Interband cascade lasers

I Vurgaftman; R Weih; M Kamp; J R Meyer; C L Canedy; C S Kim; M Kim; W W Bewley; C D Merritt; J Abell; S Höfling

doi:10.1088/0022-3727/48/12/123001

1. Introduction and historical background

The interband cascade laser (ICL) is a promising mid-infrared (IR) (defined here as λ = 3–6 μm) source that combines the relatively long upper-level recombination lifetime of a conventional diode laser with the voltage-efficient cascading scheme introduced by the quantum cascade laser ((QCL) which produces coherent light via intersubband transitions within the conduction band). The mid-IR spectral region has gained increasing relevance in recent years, because it hosts numerous 'fingerprint' spectral lines that can be used to sense trace gases such as methane, carbon dioxide, carbon monoxide, formaldehyde, etc. While practical sensing systems typically require continuous-wave (cw) emission into a single spectral mode, a low output power on the order of 1 mW is usually sufficient [1]. On the other hand, such military applications as IR countermeasures require substantial cw output powers, whereas spectral purity is generally less critical. Other potential applications for compact and efficient mid-IR sources include industrial process control, clinical breath analysis, free-space optical communications, IR scene projection, and the detection of chemical/biological threats.

Nonetheless, until 2002 no semiconductor source could offer the cw operation at ambient temperature [2] that is essential if any mid-IR technology is to be considered sufficiently practical for widespread implementation. Earlier development efforts had focused on extending the well-established diode laser architecture, with type I alignment of the conduction and valence bands [3,4], beyond the 2–3 µm spectral window by increasing the strain in GaSb-based multiple quantum wells (MQWs). However, significant challenges included rapid wavelength scalings of the Auger non-radiative decay and free-carrier absorption loss, carrier escape associated with a marginal valence-band offset in the MQWs, and the general immaturity of GaSb-based growth and processing technologies.

This view was upended by the rapid progress of QCLs grown on InP [5,6]. A modest temperature sensitivity, coupled with efficient heat dissipation, have enabled QCL output powers to reach 5 W cw in a near-diffraction-limited beam at room temperature (RT) [7]. The primary disadvantages are the QCL's high threshold current density, associated with rapid phonon-assisted depopulation of the upper lasing subband and interface roughness scattering (≈1 ps), coupled with a bias of at least ≈10 V because 30–50 stages are needed to supply sufficient gain.

Even though the ICL was invented in 1994 [8], the same year that the QCL was first demonstrated experimentally, over the intervening period it has received much less attention and far fewer resources have been devoted to its development. Within 2–3 years, the hole injector was added to the concept [9], along with replacement of the single type II interface in the active gain region with a 'W' configuration that incorporates two InAs electron QWs on both sides of the GaInSb hole QW [10]. The 'W' structure has become a standard feature of state-of-the-art ICL designs, even though its advantage is not so clear if the nonradiative recombination rate decreases with electron–hole wavefunction overlap even as the gain increases.

While the first experimental realization of an ICL was reported in 1997 [11], the initial devices operated only at cryogenic temperatures [12,13] due to several design aspects that were far from what is now known to be optimal. In particular, excessive electron injector thicknesses induced high threshold current densities and low external differential quantum efficiencies (EDQEs) per stage. Nevertheless, gradual improvements in both designs and the growth techniques led to steady improvement of the temperature performance, with T_max = 250–286 K being reached for pulsed operation in the 1998–2000 time frame [14–16]. A significant milestone was the demonstration in 2002 of pulsed operation at RT by an ICL with 18 active stages, even though the threshold current density (Jth) exceeded 6 kA cm⁻² [17]. Designs employing the 'W' active QW configuration subsequently operated at λ = 3.3 µm with much lower RT Jth (≈1 kA cm⁻²), along with cw lasing to T = 200 K [18].

In 2005, the group at Jet Propulsion Laboratory (JPL) substantially reduced the pulsed RT Jth further, to 630 A cm⁻², with only 12 active stages [19]. This eventually led to cw operation up to T_max = 264 K [20], which enabled JPL to qualify single-mode distributed feedback (DFB) ICLs for methane detection on the NASA Mars Curiosity Mission [21].

The first ICLs designed, grown, processed, and characterized at Naval Research Laboratory (NRL) were reported in 2006 [22]. In subsequent NRL designs the number of stages was reduced from 10 to 5 [23], in order to minimize the threshold power density. Simulations had determined that an ICL's gain per unit current density is actually quite high, e.g. much higher than in a QCL that typically requires 30 or more stages to minimize Jth at RT. NRL subsequently reported a series of advances in such key performance figures of merit as pulsed Jth, Pth, and EDQE, at and above RT, by employing a much thinner electron injector and two GaSb QWs in the hole injector. The result was a lower RT threshold current density of 1.15 kA cm⁻² for a 10 stage structure. An epitaxial-side-up narrow ridge from that wafer, with Au electroplating deposited on top for heat dissipation, operated cw to T_max = 269 K [24].

In 2008, an NRL ICL emitting at λ = 3.75 µm became the first to operate cw at RT [25]. That milestone was made possible by several additional design modifications, including the introduction of lightly n-doped GaSb separate-confinement layers (SCLs) to reduce the waveguide loss, decrease of the n-doping in the InAs/AlSb optical cladding layers, and reduction of the number of stages to five. The resulting pulsed Jth at RT dropped to ≈400 A cm⁻², the lowest threshold reported up to that point.

In the past 6 years, NRL has continued to explore numerous other design modifications [26–28]. While some had little or no significant impact on the laser performance, others were quite beneficial to the threshold and efficiency. This applies especially to the 'carrier rebalancing' concept that will be discussed in the next section. In parallel, the University of Würzburg (UWUERZ) and Nanoplus GmbH have reported a variety of valuable recent contributions to the ICL development [29,30]. The next section will discuss these and other state-of-the-art ICL design concepts in depth. While alternative approaches that are suitable for the longest wavelengths of the mid-IR and beyond are now being explored by Professor Rui Yang and colleagues at the University of Oklahoma [31,32], RT cw operation has not yet been reported. Those concepts are not further discussed in this review.

Although ICLs have been the subject of several previous review articles [33,34], we provide here an up-to-date description of both the physics of operation and the development status. Section 2 discusses the ICL design principles, and also identifies some ingredients that may still be missing from the current picture. The device fabrication aspects of the ICL technology, including growth and processing methods, are discussed in section 3. Section 4 reviews the operating characteristics of recent broad area, narrow-ridge, and single-spectral-mode DFB ICLs. We conclude in section 5 with some thoughts on the future prospects for these devices.

