---
title: Asymmetric GaInSb/AlGaAsSb Quantum Wells
url: https://www.emergentmind.com/topics/asymmetric-gainsb-algaassb-quantum-wells
type: topic
---

# Asymmetric GaInSb/AlGaAsSb Quantum Wells

Asymmetric GaInSb/AlGaAsSb quantum wells are GaSb-based type-I heterostructures in which the confinement potential is intentionally made nonuniform across the active region. In the reported literature, that asymmetry takes at least three distinct forms: thickness-modulated interband quantum-well stacks for broadband gain beyond \(2\,\mu\mathrm{m}\), composite or stepped GaInSb wells with ultrathin AlInSb insertions for emission near \(1.55\,\mu\mathrm{m}\), and structurally asymmetric intersubband wells for dipole-allowed polaritonic transitions in the terahertz regime. Across these implementations, the central objective is to tune quantized transition energies, reshape carrier confinement, suppress leakage channels, or lift symmetry-forbidden optical selection rules [2508.17495][2012.13369][1204.4053].

## 1. Structural realizations of asymmetry

In GaSb-based waveguide amplifiers operating beyond \(2\,\mu\mathrm{m}\), asymmetry is implemented by combining Ga\(_{0.73}\)In\(_{0.27}\)Sb wells of different thicknesses inside Al\(_{0.25}\)Ga\(_{0.75}\)As\(_{0.02}\)Sb\(_{0.98}\) barriers. The reported active-region sets include single wells with \(L = 7\)–\(13\) nm, double quantum wells composed of one 7 nm well plus one thicker well, and a multi-quantum-well “Structure C” comprising one 7 nm well plus three 13 nm wells. The wells are undoped, the inter-well barriers are 20 nm Al\(_{0.25}\)Ga\(_{0.75}\)As\(_{0.02}\)Sb\(_{0.98}\), and the active region is embedded in a separate confinement heterostructure with 130 nm Al\(_{0.25}\)Ga\(_{0.75}\)As\(_{0.02}\)Sb\(_{0.98}\) waveguide layers on each side. Refractive indices are reported as \(n \approx 3.89\) for the GaInSb wells, \(n \approx 3.62\) for the waveguide, and \(n \approx 3.487\)–\(3.489\) for the claddings, giving the index ordering \(n_\mathrm{QW} > n_\mathrm{waveguide} > n_\mathrm{cladding}\) required for vertical optical confinement [2508.17495].

Near \(1.55\,\mu\mathrm{m}\), the reported “asymmetry” is of a different kind. The outer Al\(_{0.35}\)Ga\(_{0.65}\)As\(_{0.03}\)Sb\(_{0.97}\) barriers on either side of the well have the same composition; instead, the quantum well itself is segmented by ultrathin Al\(_{0.68}\)In\(_{0.32}\)Sb insertions. Device A uses 3.6 nm Ga\(_{0.8}\)In\(_{0.2}\)Sb wells with no internal barrier insertion, Device B uses 4.8 nm wells with one 0.45 nm Al\(_{0.68}\)In\(_{0.32}\)Sb monolayer inside each well, and Device C uses 6.0 nm wells with two 0.45 nm Al\(_{0.68}\)In\(_{0.32}\)Sb monolayers inside each well. The cladding is Al\(_{0.9}\)Ga\(_{0.1}\)As\(_{0.07}\)Sb\(_{0.93}\), lattice matched to GaSb, and the design intent is to increase confinement while preserving emission near \(1.55\,\mu\mathrm{m}\) [2012.13369].

In the intersubband-polaritonic context, asymmetry is defined more abstractly by the condition \(\Delta z = z_{22} - z_{11} \neq 0\), where \(z_{11} = \langle \psi_1 | z | \psi_1 \rangle\) and \(z_{22} = \langle \psi_2 | z | \psi_2 \rangle\). The data identifies stepped wells, graded barriers, and a DC bias as routes to such asymmetry in GaInSb/AlGaAsSb systems. This establishes a useful distinction: in antimonide interband devices, asymmetry may be encoded in thickness or stepped confinement, whereas in intersubband cavity systems it is the broken parity of the envelope functions that is decisive [1204.4053].

