frequencies were listed as TO $_x$11 cm$_x$12 and LO $_x$13 cm$_x$14 for InAs, and TO $_x$15 cm$_x$16 and LO $_x$17 cm$_x$18 for InSb. In representative superlattices, the dominant LO$_x$19 feature in (001) backscattering appeared near 236.4 cm$_x$20 for 3-2295 ($_x$21), 235.1 cm$_x$22 for B1871 ($_x$23), 234.4 cm$_x$24 for 3-2287 ($_x$25), and 235.7–236.4 cm$_x$26 for B1775 ($_x$27); in (110) cross polarizations, a prominent InAs-like TO$_x$28 near 214.7–215.4 cm$_x$29 was observed for $_x$30–0.33, with a weak InSb-like TO$_x$31 near 185–189 cm$_x$32 (Liu et al., 2017).

The “forbidden” parallel-polarization LO-like feature in cleaved-edge geometry is particularly significant. It appears near the LO$_x$33 frequency and is stronger in $_x$34 than in $_x$35, a behavior attributed to an $_x$36 Fröhlich interaction channel versus a $_x$37 deformation-potential channel. Its intensity relative to TO$_x$38 increases monotonically with $_x$39, and the paper treats it as an empirical indicator of vertical modulation strength (Liu et al., 2017). This is important for device physics because the same modulation that produces zone folding and mixed phonon character also shapes electron–phonon scattering channels relevant to transport and lifetime.

The strain interpretation is quantitative. The alloy lattice constant follows Vegard’s law,

$_x$40

and the out-of-plane strain under coherent growth is written

$_x$41

Phonon shifts under biaxial strain were modeled through hydrostatic and shear components,

$_x$42

Within that framework, the observed superlattice LO$_x$43 and TO$_x$44 positions fall between the limits for strained InAs and strained InAs$_x$45Sb$_x$46, supporting the conclusion that the dominant modes are not simple alloy modes but superlattice-specific confined-to-extended vibrations (Liu et al., 2017).

6. Array-level characterization, comparison with other IR technologies, and unresolved issues

At the focal-plane-array level, the literature includes laboratory characterization of commercial LWIR type-II SLS cameras used for precision photometry. The specific detector architectures and compositions in those units were not disclosed by the vendors, and the paper treats InAs/GaSb and InAs/InAsSb as typical SLS material systems rather than identifying the tested cameras as one or the other. What is directly established is that SLS FPAs operating at 77–80 K can exhibit substantial linear dynamic range and useful temporal stability under raw-count operation. The FLIR A6750sc SLS camera, with a 640 $_x$47 512 array, 15 $_x$48m pitch, and 14-bit output, showed an 11.0 dB linear dynamic range with slope $_x$49 counts/$_x$50s, intercept $_x$51 counts, and maximum radiometric drift $_x$52 over 4 h at $_x$53s. The Telops FAST V1k, also 640 $_x$54 512 but with 25 $_x$55m pitch and 16-bit output, showed a 12.2 dB linear dynamic range from 6 to 100 $_x$56s with slope $_x$57 counts/$_x$58s, intercept $_x$59 counts, and stability within $_x$60 for approximately 30–50 min, although variation over several hours reached approximately $_x$61 (Peterson-Greenberg et al., 2018).

These measurements matter because they separate detector-material questions from calibration and ROIC questions. The study did not report device-specific QE$_x$62, cutoff wavelength, dark current, conversion gain, read noise, interpixel capacitance, or persistence parameters, but it did show that SLS arrays can be operated within well-defined linear envelopes using

$_x$63

with a 1% nonlinearity criterion and the linear correction

$_x$64

For high-precision radiometry, this places emphasis on periodic recalibration cadence and raw-response monitoring rather than on any assumed intrinsic superiority of the material platform alone (Peterson-Greenberg et al., 2018).

In comparison with HgCdTe, the cited literature states that MCT still delivers superior QE and dark current and can cover all infrared bands, whereas current SLS dark current at approximately 78 K is approaching but has not yet matched the best HgCdTe devices and appears to remain Shockley–Read–Hall limited. The same literature also states that, if SRH lifetimes can be improved from approximately 30 ns to approximately $_x$65s, theoretical SLS dark current would beat HgCdTe by more than an order of magnitude across LWIR. In comparison with InAs/GaSb T2SL, the choice of InAs/InAsSb is motivated by superior minority-carrier lifetimes, simpler interfaces, and better material controllability and uniformity (Peterson-Greenberg et al., 2018, Dehzangi, 2022).

Several unresolved issues recur across the literature. For planar implanted detectors, electrical and optical cross-talk, array-level MTF impact, and complete FPA imaging metrics such as NETD and operability were not reported, and the authors call for further study of cross-talk suppression and defect control. For atomistic modeling, alloy disorder beyond VCA, interface roughness, chemical intermixing beyond simple graded profiles, and transport-linked quantities such as effective masses and oscillator strengths remain incomplete. For optics, local-field and excitonic effects were neglected in the reported first-principles spectra. For structural diagnostics, Raman established strong modulation-sensitive phonon signatures, but did not resolve discrete zone-folded acoustic peaks from which the period could be extracted directly. Taken together, these gaps indicate that InAs/InAs$_x$66Sb$_x$67 SLS research has progressed from material definition toward process integration and predictive modeling, but the decisive link between atomistic non-idealities, dark-current pathways, and full array performance remains an open research problem (Dehzangi, 2022, Hussain et al., 2022, Dehzangi et al., 17 Jul 2025).