2. Theory

In a conventional diode laser, electrons and holes are injected from opposite sides of a p–n heterojunction, ideally populating all of the MQWs equally (although pronounced non-uniformities can occur under some circumstances). Therefore, the total threshold current density is given by MJ, where J is the threshold current density per QW, and M is the number of QWs, which implies that the parasitic voltage drop MJρ_s, where ρ_s is the series resistance-area product, scales with the number of QWs. In the mid-IR, the parasitic drop tends to be relatively more significant compared to the 'useful' voltage ħω/e, since the currents are often high, and the photon energy ħω is smaller than at shorter wavelengths.

Cascading mitigates this inefficiency by providing a practical means for connecting the MQWs 'in series', i.e. with the same current flowing through every stage and each injected carrier traversing every active QW in turn [35]. The parasitic voltage drop then decreases to Jρ_s while the 'useful' voltage increases to Mħω/e. Cascading can therefore reduce the net threshold power density considerably, provided the required recycling of carriers from valence to conduction band within each stage does not introduce much additional voltage or current [26,36]. Instead of employing an Esaki or tunnel diode to recycle the carriers [37], the ICL exploits the unusual semimetallic band alignment between the narrow-gap semiconductors InAs and Ga(In)Sb [26]. The ICL architecture thus avoids the appreciable internal optical losses (especially due to free hole absorption) that could potentially occur if every stage incorporated a heavily doped tunnel junction.

The semimetallic overlap between the conduction subbands in an InAs electron QW and the valence subbands in an adjacent Ga(In)Sb hole QW can be tuned with an applied electric field. Since at RT the conduction band minimum of bulk InAs lies ≈0.2 eV below the valence band maximum of GaSb, only a modest amount of quantum confinement is sufficient to open a small energy gap E_i. If we now apply an external field F, the lowest conduction subband is lowered with respect to the highest valence subband to create a semimetallic interface (SMIF), at which electrons populate the InAs QW and holes the Ga(In)Sb QW in quasi-thermal equilibrium. This produces a voltage-dependent band overlap E_SM(V) = F(d_e + d_h) − E_i, where d_e and d_h are the centre-of-mass distances from the SMIF of the electron and hole probability densities. At typical ICL operating voltages, this overlap is ≈100 meV. Assuming parabolic bands and a common quasi-Fermi level (QFL) across the SMIF, the generated electron and hole densities are then: n = p ≈ m_rE_SM/πħ², where m_r = m_em_h/(m_e + m_h) is the reduced effective mass.

The active core of the ICL is comprised of multiple repeated stages, each of which can be subdivided into: (1) the active QWs; (2) the hole injector; and (3) the electron injector. The active QWs can have a type II or type I band alignment. The conduction and valence band profiles in a typical type II ICL designed for emission at λ = 3.7 µm are shown in figure 1, which includes the 'W' active regions for two successive stages. The SMIF is seen to separate the hole injector, comprised of coupled GaSb/AlSb QWs, from the electron injector consisting of coupled InAs/AlSb QWs. The field-dependent band overlap between states in the hole injector and active hole QW on the one hand, and states in the electron injector and active electron QW in the next stage on the other determines the carrier densities throughout the active core. We assume a common QFL on both sides of the SMIF, since the carrier transport via direct, phonon-assisted, and other tunnelling mechanisms should be quite rapid. The relatively long carrier lifetime in the active QWs (~1 ns) then assures that the QFL is discontinuous across the active region of each stage (E_Fi − E_Fi+1 ≥ ħω ≥ E_g, with E_Fi, the electron QFL in stage i, being equal to the hole QFL in stage i − 1).

**Figure 1.** Band diagram of 1.5 stages of a typical ICL active core. Wavefunction probability densities and zone-centre energies (indicated by the wavefunction zero points) for some of the most important subbands are superimposed. The probability densities for the active electron (hole) subbands are indicated with blue (red) lines, while those for the injector-electron (hole) subbands are indicated with wine-coloured (green) lines. The blue (red) arrows indicate the positions of the QFLs in each stage.
Download figure:
Standard image High-resolution image

In the ideal design, the applied bias must separate the QFLs in successive stages (equal to the single-stage voltage drop) enough to simultaneously: (1) produce sufficient optical gain to compensate the photon loss in the cavity, and (2) internally generate a quasi-equilibrium carrier density consistent with the voltage drop (accounting for any extrinsic doping). While the first condition is a property of the active QWs only, the second follows from the design and doping of all the electron and hole QWs on both sides of the SMIF. If, for example, the QWs in the electron injector are thicker than optimal, excess carriers will be generated that induce unnecessary free-carrier absorption. Conversely, if the injector QWs are too thin the threshold voltage will exceed that needed to induce the ideal QFL separation of E_Fi − E_Fi+1 ≈ ħω (or somewhat larger if the loss is very high).

Since many of the states on both sides of the SMIF reside in the injectors rather than the active QWs, only some fraction of the carriers generated internally at the SMIF, or introduced by extrinsic doping, contribute to the optical gain. Therefore, the electron and hole injectors must be carefully designed so as to maximize the fraction of injected carriers populating the active states, while retaining sufficient carrier transport through the injectors to maintain a single QFL throughout the stage.

The current understanding based on extensive empirical testing is that the energy subbands in the two hole injector QWs should lie substantially below the topmost active hole subband, while the states of the thicker electron injector should be located in the vicinity of the SMIF and energetically lower than the active electron QW state. These constraints imply that most of the electrons populating a given stage reside in the injector states rather than the active region, whereas nearly all of the holes transfer successfully to the active hole QW. This is illustrated by the dashed curves in figure 2, which plot the electron (wine-coloured) and hole (green) densities at threshold for the ICL structure of figure 1. The consequence is that if no extrinsic doping is introduced (dashed curves) to supplement the carriers generated at the SMIF, the active hole density will substantially exceed the active electron density at the lasing threshold (pth >> nth) [27]. Unless nnp (multi-electron) Auger processes strongly dominate the carrier lifetime, the large hole/electron density ratio will tend to reduce the gain per unit current density, and also increase the internal loss unless the cross section for free hole absorption is negligible.

**Figure 2.** Calculated quasi-equilibrium electron (wine-coloured) and hole (green) density distributions for the ICL designs without (dashed) and with (solid) carrier rebalancing. Rebalancing is achieved by increasing the extrinsic Si doping level in four of the InAs electron injector QWs from 4 × 10¹⁷cm⁻³ to 4 × 10¹⁸cm⁻³. The blue and red lines indicate the conduction- and valence-band bulk band edges, respectively.
Download figure:
Standard image High-resolution image

To investigate this potential issue, ICL structures emitting at λ = 3.6–3.9 µm were grown with a series of heavy n-doping levels in the electron injectors. While many of the additional electrons remained behind in the injector, NRL simulations predicted that a sufficient fraction would be transferred to the active QWs to substantially alter the hole/electron density ratio there (solid curves in figure 2). The total sheet doping, n_s, was varied from 4.8 × 10¹¹ to 7.4 × 10¹² cm⁻², with 2–4 of the injector QWs receiving the additional doping. This study found that Jth at RT was minimized at n_s ≈ 5 × 10¹² cm⁻², which according to the simulations corresponded to nth/pth only slightly larger than unity. Note that the required sheet doping density substantially exceeds nth and pth, due to the inefficient electron transfer from the doped injector. Another conclusion, from the dependence on doping level, is that the nnp and ppn Auger coefficients most likely have comparable values [27].