## 2. Band structure, strain, and symmetry breaking

The GaInSb/AlGaAsSb material system is reported as type-I at \(\Gamma\), with both electrons and holes confined in the GaInSb wells. In the broadband \(>2\,\mu\mathrm{m}\) structures, Ga\(_{0.73}\)In\(_{0.27}\)Sb wells are compressively strained on GaSb, while the AlGaAsSb barriers and waveguide are lattice matched and relatively strain-neutral. Compressive strain lifts the heavy-hole band above the light-hole band, so the TE-polarized \(E1\)–\(HH1\) transition dominates at moderate carrier densities, while the \(E1\)–\(LH1\) transition appears at higher densities and adds shorter-wavelength gain components. The reported idealized subband scaling is

$$
E_n \approx \frac{\hbar^2 \pi^2 n^2}{2 m^\ast L^2},
$$

with finite barriers treated self-consistently in the “Harold” solver. The lowest TE-dominant interband transition is written as

$$
E_{cv} \approx E_g^{(\mathrm{well})} + E_{e,1} + E_{hh,1} - \Delta_{\mathrm{strain}},
$$

and the two-dimensional density of states per subband is

$$
g_{2D} = \frac{m^\ast}{\pi \hbar^2}.
$$

These relations explain the thickness dependence reported for the wells: a thinner well at \(L \approx 7\) nm produces larger confinement energies and a shorter \(\lambda_\mathrm{peak}\) near \(1.97\)–\(1.99\,\mu\mathrm{m}\), whereas a thicker well at \(L \approx 13\) nm produces a longer \(\lambda_\mathrm{peak}\) near \(2.09\)–\(2.10\,\mu\mathrm{m}\) [2508.17495].

The \(1.55\,\mu\mathrm{m}\) CQW lasers show the same basic strain logic. Ga\(_{0.8}\)In\(_{0.2}\)Sb wells are compressively strained on GaSb; Al\(_{0.35}\)Ga\(_{0.65}\)As\(_{0.03}\)Sb\(_{0.97}\) barriers and Al\(_{0.9}\)Ga\(_{0.1}\)As\(_{0.07}\)Sb\(_{0.93}\) claddings are lattice matched to GaSb. The conduction-band offsets and intervalley separations relative to the electron ground state \(e1\) are reported at room temperature and ambient pressure as \(e1\)–\(\Gamma_\mathrm{barrier} = 218\)–223 meV, \(e1\)–\(X_\mathrm{barrier} = 275\)–280 meV, \(e1\)–\(L_\mathrm{barrier} = 299\)–304 meV, and \(e1\)–cladding \(\Gamma > 439\) meV. The same compressive strain raises the heavy-hole band, increases HH–LH splitting, and favors TE polarization, while TM is suppressed [2012.13369].

In the intersubband strong-coupling formulation, symmetry is central rather than incidental. In a centro-symmetric well, \(z_{11} = z_{22}\) and therefore \(\Delta z = 0\); the upper-polariton to lower-polariton transition is then dipole forbidden. In an asymmetric well, \(\Delta z \neq 0\), and the interbranch dipole becomes

$$
d_{+-} = y_{\mathrm{UP}} y_{\mathrm{LP}} \Delta z,
$$

where \(y_{\mathrm{UP}}\) and \(y_{\mathrm{LP}}\) are the matter fractions of the upper and lower polaritons. This is the formal statement of how structural asymmetry lifts the selection rule that forbids electric-dipole transitions between different cavity polariton branches in centro-symmetric systems [1204.4053].

## 3. Broadband interband gain beyond \(2\,\mu\mathrm{m}\)

The principal reported application of asymmetric GaInSb/AlGaAsSb wells is broadband interband gain in GaSb-based semiconductor amplifiers and superluminescent diodes beyond \(2\,\mu\mathrm{m}\). The design strategy is to combine a thin well and one or more thick wells so that the \(E1\)–\(HH1\) transitions of the different wells overlap spectrally, and then, at higher injection, to admit an \(E1\)–\(LH1\) contribution from the thick wells. In the reported analysis, this produces a flat and wide gain spectrum within the \(2\)–\(2.2\,\mu\mathrm{m}\) window, with extension toward shorter wavelength at higher carrier density [2508.17495].

