Distributed-feedback interband cascade laser

Abstract

The invention relates to distributed-feedback, interband cascade lasers. To reduce unwanted, higher- order transverse optical modes without the introduction of complex grating architectures a waveguide heterostructure with an active part, herein referred as "core" is suggested, which is sandwiched between two sets of semiconductor layers, herein referred as "passive parts". The passive parts are located between the core and an electrode, and between a substrate and the core and comprise at least one, thinner, higher-doped separate confinement layer and one, thicker, lower-doped separate confinement layer.

Description

[0001] Distributed-Feedback Interband Cascade Laser

[0002] FIELD OF THE INVENTION

[0003] This invention relates to the operation of distributed-feedback, interband cascade lasers, in particular to a waveguide design to achieve the reduction in unwanted, higher-order transverse optical modes - preferentially selecting the fundamental transverse mode, without the introduction of complex grating architectures nor the need for demanding fabrication processes as described in prior art.

[0004] BACKGROUND OF THE INVENTION

[0005] Interband Cascade Lasers (ICLs) are semiconductor lasers which rely on a cascaded active region similar to Quantum Cascade Lasers (QCLs). In contrast to QCLs, in ICLs the radiative transitions occur between confined states of the conduction and valence bands in a multi-quantum well heterostructure with a “type-II” band alignment as described by J. Meyer, W. Bewley, C. Canedy, C. Kim, M. Kim, C. Merrit and I. Vurgaftman in "The Interband Cascade Laser," Photonics 7(3), 75 (2020). These devices can be grown either on GaSb or on InAs substrates. GaSb-based ICLs operate in continuous wave at room temperature in the mid-infrared spectral range between 3-6 μm.

[0006] The main application of these lasers is chemical sensing, particularly high-sensitivity detection of methane, formaldehyde and other gases. Trace gas sensing based on optical spectroscopy requires lasers sources with a single-frequency emission spectrum. This is typically achieved using the distributed feedback (DFB) architecture in which the modal refractive index is periodically modulated by patterning a grating in the waveguide structure. Such discloses US 9 923 338 B2. While the DFB grating selects a single longitudinal mode, the waveguide cross section must be appropriately designed so that only one transverse mode, preferably the fundamental one, can reach threshold. If several transverse modes lase simultaneously then the spectrum contains multiple frequencies which hinders the usefulness of the device for chemical sensing applications.

[0007] The active region of ICLs typically contains 5-10 gain stages and is ~400-800 nm thick. Since the active region is significantly thinner than the wavelength of the emitted light, separate confinement layers (SCLs) are grown below and above active region in order to enhance the confinement factor, defined as the overlap of the optical mode with the active region, and hence to reduce the lasing threshold. GaSb, which has a higher refractive index than the active region and is lattice-matched to the substrate, is the natural material choice for SCLs. Ill the prior art, the lower and upper SCLs are symmetric and have thicknesses of 500-800 nm. To achieve single transverse mode operation in a ridge waveguide ICL emitting at 3.35 μm with such symmetric SCLs requires a waveguide width w < 3 μm. The fabrication of ICLs with such a narrow waveguide width is not straightforward because the structure has to be etched down to the lower SCL to avoid current spreading in the active region, i.e. an etch depth of ~2 μm. If the sidewalls are not vertical, the width of each layer above the active region, i.e SCL, cladding, grating layers, become narrower as the top of the stack is reached, so narrow that due to photolithography resolution limits and available contact area, it is difficult to form a reliable metallic contact for current injection.

[0008] Therefore, it is desirable to modify the waveguide design so that it doesn’t have to be < 3 μm for single mode operation, in order to relax the requirements on the fabrication accuracy. In addition, increasing the maximum width of single mode operation allows to inject higher current to increase the output power. Several waveguide designs have been proposed and demonstrated to address this issue:

[0009] Von Edlinger et al. developed DFB ICLs with 4thorder vertical sidewall gratings [M. von Edlinger, J. Scheuermann, R. Weih, C. Zimmermann, L. Nahle, M. Fischer, J. Koeth, S. Hofling and M. Kamp, "Monomode Interband Cascade Lasers at 5.2 μm for Nitric Oxide Sensing" IEEE Photonics Technology Letters 26(5), 480-482 (2014)]. In this approach, the sidewall gratings have two functions. They provide a wavelength-selective distributed feedback for longitudinal mode selection and they selectively increase the losses of high order transverse modes so that only the fundamental mode can reach threshold. This approach requires a challenging fabrication. The gratings have to be etched all the way through the top cladding, top SCL, and active region, down to the lower SCL, i.e. a total depth > 2 μm with a good aspect ratio. Then, an insulating layer, typically SiOi or SisN4 and a metallic layer have to be deposited on top of the patterned sidewalls.