Whereas empirical studies showed it advantageous to employ two GaSb QWs (rather than one) in the hole injector, nominally to suppress electron tunnelling leakage from the active region to the electron injector, adding still more QWs to the hole injector is unlikely to prove beneficial. Since the thicknesses of the two hole QWs are adjusted so as to roughly align their topmost subbands at the threshold field, those wells become strongly coupled with two closely spaced subbands. The GaSb QW thicknesses are chosen so as to place the two subbands ≈80–100 meV below the maximum in the active hole QW. This empirical rule should ensure very low occupation of the hole injector states in quasi-equilibrium, as illustrated in figure 2. The thicknesses of the AlSb barriers separating the active and hole-injector QWs are typically ≈10–12 Å, in order to assure unencumbered hole transport. However, somewhat thicker barriers degrade the ICL performance only gradually.

The QW layer thicknesses in the electron injector should be adjusted so as to produce the required threshold carrier densities, nth and pth, when the heavy extrinsic doping is taken into account. In practice, this occurs when the first electron injector QW has a thickness between 40 and 50 Å. The electric field in the active core is then ≈70–90 kV cm⁻¹ at the lasing threshold. The thickness of each subsequent well is reduced (chirped) so as to maintain coupling between adjacent QWs. However, the electron injector subbands do not actually form a miniband with a common energy level and similar occupation probabilities in all the QWs. Figures 1 and 2 indicate that the lower-energy subbands concentrated near the SMIF are more heavily populated than the higher-energy subbands adjacent to the active region of the next stage. This arrangement reduces the density of states (DOS) at lower energies, while still allowing sufficient electron transport from the SMIF to the active electron QWs.

Like the barriers in the hole injector, the AlSb barriers in the electron injector are relatively thin, e.g. 12–14 Å. However, two of the barriers in each stage are much thicker, at 25–30 Å. One separates the electron and hole injectors at the SMIF, in order to minimize parasitic interband absorption across that interface. The other separates the electron injector from the first active electron QW, so as to isolate the active electron subband from the injector states and prevent significant hybridization while still permitting sufficient electron tunnelling.

We finally consider the design of the active QWs. It was mentioned above that state-of-the-art ICLs employ the 'W' active region, in which two InAs electron wells sandwich a single GaInSb hole well. The strain induced at the typical Ga_1 − xIn_xSb composition of x ≈ 0.35 approaches the maximum that can be tolerated while maintaining coherent layer growth. Although hole-well thicknesses ranging from 20 to 30 Å have been employed, no significant trends were observed in the experimental laser thresholds or slope efficiencies.

The 'W' configuration of the active QWs substantially enhances the electron–hole wavefunction overlap in comparison to that for an InAs/GaInSb QW with a single type II interface. For example, when λ = 3.2 µm and the GaInSb thickness is 20 Å in both structures, the square of the overlap peaks at 42% for the 'W' structure but only 17% for the single-interface structure. The two InAs QW thicknesses are chosen so as to dial in the laser emission wavelength via quantum confinement, while maintaining roughly equal electron wavefunction probabilities in the two wells.

The goal in designing the optical waveguide is to simultaneously minimize both the internal loss and the material gain required to reach transparency. While the ICL's active core is unique among semiconductor lasers, the waveguide design follows well-known general principles, apart from the beneficial feature compared to standard diode lasers that both cladding layers are n-type (with lower free-carrier absorption loss) because the holes are all generated internally. The ICL waveguide is constructed from the following basic building blocks: (1) the active core, (2) the n-doped optical claddings, most commonly comprised of InAs/AlSb short-period superlattices (SLs), although lattice-matched bulk alloys such as AlGaAsSb provide an alternative; [38] (3) the lightly n-doped GaSb SCLs; and (4) various transition SLs that separate the other three regions from each other, and from the GaSb substrate/buffer and n⁺-InAs(Sb) top contact. The transition SLs reduce the parasitic voltages that would be associated with abrupt heterointerfaces between adjacent regions with very different conduction-band offsets.

Figure 3 shows the waveguide intensity profile for the fundamental transverse-electric (TE) mode in a 7 stage ICL with two 800 nm-thick SCLs (λ = 3.7 µm). The simulation assumes n_a = 3.45 for the active core, n_c = 3.30 for the SL cladding and transition regions, and n_s = 3.78 for GaSb. The largest uncertainty is in the active core's average index, although its effect on the modal index is minor because the mode is dominated by the GaSb SCLs. While a higher core index would imply a larger confinement factor, Γ_a, and more gain per unit current density, cavity-length studies [39] find that theory tends to overestimate the gain. The index for an AlAs_0.08Sb_0.92 cladding is expected to be only slightly lower (by ≈ 0.1) than that of the SL claddings.

The optical mode overlap with the SCLs is ≈77% for the ICL in figure 3. In NRL structures, the centre of each SCL is n-doped at N_l = 5 × 10¹⁵ cm⁻³, whereas 50 nm-thick layers at the edge of each SCL (containing Γ_h ≈ 9% of the mode) are doped at N_h = 1 × 10¹⁷ cm⁻³ to reduce the voltage drop at the heterointerfaces. Using the data in [40], the free-carrier contribution to the internal loss from the GaSb SCLs is expected to be ≈0.5 cm⁻¹ for the structure of figure 3.

The clad doping is also non-uniform, in that the inner regions closest to the SCL are doped to N_i = 7.5 × 10¹⁶ cm⁻³, while the outer regions with less mode overlap are doped more heavily, to N_o = 5 × 10¹⁷ cm⁻³. The objective is to trade minimization of the series resistance against minimization of the free-carrier absorption loss. While the total mode overlap with the claddings is estimated to be ≈6%, the overlap with the more heavily doped outer regions nearly vanishes. The confinement factor calculated for the active region (including the electron and hole injectors as well as the active QWs) is Γ_a ≈ 14%. The remainder of the mode (≈3%) resides in the InAs/AlSb SL transition regions, which are typically doped to N_t = 2 × 10¹⁷ cm⁻³.

Figure 4 shows that the optical confinement factors in the active core and cladding layers decrease as the SCL thicknesses increase, as expected. Eventually, the SCLs can become so thick that the TE₂ mode, with two nodes along the growth direction (and an antinode at the active core), can have a greater active-core confinement factor than the TE₀ mode. In state-of-the-art ICLs emitting at λ ≈ 4 µm, the optimal SCL thickness determined by these trade-offs is t_s ≈ 800 nm.