For the double-quantum-well case consisting of one 7 nm well and one 13 nm well at \(20^\circ\)C, balanced flat gain is reported at \(N \approx 1.2 \times 10^{18}\,\mathrm{cm}^{-3}\), corresponding to \(\approx 0.34\) A in a \(2\) mm \(\times\) \(80\,\mu\)m broad-area device. The full-width at half-maximum is \(\approx 256\) nm. A high, nearly constant gain of \(\approx 710\,\mathrm{cm}^{-1}\) is maintained from \(\sim 1953\) to 2140 nm, with variation \(\sim 70\,\mathrm{cm}^{-1}\) in the overlap region. The reported peak positions are \(\sim 1980\) nm for the 7 nm well and \(\sim 2090\)–2100 nm for the 13 nm well.

The broader multi-quantum-well “Structure C,” comprising one 7 nm well and three 13 nm wells, is reported at \(20^\circ\)C to reach a flat gain across \(\sim 1950\)–2135 nm at \(N \approx 1.73 \times 10^{18}\,\mathrm{cm}^{-3}\), corresponding to \(\approx 2.1\) A. The flatness window is \(\sim 185\) nm, the full-width at half-maximum is \(\approx 342\) nm, the maximum material gain is \(\approx 1253\,\mathrm{cm}^{-1}\), and the double peaks occur at \(\sim 1987\) nm and \(\sim 2096\) nm. At higher density, \(N \approx 2.14 \times 10^{18}\,\mathrm{cm}^{-3}\), an additional shorter-wavelength peak appears near \(\sim 1800\) nm, identified as \(E1\)–\(LH1\) of the 13 nm well; this broadens the spectrum but reduces flatness.

The modelling framework is a self-consistent solver combining Poisson, drift-diffusion continuity, capture/escape balance in quantum wells, photon rate, and heat flow under isothermal pulsed operation. The reported material-gain expression is

$$
g(\omega) = \frac{\pi e^2}{n_r c m_0 \epsilon_0} \sum_i |M_i|^2 \left[f_c(E_{c,i}) - f_v(E_{v,i})\right] L(\omega - \omega_i),
$$

and the modal gain is

$$
G_m = \Gamma g - \alpha_i.
$$

Calibration values reported for the validated simulations include scattering loss \(\alpha_i \approx 4.5\,\mathrm{cm}^{-1}\), mirror reflectivities \(\sim 10^{-7}\), spontaneous-emission coupling to the guided mode \(\sim 10^{-4}\), a “gain factor” of 0.6, \(\tau_\mathrm{in} \approx 1 \times 10^{-13}\) s, \(\tau_{\mathrm{cap},e} \approx 5 \times 10^{-12}\) s, \(\tau_{\mathrm{cap},h} \approx 3 \times 10^{-13}\) s, and \(\tau_\mathrm{SRH,e} = \tau_\mathrm{SRH,h} \approx 1 \times 10^{-7}\) s.

The same study reports that cavity-loss engineering can further broaden the accessible spectrum. In “Structure C#,” a 225 \(\mu\)m unpumped end section is introduced, leaving a 1775 \(\mu\)m pumped active section and imposing \(\sim 95\,\mathrm{cm}^{-1}\) effective mirror loss. At \(N \approx 2.412 \times 10^{18}\,\mathrm{cm}^{-3}\), corresponding to \(\approx 4.4\) A, three pronounced peaks appear: \(\sim 2100\) nm \((E1\)–\(HH1\), 13 nm, \(\sim 1600\,\mathrm{cm}^{-1})\), \(\sim 1980\) nm \((E1\)–\(HH1\), 7 nm, \(\sim 1733\,\mathrm{cm}^{-1})\), and \(\sim 1800\) nm \((E1\)–\(LH1\), 13 nm, \(\sim 1670\,\mathrm{cm}^{-1})\). The broadband window then extends from \(\sim 1760\) to 2150 nm with nonuniformity \(\sim 278\,\mathrm{cm}^{-1}\). Temperature broadens the gain further: for Structure C, increasing \(T\) from \(10^\circ\)C to \(100^\circ\)C red-shifts the thin-well peak from \(\sim 1970\) to 2113 nm and the thick-well peak from \(\sim 2078\) to 2237 nm, widens the FWHM from \(\sim 335\) to \(\sim 400\) nm, and reduces the maximum gain by \(\sim 195\,\mathrm{cm}^{-1}\).