[0010] Forouhar et al. developed a double-ridge waveguide design and fabrication process for DFB ICLs [S. Forouhar, C. Borgentun, C. Frez, R. M. Briggs, M. Bagheri, C. L. Canedy, C. S. Kim, M. Kim, W. W. Bewley, C. D. Merritt, J. Abell, I. Vurgaftman and J. R. Meyer, "Reliable mid-infrared laterally- coupled distributed-feedback interband cascade lasers" Applied Physics Lett (5), 051110 (2014)]. A narrow ridge is first etched in the top cladding to control the optical mode. Then a wider ridge is etched through the active region to limit current spreading. While it relaxes the requirements on fabrication accuracy, particularly on the verticality of the sidewalls, this approach has the drawbacks that it increases the number of fabrication steps and it results in a lower electrical-to-

[0011] 2 optical power conversion efficiency because current is injected in a region that is larger than the optical mode.

[0012] More recently, Xie et al. demonstrated DFB ICLs with a top grating and corrugated sidewalls [F. Xie, M. Stocker, J. Pham, F. Towner, K. Shen, J. Wang and K. Lascola, "Distributed feedback interband cascade lasers with top grating and corrugated sidewalls" Appl. Phys. Lett. 112(13), 131102 (2018)]. Compared to the two previously discussed approaches, this design has the advantage that it relies on a top grating similar to aforementioned US 9923 338. In this configuration the grating is not as deep, only a few hundreds of nanometers, and hence the aspect ratio is not as critical. Furthermore, the ridge waveguide is etched in a single step. The corrugated sidewalls selectively increase the propagation losses of the high-order transverse modes. One drawback of this approach is that, since the modal refractive index varies with waveguide width, the grating cannot be patterned on the corrugated regions, otherwise the Bragg wavelength would not be constant. Therefore, the waveguide has to contain straight regions with a DFB grating, and corrugated regions with varying width for mode control with no grating. Therefore, the DFB grating can only be patterned over a fraction of the overall length of the device.

[0013] Nauschutz et al. demonstrated DFB ICLs at a longer wavelength of 6 μm using lateral metal gratings [J. Nauschutz, J. Scheuermann, R. Weih, J. Koeth, B. Schwarz and S. Hofling, "Room temperature operation of single mode GaSb-based DFB interband cascade lasers beyond 6.1 μm" Electronics Letters 59(19), el2968 (2023)]. In this configuration, the gratings do not have to be etched as deep as for the vertical sidewall gratings. However, in an evaluation, fabrication is still more challenging than the top grating design because the gratings have to be patterned on a non-flat surface and the position of the gratings with respect to the active region can vary between fabrication runs because it depends on the tech depth of the ridge waveguide. In addition, the grating coupling constant varies strongly with the ridge width because the optical mode is only coupled to the grating on its sides and not over its entire width as for the top grating.

[0014] SUMMARY OF THE INVENTION

[0015] To achieve the desired goal of single-transverse-mode operation in wider waveguides while circumventing the drawbacks of the prior art designs discussed in the previous section, a novel waveguide design with a modified vertical structure is proposed. In particular, an asymmetric SCL configuration is used in which the upper SCL is significantly thinner than the lower SCL. The design according to the invention which is shown in figures 1 and 2 is based on a top grating and the ridge waveguide is etched in a single step.