While the internal loss in the active core is not known precisely, it likely dominates the total loss in state-of-the-art designs. While this suggests it may be beneficial to reduce the active confinement factor Γ_a, eventually the available optical gain becomes insufficient to overcome the losses originating elsewhere in the waveguide. Further studies are needed to fully understand the variation of the internal loss in ICLs with various design parameters. The top cladding layer must be thick enough to minimize mode overlap with the lossy top-contact metallization, while the bottom cladding must be somewhat thicker to avoid mode leakage into the high-index GaSb substrate [41]. These guidelines imply typical top and bottom SL cladding thicknesses of ≈1.5 µm and ≈3.0 µm, respectively, for state-of-the-art ICLs emitting at λ = 4 µm.

3. Technology

3.1. Growth

ICLs are typically grown on epi-ready Te-doped GaSb wafers in a solid source molecular beam epitaxy (MBE) system equipped with conventional effusion cells for the group-III elements (Al, Ga, In), as well as dopants (Te, Si) and valved cracker cells for the group-V elements (As, Sb). Even though ICLs have been grown using a conventional effusion cell for Sb and an As-cracker operated with a low cracking-zone temperature that primarily provide Sb₄ and As₄ tetramers [42], valved crackers that provide Sb₂ and As₂ reduce the group-V cross incorporation into neighbouring layers, and provide adequate flux within the layers grown at different growth rates. In order to provide a smooth surface for the subsequent layers, residual oxides are usually removed thermally in a Sb-rich atmosphere and a GaSb buffer layer is grown.

The low-refractive-index cladding layers, comprised of InAs/AlSb SLs, are grown next. To avoid reabsorption of the emitted photons, a short period (~5 nm) that can be adapted to different emission wavelengths is typically chosen for increasing the effective bandgap by quantum confinement. The typical growth temperature for these cladding layers is in the 400 to 450 °C range. The SL design benefits from the advantageous lattice constant constellation within the 6.1 Å semiconductor family. While the RT lattice constant of InAs (6.0583 Å) is slightly smaller than that of GaSb (6.0959 Å), the lattice constant of AlSb (6.1355 Å) is slightly larger with nearly equal mismatch. Due to the change of both anion and cation at each interface within the InAs/AlSb SL, the interfaces have a significant effect on the morphological, electrical and optical properties [43–45]. By introducing so-called soak times, either AlAs or InSb interfaces can be forced. SLs grown with InSb-like interfaces generally exhibit superior properties to those with AlAs-like interfaces. Using the more reactive As₂ dimers, it is also possible to grow strain-compensated SLs having mixed interfaces without additional soak times [46]. The interface composition can then be influenced via the group-V flux ratio during the SL growth, and thus the residual group-V concentration of the previous layer. The InAs layers are usually Si-doped, with decreasing concentration towards the active core to reduce internal losses.

The lower cladding layer is followed by the lower GaSb SCL. Since the GaSb layer quality is improved at temperatures higher than 450 °C [47], ramping the substrate temperature to about 480 °C for this growth provides a smoother surface for the subsequent cascaded active region, whose optimal growth conditions are determined from photoluminescence (PL) test samples. The PL emission intensity is generally maximized and the linewidth minimized at active-region growth temperatures in the 430–480 °C range [48–52], with the most recent studies suggesting temperatures toward the lower end of that range. Another critical parameter is the As flux during the active QW growth, since the presence of a high As background in the growth chamber tends to roughen the InAs/GaInSb interface and decrease the oscillator strength. The negative consequences of As incorporation at the interfaces have also been verified via Fourier-transformed PL, photoreflectance, and high angle annular dark field scanning tunnelling transmission microscopy [53]. An As flux just sufficient to stabilize the As-rich (2 × 4) reconstruction during the InAs layer growth is recommended.

After the active stages, the upper SCL is grown, with the growth performed at an elevated substrate temperature for the samples grown at the University of Würzburg and at the same temperature for the samples grown at NRL, followed by the upper InAs/AlSb SL cladding for which the temperature is ramped down again for the University of Würzburg samples. For substrate temperatures below 480 °C, no degradation in the active stages, e.g. due to intermixing, has been noted. The structure is completed with a highly Si-doped (~1 × 10¹⁹ cm⁻³) InAs cap layer that forms a low resistive ohmic contact. In addition, graded transition SLs are inserted at the boundaries between the various regions of the full ICL structure, in order to avoid parasitic voltage drops as discussed above.

Figure 5(a) presents a scanning electron microscopy (SEM) micrograph of a 10 stage ICL. Figure 5(b) shows a typical high-resolution x-ray diffraction (HR-XRD) curve of a 5 stage ICL. The central peak for the SL cladding layers is almost coincident with the substrate peak, indicating well-lattice-matched growth. The small higher-order active region peaks also confirm high growth quality.

3.2. Processing

The ridges for edge-emitting ICLs are generally defined using reactive ion etching based on a Cl- and Ar- plasma. The dry etch is typically followed by a clean-up phosphoric wet etch step that smooths the sidewalls and removes contaminants. Due to highly anisotropic transport in the SL cladding and active regions of an ICL, especially below threshold where the active core has especially high impedance along the vertical axis, the etch must proceed to below the active core to avoid excessive current spreading. Otherwise, the threshold current and efficiency vary strongly with etch depth, ridge width, and stage multiplicity. For example, Jth for a 100 µm-wide broad area 5 stage ICL emitting at 3.75 µm was 232 A cm⁻² when the etch proceeded through the active region to the lower SCL, whereas it increased to 286 A cm⁻² when the etch extended only into the 200 nm-thick upper SCL. When the etch was only 1.4 µm deep, into the 2.2 µm-thick upper cladding region, Jth increased even further, to 350 A cm⁻² (corresponding to a 50% larger pumped area). For much narrower ridges with 4.5 µm width, a seven-fold increase in the threshold current density was observed when the etch was stopped above the active stages [54].

Broad-area lasers are suitable for measuring such basic characteristics as the emission wavelength of an MBE-grown ICL wafer. These devices are highly reproducible and fast to fabricate. No sidewall passivation is necessary if the etch is shallow, although the threshold current density is then overestimated and the slope efficiency underestimated due to current spreading. It is also possible to process broad-area lasers for pulsed testing using photolithography and wet chemical etching to the bottom SCL layer, without any passivation of the sidewalls, as long as the ridge is wide enough to make a metal contact on top. This process is also relatively quick and convenient, and minimizes the effect of current spreading on the lasing thresholds and slope efficiencies [55].