A recurring design conclusion is that symmetric stacks center the gain around one transition energy and provide narrower FWHM and less flatness, whereas asymmetric stacks intentionally spread the transition energies of \(E1\)–\(HH1\) in thin and thick wells and, at higher injection, add \(E1\)–\(LH1\) in the thicker wells. Fabrication accuracy is reported to favor thickness tuning over alloy-composition changes.

## 4. Composite wells near \(1.55\,\mu\mathrm{m}\): thermal performance and leakage physics

In the \(1.55\,\mu\mathrm{m}\) regime, the principal issue is not broadband flattening but thermal performance. The reported GaInSb/AlGaAsSb composite quantum-well lasers show room-temperature threshold current densities of \(J_\mathrm{th} \approx 1092\,\mathrm{A/cm}^2\) for Device A, \(654\,\mathrm{A/cm}^2\) for Device B, and \(471\,\mathrm{A/cm}^2\) for Device C under 500 ns, 10 kHz pulsed injection. The corresponding radiative threshold components are \(J_\mathrm{rad} \approx 170 \pm 6\,\mathrm{A/cm}^2\), \(147 \pm 10\,\mathrm{A/cm}^2\), and \(189 \pm 5\,\mathrm{A/cm}^2\), giving \(J_\mathrm{rad}/J_\mathrm{th} \approx 15 \pm 1\%\), \(23 \pm 2\%\), and \(40 \pm 1\%\), respectively. Increasing composite well thickness from 3.6 to 4.8 to 6.0 nm, together with the AlInSb insertions, therefore reduces \(J_\mathrm{th}\) by increasing gain volume and optical confinement and lowering \(n_\mathrm{th}\) [2012.13369].

The temperature dependence is reported to change character around \(T \sim 150\) K. Below this temperature, \(J_\mathrm{th}\) is dominated by radiative recombination and defect-related recombination is negligible. Above \(\sim 150\) K, \(J_\mathrm{th}\) rises super-linearly because of non-radiative processes. The decomposition used in the analysis is

$$
J_\mathrm{th}(T) = J_\mathrm{rad}(T) + J_\mathrm{Auger}(T) + J_\mathrm{leak}(T) + J_\mathrm{other}(T),
$$

with \(J_\mathrm{rad} \propto n^2\), \(J_\mathrm{Auger} \propto n^3\), and \(J_\mathrm{leak}\) described by an Arrhenius-like activated escape. Both leakage-dominated and Auger-dominated models fit \(T_0(T)\) and \(J_\mathrm{th}(T)\) reasonably, but slope-efficiency analysis shows that \(d\eta_i/dT\) dominates and implicates leakage as the necessary contributor.

High-pressure spectroscopy provides the central discrimination. Hydrostatic pressure up to 400 MPa was applied, and the lasing energy pressure coefficient is reported as \(dE_\mathrm{lase}/dP \approx 11\,\mathrm{meV/kbar}\) for all devices. The measured increase of \(J_\mathrm{th}(P)\) is stronger than expected for leakage into \(L\) states and is instead consistent with leakage to barrier \(X\) states. Pressure-dependent fits attribute up to 43% of the room-temperature threshold current to leakage into the barrier \(X\) valley. This is a significant corrective to a common oversimplification: in these GaInSb/AlGaAsSb CQW lasers, thermal degradation is not adequately described by Auger recombination alone.

The design recommendations are correspondingly specific. The study concludes that carrier leakage to barrier \(X\) can be reduced by increasing the activation energy of the leakage paths through a small increase in the lattice-matched barrier Al and As fractions, without compromising optical confinement by the cladding. It also recommends continued use of 0.45 nm AlInSb barriers within GaInSb wells to maintain \(\sim 1.55\,\mu\mathrm{m}\) emission while enabling thicker wells that reduce \(n_\mathrm{th}\). The data additionally notes that off-center placement of internal barriers to tailor wavefunction localization away from interfaces prone to \(\Gamma \rightarrow X\) coupling was not demonstrated in the paper, but is a plausible strategy.