[0016] BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The figures in the atached drawings show:

[0018] Figure 1 a schematic design of separate confinement layers (SCLs) according to the invention with a separate confinement layer;

[0019] Figure 2 an example of a ridge-waveguide interband cascade laser structure, based on the schematic of figure 1, with an insulating, dielectric-passivation, layer and epitaxial- and substrate-side electrodes, according to the invention;

[0020] Figure 3 a lower-nominally-doped 2 x 1016cm'3SCL layer combinations investigated. An “X” in dark gray boxes denotes previously investigated symmetric SCH cases. A in light gray boxes denotes combinations investigated during the process of this invention. Empty grey boxes are combinations that have been omited from the study as further computation diverged from desirable result within this regime;

[0021] Figures 4 computed TEoo (figs. 4a, 4c, 4e, 4g, 4i) and TEw (figs. 4b, 4d, 4f, 4h, 4j) mode profiles of a DFB ICL, emiting at 3.5 μm, for ridge widths of 5.0 μm (figures 4a, 4b), 4.5 μm (figures 4c, 4d), 4.0 μm (figures 4e, 4f), 3.5 μm (figures 4g, 4h) and 3.0 μm (figures 4i, 4j) for an optimized, confinement-layer-engineered, DFB-ICL for preferential fundamental transverse mode operation;

[0022] Figure 5 computed confinement factor T of the two modes as function of ridge width, for an optimized, confinement-layer-engineered, DFB-ICL for preferential fundamental transverse mode operation. Please make note of the symmetric SCL references as a baseline; and

[0023] Figure 6 calculated gain margin between modes TEio and TEoo function of ridge width, for an optimized, confinement-layer-engineered, DFB-ICL for preferential fundamental transverse mode operation. Please make note of the symmetric SCL reference as a baseline.

[0024] DETAILED DESCRIPTION OF THE INVENTION

[0025] As shown schematically in Figure 1, the following waveguide layers are grown on a n-GaSb substrate in the following epitaxial growth steps: n-GaSb buffer layer, n-InAs / AlSb superlattice grading layer, n-InAs / AlSb superlattice lower cladding layer, n-InAs / AlSb superlattice grading layer, n-GaSb lower separate confinement layer, GalnSb / InAs / AlSb superlatice active region, n-InAs / AlSb superlattice grading layer, n-GaSb upper separate confinement layer, n-InAs / AlSb superlattice grading layer, n- InAs / AlSb superlattice upper cladding layer, n-InAs / AlSb superlattice transition layers, and capped with an n-InAsSb grating host layer. The superlattice grading layers, cladding, layers, confinement layers, transition layers, and grating host layers have various dopings, throughout. The transition and grading layers are implemented in the epitaxial structure for electronic transport purposes, and are not necessarily critical to the invention. The separate confinement layers (also known as SCLs), are paramount to realize the invention. Generally, the purpose of the SCLs is to increase the overlap of the optical modes of interest with the active region, by creating a strong waveguide. For the invention preferably an asymmetry with respect to the SCL layer thicknesses is introduced to manipulate the optical modes of interest. Specifically, by strategically employing an asymmetric SCL approach to the waveguide design, an overlap of the first-order transverse optical mode can be effectively reduced while maintaining the laser performance parameters of interest of a conventional symmetric architecture (i.e., such as the confinement of the fundamental mode, the waveguide losses of the fundamental mode, amongst others). Finally, according to embodiments of the invention the SCL is comprised of a thin, higher-doped SCL and a thicker, lower-doped SCL, which is solely implemented to enhance the electronic transport of the epitaxial structure, similar to the transition and grading layers aforementioned. This is done for optimal electronic transport purposes. Fundamentally, the doping and subsequently the electronic transport, play a negligible role effecting the optical mode. Thus, it should be noted that the SCLs do not need to have different dopings, throughout. The invention could also be embodied with different doping modulation profiles in the SCLs or without any doping modulation.

[0026] The doping of the sub-layers within each SCL (e.g., “thicker, low-doped semiconductor layer”, “Lower-Doped Upper SCL”, “Lower-Doped Lower SCL”, “thinner, higher-doped semiconductor layer”, “Higher-Doper Upper SCL”, and “Higher-Doped Lower SCL”) can be defined in any fashion, as the doping concentration only plays a role in the electronic transport and has negligible effects on the optical mode profile.