To mitigate leakage or oxidation at the sidewalls of an ICL that is etched through the active region, passivation is typically employed to suppress excessive short-circuiting currents. State-of-the-art ICLs use a combination of sputtered Si₃N₄ and SiO₂ [56], with an overall thickness of approximately 300–500 nm. Al₂O₃ and MgO may be suitable alternatives with better thermal conductivity, although no comparative studies have yet been carried out. Thermal management is quite critical to the ICL performance, since the thermal conductivities of the active core and short-period InAs/AlSb SL cladding regions are low and highly anisotropic. Using the 3ω method, the in-plane and cross-plane thermal conductivities were measured recently [57]. This showed the cross-plane thermal conductivity of the SL cladding layers to be particularly low, in the 1–3 W mK⁻¹ range. To expedite heat removal from the active region, and thereby increase the maximum operating temperature and output power, ≈5 µm of gold is typically electroplated on top of ICL narrow ridges [58]. Lanes of width ≈50 µm are left unplated, however, to assure high facet quality following the cleave. Figure 6 shows an SEM micrograph of a 11.8 µm-wide ridge with electroplated gold on top. The dark areas depict the insulating passivation layer which is opened on top of the ridge for contacting. The thermal dissipation can be enhanced further by mounting the ICL ridges epitaxial side down [59].Heat is then directly transferred from the ridge to the heat sink, without passing through the substrate.

**Figure 6.** SEM micrograph of a processed 11.8 µm wide ICL rigde with a 5 µm thick electroplated gold layer. The etch proceeded right through the lower SCL. In the highest magnification, the SCL layers and the active region can be identified.
Download figure:
Standard image High-resolution image

Bauer et al discuss several DFB laser concepts [60] (see also section 4.4) that require additional processing steps. In the case of vertical sidewall gratings, the periodic modulation (typically 4th order) can be produced in the same step as the ridge via optical lithography or electron-beam lithography. Other DFB concepts employ 1st-order gratings and require a greater processing effort: In the case of a top grating a 200 nm-thick germanium layer is deposited on top of the ridge and periodically patterned with electron-beam lithography [61]. After a lift-off step, the top contact is evaporated. For the processing of loss-coupled DFBs based on lateral metal gratings [62], a 30 nm Si₃N₄ layer is sputtered on the ridge to prevent short circuiting of the active region. Afterwards the metal grating is defined via electron-beam lithography in Poly(methyl methacrylate) (PMMA) resist, with the subsequent evaporation of chromium and a lift-off step. The final steps, including passivation, contact evaporation, and gold electroplating, are the same as in the processing of narrow ridges.

4. Experimental status

4.1. Broad area laser performance (pulsed)

The intrinsic performance of a grown ICL structure may be estimated from measurements in pulsed mode at low duty cycle, for which the device heating due to current injection is avoided. Typical pulses are 100–300 ns at a repetition rate in the 1–10 kHz range. Additionally, ridge widths >100 µm are generally used to minimize the influence of leakage currents and optical losses at the etched sidewalls. Unless stated otherwise, the results presented in this section are based on 2 mm-long and 150 µm-wide devices with untreated facets. Figure 7 shows wavelength-dependent threshold current densities for broad area ICLs with different stage multiplicities operated at 300 K. The structures were grown at NRL and UWUERZ, and employed designs with carrier rebalancing in order to minimize Jth. Threshold current densities below 500 A cm⁻² were achieved throughout the 2.8–5.2 µm wavelength range.

**Figure 7.** Pulsed threshold current densities for broad area ICLs with various stage numbers at 300 K.
Download figure:
Standard image High-resolution image

Much of the work to date on optimizing the ICL designs for high performance has concentrated on the 3–4 µm wavelength region, for which a threshold current density as low as 134 A cm⁻² at 300 K was obtained for a 5 stage device emitting at λ = 3.6 µm. At wavelengths shorter than 3.1 µm, Jth increases somewhat to above 200 A cm⁻². The threshold also increases at wavelengths longer than 4.1 µm, where Jth ~ 400 A cm⁻² at 4.7 µm and 650 A cm⁻² at 5.5 µm [63][. Since only a few ICLs have been grown in the wavelength regime greater than 4 µm, the active core and waveguide designs of those structures may not yet be fully optimized. The temperature increase of Jth is represented by the characteristic temperature T₀, which typically ranges from 45–60 K in the temperature range from 290 to 350 K. For wavelengths longer than 4 µm, T₀ tends to decrease to 40–45 K. The lowest threshold current density for an ICL (114 A cm⁻² at 300 K) was reached with a 10 stage device emitting at 3.65 µm [64]. Furthermore, a 10 stage ICL emitting at 5.2 µm exhibited considerably lower Jth than any of the devices with fewer stages and emitting at wavelengths as much as 1 µm shorter. The general increase of the external quantum efficiency with increasing stage multiplicity [55] also makes such devices interesting for high-power applications.

To further investigate the influence of the number of stages (M) on the lasing characteristics, a series of ICLs with stage multiplicities varying from 1 to 12 was grown within a period of less than two weeks. The fixed active region was similar to the one used in [27], since that design had already performed well in 5 stage devices. For the purpose of carrier rebalancing, the inner 4 of the 6 InAs electron injector wells were heavily doped with Si to a density of 5 × 10¹⁸cm⁻³. The 'W' active quantum wells were designed to emit at a wavelength of 3.6 µm. The inset of figure 10 (below) shows the emission spectrum from a 4 stage broad area ICL when operated at RT in pulsed mode (1 kHz repetition rate, 200 ns pulse width). All of the devices in this series with varying M emitted at wavelengths between 3.54 and 3.68 µm. This indicates a highly stable In effusion cell flux, since the emission wavelength is extremely sensitive to the InAs layer thicknesses in the active QWs [65].

The active stages were embedded in 200 nm-thick GaSb SCLs doped with Te to a level of 3 × 10¹⁷cm⁻³ on both sides. The lower and upper cladding thicknesses were 3.5 µm and 2.0 µm, respectively. Since the active region thickness varied with the number of stages, the optical mode profiles change. The left panel of figure 8 shows calculated optical mode profiles for M = 1, 2, 4, 6, 8, 10 and 12. In this figure the mode and refractive index profiles are shifted to the right in order to centre all modes at the same position, even though the lower cladding has the same thickness for all stage numbers. For higher stage multiplicities a greater fraction of the optical mode overlaps the active core, whereas the modal overlap with the SCL and SL-cladding layers decreases. Figure 8 (right) plots the modal confinement of the different regions as a function of stage number. The overlap with the active region increases from 0.095 for one stage to 0.404 for 12 stages, while the overlap with the SCLs and cladding are reduced from 0.374 to 0.249 and from 0.532 to 0.347, respectively. It should be noted that the active region confinement in [64] was underestimated since the refractive index of GaSb was taken to be 3.73, whereas ellipsometry measurements have shown 3.78 to be more accurate. The increased modal confinement in the active region for larger stage numbers implies that a higher modal gain and thus lower threshold current density should be expected. Notice also the different (single-humped) mode profiles obtained when 200 nm-thick SCLs are employed here as opposed to the 800 nm-thick SCLs employed in generating figure 3, as well as the much lower confinement in the SCLs and higher confinement in the active and clad regions as compared to the thicker SCLs in figure 4.