## 5. Intersubband cavity polaritons and terahertz emission

A separate, theoretically oriented branch of the subject concerns asymmetric quantum wells in semiconductor microcavities, where asymmetry enables a transition that is forbidden in centro-symmetric systems. The starting point is the first two conduction subbands of a doped quantum well with intersubband transition energy \(\hbar \omega_{12}\), strongly coupled to a planar microcavity TM mode at frequency \(\omega_\mathrm{ph}(q)\). In the rotating-wave approximation, the resulting upper and lower intersubband cavity polaritons have eigenfrequencies

$$
\omega_{\mathrm{UP/LP}}(q)=\frac{\omega_{\mathrm{ph}}(q)+\omega_{12}}{2}
\pm \frac{1}{2}\sqrt{\left[\omega_{\mathrm{ph}}(q)-\omega_{12}\right]^2+4\Omega(q)^2},
$$

where \(\Omega(q) = \sqrt{N}\,\chi(q)\) is the collective vacuum Rabi frequency. In symmetric wells the upper-to-lower polariton transition is dipole forbidden; in asymmetric wells with \(\Delta z \neq 0\), the transition becomes allowed and can radiate in the terahertz range [1204.4053].

The spontaneous scattering rate from upper polariton to lower polariton plus a THz photon is reported to scale with \(\omega_\mathrm{THz}\), \((e \Delta z y_\mathrm{UP} y_\mathrm{LP})^2\), and the spectral overlap with the lower-polariton linewidth. In the stimulated regime, the rate acquires bosonic occupation factors and a nonbosonicity correction \(B\), which approaches 1 in the dilute limit. The threshold condition for stimulated THz emission is written as

$$
\frac{N_{\mathrm{UP}}}{S}
=
\frac{\Gamma_{\mathrm{THz}}}{\omega_{\mathrm{THz}}}
\frac{\hbar \epsilon_0 \epsilon_r L_{\mathrm{cav}} \Gamma_{\mathrm{LP}}}
{2 (e\,\Delta z\, y_{\mathrm{UP}}\, y_{\mathrm{LP}})^2}.
$$

The internal conversion efficiency is given as

$$
\eta=
\frac{x_{\mathrm{UP}}^2\,\max\!\left[1-\frac{\Gamma_{\mathrm{UP}}\Gamma_{\mathrm{THz}}}{\Xi_s P},\,0\right]}
{1+\Gamma_{\mathrm{UP}}/\Gamma_{\mathrm{LP}}},
$$

with \(x_\mathrm{UP}^2\) the photonic fraction of the upper polariton.

For GaInSb/AlGaAsSb, the provided design-oriented details identify stepped wells, graded barriers, or an external bias as practical ways to realize \(\Delta z \neq 0\). The asymmetry target is stated as \(\Delta z \approx z_{12} \approx 0.05\)–0.15 \(L_\mathrm{QW}\), and for antimonide realizations the indicated ranges are \(L_\mathrm{QW} \approx 20\)–35 nm, \(\hbar \omega_{12} \approx 50\)–150 meV, \(n_\mathrm{QW} \approx 20\)–60 wells, and \(N_\mathrm{2DEG} \approx (0.5\)–\(2)\times 10^{12}\,\mathrm{cm}^{-2}\) per well. A more specific target set is \(L_\mathrm{QW} \approx 25\)–30 nm with stepped asymmetry \(D \approx 0.3\)–0.5 \(L_\mathrm{QW}\), \(V \approx 100\)–200 meV, \(z_{12} \approx 1\)–3 nm, \(\Delta z \approx 1\)–3 nm, and \(\Omega/\omega_{12} \approx 0.05\)–0.1, which the data states should produce THz emission across \(\approx 0.5\)–4 THz. Because these antimonide-specific values are framed as design guidance rather than as direct device measurements, a cautious reading is warranted: they indicate applicability to GaInSb/AlGaAsSb, not a completed experimental demonstration.