[0027] In the following, a particular embodiment of the invention in the form of a distributed-feedback interband cascade lasers (DFB-ICL) emitting at a center wavelength of 3.5 μm is described. It is understood that the present invention can be embodied in various forms, that the embodiments described here are only examples and that one skilled in the art may utilize other embodiments without departing from the scope of the present invention, for example: at different wavelengths and / or for different strengths of gain margin. As outlined in previous sections, the aim of the invention is to create a gain margin large enough to mitigate the need for complex processing techniques (i.e. grating profiles) to suppress higher-order modes, where the fundamental optical mode is preferential. Generally-speaking, as a rule of thumb for interband cascade lasers, the fundamental mode is preferential when the emitter’s waveguide dimension is approximately equal to or less than the emission wavelength:

[0028] ^emitter ~ ^-emission

[0029] By manipulating the epitaxial layers via an iterative optimization scheme the same result can be achieved without the need for narrow emitter widths as previously identified.

[0030] This reduces the need for complex wafer processing techniques, improving the manufacturability (such as reducing additional processing risks, improving yield across the wafer, reducing operational costs, ultimately resulting in significant cost-savings).

[0031] The gain margin, known as the difference in the Figure of Merit (AFoM) between the first-order transverse optical mode, TEio, and the fundamental optical mode, TEoo, is defined as:

[0032] • The waveguide loss, aw, is calculated using the imaginary component of the effective index provided by COMSOL (or representative simulation tool) for a given geometry:

[0033] • The mirror loss, am, is defined by where L is the cavity length of the given device, and Re-ont and Rback are the front- and back- facet reflectivity of the device. This can be simplified further given that the reflectivity of the front and back facets of the laser cavity are equal in the uncoated case, where R is the reflectivity. • The overlap factor, r, of the optical mode with active region, which is calculated as the integral of |EX|2over the active region normalized to the integral of |EX|2everywhere.

[0034] The design of the waveguide structure was motivated by the previous works completed by different groups on suppressing higher order modes in order to achieve fundamental, transverse mode operation - where most groups have utilized advanced grating techniques for mode selection. Table 1 highlights the epitaxial structure from the substrate layer to the grating host layer, which encompasses the holistic epitaxial structure. The following figures and tables are relevant to this epitaxial structure and congruent with the simulation effort which led to the invention.

[0035] Table 1. Thicknesses and doping levels of the main waveguide layers. It should be noted that the Upper and Lower SCLs comprise of several sub-layers of different dopings. This is for electronic transport purposes and plays a negligible role in the optical mode profile.

[0036] Mapping the space (i.e., via an iterative, parametric sweep) consisted of twenty-nine asymmetric thickness arrangements which are highlighted in Figure 3. In each case, the transverse lateral modes TEoo and TEio were investigated as a function of emitter width (farther highlighted in subsequent figures and tables).

[0037] Figures 4 show computed TEoo (figs. 4a, 4c, 4e, 4g, 4i) and TEio (figs. 4b, 4d, 4f, 4h, 4j) mode profiles of a DFB ICL, emitting at 3.5 μm, for ridge widths of 5.0 μm (figures 4a, 4b), 4.5 μm (figures 4c, 4d), 4.0 μm (figures 4e, 4f), 3.5 μm (figures 4g, 4h) and 3.0 μm (figures 4i, 4j) for an optimized, confinement-layer-engineered, DFB-ICL for preferential fundamental transverse mode operation.

[0038] Table 2: Computed waveguide losses, mirror losses, confinement factor and figure of merit of modes TEoo and TEio as function of ridge width, for an optimized, confinement-layer-engineered, DFB-ICL for preferential fundamental transverse mode operation.

[0039] As can be seen in Figure 5, the gain margin has mcreas^ approximately a factor of -2.5 with respect to emitter width for the optimal asymmetric case, in comparison to the original symmetric case. For example, for a 4.5 pm-wide emiter, it can be expected to yield a gain margin of ~ 18 cm'1.

[0040] At very narrow emitter widths (< 3 μm- wide) the factor of improvement is beyond 2.5.