**Figure 8.** Optical mode profiles (left) and confinement factors in the claddings, SCL and active regions (right), for ICL waveguides with stage numbers from 1 to 12.
Download figure:
Standard image High-resolution image

The threshold voltage should scale almost linearly with M, since additional stages require that more bias (at least ħω per stage) be applied to tilt the band structure. Figure 9 shows the measured I–V characteristics for broad-area lasers with different stage numbers at RT. At low currents, the resistance remains very high until the electric field becomes large enough to properly align the energy levels within the active stages. However, once the highly conductive regime is reached, the series resistance decreases to <1 Ω. The inset of figure 9 plots the dependence on M of the voltage at which current starts to flow through the structure (V₀) and the threshold voltage (Vth). The broken line indicates that V₀ increases almost linearly with M. That the curve intercepts the ordinate just above zero, at 0.16 V, indicates that most of the bias drops across the cascaded stages while only a low residual fraction can be attributed to the contacts and transition regions between the SCLs and claddings. For M > 6, Vth behaves similarly to V₀, but is slightly higher due to the series resistance and contributions of order M × k_BT that are needed to overcome losses. In the case of lower stage multiplicities, Vth deviates more strongly from the linear trend because the threshold current density is higher. In the limit M = 1, Vth could not be investigated since laser operation was not observed at T = 20 °C.

**Figure 9.** *I–V* characteristics of 150 µm-wide and 2 mm-long broad area devices with different stage multiplicities at T = 20 °C. The inset shows the corresponding values for the intercept of the linear part of the *I–V* curve (V₀) and threshold voltage Vth.
Download figure:
Standard image High-resolution image

To enhance the statistics for the threshold values, at least five 2 mm-long lasers with each M were characterized. Figure 10 plots the resulting average values of Jth and Pth at T = 20 °C. A general trend towards lower threshold current densities for higher stage numbers is observed. The highest value of 322 A cm⁻² was measured for the two-stage laser, whereas Jth for the 4 stage device already drops to 170 A cm⁻², which can be attributed to the strong increase of modal gain and reduced loss. With further increase of the stage number, Jth continues to decrease modestly due to the higher modal gain. The lowest average value of 106 A cm⁻² was achieved for the 10 stage ICL, and several lasers with M = 10 exhibited Jth slightly below 100 A cm⁻². The threshold for the 12 stage laser increased slightly, to 127 A cm⁻². However, the reduction of Jth at high M comes at the price of a monotonically increasing Vth (figure 8). This leads to a very different picture for the M dependence of the threshold power density, an important figure of merit in applications requiring low power consumption. The lowest Pth of 326 W cm⁻² was measured for the 4 stage device.

Figure 11 plots the total EDQE (from all stages) versus M for the series of broad area lasers when operated in pulsed mode at T = 20 °C. The EDQE scales with M, as expected. Also plotted is the EDQE per stage, which stays nearly constant in the range 24–30%, and decreases only gradually at higher stage numbers. This indicates that the overall loss is nearly independent of M, which in turn implies that the losses contributed by the three main parts of the waveguide (active region, SCL, and cladding) are weighted roughly equally. It was shown in [55] that the slope efficiency can exceed the stage number proportionality when the overall loss is reduced, in particular by using thick and very-low-doped SCLs, along with careful design optimization of the optical mode overlap with the different waveguide parts. This led to improvement of the slope efficiency (from 2 facets) for broad area devices from 620–680 mW A⁻¹ for 5 stage devices to values near 1200 mW A⁻¹ for 7 stage ICLs. That the increase is larger than 7/5 indicates that the net loss was reduced, due to the shift of a portion of the optical mode out of the cladding layers and into the low-doped SCLs.

The results of the cascade variation study clearly show that there is still room for optimization. However, the optimal stage multiplicity also depends on which figure of merit is given the greatest weight. While mobile applications will benefit from the low threshold powers of devices with low M (figure 10), high-power applications will benefit from the higher net EDQE associated with a larger M (figure 11).

4.2. Ridge waveguide laser performance (cw)

To ensure sufficient heat dissipation, and to avoid the occurrence of higher-order lateral modes, narrow ridges have been fabricated for operation in cw mode. After chip fabrication the lasers are mounted on c-mounts, either epitaxial side up or down. In the first case, heat dissipation is mainly achieved by lateral heat flow in the electroplated gold layer on top of the structure, whereas in the latter case heat is more efficiently transferred directly to the copper mount. The advantage of epi-down mounting can be quantified by the thermal impedance-area product (R_tA), which can be extracted from the P–I–V characteristics. While the values measured for epi-up-mounted devices (ridge width varying from 7.7 to 15.7 µm) were in the range 5.1–8.6 K (kW/cm²)⁻¹, the R_tA for epi-down-mounted devices with the same dimensions decreased to 3.3–5.2 K (kW/cm²)⁻¹ due to better heat removal. To enhance the output power and wallplug efficiency (WPE), the facets are commonly coated with high reflection (HR) and anti-reflection (AR) layers having reflectivities of >95% and ≈1–2%, respectively. The first carrier rebalanced and epi-up-mounted 5 stage narrow ridge devices lased in cw mode up to a maximum operation temperature (T_max) of 107 °C [27], with typical output powers of ≈50 mW at RT. The threshold power for a 0.5 mm-long laser was only 29 mW [27], by far the lowest for a semiconductor laser emitting in the 3–4 µm range.

Epi-down mounting substantially improves the maximum output power ( $P_{\text{max}}^{\text{cw}}$ ) and WPE [66]. The left side of figure 12 shows the P–I–V characteristic of an epi-down-mounted 15.7 µm-wide and 4 mm-long device operating at 25 °C. The low Jth of 156 A cm⁻² and Vth of 2.3 V are comparable to the values measured for broad-area devices processed from the same wafer, indicating that the degradation due to processing and mounting was minimal. At an injection current of 1.2 A, the maximum output power was 253 mW. The beam quality determined from the far-field profiles illustrated on the right side of figure 12 is represented by M² = 2.2 at I = 0.4 A and M² = 2.7 at I = 1.2 A.

Narrowing the ridge width generally improves the beam quality, but at the expense of higher Jth and lower output power. For example, a 7.7 µm-wide ridge showed a slightly elevated Jth of 184 A cm⁻² due to sidewall carrier recombination, but operated in cw mode up to $T_{\text{max}}^{\text{cw}}=118$ °C [64]. The P–I characteristics at different operating temperatures are shown in figure 13 (left). The device emitted a nearly ideal Gaussian beam at all injection currents, e.g. M² = 1.1 at 200 mA and M² = 1.3 at 600 mA. The WPE tends to be maximized in short cavities for which the internal loss is comparable to the mirror loss. Figure 14 shows the P–I curves of two short cavities, along with their WPEs versus current. A WPE of nearly 15% is reached for the 0.5 mm-long cavity at 76 mA, which compares to the highest value of 21% reported for QCLs (λ = 4.8–4.9 µm) at RT [67]. Very recently, NRL ICLs with 1 mm-long cavities and both 7 and 10 stages attained 18% WPE [68].