## 6. Design rules, misconceptions, applications, and open problems

Several design rules recur across the reported work. For broadband interband gain, the recommended asymmetric pair is \(\sim 7\) nm and \(\sim 13\) nm Ga\(_{0.73}\)In\(_{0.27}\)Sb wells separated by \(\sim 20\) nm Al\(_{0.25}\)Ga\(_{0.75}\)As\(_{0.02}\)Sb\(_{0.98}\) barriers, embedded in 130 nm Al\(_{0.25}\)Ga\(_{0.75}\)As\(_{0.02}\)Sb\(_{0.98}\) SCH layers on each side and Al\(_{0.5}\)Ga\(_{0.5}\)As\(_{0.04}\)Sb\(_{0.96}\) claddings with graded doping. The strain-times-thickness budget is reported as \(\lesssim 100\ \%\times\mathrm{nm}\), and the \(1\times 7\) nm plus \(3\times 13\) nm stack has a total QW thickness of 46 nm, chosen to remain below the estimated relaxation threshold. The recommended carrier-density range for flat spectra beyond \(2\,\mu\mathrm{m}\) is \(N \approx (1.2\)–\(1.8)\times 10^{18}\,\mathrm{cm}^{-3}\), while higher \(N\) accesses LH contributions and broadens the spectrum at the expense of flatness [2508.17495].

For the \(1.55\,\mu\mathrm{m}\) CQW lasers, the corresponding optimization principle is to maximize \(\Delta E_{\Gamma X}\) while preserving lattice match and optical confinement. The reported actionable recommendation is to slightly increase Al and As in the AlGaAsSb barriers, keeping the structure lattice matched to GaSb, in order to increase the activation energy of the leakage path into \(X\). In parallel, the use of one or two 0.45 nm Al\(_{0.68}\)In\(_{0.32}\)Sb insertions inside GaInSb wells is retained to support thicker wells, larger gain volume, and lower \(J_\mathrm{th}\) without losing the target emission wavelength [2012.13369].

Two misconceptions are directly addressed by the data. First, asymmetry in GaInSb/AlGaAsSb quantum wells does not necessarily mean dissimilar outer barriers; in the CQW lasers, the outer barriers are compositionally identical and the asymmetry is created inside the well by AlInSb insertions. Second, thermal degradation in these antimonide lasers is not reducible to Auger recombination by default; the pressure-dependent analysis identifies leakage to the barrier \(X\) minima as accounting for up to 43% of the threshold current at room temperature. A related misconception in the broadband \(>2\,\mu\mathrm{m}\) regime is that broad spectra must be obtained by alloy-composition changes; the reported study instead emphasizes thickness tuning because fabrication accuracy favors it.

The application space reported in the literature is correspondingly broad. Beyond \(2\,\mu\mathrm{m}\), GaSb-based SOAs and SLDs benefit from broadband, flat gain chips for widely tunable lasers with reduced output variation across 1.95–2.15 \(\mu\mathrm{m}\), and up to \(\sim 1.76\)–2.15 \(\mu\mathrm{m}\) at higher current. The same sources are identified for absorption spectroscopy, including 1.95–2.45 \(\mu\mathrm{m}\) molecular bands, for 2 \(\mu\mathrm{m}\) OCT with reduced scattering and higher axial resolution, and for free-space communications and FMCW LiDAR in the thulium fiber transmission window. Near \(1.55\,\mu\mathrm{m}\), the reported GaInSb/AlGaAsSb CQW results are explicitly framed as design insights for monolithic integration of GaSb-based lasers on silicon. In the THz-polaritonic setting, a plausible implication is that the same antimonide platform could combine flexible band engineering with structurally tunable asymmetry to realize electrically or structurally tunable THz emission, provided that free-carrier absorption, interface roughness, and dephasing are controlled [1204.4053].

Open problems are stated with unusual clarity. For broadband \(>2\,\mu\mathrm{m}\) devices, they include quantifying and mitigating Auger recombination and intervalence band absorption beyond \(\sim 2.2\,\mu\mathrm{m}\), improving carrier-distribution uniformity across many wells, and extending bandwidth toward \(\geq 500\) nm while controlling strain, thermal roll-off, and loss. For the \(1.55\,\mu\mathrm{m}\) CQW lasers, the unresolved task is to suppress \(\Gamma \rightarrow X\) leakage without sacrificing optical confinement or lattice matching. For polaritonic THz devices, the challenge is to translate the asymmetric-well formalism into antimonide microcavities with sufficiently low \(\Gamma_\mathrm{LP}\), high \(Q_\mathrm{THz}\), and controlled THz mode overlap with doped layers. Collectively, these results define asymmetric GaInSb/AlGaAsSb quantum wells not as a single device class, but as a heterostructure design methodology for engineering confinement, transition energies, and symmetry properties across interband and intersubband photonics.

Source: https://www.emergentmind.com/topics/asymmetric-gainsb-algaassb-quantum-wells