[0041] It is important to note that for this particular case, the gain margin is maximized, while minimizing any significant change to the epitaxial architecture in order to preserve the lasing parameters of import; i.e., any case where an attempt to optimize the gain margin is made, a trade-off elsewhere may or may not have to be considered (e.g., optical confinement, waveguide loss, etc.). Here a significant increase in the gain margin while maintaining all other parameters within several percent (~10% or less) of their original values can be obtained. This is of great interest for several reasons: 1) Although the penalties are slight, they are made up by suppressing higher-order modes while having wider emitter widths, ultimately improving the power density all the while obtaining single, transverse mode operation. 2) Since ICLs operate at lower voltages than their QCL predecessor, there is less concern for thermal dissipation / management, i.e., wider emitters do not pose a concern and are preferential if single-mode operation can be achieved. Furthermore, it is widely known that ICLs operating between ~3 - 4 μm do so at higher wall-plug efficiencies in both pulsed and CW operation. Thus, this waveguide architecture allows to achieve state-of-the-art performance within that spectral range.

[0042] In conclusion, a novel waveguide design for the suppression of higher-order, transverse, lateral modes of DFB-ICLs without the introduction of complex wafer process grating techniques is presented. The active region of the DFB-ICL is stacked between two, specifically tailored, confinement layers which are engineered in an asymmetric manner to preferentially increase the gain margin between the fundamental and first-order optical inodes. This allows for improved flexibility and control of the optical modes with respect to the waveguide’s emitter width, improves the power density of a given laser, all the while keeping previous lasing parameters of interest within acceptable ranges. Ultimately, this waveguide design approach allows to achieve state-of-the-art performance for DFB- ICLs within the ~3 - 4 μm-emission regime.

Claims

CLAIMS1. A waveguide heterostructure for a semiconductor laser grown using compound semiconductor materials, with an active part, herein referred as “core”, comprising an active region depending on the type of semiconductor used, which is sandwiched between two sets of semiconductor layers, herein referred as “passive parts”, where the passive parts are located between the core and an electrode, and between a substrate and the core, wherein the passive parts, comprising at least one, thinner separate confinement layer and one thicker separate confinement layer, herein referred as “thin, upper SCL” and “thick, lower SCL”, respectively.

2. The waveguide structure according to claim 1, wherein the layers of the passive parts are such that, at the emission wavelength of the laser, the refractive indexes of higher-doped and lower- doped separate confinement layers are higher than the refractive index of the core.

3. The waveguide heterostructure according to claim 1 or 2, wherein the passive parts, comprising at least one, thicker, low-doped (carrier concentration ~2 x 1016cm'3) semiconductor layer, herein referred as “lower-doped separate confinement layer”, and at least two, thinner, higher-doped semiconductor layers (carrier concentration ~1 x 1017cm'3), herein referred as “higher-doped separate confinement layers”.

4. The waveguide heterostructure, according to one or more of claims 1, 2, 3 or 6, where the doping of the sub-layers within each SCL (e.g., “thicker, low-doped semiconductor layer”, “Lower- Doped Upper SCL”, “Lower-Doped Lower SCL”, “thinner, higher-doped semiconductor layer”, “Higher-Doper Upper SCL”, and “Higher-Doped Lower SCL”) can be defined in any fashion.

5. The waveguide heterostructure, according to one or more of claims 1, 2, 3 or 4, where the “Upper SCL” is thicker than the lower “SCL”, also obtaining an asymmetric waveguide geometry for single-transverse mode operation (i.e., higher-order mode suppression).

6. The waveguide heterostructure according to one or more of claims 1 to 3 , wherein an active region layer is built of an assortment of active layers of an interband cascade laser.7, The waveguide heterostructure according to one or more of the preceding claims, wherein the composition of the waveguide heterostructure is chosen according to:

8. The waveguide according to one or more of the preceding claims, wherein the core is built of an interband cascade structure, a quantum cascade structure, a quantum well structure, a quantum dash structure, a quantum dot structure, or any other possible structure providing gain in semiconductor lasers.The waveguide heterostructure according to one or more of the preceding claims, wherein the upper passive parts are significantly thinner than the lower passive parts, also compared with a symmetrical design, reducing the distance between core and the metallic contact, thereby improving the thermal evacuation of heat from the core through the upper passive part to the metallic contact and to electro-plated Au layers for heat dissipation, either by lateral heat flow for epitaxial-side-up mounting, or to a thermoelectric cooler for the epitaxial-side-down mounting, ultimately lowering the operating temperature of the laser, improving output power and wall plug efficiency, allowing an asymmetrical laser to operate at higher power compared to symmetrical designs.

Images