**Figure 13.** *P–I–V* characteristics of an epi-down-mounted 7.7 µm-wide, 4 mm-long ICL with HR and AR facet coatings operating in cw mode at various operation temperatures (left). Reproduced with permission from [66], Copyright 2012 AIP Publishing LLC. Far-field patterns at different injection currents for a 15.7 µm-wide and 4 mm-long device with HR and AR facet coatings mounted epi-side-down (right).
Download figure:
Standard image High-resolution image

**Figure 14.** Output power and WPE versus current for two 15.7 µm-wide devices. The 0.5 mm long cavity had one HR and one uncoated facet, while the 1 mm long cavity had an HR-coated back facet and AR-coated front facet. Reproduced with permission from [66], Copyright 2012 AIP Publishing LLC.
Download figure:
Standard image High-resolution image

The cw operation of ICLs at RT is not limited to the 3–4 µm wavelength range. A maximum operation temperature of 60 °C was obtained for a 4 mm-long and 10.9 µm-wide ridge emitting at 4.8 µm. At RT, more than 15 mW of cw output power was measured. Another device, with the same dimensions and processed from a different wafer, lased in cw mode up to 48 °C at λ = 5.6 µm with similar output powers [63].

4.3. Power scaling

An important class of potential applications requires high output powers. The maximum output power for ICLs generally scales with the ridge width, since the threshold power density and thermal impedance tend to decrease for wider ridges. However, this can break down when the lower lateral thermal dissipation leads to excessive heating caused by a high threshold power. A more fundamental drawback of power scaling via ridge width is the degradation of beam quality that results from lasing in higher-order lateral modes. To combine high ICL output power with nearly diffraction limited beam quality, two distinct concepts have been investigated to date. The results may be quantified in terms of the 'brightness' figure of merit (B) defined as P_out/M².

The first concept introduced corrugations into the etched ridge sidewalls. Since higher-order modes have greater overlap with the sidewalls, they experience increased scattering losses and are hence selectively suppressed relative to the fundamental mode. At the same time, it is advantageous to minimize the overlap of the fundamental mode with the corrugations, in order to reduce the impact on the laser efficiency. Ridge waveguide lasers with 5 and 7 stages were processed with corrugated sidewalls (typical peak-to-valley amplitudes of 1.4 µm and periods of 2.0 µm). Generally the 7 stage devices showed higher output power, due in part to their higher differential output power but also because of a lower internal loss (see the discussion near the end of section 4.1). The left side of figure 15 shows the RT P–I characteristics of a 22.0 µm-wide and 3 mm-long 7 stage device with HR/AR coated facets. The output power reached 384 mW with high beam quality (M² = 2.6), and with a WPE of 12.4% at $P_{\text{max}}^{\text{cw}}$ . While minor deviations of the far-field peak from zero may be due to alignment imperfections, the larger asymmetries shown in figure 15 may reflect phase locking of the lateral modes [69].

**Figure 15.** Left: Light-current characteristics in cw mode for a 22.0 µm-wide, 3 mm-long 7 stage ICL mounted epi-side-down with an HR coating on the back facet and an AR coating on the front facet at a temperature of 25 °C. Right: Far-field profiles of the same ridge at different current injection levels.
Download figure:
Standard image High-resolution image

A second option for increasing the output power while maintaining high beam quality is to taper the ridges [70,71]. When this concept is applied to near-IR lasers, a narrow waveguide that supports only one lateral mode is combined with a gain-guided tapered section (typical half angle ≈ 3°) that increases the pumped area. As described above, the ICL's extensive lateral current spreading makes it mandatory to etch through the active core, although index-guided structures have been fabricated from 5 stage IC material. Figure 16 (left) shows the P–I characteristics of a tapered ICL [72]. The device was 4 mm long, with a back facet aperture of 5.5 µm, front facet aperture of 63 µm, and taper half angle or 0.42° (resulting in a net pumped area of 0.00139 cm²). Sidewall corrugations were applied to the wider part of the ridge to further suppress lasing in higher-order later modes. The device had Jth = 210 A cm⁻² and emitted up to 403 mW of cw output power at I = 2 A. The right panel of figure 16 show the normalized far-field profiles of that device. The measured beam quality factors ranged from M² = 1.4 at I = 0.5 A to M² = 2.3 at I = 2.0 A. The resulting brightness of B ≈ 175 mW was 28% higher than the best previous ICL result (for a 5 stage ICL with 25 µm wide ridge and corrugated sidewalls) [66]. Collected results for the output power, beam quality and brightness of various high-power ICLs are presented in table 1.

**Figure 16.** (a): Cw *P–I* characteristics for a tapered-ridge ICL at T = 25 °C. (b) Cw far-field emission profiles at a series of injection currents. The extracted M² values range from 1.4 at I = 0.5 A to 2.3 at I = 2 A. Reproduced with permission from [72], Copyright 2013 AIP Publishing LLC.
Download figure:
Standard image High-resolution image

Table 1. Maximum cw-output powers, M² values, and brightnesses for various high-power ICLs, including structures with variable ridge width, corrugated sidewalls, and tapered ridges with 5 and 7 stages.

Number of stages	Ridge width (μm)	Sidewalls	Output power (mW)	Effective M²	Brightness = Power/M²
5	7.7	Straight	138	1.3	106
5	10.8	Straight	198	1.8	110
5	15.7	Straight	254	2.7	94
5	13.2	Corrugated	102	1.2	85
5	18.2	Corrugated	205	1.6	128
5	25.1	Corrugated	274	2.2	125
5	25.2	Corrugated	291	2.2	132
5	Tapered (5.5/63)	Corrugated	403	2.3	175
5	Tapered (5/57)	Corrugated	264	1.8	155
7	22	Corrugated	383	2.4	160
7	32	Corrugated	592	3.7	160

4.4. Single mode devices

In addition to compactness, robustness and low power consumption, spectroscopic applications demand a spectrally narrow line width, and thus emission in a single longitudinal mode [60]. Several concepts have been studied to allow ICLs to emit in a single mode for cw operation at RT. In 2008, a photonic-crystal DFB ICL emitted 67 mW of cw output power in a single mode (λ = 3.27 µm) at a temperature of 78 K [73]. To confine current and the optical mode, the region outside the gain section was bombarded with helium ions. One year later, the first cw DFB ICL emission at RT was achieved by employing a corrugated sidewall grating that on the one hand served to suppress higher-order lateral modes and on the other provided DFB for selection of a single longitudinal mode. With a 4th-order grating etched into the ridges sidewalls [74], cw emission in a single mode was achieved up to T = 40 °C. More recently, a corrugated-sidewall DFB ridge with 13.2 µm width and 4 mm length emitted up to 55 mW in a single spectral mode at 25 °C [26]. In 2014, a DFB ICL of this kind was realized in the 5.2 µm wavelength region for sensing nitric oxide sensing with a room-temperature drive power of only 138 mW [75]. An advantage of the vertical sidewall grating approach is that it can employ optical rather than electron-beam lithography. However, it is not very robust since the coupling strength is quite sensitive to ridge width and the fill factor of the grating. The use of a third-order rather than four-order grating would solve the latter issue, but would be more challenging to fabricate with optical lithography.

The option of etching the DFB grating into a Ge layer deposited on top of the ridge provides a coupling coefficient that does not depend significantly on the ridge width. That approach was applied to ICL gain material in 2007 [76], and more recently produced robust cw emission in a single mode at λ ≈ 3.8 µm for temperatures up to 80 °C [77]. Figure 17 shows an SEM micrograph of a ridge with a 1st-order Ge grating on top. Figure 18 illustrates the single-mode emission spectra at a series of temperatures and a fixed drive current of 170 mA (left), together with P–I–V characteristics for the 2 mm-long and 7.4 µm-wide device (right). Up to 27 mW of single mode output power was realized at T = 40 °C. Typical temperature and power tuning rates were 0.39–0.40 nm °C⁻¹ and 12 nm W⁻¹, respectively, with a total tuning range of >25 nm. Even though the top grating approach is very robust, its primary disadvantage is a higher optical loss resulting from the modal overlap with the top contact metallization. Since the grating coupling coefficient is also proportional to this overlap, it is inevitable that the threshold power density will increase and the slope efficiency decrease.

**Figure 17.** False-colour SEM micrograph of a 7.4 µm wide ridge with a first order Ge top grating. Reproduced with permission from [77], Copyright 2012 AIP Publishing LLC.
Download figure:
Standard image High-resolution image

**Figure 18.** Left: Emission spectra of a 7.4 µm-wide and 2 mm-long HR/AR-coated ICL ridge topped by a Ge DFB grating, at a drive current of 170 mA and a series of temperatures. The first-order grating period was 531 nm. Right: *P–I–V* characteristics of the device at several temperatures. The solid and dashed portions of the power curves correspond to single- and multi-mode emission, respectively. Reproduced with permission from [77], Copyright 2012 AIP Publishing LLC.
Download figure:
Standard image High-resolution image

In 2014, two additional concepts for achieving single mode emission from an ICL were demonstrated. In the first, a laterally coupled DFB was fabricated by etching the ridge in two steps [54]. These devices produced up to 18 mW of cw single-mode output at λ = 3.57 µm and a temperature of 46 °C. Lifetime measurements at T = 40 °C found negligible degradation of the output when four devices were operated for more than 8800 h. A second, loss-coupled, approach employed metal gratings deposited next to the ridge [78]. That approach also provided single-mode cw emission at temperatures above ambient.

5. Future prospects

ICLs have proven to be a practical mid-IR source for a variety of applications, and are especially attractive when very low power consumption is required. Most of the work to date has concentrated on the wavelength region between 3 and 4 µm, where the threshold current and power densities appear to be approaching the theoretical limits. Moreover, it may be difficult to substantially further reduce the optical losses that have already come down to a few cm⁻¹. Nevertheless, additional optimization of the active region design may lead to somewhat better performance.

On the other hand, there is clearly room for improving the performance in spectral regions outside the ICL's established central 'sweet spot', i.e. at λ < 3 µm and > 4 µm where little optimization has been attempted to date. Both the cascade designs and configuration of the waveguides for higher gain and/or lower loss need to be investigated. Further work will also help to optimize the number of cascades versus wavelength for applications requiring either low power consumption or high output power as the primary figure of merit. While to minimize the drive power it is presumably beneficial to employ a smaller number of active stages, even in those applications the optimal M may increase somewhat with increasing wavelength (since more gain may be required as the losses increase).

An additional concept, which was introduced as early as 1996 [9] and expanded upon theoretically more recently [79], is to use type I active transitions in an ICL that benefits from spatially direct radiative emission. Following a very early experimental demonstration in 1998 [80], much better performance was obtained in 2013 when a cw output power of 590 mW was generated by a 100 µm-wide ridge at RT for an emission wavelength of almost 3 µm [81]. While this paves the way to the realization of devices emitting to at least 3.4 µm, it remains to be seen how future type I ICLs will compare with the state-of-the-art type II devices operating in the same spectral region.

Although the interband nature of the ICL's lasing transition results in the strong absorption on the short-wavelength side of the gain peak that is not present in a QCL [82,83], the ICLs have nonetheless demonstrated wavelength tunability over hundreds of nanometres [84] as well as low-temperature emission at two wavelengths separated by ≈1 µm [85]. This makes them promising for widely tunable mid-IR single-mode devices such as longitudinally coupled DFBs [86, 87], lateral DFB arrays [88], coupled cavities [89] and sampled gratings [90]. The interband cascade technology should also be adaptable to miniature lab-on-a-chip platforms, on which ICLs and interband cascade detectors [91] are integrated in a monolithic device, as has already been pursued using quantum cascade structures [92].

Finally, in contrast to QCLs, the active transitions couple to TE-polarized photons. This makes the ICL compatible with vertical emission without requiring the assistance of a grating. Whereas to date no electrically pumped vertical-cavity surface-emitting laser (VCSEL) has emitted at a wavelength beyond 2.6 µm [93], interband cascade gain materials could extend that capability throughout the mid-IR spectral band. Interband cascade light-emitting devices (ICLEDs) with 15 stages, but no optical cavity to provide feedback, recently produced >1 mW cw of non-coherent vertical emission for room-temperature operation at a wavelength beyond 3 µm for the first time [94].

Due to its very low drive power requirement, the ICL is increasingly viewed as the laser of choice for mid-IR (meaning the entire λ = 3–6 µm spectral band) laser spectroscopy applications that do not require high output power but need to be hand-portable and/or battery operated. With ICLs already beginning to find use in fielded systems to detect methane, formaldehyde, and other trace chemicals, it is anticipated that this role will expand substantially over the coming decade.

Acknowledgments

NRL acknowledges support from the Office of Naval Research. UWUERZ is grateful to the European Union for financial support of this work within the FP7 project 'WideLase' (No. 318798). We also thank S. Kuhn, M. Wagenbrenner, S. Handel, and T. Steinl for technical assistance.

Interband cascade lasers

Article metrics

Submit

Permissions

Author e-mails

Author affiliations

Dates

Peer review information

Abstract

1. Introduction and historical background

2. Theory